From 61385345280d713dc9d94dddd8ba27c23bb4a1aa Mon Sep 17 00:00:00 2001 From: Patrick Steinhardt Date: Tue, 8 Oct 2024 14:15:53 +0200 Subject: [PATCH 001/196] builtin/gc: fix crash when running `git maintenance start` It was reported on the mailing list that running `git maintenance start` immediately segfaults starting with b6c3f8e12c (builtin/maintenance: fix leak in `get_schedule_cmd()`, 2024-09-26). And indeed, this segfault is trivial to reproduce up to a point where one is scratching their head why we didn't catch this regression in our test suite. The root cause of this error is `get_schedule_cmd()`, which does not populate the `out` parameter in all cases anymore starting with the mentioned commit. Callers do assume it to always be populated though and will e.g. call `strvec_split()` on the returned value, which will of course segfault when the variable is uninitialized. So why didn't we catch this trivial regression? The reason is that our tests always set up the "GIT_TEST_MAINT_SCHEDULER" environment variable via "t/test-lib.sh", which allows us to override the scheduler command with a custom one so that we don't accidentally modify the developer's system. But the faulty code where we don't set the `out` parameter will only get hit in case that environment variable is _not_ set, which is never the case when executing our tests. Fix the regression by again unconditionally allocating the value in the `out` parameter, if provided. Add a test that unsets the environment variable to catch future regressions in this area. Reported-by: Shubham Kanodia Signed-off-by: Patrick Steinhardt Signed-off-by: Derrick Stolee --- builtin/gc.c | 7 +++++-- t/t7900-maintenance.sh | 16 ++++++++++++++++ 2 files changed, 21 insertions(+), 2 deletions(-) diff --git a/builtin/gc.c b/builtin/gc.c index 6a7a2da006eeee..d52735354c9f87 100644 --- a/builtin/gc.c +++ b/builtin/gc.c @@ -1832,7 +1832,7 @@ static const char *get_extra_launchctl_strings(void) { * | Input | Output | * | *cmd | return code | *out | *is_available | * +-------+-------------+-------------------+---------------+ - * | "foo" | false | NULL | (unchanged) | + * | "foo" | false | "foo" (allocated) | (unchanged) | * +-------+-------------+-------------------+---------------+ * * GIT_TEST_MAINT_SCHEDULER set to “foo:./mock_foo.sh,bar:./mock_bar.sh” @@ -1850,8 +1850,11 @@ static int get_schedule_cmd(const char *cmd, int *is_available, char **out) struct string_list_item *item; struct string_list list = STRING_LIST_INIT_NODUP; - if (!testing) + if (!testing) { + if (out) + *out = xstrdup(cmd); return 0; + } if (is_available) *is_available = 0; diff --git a/t/t7900-maintenance.sh b/t/t7900-maintenance.sh index a66d0e089d2d3a..4008e4e45e622e 100755 --- a/t/t7900-maintenance.sh +++ b/t/t7900-maintenance.sh @@ -646,6 +646,22 @@ test_expect_success !MINGW 'register and unregister with regex metacharacters' ' maintenance.repo "$(pwd)/$META" ' +test_expect_success 'start without GIT_TEST_MAINT_SCHEDULER' ' + test_when_finished "rm -rf crontab.log script repo" && + mkdir script && + write_script script/crontab <<-EOF && + echo "\$*" >>"$(pwd)"/crontab.log + EOF + git init repo && + ( + cd repo && + sane_unset GIT_TEST_MAINT_SCHEDULER && + PATH="$(pwd)/../script:$PATH" git maintenance start --scheduler=crontab + ) && + test_grep -- -l crontab.log && + test_grep -- git_cron_edit_tmp crontab.log +' + test_expect_success 'start --scheduler=' ' test_expect_code 129 git maintenance start --scheduler=foo 2>err && test_grep "unrecognized --scheduler argument" err && From a3d7a50364caf2f030d47a62d60bf16c21e9eadb Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Tue, 16 Jul 2024 09:17:15 -0400 Subject: [PATCH 002/196] t: remove advice from some tests These seem to be custom tests to microsoft/git as they break without these changes, but these changes are not needed upstream. Signed-off-by: Derrick Stolee --- t/t1091-sparse-checkout-builtin.sh | 1 + t/t7002-mv-sparse-checkout.sh | 3 +++ 2 files changed, 4 insertions(+) diff --git a/t/t1091-sparse-checkout-builtin.sh b/t/t1091-sparse-checkout-builtin.sh index 8c5cd651b4b80b..1d87cf611c0b7a 100755 --- a/t/t1091-sparse-checkout-builtin.sh +++ b/t/t1091-sparse-checkout-builtin.sh @@ -702,6 +702,7 @@ test_expect_success 'pattern-checks: contained glob characters' ' test_expect_success BSLASHPSPEC 'pattern-checks: escaped characters' ' git clone repo escaped && + git -C escaped config advice.sparseIndexExpanded false && TREEOID=$(git -C escaped rev-parse HEAD:folder1) && NEWTREE=$(git -C escaped mktree <<-EOF $(git -C escaped ls-tree HEAD) diff --git a/t/t7002-mv-sparse-checkout.sh b/t/t7002-mv-sparse-checkout.sh index 57969ce805a548..044038ae40a92b 100755 --- a/t/t7002-mv-sparse-checkout.sh +++ b/t/t7002-mv-sparse-checkout.sh @@ -156,6 +156,9 @@ test_expect_success 'mv refuses to move sparse-to-non-sparse' ' test_expect_success 'recursive mv refuses to move (possible) sparse' ' test_when_finished rm -rf b c e sub2 && + + git config advice.sparseIndexExpanded false && + git reset --hard && # Without cone mode, "sub" and "sub2" do not match git sparse-checkout set sub/dir sub2/dir && From 394eed4843eb28eb52ea9945f60f7578f9940401 Mon Sep 17 00:00:00 2001 From: Victoria Dye Date: Tue, 19 Sep 2023 18:20:32 -0700 Subject: [PATCH 003/196] sparse-index.c: fix use of index hashes in expand_index In ac8acb4f2c7 (sparse-index: complete partial expansion, 2022-05-23), 'expand_index()' was updated to expand the index to a given pathspec. However, the 'path_matches_pattern_list()' method used to facilitate this has the side effect of initializing or updating the index hash variables ('name_hash', 'dir_hash', and 'name_hash_initialized'). This operation is performed on 'istate', though, not 'full'; as a result, the initialized hashes are later overwritten when copied from 'full'. To ensure the correct hashes are in 'istate' after the index expansion, change the arg used in 'path_matches_pattern_list()' from 'istate' to 'full'. Note that this does not fully solve the problem. If 'istate' does not have an initialized 'name_hash' when its contents are copied to 'full', initialized hashes will be copied back into 'istate' but 'name_hash_initialized' will be 0. Therefore, we also need to copy 'full->name_hash_initialized' back to 'istate' after the index expansion is complete. Signed-off-by: Victoria Dye --- sparse-index.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/sparse-index.c b/sparse-index.c index 3d7f2164e25ee5..62b2393f5fb117 100644 --- a/sparse-index.c +++ b/sparse-index.c @@ -402,7 +402,7 @@ void expand_index(struct index_state *istate, struct pattern_list *pl) if (pl && path_matches_pattern_list(ce->name, ce->ce_namelen, NULL, &dtype, - pl, istate) == NOT_MATCHED) { + pl, full) == NOT_MATCHED) { set_index_entry(full, full->cache_nr++, ce); continue; } @@ -430,6 +430,7 @@ void expand_index(struct index_state *istate, struct pattern_list *pl) } /* Copy back into original index. */ + istate->name_hash_initialized = full->name_hash_initialized; memcpy(&istate->name_hash, &full->name_hash, sizeof(full->name_hash)); memcpy(&istate->dir_hash, &full->dir_hash, sizeof(full->dir_hash)); istate->sparse_index = pl ? INDEX_PARTIALLY_SPARSE : INDEX_EXPANDED; From 7bcd46a8d6b7f8be906d0312448ae7b7ed9f74b7 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 25 Aug 2023 09:58:27 -0400 Subject: [PATCH 004/196] t5300: confirm failure of git index-pack when non-idx suffix requested Add test case to demonstrate that `git index-pack -o pack-path` fails if does not end in ".idx" when `--rev-index` is enabled. In e37d0b8730b (builtin/index-pack.c: write reverse indexes, 2021-01-25) we learned to create `.rev` reverse indexes in addition to `.idx` index files. The `.rev` file pathname is constructed by replacing the suffix on the `.idx` file. The code assumes a hard-coded "idx" suffix. In a8dd7e05b1c (config: enable `pack.writeReverseIndex` by default, 2023-04-12) reverse indexes were enabled by default. If the `-o ` argument is used, the index file may have a different suffix. This causes an error when it tries to create the reverse index pathname. The test here demonstrates the failure. (The test forces `--rev-index` to avoid interaction with `GIT_TEST_NO_WRITE_REV_INDEX` during CI runs.) Signed-off-by: Jeff Hostetler --- t/t5300-pack-object.sh | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) diff --git a/t/t5300-pack-object.sh b/t/t5300-pack-object.sh index db345c4e997e22..95dbf634811820 100755 --- a/t/t5300-pack-object.sh +++ b/t/t5300-pack-object.sh @@ -355,6 +355,31 @@ test_expect_success 'build pack index for an existing pack' ' : ' +# The `--rev-index` option of `git index-pack` is now the default, so +# a `foo.rev` REV file will be created when a `foo.idx` IDX file is +# created. Normally, these pathnames are based upon the `foo.pack` +# PACK file pathname. +# +# However, the `-o` option lets you set the pathname of the IDX file +# indepdent of the PACK file. +# +# Verify what happens if these suffixes are changed. +# +test_expect_success 'complain about index name' ' + # Normal case { .pack, .idx, .rev } + cat test-1-${packname_1}.pack >test-complain-0.pack && + git index-pack -o test-complain-0.idx --rev-index test-complain-0.pack && + test -f test-complain-0.idx && + test -f test-complain-0.rev && + + # Non .idx suffix + cat test-1-${packname_1}.pack >test-complain-1.pack && + test_must_fail git index-pack -o test-complain-1.idx-suffix --rev-index test-complain-1.pack 2>err && + grep "does not end" err && + ! test -f test-complain-1.idx-suffix && + ! test -f test-complain-1.rev +' + test_expect_success 'unpacking with --strict' ' for j in a b c d e f g From f2db492cc617269ea3170e647c41ec4b737cbcbd Mon Sep 17 00:00:00 2001 From: Victoria Dye Date: Wed, 20 Sep 2023 13:12:30 -0700 Subject: [PATCH 005/196] t1092: add test for untracked files and directories Add a test verifying that sparse-checkout (with and without sparse index enabled) treat untracked files & directories correctly when changing sparse patterns. Specifically, it ensures that 'git sparse-checkout set' * deletes empty directories outside the sparse cone * does _not_ delete untracked files outside the sparse cone Signed-off-by: Victoria Dye --- t/t1092-sparse-checkout-compatibility.sh | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/t/t1092-sparse-checkout-compatibility.sh b/t/t1092-sparse-checkout-compatibility.sh index 6e230b54876061..e436cb67dfca26 100755 --- a/t/t1092-sparse-checkout-compatibility.sh +++ b/t/t1092-sparse-checkout-compatibility.sh @@ -317,6 +317,22 @@ test_expect_success 'root directory cannot be sparse' ' test_cmp expect actual ' +test_expect_success 'sparse-checkout with untracked files and dirs' ' + init_repos && + + # Empty directories outside sparse cone are deleted + run_on_sparse mkdir -p deep/empty && + test_sparse_match git sparse-checkout set folder1 && + test_must_be_empty sparse-checkout-err && + run_on_sparse test_path_is_missing deep && + + # Untracked files outside sparse cone are not deleted + run_on_sparse touch folder1/another && + test_sparse_match git sparse-checkout set folder2 && + grep "directory ${SQ}folder1/${SQ} contains untracked files" sparse-checkout-err && + run_on_sparse test_path_exists folder1/another +' + test_expect_success 'status with options' ' init_repos && test_sparse_match ls && From 69b5eb8e25cf12a0c64120b894660079e66770fb Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 25 Aug 2023 11:06:28 -0400 Subject: [PATCH 006/196] index-pack: disable rev-index if index file has non .idx suffix Teach index-pack to silently omit the reverse index if the index file does not have the standard ".idx" suffix. In e37d0b8730b (builtin/index-pack.c: write reverse indexes, 2021-01-25) we learned to create `.rev` reverse indexes in addition to `.idx` index files. The `.rev` file pathname is constructed by replacing the suffix on the `.idx` file. The code assumes a hard-coded "idx" suffix. In a8dd7e05b1c (config: enable `pack.writeReverseIndex` by default, 2023-04-12) reverse indexes were enabled by default. If the `-o ` argument is used, the index file may have a different suffix. This causes an error when it tries to create the reverse index pathname. Since we do not know why the user requested a non-standard suffix for the index, we cannot guess what the proper corresponding suffix should be for the reverse index. So we disable it. The t5300 test has been updated to verify that we no longer error out and that the .rev file is not created. TODO We could warn the user that we skipped it (perhaps only if they TODO explicitly requested `--rev-index` on the command line). TODO TODO Ideally, we should add an `--rev-index-path=` argument TODO or change `--rev-index` to take a pathname. TODO TODO I'll leave these questions for a future series. Signed-off-by: Jeff Hostetler --- builtin/index-pack.c | 4 ++++ t/t5300-pack-object.sh | 7 +++---- 2 files changed, 7 insertions(+), 4 deletions(-) diff --git a/builtin/index-pack.c b/builtin/index-pack.c index e228c56ff2701b..d2073c8a84f26c 100644 --- a/builtin/index-pack.c +++ b/builtin/index-pack.c @@ -1738,6 +1738,7 @@ int cmd_index_pack(int argc, unsigned foreign_nr = 1; /* zero is a "good" value, assume bad */ int report_end_of_input = 0; int hash_algo = 0; + int dash_o = 0; /* * index-pack never needs to fetch missing objects except when @@ -1831,6 +1832,7 @@ int cmd_index_pack(int argc, if (index_name || (i+1) >= argc) usage(index_pack_usage); index_name = argv[++i]; + dash_o = 1; } else if (starts_with(arg, "--index-version=")) { char *c; opts.version = strtoul(arg + 16, &c, 10); @@ -1882,6 +1884,8 @@ int cmd_index_pack(int argc, repo_set_hash_algo(the_repository, GIT_HASH_SHA1); opts.flags &= ~(WRITE_REV | WRITE_REV_VERIFY); + if (rev_index && dash_o && !ends_with(index_name, ".idx")) + rev_index = 0; if (rev_index) { opts.flags |= verify ? WRITE_REV_VERIFY : WRITE_REV; if (index_name) diff --git a/t/t5300-pack-object.sh b/t/t5300-pack-object.sh index 95dbf634811820..2da5b5d9a2c366 100755 --- a/t/t5300-pack-object.sh +++ b/t/t5300-pack-object.sh @@ -372,11 +372,10 @@ test_expect_success 'complain about index name' ' test -f test-complain-0.idx && test -f test-complain-0.rev && - # Non .idx suffix + # Non .idx suffix -- implicitly omits the .rev cat test-1-${packname_1}.pack >test-complain-1.pack && - test_must_fail git index-pack -o test-complain-1.idx-suffix --rev-index test-complain-1.pack 2>err && - grep "does not end" err && - ! test -f test-complain-1.idx-suffix && + git index-pack -o test-complain-1.idx-suffix --rev-index test-complain-1.pack && + test -f test-complain-1.idx-suffix && ! test -f test-complain-1.rev ' From fd1e3c24ad699903a01674880dd4673bc1e17219 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 26 Jun 2024 12:27:41 -0400 Subject: [PATCH 007/196] trace2: prefetch value of GIT_TRACE2_DST_DEBUG at startup Prefetch the value of GIT_TRACE2_DST_DEBUG during startup and before we try to open any Trace2 destination pathnames. Normally, Trace2 always silently fails if a destination target cannot be opened so that it doesn't affect the execution of a Git command. The command should run normally, but just not generate any trace data. This can make it difficult to debug a telemetry setup, since the user doesn't know why telemetry isn't being generated. If the environment variable GIT_TRACE2_DST_DEBUG is true, the Trace2 startup will print a warning message with the `errno` to make debugging easier. However, on Windows, looking up the env variable resets `errno` so the warning message always ends with `...tracing: No error` which is not very helpful. Prefetch the env variable at startup. This avoids the need to update each call-site to capture `errno` in the usual `saved-errno` variable. Signed-off-by: Jeff Hostetler --- trace2.c | 10 ++++++++++ trace2/tr2_dst.c | 2 +- trace2/tr2_dst.h | 12 ++++++++++++ 3 files changed, 23 insertions(+), 1 deletion(-) diff --git a/trace2.c b/trace2.c index f894532d05331c..0a4ff698f607ab 100644 --- a/trace2.c +++ b/trace2.c @@ -225,6 +225,16 @@ void trace2_initialize_fl(const char *file, int line) if (!tr2_tgt_want_builtins()) return; trace2_enabled = 1; + + /* + * getenv() on Windows stomps on `errno` and the code in + * tr2_dst.c verifies that warnings are enabled before + * formatting the warning message (and calling strerror()). + * So prefetch the value from the environment before we need + * it. + */ + tr2_dst_want_warning(); + if (!git_env_bool("GIT_TRACE2_REDACT", 1)) trace2_redact = 0; diff --git a/trace2/tr2_dst.c b/trace2/tr2_dst.c index 5be892cd5cdefa..61579f24bdbde3 100644 --- a/trace2/tr2_dst.c +++ b/trace2/tr2_dst.c @@ -24,7 +24,7 @@ */ static int tr2env_max_files = 0; -static int tr2_dst_want_warning(void) +int tr2_dst_want_warning(void) { static int tr2env_dst_debug = -1; diff --git a/trace2/tr2_dst.h b/trace2/tr2_dst.h index b1a8c144e073ba..4166539eb9e100 100644 --- a/trace2/tr2_dst.h +++ b/trace2/tr2_dst.h @@ -35,4 +35,16 @@ int tr2_dst_trace_want(struct tr2_dst *dst); */ void tr2_dst_write_line(struct tr2_dst *dst, struct strbuf *buf_line); +/* + * Return true if we want warning messages when trying to open a + * destination. + * + * (Trace2 always silently fails if a target cannot be opened so that + * we don't affect the execution of the Git command, but it is helpful + * for debugging telemetry configuration if we log warning messages + * when trying to open a target. This is controlled by another config + * value.) + */ +int tr2_dst_want_warning(void); + #endif /* TR2_DST_H */ From 7715abe8da26dcc4c467ce80a6198367fa31c984 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 29 Apr 2024 12:47:27 -0400 Subject: [PATCH 008/196] survey: calculate more stats on refs Calculate the number of symrefs, loose vs packed, and the maximal/accumulated length of local vs remote branches. Signed-off-by: Jeff Hostetler Signed-off-by: Johannes Schindelin --- builtin/survey.c | 82 +++++++++++++++++++++++++++++++++++++++++-- t/t8100-git-survey.sh | 21 +++++++---- 2 files changed, 94 insertions(+), 9 deletions(-) diff --git a/builtin/survey.c b/builtin/survey.c index 34850399fd601a..826c46a60f155a 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -15,9 +15,7 @@ #include "revision.h" #include "strbuf.h" #include "strvec.h" -#include "tag.h" #include "trace2.h" -#include "color.h" static const char * const survey_usage[] = { N_("(EXPERIMENTAL!) git survey "), @@ -53,6 +51,22 @@ struct survey_report_ref_summary { size_t tags_annotated_nr; size_t others_nr; size_t unknown_nr; + + size_t cnt_symref; + + size_t cnt_packed; + size_t cnt_loose; + + /* + * Measure the length of the refnames. We can look for + * potential platform limits. The partial sums may help us + * estimate the size of a haves/wants conversation, since each + * refname and a SHA must be transmitted. + */ + size_t len_max_local_refname; + size_t len_sum_local_refnames; + size_t len_max_remote_refname; + size_t len_sum_remote_refnames; }; struct survey_report_object_summary { @@ -380,6 +394,42 @@ static void survey_report_plaintext_refs(struct survey_context *ctx) free(fmt); } + /* + * SymRefs are somewhat orthogonal to the above classification (e.g. + * "HEAD" --> detached and "refs/remotes/origin/HEAD" --> remote) so the + * above classified counts will already include them, but it is less + * confusing to display them here than to create a whole new section. + */ + if (ctx->report.refs.cnt_symref) { + char *fmt = xstrfmt("%"PRIuMAX"", (uintmax_t)refs->cnt_symref); + insert_table_rowv(&table, _("Symbolic refs"), fmt, NULL); + free(fmt); + } + + if (ctx->report.refs.cnt_loose || ctx->report.refs.cnt_packed) { + char *fmt = xstrfmt("%"PRIuMAX"", (uintmax_t)refs->cnt_loose); + insert_table_rowv(&table, _("Loose refs"), fmt, NULL); + free(fmt); + fmt = xstrfmt("%"PRIuMAX"", (uintmax_t)refs->cnt_packed); + insert_table_rowv(&table, _("Packed refs"), fmt, NULL); + free(fmt); + } + + if (ctx->report.refs.len_max_local_refname || ctx->report.refs.len_max_remote_refname) { + char *fmt = xstrfmt("%"PRIuMAX"", (uintmax_t)refs->len_max_local_refname); + insert_table_rowv(&table, _("Max local refname length"), fmt, NULL); + free(fmt); + fmt = xstrfmt("%"PRIuMAX"", (uintmax_t)refs->len_sum_local_refnames); + insert_table_rowv(&table, _("Sum local refnames length"), fmt, NULL); + free(fmt); + fmt = xstrfmt("%"PRIuMAX"", (uintmax_t)refs->len_max_remote_refname); + insert_table_rowv(&table, _("Max remote refname length"), fmt, NULL); + free(fmt); + fmt = xstrfmt("%"PRIuMAX"", (uintmax_t)refs->len_sum_remote_refnames); + insert_table_rowv(&table, _("Sum remote refnames length"), fmt, NULL); + free(fmt); + } + print_table_plaintext(&table); clear_table(&table); } @@ -637,6 +687,7 @@ static void survey_phase_refs(struct survey_context *ctx) for (size_t i = 0; i < ctx->ref_array.nr; i++) { unsigned long size; struct ref_array_item *item = ctx->ref_array.items[i]; + size_t len = strlen(item->refname); switch (item->kind) { case FILTER_REFS_TAGS: @@ -663,6 +714,33 @@ static void survey_phase_refs(struct survey_context *ctx) ctx->report.refs.unknown_nr++; break; } + + /* + * SymRefs are somewhat orthogonal to the above + * classification (e.g. "HEAD" --> detached + * and "refs/remotes/origin/HEAD" --> remote) so + * our totals will already include them. + */ + if (item->flag & REF_ISSYMREF) + ctx->report.refs.cnt_symref++; + + /* + * Where/how is the ref stored in GITDIR. + */ + if (item->flag & REF_ISPACKED) + ctx->report.refs.cnt_packed++; + else + ctx->report.refs.cnt_loose++; + + if (item->kind == FILTER_REFS_REMOTES) { + ctx->report.refs.len_sum_remote_refnames += len; + if (len > ctx->report.refs.len_max_remote_refname) + ctx->report.refs.len_max_remote_refname = len; + } else { + ctx->report.refs.len_sum_local_refnames += len; + if (len > ctx->report.refs.len_max_local_refname) + ctx->report.refs.len_max_local_refname = len; + } } trace2_region_leave("survey", "phase/refs", ctx->repo); diff --git a/t/t8100-git-survey.sh b/t/t8100-git-survey.sh index 8c6edfcae0c6c2..0d35dfcf311827 100755 --- a/t/t8100-git-survey.sh +++ b/t/t8100-git-survey.sh @@ -59,13 +59,20 @@ test_expect_success 'git survey (default)' ' ----------------------------------------------------- REFERENCES SUMMARY - ======================== - , Ref Type | Count - -----------------+------ - , Branches | 1 - Remote refs | 0 - Tags (all) | 2 - Tags (annotated) | 2 + ================================== + , Ref Type | Count + ---------------------------+------ + , Branches | 1 + , Remote refs | 0 + , Tags (all) | 2 + , Tags (annotated) | 2 + , Symbolic refs | 1 + , Loose refs | 4 + , Packed refs | 0 + Max local refname length | 15 + Sum local refnames length | 46 + Max remote refname length | 0 + Sum remote refnames length | 0 REACHABLE OBJECT SUMMARY ======================== From 00f402cea0e239c682689c7594ede0a3194258d2 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 7 Oct 2024 15:24:09 +0200 Subject: [PATCH 009/196] amend! survey: add --top= option and config survey: add --top= option and config The 'git survey' builtin provides several detail tables, such as "top files by on-disk size". The size of these tables defaults to 10, currently. Allow the user to specify this number via a new --top= option or the new survey.top config key. Signed-off-by: Derrick Stolee Signed-off-by: Johannes Schindelin --- builtin/survey.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/builtin/survey.c b/builtin/survey.c index 826c46a60f155a..26bc38670eaff6 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -947,7 +947,7 @@ int cmd_survey(int argc, const char **argv, const char *prefix, struct repositor .opts = { .verbose = 0, .show_progress = -1, /* defaults to isatty(2) */ - .top_nr = 100, + .top_nr = 10, .refs.want_all_refs = -1, From 01146de194e480bdab695fec2399843ed50e9bf6 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 29 Apr 2024 15:40:00 -0400 Subject: [PATCH 010/196] survey: show some commits/trees/blobs histograms With this commit, we gather statistics about the sizes of commits, trees, and blobs in the repository, and then present them in the form of "hexbins", i.e. log(16) histograms that show how many objects fall into the 0..15 bytes range, the 16..255 range, the 256..4095 range, etc. For commits, we also show the total count grouped by the number of parents, and for trees we additionally show the total count grouped by number of entries in the form of "qbins", i.e. log(4) histograms. Signed-off-by: Jeff Hostetler Signed-off-by: Johannes Schindelin --- builtin/survey.c | 338 ++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 334 insertions(+), 4 deletions(-) diff --git a/builtin/survey.c b/builtin/survey.c index 26bc38670eaff6..9722f7a60d3761 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -16,6 +16,8 @@ #include "strbuf.h" #include "strvec.h" #include "trace2.h" +#include "tree.h" +#include "tree-walk.h" static const char * const survey_usage[] = { N_("(EXPERIMENTAL!) git survey "), @@ -69,11 +71,162 @@ struct survey_report_ref_summary { size_t len_sum_remote_refnames; }; +/* + * HBIN -- hex binning (histogram bucketing). + * + * We create histograms for various counts and sums. Since we have a + * wide range of values (objects range in size from 1 to 4G bytes), a + * linear bucketing is not interesting. Instead, lets use a + * log16()-based bucketing. This gives us a better spread on the low + * and middle range and a coarse bucketing on the high end. + * + * The idea here is that it doesn't matter if you have n 1GB blobs or + * n/2 1GB blobs and n/2 1.5GB blobs -- either way you have a scaling + * problem that we want to report on. + */ +#define HBIN_LEN (sizeof(unsigned long) * 2) +#define HBIN_MASK (0xF) +#define HBIN_SHIFT (4) + +static int hbin(unsigned long value) +{ + for (int k = 0; k < HBIN_LEN; k++) { + if ((value & ~(HBIN_MASK)) == 0) + return k; + value >>= HBIN_SHIFT; + } + + return 0; /* should not happen */ +} + +/* + * QBIN -- base4 binning (histogram bucketing). + * + * This is the same idea as the above, but we want better granularity + * in the low end and don't expect as many large values. + */ +#define QBIN_LEN (sizeof(unsigned long) * 4) +#define QBIN_MASK (0x3) +#define QBIN_SHIFT (2) + +static int qbin(unsigned long value) +{ + for (int k = 0; k < QBIN_LEN; k++) { + if ((value & ~(QBIN_MASK)) == 0) + return k; + value >>= (QBIN_SHIFT); + } + + return 0; /* should not happen */ +} + +/* + * histogram bin for objects. + */ +struct obj_hist_bin { + uint64_t sum_size; /* sum(object_size) for all objects in this bin */ + uint64_t sum_disk_size; /* sum(on_disk_size) for all objects in this bin */ + uint32_t cnt_seen; /* number seen in this bin */ +}; + +static void incr_obj_hist_bin(struct obj_hist_bin *pbin, + unsigned long object_length, + off_t disk_sizep) +{ + pbin->sum_size += object_length; + pbin->sum_disk_size += disk_sizep; + pbin->cnt_seen++; +} + +/* + * Common fields for any type of object. + */ +struct survey_stats_base_object { + uint32_t cnt_seen; + + uint32_t cnt_missing; /* we may have a partial clone. */ + + /* + * Number of objects grouped by where they are stored on disk. + * This is a function of how the ODB is packed. + */ + uint32_t cnt_cached; /* see oi.whence */ + uint32_t cnt_loose; /* see oi.whence */ + uint32_t cnt_packed; /* see oi.whence */ + uint32_t cnt_dbcached; /* see oi.whence */ + + uint64_t sum_size; /* sum(object_size) */ + uint64_t sum_disk_size; /* sum(disk_size) */ + + /* + * A histogram of the count of objects, the observed size, and + * the on-disk size grouped by the observed size. + */ + struct obj_hist_bin size_hbin[HBIN_LEN]; +}; + +/* + * PBIN -- parent vector binning (histogram bucketing). + * + * We create a histogram based upon the number of parents + * in a commit. This is a simple linear vector. It starts + * at zero for "initial" commits. + * + * If a commit has more parents, just put it in the last bin. + */ +#define PBIN_VEC_LEN (32) + +struct survey_stats_commits { + struct survey_stats_base_object base; + + /* + * Count of commits with k parents. + */ + uint32_t parent_cnt_pbin[PBIN_VEC_LEN]; +}; + +/* + * Stats for reachable trees. + */ +struct survey_stats_trees { + struct survey_stats_base_object base; + + /* + * In the following, nr_entries refers to the number of files or + * subdirectories in a tree. We are interested in how wide the + * tree is and if the repo has gigantic directories. + */ + uint64_t max_entries; /* max(nr_entries) -- the width of the largest tree */ + + /* + * Computing the sum of the number of entries across all trees + * is probably not that interesting. + */ + uint64_t sum_entries; /* sum(nr_entries) -- sum across all trees */ + + /* + * A histogram of the count of trees, the observed size, and + * the on-disk size grouped by the number of entries in the tree. + */ + struct obj_hist_bin entry_qbin[QBIN_LEN]; +}; + +/* + * Stats for reachable blobs. + */ +struct survey_stats_blobs { + struct survey_stats_base_object base; +}; + struct survey_report_object_summary { size_t commits_nr; size_t tags_nr; size_t trees_nr; size_t blobs_nr; + + struct survey_stats_commits commits; + struct survey_stats_trees trees; + struct survey_stats_blobs blobs; }; /** @@ -363,6 +516,98 @@ static void print_table_plaintext(struct survey_table *table) free(column_widths); } +static void pretty_print_bin_table(const char *title_caption, + const char *bucket_header, + struct obj_hist_bin *bin, + uint64_t bin_len, int bin_shift, uint64_t bin_mask) +{ + struct survey_table table = SURVEY_TABLE_INIT; + struct strbuf bucket = STRBUF_INIT, cnt_seen = STRBUF_INIT; + struct strbuf sum_size = STRBUF_INIT, sum_disk_size = STRBUF_INIT; + uint64_t lower = 0; + uint64_t upper = bin_mask; + + table.table_name = title_caption; + strvec_pushl(&table.header, bucket_header, "Count", "Size", "Disk Size", NULL); + + for (int k = 0; k < bin_len; k++) { + struct obj_hist_bin *p = bin + k; + uint64_t lower_k = lower; + uint64_t upper_k = upper; + + lower = upper+1; + upper = (upper << bin_shift) + bin_mask; + + if (!p->cnt_seen) + continue; + + strbuf_reset(&bucket); + strbuf_addf(&bucket, "%"PRIu64"..%"PRIu64, lower_k, upper_k); + + strbuf_reset(&cnt_seen); + strbuf_addf(&cnt_seen, "%"PRIu64, (uintmax_t)p->cnt_seen); + + strbuf_reset(&sum_size); + strbuf_addf(&sum_size, "%"PRIu64, (uintmax_t)p->sum_size); + + strbuf_reset(&sum_disk_size); + strbuf_addf(&sum_disk_size, "%"PRIu64, (uintmax_t)p->sum_disk_size); + + insert_table_rowv(&table, bucket.buf, + cnt_seen.buf, sum_size.buf, sum_disk_size.buf); + } + strbuf_release(&bucket); + strbuf_release(&cnt_seen); + strbuf_release(&sum_size); + strbuf_release(&sum_disk_size); + + print_table_plaintext(&table); + clear_table(&table); +} + +static void survey_report_hbin(const char *title_caption, + struct obj_hist_bin *bin) +{ + pretty_print_bin_table(title_caption, + "Byte Range", + bin, + HBIN_LEN, HBIN_SHIFT, HBIN_MASK); +} + +static void survey_report_tree_lengths(struct survey_context *ctx) +{ + pretty_print_bin_table(_("TREE HISTOGRAM BY NUMBER OF ENTRIES"), + "Entry Range", + ctx->report.reachable_objects.trees.entry_qbin, + QBIN_LEN, QBIN_SHIFT, QBIN_MASK); +} + +static void survey_report_commit_parents(struct survey_context *ctx) +{ + struct survey_stats_commits *psc = &ctx->report.reachable_objects.commits; + struct survey_table table = SURVEY_TABLE_INIT; + struct strbuf parents = STRBUF_INIT, counts = STRBUF_INIT; + + table.table_name = _("HISTOGRAM BY NUMBER OF COMMIT PARENTS"); + strvec_pushl(&table.header, "Parents", "Counts", NULL); + + for (int k = 0; k < PBIN_VEC_LEN; k++) + if (psc->parent_cnt_pbin[k]) { + strbuf_reset(&parents); + strbuf_addf(&parents, "%02d", k); + + strbuf_reset(&counts); + strbuf_addf(&counts, "%14"PRIuMAX, (uintmax_t)psc->parent_cnt_pbin[k]); + + insert_table_rowv(&table, parents.buf, counts.buf, NULL); + } + strbuf_release(&parents); + strbuf_release(&counts); + + print_table_plaintext(&table); + clear_table(&table); +} + static void survey_report_plaintext_refs(struct survey_context *ctx) { struct survey_report_ref_summary *refs = &ctx->report.refs; @@ -515,6 +760,19 @@ static void survey_report_plaintext(struct survey_context *ctx) ctx->report.by_type, REPORT_TYPE_COUNT); + survey_report_commit_parents(ctx); + + survey_report_hbin(_("COMMITS HISTOGRAM BY SIZE IN BYTES"), + ctx->report.reachable_objects.commits.base.size_hbin); + + survey_report_tree_lengths(ctx); + + survey_report_hbin(_("TREES HISTOGRAM BY SIZE IN BYTES"), + ctx->report.reachable_objects.trees.base.size_hbin); + + survey_report_hbin(_("BLOBS HISTOGRAM BY SIZE IN BYTES"), + ctx->report.reachable_objects.blobs.base.size_hbin); + survey_report_plaintext_sorted_size( &ctx->report.top_paths_by_count[REPORT_TYPE_TREE]); survey_report_plaintext_sorted_size( @@ -783,6 +1041,8 @@ static void increment_totals(struct survey_context *ctx, unsigned long object_length = 0; off_t disk_sizep = 0; enum object_type type; + struct survey_stats_base_object *base; + int hb; oi.typep = &type; oi.sizep = &object_length; @@ -791,11 +1051,81 @@ static void increment_totals(struct survey_context *ctx, if (oid_object_info_extended(ctx->repo, &oids->oid[i], &oi, oi_flags) < 0) { summary->num_missing++; - } else { - summary->nr++; - summary->disk_size += disk_sizep; - summary->inflated_size += object_length; + continue; + } + + summary->nr++; + summary->disk_size += disk_sizep; + summary->inflated_size += object_length; + + switch (type) { + case OBJ_COMMIT: { + struct commit *commit = lookup_commit(ctx->repo, &oids->oid[i]); + unsigned k = commit_list_count(commit->parents); + + if (k >= PBIN_VEC_LEN) + k = PBIN_VEC_LEN - 1; + + ctx->report.reachable_objects.commits.parent_cnt_pbin[k]++; + base = &ctx->report.reachable_objects.commits.base; + break; } + case OBJ_TREE: { + struct tree *tree = lookup_tree(ctx->repo, &oids->oid[i]); + if (tree) { + struct survey_stats_trees *pst = &ctx->report.reachable_objects.trees; + struct tree_desc desc; + struct name_entry entry; + int nr_entries; + int qb; + + parse_tree(tree); + init_tree_desc(&desc, &oids->oid[i], tree->buffer, tree->size); + nr_entries = 0; + while (tree_entry(&desc, &entry)) + nr_entries++; + + pst->sum_entries += nr_entries; + + if (nr_entries > pst->max_entries) + pst->max_entries = nr_entries; + + qb = qbin(nr_entries); + incr_obj_hist_bin(&pst->entry_qbin[qb], object_length, disk_sizep); + } + base = &ctx->report.reachable_objects.trees.base; + break; + } + case OBJ_BLOB: + base = &ctx->report.reachable_objects.blobs.base; + break; + default: + continue; + } + + switch (oi.whence) { + case OI_CACHED: + base->cnt_cached++; + break; + case OI_LOOSE: + base->cnt_loose++; + break; + case OI_PACKED: + base->cnt_packed++; + break; + case OI_DBCACHED: + base->cnt_dbcached++; + break; + default: + break; + } + + base->sum_size += object_length; + base->sum_disk_size += disk_sizep; + + hb = hbin(object_length); + incr_obj_hist_bin(&base->size_hbin[hb], object_length, disk_sizep); + } } From 56703e74a3c3c0d6a3c77b063820b5616ceb0697 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 1 May 2024 12:56:38 -0400 Subject: [PATCH 011/196] survey: add vector of largest objects for various scaling dimensions Create `struct large_item` and `struct large_item_vec` to capture the n largest commits, trees, and blobs under various scaling dimensions, such as size in bytes, number of commit parents, or number of entries in a tree. Each of these have a command line option to set them independently. Signed-off-by: Jeff Hostetler --- Documentation/config/survey.txt | 29 +++++ Documentation/git-survey.txt | 31 +++++ builtin/survey.c | 222 +++++++++++++++++++++++++++++++- 3 files changed, 276 insertions(+), 6 deletions(-) diff --git a/Documentation/config/survey.txt b/Documentation/config/survey.txt index 9e594a2092f225..fd2d7e153b38ce 100644 --- a/Documentation/config/survey.txt +++ b/Documentation/config/survey.txt @@ -11,4 +11,33 @@ survey.*:: top:: This integer value implies `--top=`, specifying the number of entries in the detail tables. + showBlobSizes:: + A non-negative integer value. Requests details on the + largest file blobs by size in bytes. Provides a + default value for `--blob-sizes=` in + linkgit:git-survey[1]. + showCommitParents:: + A non-negative integer value. Requests details on the + commits with the most number of parents. Provides a + default value for `--commit-parents=` in + linkgit:git-survey[1]. + showCommitSizes:: + A non-negative integer value. Requests details on the + largest commits by size in bytes. Generally, these + are the commits with the largest commit messages. + Provides a default value for `--commit-sizes=` in + linkgit:git-survey[1]. + showTreeEntries:: + A non-negative integer value. Requests details on the + trees (directories) with the most number of entries + (files and subdirectories). Provides a default value + for `--tree-entries=` in linkgit:git-survey[1]. + showTreeSizes:: + A non-negative integer value. Requests details on the + largest trees (directories) by size in bytes. This + will set will usually be equal to the + `survey.showTreeEntries` set, but may be skewed by very + long file or subdirectory entry names. Provides a + default value for `--tree-sizes=` in + linkgit:git-survey[1]. -- diff --git a/Documentation/git-survey.txt b/Documentation/git-survey.txt index 894c7be3053eb9..e47924d0351ca1 100644 --- a/Documentation/git-survey.txt +++ b/Documentation/git-survey.txt @@ -59,6 +59,32 @@ only refs for the given options are added. --other:: Add notes (`refs/notes/`) and stashes (`refs/stash/`) to the set. +Large Item Selection +~~~~~~~~~~~~~~~~~~~~ + +The following options control the optional display of large items under +various dimensions of scale. The OID of the largest `n` objects will be +displayed in reverse sorted order. For each, `n` defaults to 10. + +--commit-parents:: + Shows the OIDs of the commits with the most parent commits. + +--commit-sizes:: + Shows the OIDs of the largest commits by size in bytes. This is + usually the ones with the largest commit messages. + +--tree-entries:: + Shows the OIDs of the trees with the most number of entries. These + are the directories with the most number of files or subdirectories. + +--tree-sizes:: + Shows the OIDs of the largest trees by size in bytes. This set + will usually be the same as the vector of number of entries unless + skewed by very long entry names. + +--blob-sizes:: + Shows the OIDs of the largest blobs by size in bytes. + OUTPUT ------ @@ -78,6 +104,11 @@ Reachable Object Summary The reachable object summary shows the total number of each kind of Git object, including tags, commits, trees, and blobs. +CONFIGURATION +------------- + +include::config/survey.txt[] + GIT --- Part of the linkgit:git[1] suite diff --git a/builtin/survey.c b/builtin/survey.c index 9722f7a60d3761..7a93c9c991f8b9 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -41,6 +41,15 @@ static struct survey_refs_wanted default_ref_options = { struct survey_opts { int verbose; int show_progress; + + int show_largest_commits_by_nr_parents; + int show_largest_commits_by_size_bytes; + + int show_largest_trees_by_nr_entries; + int show_largest_trees_by_size_bytes; + + int show_largest_blobs_by_size_bytes; + int top_nr; struct survey_refs_wanted refs; }; @@ -138,6 +147,87 @@ static void incr_obj_hist_bin(struct obj_hist_bin *pbin, pbin->cnt_seen++; } +/* + * Remember the largest n objects for some scaling dimension. This + * could be the observed object size or number of entries in a tree. + * We'll use this to generate a sorted vector in the output for that + * dimension. + */ +struct large_item { + uint64_t size; + struct object_id oid; +}; + +struct large_item_vec { + char *dimension_label; + char *item_label; + uint64_t nr_items; + struct large_item items[FLEX_ARRAY]; /* nr_items */ +}; + +static struct large_item_vec *alloc_large_item_vec(const char *dimension_label, + const char *item_label, + uint64_t nr_items) +{ + struct large_item_vec *vec; + size_t flex_len = nr_items * sizeof(struct large_item); + + if (!nr_items) + return NULL; + + vec = xcalloc(1, (sizeof(struct large_item_vec) + flex_len)); + vec->dimension_label = strdup(dimension_label); + vec->item_label = strdup(item_label); + vec->nr_items = nr_items; + + return vec; +} + +static void free_large_item_vec(struct large_item_vec *vec) +{ + if (!vec) + return; + + free(vec->dimension_label); + free(vec->item_label); + free(vec); +} + +static void maybe_insert_large_item(struct large_item_vec *vec, + uint64_t size, + struct object_id *oid) +{ + size_t rest_len; + size_t k; + + if (!vec || !vec->nr_items) + return; + + /* + * Since the odds an object being among the largest n + * is small, shortcut and see if it is smaller than + * the smallest one in our set and quickly reject it. + */ + if (size < vec->items[vec->nr_items - 1].size) + return; + + for (k = 0; k < vec->nr_items; k++) { + if (size < vec->items[k].size) + continue; + + /* push items[k..] down one and insert it here */ + + rest_len = (vec->nr_items - k - 1) * sizeof(struct large_item); + if (rest_len) + memmove(&vec->items[k + 1], &vec->items[k], rest_len); + + memset(&vec->items[k], 0, sizeof(struct large_item)); + vec->items[k].size = size; + oidcpy(&vec->items[k].oid, oid); + return; + } +} + /* * Common fields for any type of object. */ @@ -183,6 +273,9 @@ struct survey_stats_commits { * Count of commits with k parents. */ uint32_t parent_cnt_pbin[PBIN_VEC_LEN]; + + struct large_item_vec *vec_largest_by_nr_parents; + struct large_item_vec *vec_largest_by_size_bytes; }; /* @@ -192,11 +285,18 @@ struct survey_stats_trees { struct survey_stats_base_object base; /* - * In the following, nr_entries refers to the number of files or - * subdirectories in a tree. We are interested in how wide the - * tree is and if the repo has gigantic directories. + * Keep a vector of the trees with the most number of entries. + * This gives us a feel for the width of a tree when there are + * gigantic directories. */ - uint64_t max_entries; /* max(nr_entries) -- the width of the largest tree */ + struct large_item_vec *vec_largest_by_nr_entries; + + /* + * Keep a vector of the trees with the largest size in bytes. + * The contents of this may or may not match items in the other + * vector, since entryname length can alter the results. + */ + struct large_item_vec *vec_largest_by_size_bytes; /* * Computing the sum of the number of entries across all trees @@ -216,6 +316,11 @@ struct survey_stats_trees { */ struct survey_stats_blobs { struct survey_stats_base_object base; + + /* + * Remember the OIDs of the largest n blobs. + */ + struct large_item_vec *vec_largest_by_size_bytes; }; struct survey_report_object_summary { @@ -396,6 +501,12 @@ struct survey_context { static void clear_survey_context(struct survey_context *ctx) { + free_large_item_vec(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents); + free_large_item_vec(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes); + free_large_item_vec(ctx->report.reachable_objects.trees.vec_largest_by_nr_entries); + free_large_item_vec(ctx->report.reachable_objects.trees.vec_largest_by_size_bytes); + free_large_item_vec(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes); + ref_array_clear(&ctx->ref_array); strvec_clear(&ctx->refs); } @@ -608,6 +719,32 @@ static void survey_report_commit_parents(struct survey_context *ctx) clear_table(&table); } +static void survey_report_largest_vec(struct large_item_vec *vec) +{ + struct survey_table table = SURVEY_TABLE_INIT; + struct strbuf size = STRBUF_INIT; + + if (!vec || !vec->nr_items) + return; + + table.table_name = vec->dimension_label; + strvec_pushl(&table.header, "Size", "OID", NULL); + + for (int k = 0; k < vec->nr_items; k++) { + struct large_item *pk = &vec->items[k]; + if (!is_null_oid(&pk->oid)) { + strbuf_reset(&size); + strbuf_addf(&size, "%"PRIuMAX, (uintmax_t)pk->size); + + insert_table_rowv(&table, size.buf, oid_to_hex(&pk->oid), NULL); + } + } + strbuf_release(&size); + + print_table_plaintext(&table); + clear_table(&table); +} + static void survey_report_plaintext_refs(struct survey_context *ctx) { struct survey_report_ref_summary *refs = &ctx->report.refs; @@ -787,6 +924,12 @@ static void survey_report_plaintext(struct survey_context *ctx) &ctx->report.top_paths_by_inflate[REPORT_TYPE_TREE]); survey_report_plaintext_sorted_size( &ctx->report.top_paths_by_inflate[REPORT_TYPE_BLOB]); + + survey_report_largest_vec(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents); + survey_report_largest_vec(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes); + survey_report_largest_vec(ctx->report.reachable_objects.trees.vec_largest_by_nr_entries); + survey_report_largest_vec(ctx->report.reachable_objects.trees.vec_largest_by_size_bytes); + survey_report_largest_vec(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes); } /* @@ -858,6 +1001,27 @@ static int survey_load_config_cb(const char *var, const char *value, ctx->opts.show_progress = git_config_bool(var, value); return 0; } + if (!strcmp(var, "survey.showcommitparents")) { + ctx->opts.show_largest_commits_by_nr_parents = git_config_ulong(var, value, cctx->kvi); + return 0; + } + if (!strcmp(var, "survey.showcommitsizes")) { + ctx->opts.show_largest_commits_by_size_bytes = git_config_ulong(var, value, cctx->kvi); + return 0; + } + + if (!strcmp(var, "survey.showtreeentries")) { + ctx->opts.show_largest_trees_by_nr_entries = git_config_ulong(var, value, cctx->kvi); + return 0; + } + if (!strcmp(var, "survey.showtreesizes")) { + ctx->opts.show_largest_trees_by_size_bytes = git_config_ulong(var, value, cctx->kvi); + return 0; + } + if (!strcmp(var, "survey.showblobsizes")) { + ctx->opts.show_largest_blobs_by_size_bytes = git_config_ulong(var, value, cctx->kvi); + return 0; + } if (!strcmp(var, "survey.top")) { ctx->opts.top_nr = git_config_bool(var, value); return 0; @@ -1068,6 +1232,9 @@ static void increment_totals(struct survey_context *ctx, ctx->report.reachable_objects.commits.parent_cnt_pbin[k]++; base = &ctx->report.reachable_objects.commits.base; + + maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents, k, &commit->object.oid); + maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes, object_length, &commit->object.oid); break; } case OBJ_TREE: { @@ -1087,8 +1254,8 @@ static void increment_totals(struct survey_context *ctx, pst->sum_entries += nr_entries; - if (nr_entries > pst->max_entries) - pst->max_entries = nr_entries; + maybe_insert_large_item(pst->vec_largest_by_nr_entries, nr_entries, &tree->object.oid); + maybe_insert_large_item(pst->vec_largest_by_size_bytes, object_length, &tree->object.oid); qb = qbin(nr_entries); incr_obj_hist_bin(&pst->entry_qbin[qb], object_length, disk_sizep); @@ -1098,6 +1265,8 @@ static void increment_totals(struct survey_context *ctx, } case OBJ_BLOB: base = &ctx->report.reachable_objects.blobs.base; + + maybe_insert_large_item(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes, object_length, &oids->oid[i]); break; default: continue; @@ -1304,6 +1473,14 @@ int cmd_survey(int argc, const char **argv, const char *prefix, struct repositor OPT_BOOL_F(0, "detached", &ctx.opts.refs.want_detached, N_("include detached HEAD"), PARSE_OPT_NONEG), OPT_BOOL_F(0, "other", &ctx.opts.refs.want_other, N_("include notes and stashes"), PARSE_OPT_NONEG), + OPT_INTEGER_F(0, "commit-parents", &ctx.opts.show_largest_commits_by_nr_parents, N_("show N largest commits by parent count"), PARSE_OPT_NONEG), + OPT_INTEGER_F(0, "commit-sizes", &ctx.opts.show_largest_commits_by_size_bytes, N_("show N largest commits by size in bytes"), PARSE_OPT_NONEG), + + OPT_INTEGER_F(0, "tree-entries", &ctx.opts.show_largest_trees_by_nr_entries, N_("show N largest trees by entry count"), PARSE_OPT_NONEG), + OPT_INTEGER_F(0, "tree-sizes", &ctx.opts.show_largest_trees_by_size_bytes, N_("show N largest trees by size in bytes"), PARSE_OPT_NONEG), + + OPT_INTEGER_F(0, "blob-sizes", &ctx.opts.show_largest_blobs_by_size_bytes, N_("show N largest blobs by size in bytes"), PARSE_OPT_NONEG), + OPT_END(), }; @@ -1327,6 +1504,39 @@ int cmd_survey(int argc, const char **argv, const char *prefix, struct repositor fixup_refs_wanted(&ctx); + if (ctx.opts.show_largest_commits_by_nr_parents) + ctx.report.reachable_objects.commits.vec_largest_by_nr_parents = + alloc_large_item_vec( + "largest_commits_by_nr_parents", + "nr_parents", + ctx.opts.show_largest_commits_by_nr_parents); + if (ctx.opts.show_largest_commits_by_size_bytes) + ctx.report.reachable_objects.commits.vec_largest_by_size_bytes = + alloc_large_item_vec( + "largest_commits_by_size_bytes", + "size", + ctx.opts.show_largest_commits_by_size_bytes); + + if (ctx.opts.show_largest_trees_by_nr_entries) + ctx.report.reachable_objects.trees.vec_largest_by_nr_entries = + alloc_large_item_vec( + "largest_trees_by_nr_entries", + "nr_entries", + ctx.opts.show_largest_trees_by_nr_entries); + if (ctx.opts.show_largest_trees_by_size_bytes) + ctx.report.reachable_objects.trees.vec_largest_by_size_bytes = + alloc_large_item_vec( + "largest_trees_by_size_bytes", + "size", + ctx.opts.show_largest_trees_by_size_bytes); + + if (ctx.opts.show_largest_blobs_by_size_bytes) + ctx.report.reachable_objects.blobs.vec_largest_by_size_bytes = + alloc_large_item_vec( + "largest_blobs_by_size_bytes", + "size", + ctx.opts.show_largest_blobs_by_size_bytes); + survey_phase_refs(&ctx); survey_phase_objects(&ctx); From 6e58ef526fd0651c97b9d0218f2cf6cc0660469f Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 15 May 2024 15:56:36 -0400 Subject: [PATCH 012/196] survey: add pathname of blob or tree to large_item_vec Include the pathname of each blob or tree in the large_item_vec to help identify the file or directory associated with the OID and size information. This pathname is computed during the path walk, so it reflects the first observed pathname seen for that OID during the traversal over all of the refs. Since the file or directory could have moved (without being modified), there may be multiple "correct" pathnames for a particular OID. Since we do not control the ref traversal order, we should consider it to be a "suggested pathname" for the OID. Signed-off-by: Jeff Hostetler --- builtin/survey.c | 43 ++++++++++++++++++++++++++++++++----------- 1 file changed, 32 insertions(+), 11 deletions(-) diff --git a/builtin/survey.c b/builtin/survey.c index 7a93c9c991f8b9..2890e11e635f1f 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -156,6 +156,7 @@ static void incr_obj_hist_bin(struct obj_hist_bin *pbin, struct large_item { uint64_t size; struct object_id oid; + struct strbuf name; }; struct large_item_vec { @@ -171,6 +172,7 @@ static struct large_item_vec *alloc_large_item_vec(const char *dimension_label, { struct large_item_vec *vec; size_t flex_len = nr_items * sizeof(struct large_item); + size_t k; if (!nr_items) return NULL; @@ -180,6 +182,9 @@ static struct large_item_vec *alloc_large_item_vec(const char *dimension_label, vec->item_label = strdup(item_label); vec->nr_items = nr_items; + for (k = 0; k < nr_items; k++) + strbuf_init(&vec->items[k].name, 0); + return vec; } @@ -188,6 +193,9 @@ static void free_large_item_vec(struct large_item_vec *vec) if (!vec) return; + for (size_t k = 0; k < vec->nr_items; k++) + strbuf_release(&vec->items[k].name); + free(vec->dimension_label); free(vec->item_label); free(vec); @@ -195,7 +203,8 @@ static void free_large_item_vec(struct large_item_vec *vec) static void maybe_insert_large_item(struct large_item_vec *vec, uint64_t size, - struct object_id *oid) + struct object_id *oid, + const char *name) { size_t rest_len; size_t k; @@ -215,7 +224,14 @@ static void maybe_insert_large_item(struct large_item_vec *vec, if (size < vec->items[k].size) continue; - /* push items[k..] down one and insert it here */ + /* + * The last large_item in the vector is about to be + * overwritten by the previous one during the shift. + * Steal its allocated strbuf and reuse it. + */ + strbuf_release(&vec->items[vec->nr_items - 1].name); + + /* push items[k..] down one and insert data for this item here */ rest_len = (vec->nr_items - k - 1) * sizeof(struct large_item); if (rest_len) @@ -224,6 +240,10 @@ static void maybe_insert_large_item(struct large_item_vec *vec, memset(&vec->items[k], 0, sizeof(struct large_item)); vec->items[k].size = size; oidcpy(&vec->items[k].oid, oid); + strbuf_init(&vec->items[k].name, 0); + if (name && *name) + strbuf_addstr(&vec->items[k].name, name); + return; } } @@ -728,7 +748,7 @@ static void survey_report_largest_vec(struct large_item_vec *vec) return; table.table_name = vec->dimension_label; - strvec_pushl(&table.header, "Size", "OID", NULL); + strvec_pushl(&table.header, "Size", "OID", "Name", NULL); for (int k = 0; k < vec->nr_items; k++) { struct large_item *pk = &vec->items[k]; @@ -736,7 +756,7 @@ static void survey_report_largest_vec(struct large_item_vec *vec) strbuf_reset(&size); strbuf_addf(&size, "%"PRIuMAX, (uintmax_t)pk->size); - insert_table_rowv(&table, size.buf, oid_to_hex(&pk->oid), NULL); + insert_table_rowv(&table, size.buf, oid_to_hex(&pk->oid), pk->name.buf, NULL); } } strbuf_release(&size); @@ -1197,7 +1217,8 @@ static void increment_object_counts( static void increment_totals(struct survey_context *ctx, struct oid_array *oids, - struct survey_report_object_size_summary *summary) + struct survey_report_object_size_summary *summary, + const char *path) { for (size_t i = 0; i < oids->nr; i++) { struct object_info oi = OBJECT_INFO_INIT; @@ -1233,8 +1254,8 @@ static void increment_totals(struct survey_context *ctx, ctx->report.reachable_objects.commits.parent_cnt_pbin[k]++; base = &ctx->report.reachable_objects.commits.base; - maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents, k, &commit->object.oid); - maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes, object_length, &commit->object.oid); + maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents, k, &commit->object.oid, NULL); + maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes, object_length, &commit->object.oid, NULL); break; } case OBJ_TREE: { @@ -1254,8 +1275,8 @@ static void increment_totals(struct survey_context *ctx, pst->sum_entries += nr_entries; - maybe_insert_large_item(pst->vec_largest_by_nr_entries, nr_entries, &tree->object.oid); - maybe_insert_large_item(pst->vec_largest_by_size_bytes, object_length, &tree->object.oid); + maybe_insert_large_item(pst->vec_largest_by_nr_entries, nr_entries, &tree->object.oid, path); + maybe_insert_large_item(pst->vec_largest_by_size_bytes, object_length, &tree->object.oid, path); qb = qbin(nr_entries); incr_obj_hist_bin(&pst->entry_qbin[qb], object_length, disk_sizep); @@ -1266,7 +1287,7 @@ static void increment_totals(struct survey_context *ctx, case OBJ_BLOB: base = &ctx->report.reachable_objects.blobs.base; - maybe_insert_large_item(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes, object_length, &oids->oid[i]); + maybe_insert_large_item(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes, object_length, &oids->oid[i], path); break; default: continue; @@ -1306,7 +1327,7 @@ static void increment_object_totals(struct survey_context *ctx, struct survey_report_object_size_summary *total; struct survey_report_object_size_summary summary = { 0 }; - increment_totals(ctx, oids, &summary); + increment_totals(ctx, oids, &summary, path); switch (type) { case OBJ_COMMIT: From d012ace0925e9412387404085b73a422e3bdade1 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 15 May 2024 17:44:41 -0400 Subject: [PATCH 013/196] survey: add commit-oid to large_item detail Signed-off-by: Jeff Hostetler --- builtin/survey.c | 35 +++++++++++++++++++++++++++-------- 1 file changed, 27 insertions(+), 8 deletions(-) diff --git a/builtin/survey.c b/builtin/survey.c index 2890e11e635f1f..0d263a633c3ee7 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -156,7 +156,21 @@ static void incr_obj_hist_bin(struct obj_hist_bin *pbin, struct large_item { uint64_t size; struct object_id oid; + + /* + * For blobs and trees the name field is the pathname of the + * file or directory. Root trees will have a zero-length + * name. The name field is not currenly used for commits. + */ struct strbuf name; + + /* + * For blobs and trees remember the transient commit from + * the treewalk so that we can say that this large item + * first appeared in this commit (relative to the treewalk + * order). + */ + struct object_id containing_commit_oid; }; struct large_item_vec { @@ -204,7 +218,8 @@ static void free_large_item_vec(struct large_item_vec *vec) static void maybe_insert_large_item(struct large_item_vec *vec, uint64_t size, struct object_id *oid, - const char *name) + const char *name, + const struct object_id *containing_commit_oid) { size_t rest_len; size_t k; @@ -240,6 +255,7 @@ static void maybe_insert_large_item(struct large_item_vec *vec, memset(&vec->items[k], 0, sizeof(struct large_item)); vec->items[k].size = size; oidcpy(&vec->items[k].oid, oid); + oidcpy(&vec->items[k].containing_commit_oid, containing_commit_oid ? containing_commit_oid : null_oid()); strbuf_init(&vec->items[k].name, 0); if (name && *name) strbuf_addstr(&vec->items[k].name, name); @@ -748,7 +764,7 @@ static void survey_report_largest_vec(struct large_item_vec *vec) return; table.table_name = vec->dimension_label; - strvec_pushl(&table.header, "Size", "OID", "Name", NULL); + strvec_pushl(&table.header, "Size", "OID", "Name", "Commit", NULL); for (int k = 0; k < vec->nr_items; k++) { struct large_item *pk = &vec->items[k]; @@ -756,7 +772,10 @@ static void survey_report_largest_vec(struct large_item_vec *vec) strbuf_reset(&size); strbuf_addf(&size, "%"PRIuMAX, (uintmax_t)pk->size); - insert_table_rowv(&table, size.buf, oid_to_hex(&pk->oid), pk->name.buf, NULL); + insert_table_rowv(&table, size.buf, oid_to_hex(&pk->oid), pk->name.buf, + is_null_oid(&pk->containing_commit_oid) ? + "" : oid_to_hex(&pk->containing_commit_oid), + NULL); } } strbuf_release(&size); @@ -1254,8 +1273,8 @@ static void increment_totals(struct survey_context *ctx, ctx->report.reachable_objects.commits.parent_cnt_pbin[k]++; base = &ctx->report.reachable_objects.commits.base; - maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents, k, &commit->object.oid, NULL); - maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes, object_length, &commit->object.oid, NULL); + maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents, k, &commit->object.oid, NULL, &commit->object.oid); + maybe_insert_large_item(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes, object_length, &commit->object.oid, NULL, &commit->object.oid); break; } case OBJ_TREE: { @@ -1275,8 +1294,8 @@ static void increment_totals(struct survey_context *ctx, pst->sum_entries += nr_entries; - maybe_insert_large_item(pst->vec_largest_by_nr_entries, nr_entries, &tree->object.oid, path); - maybe_insert_large_item(pst->vec_largest_by_size_bytes, object_length, &tree->object.oid, path); + maybe_insert_large_item(pst->vec_largest_by_nr_entries, nr_entries, &tree->object.oid, path, NULL); + maybe_insert_large_item(pst->vec_largest_by_size_bytes, object_length, &tree->object.oid, path, NULL); qb = qbin(nr_entries); incr_obj_hist_bin(&pst->entry_qbin[qb], object_length, disk_sizep); @@ -1287,7 +1306,7 @@ static void increment_totals(struct survey_context *ctx, case OBJ_BLOB: base = &ctx->report.reachable_objects.blobs.base; - maybe_insert_large_item(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes, object_length, &oids->oid[i], path); + maybe_insert_large_item(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes, object_length, &oids->oid[i], path, NULL); break; default: continue; From 65e49550c2be8faf9c25a6983a434fb005de6cb4 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 20 May 2024 17:23:39 -0400 Subject: [PATCH 014/196] survey: add commit name-rev lookup to each large_item Signed-off-by: Jeff Hostetler --- builtin/survey.c | 90 ++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 87 insertions(+), 3 deletions(-) diff --git a/builtin/survey.c b/builtin/survey.c index 0d263a633c3ee7..a4f99e87361142 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -13,6 +13,7 @@ #include "ref-filter.h" #include "refs.h" #include "revision.h" +#include "run-command.h" #include "strbuf.h" #include "strvec.h" #include "trace2.h" @@ -171,6 +172,12 @@ struct large_item { * order). */ struct object_id containing_commit_oid; + + /* + * Lookup `containing_commit_oid` using `git name-rev`. + * Lazy allocate this post-treewalk. + */ + struct strbuf name_rev; }; struct large_item_vec { @@ -207,8 +214,10 @@ static void free_large_item_vec(struct large_item_vec *vec) if (!vec) return; - for (size_t k = 0; k < vec->nr_items; k++) + for (size_t k = 0; k < vec->nr_items; k++) { strbuf_release(&vec->items[k].name); + strbuf_release(&vec->items[k].name_rev); + } free(vec->dimension_label); free(vec->item_label); @@ -243,6 +252,9 @@ static void maybe_insert_large_item(struct large_item_vec *vec, * The last large_item in the vector is about to be * overwritten by the previous one during the shift. * Steal its allocated strbuf and reuse it. + * + * We can ignore .name_rev because it will not be + * allocated until after the treewalk. */ strbuf_release(&vec->items[vec->nr_items - 1].name); @@ -764,7 +776,7 @@ static void survey_report_largest_vec(struct large_item_vec *vec) return; table.table_name = vec->dimension_label; - strvec_pushl(&table.header, "Size", "OID", "Name", "Commit", NULL); + strvec_pushl(&table.header, "Size", "OID", "Name", "Commit", "Name-Rev", NULL); for (int k = 0; k < vec->nr_items; k++) { struct large_item *pk = &vec->items[k]; @@ -775,6 +787,7 @@ static void survey_report_largest_vec(struct large_item_vec *vec) insert_table_rowv(&table, size.buf, oid_to_hex(&pk->oid), pk->name.buf, is_null_oid(&pk->containing_commit_oid) ? "" : oid_to_hex(&pk->containing_commit_oid), + pk->name_rev.len ? pk->name_rev.buf : "", NULL); } } @@ -1125,6 +1138,73 @@ static void do_load_refs(struct survey_context *ctx, ref_sorting_release(sorting); } +/* + * Try to run `git name-rev` on each of the containing-commit-oid's + * in this large-item-vec to get a pretty name for each OID. Silently + * ignore errors if it fails because this info is nice to have but not + * essential. + */ +static void large_item_vec_lookup_name_rev(struct survey_context *ctx, + struct large_item_vec *vec) +{ + struct child_process cp = CHILD_PROCESS_INIT; + struct strbuf in = STRBUF_INIT; + struct strbuf out = STRBUF_INIT; + const char *line; + size_t k; + + if (!vec || !vec->nr_items) + return; + + ctx->progress_total += vec->nr_items; + display_progress(ctx->progress, ctx->progress_total); + + for (k = 0; k < vec->nr_items; k++) + strbuf_addf(&in, "%s\n", oid_to_hex(&vec->items[k].containing_commit_oid)); + + cp.git_cmd = 1; + strvec_pushl(&cp.args, "name-rev", "--name-only", "--annotate-stdin", NULL); + if (pipe_command(&cp, in.buf, in.len, &out, 0, NULL, 0)) { + strbuf_release(&in); + strbuf_release(&out); + return; + } + + line = out.buf; + k = 0; + while (*line) { + const char *eol = strchrnul(line, '\n'); + + strbuf_init(&vec->items[k].name_rev, 0); + strbuf_add(&vec->items[k].name_rev, line, (eol - line)); + + line = eol + 1; + k++; + } + + strbuf_release(&in); + strbuf_release(&out); +} + +static void do_lookup_name_rev(struct survey_context *ctx) +{ + if (ctx->opts.show_progress) { + ctx->progress_total = 0; + ctx->progress = start_progress(_("Resolving name-revs..."), 0); + } + + large_item_vec_lookup_name_rev(ctx, ctx->report.reachable_objects.commits.vec_largest_by_nr_parents); + large_item_vec_lookup_name_rev(ctx, ctx->report.reachable_objects.commits.vec_largest_by_size_bytes); + + large_item_vec_lookup_name_rev(ctx, ctx->report.reachable_objects.trees.vec_largest_by_nr_entries); + large_item_vec_lookup_name_rev(ctx, ctx->report.reachable_objects.trees.vec_largest_by_size_bytes); + + large_item_vec_lookup_name_rev(ctx, ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes); + + if (ctx->opts.show_progress) + stop_progress(&ctx->progress); +} + /* * The REFS phase: * @@ -1478,7 +1558,11 @@ static void survey_phase_objects(struct survey_context *ctx) release_revisions(&revs); trace2_region_leave("survey", "phase/objects", ctx->repo); -} + + + trace2_region_enter("survey", "phase/namerev", the_repository); + do_lookup_name_rev(ctx); + trace2_region_enter("survey", "phase/namerev", the_repository);} int cmd_survey(int argc, const char **argv, const char *prefix, struct repository *repo) { From 9d157fdd4057fba313ae02c2269a7a8768223ec7 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 4 Jun 2024 10:37:23 -0400 Subject: [PATCH 015/196] survey: add --no-name-rev option Computing `git name-rev` on each commit, tree, and blob in each of the various large_item_vec can be very expensive if there are too many refs, especially if the user doesn't need the result. Lets make it optional. The `--no-name-rev` option can save 50 calls to `git name-rev` since we have 5 large_item_vec's and each defaults to 10 items. Signed-off-by: Jeff Hostetler --- Documentation/config/survey.txt | 4 ++++ Documentation/git-survey.txt | 4 ++++ builtin/survey.c | 39 +++++++++++++++++++++++---------- 3 files changed, 35 insertions(+), 12 deletions(-) diff --git a/Documentation/config/survey.txt b/Documentation/config/survey.txt index fd2d7e153b38ce..f3ae768933fe1b 100644 --- a/Documentation/config/survey.txt +++ b/Documentation/config/survey.txt @@ -4,6 +4,10 @@ survey.*:: background with these options. + -- + survey.namerev:: + Boolean to show/hide `git name-rev` information for each + reported commit and the containing commit of each + reported tree and blob. verbose:: This boolean value implies the `--[no-]verbose` option. progress:: diff --git a/Documentation/git-survey.txt b/Documentation/git-survey.txt index e47924d0351ca1..cb1fb3c37c3706 100644 --- a/Documentation/git-survey.txt +++ b/Documentation/git-survey.txt @@ -32,6 +32,10 @@ OPTIONS --progress:: Show progress. This is automatically enabled when interactive. +--[no-]name-rev:: + Print `git name-rev` output for each commit, tree, and blob. + Defaults to true. + Ref Selection ~~~~~~~~~~~~~ diff --git a/builtin/survey.c b/builtin/survey.c index a4f99e87361142..db6fd0956b5337 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -42,6 +42,7 @@ static struct survey_refs_wanted default_ref_options = { struct survey_opts { int verbose; int show_progress; + int show_name_rev; int show_largest_commits_by_nr_parents; int show_largest_commits_by_size_bytes; @@ -767,7 +768,7 @@ static void survey_report_commit_parents(struct survey_context *ctx) clear_table(&table); } -static void survey_report_largest_vec(struct large_item_vec *vec) +static void survey_report_largest_vec(struct survey_context *ctx, struct large_item_vec *vec) { struct survey_table table = SURVEY_TABLE_INIT; struct strbuf size = STRBUF_INIT; @@ -776,7 +777,7 @@ static void survey_report_largest_vec(struct large_item_vec *vec) return; table.table_name = vec->dimension_label; - strvec_pushl(&table.header, "Size", "OID", "Name", "Commit", "Name-Rev", NULL); + strvec_pushl(&table.header, "Size", "OID", "Name", "Commit", ctx->opts.show_name_rev ? "Name-Rev" : NULL, NULL); for (int k = 0; k < vec->nr_items; k++) { struct large_item *pk = &vec->items[k]; @@ -787,7 +788,7 @@ static void survey_report_largest_vec(struct large_item_vec *vec) insert_table_rowv(&table, size.buf, oid_to_hex(&pk->oid), pk->name.buf, is_null_oid(&pk->containing_commit_oid) ? "" : oid_to_hex(&pk->containing_commit_oid), - pk->name_rev.len ? pk->name_rev.buf : "", + !ctx->opts.show_name_rev ? NULL : pk->name_rev.len ? pk->name_rev.buf : "", NULL); } } @@ -977,11 +978,11 @@ static void survey_report_plaintext(struct survey_context *ctx) survey_report_plaintext_sorted_size( &ctx->report.top_paths_by_inflate[REPORT_TYPE_BLOB]); - survey_report_largest_vec(ctx->report.reachable_objects.commits.vec_largest_by_nr_parents); - survey_report_largest_vec(ctx->report.reachable_objects.commits.vec_largest_by_size_bytes); - survey_report_largest_vec(ctx->report.reachable_objects.trees.vec_largest_by_nr_entries); - survey_report_largest_vec(ctx->report.reachable_objects.trees.vec_largest_by_size_bytes); - survey_report_largest_vec(ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes); + survey_report_largest_vec(ctx, ctx->report.reachable_objects.commits.vec_largest_by_nr_parents); + survey_report_largest_vec(ctx, ctx->report.reachable_objects.commits.vec_largest_by_size_bytes); + survey_report_largest_vec(ctx, ctx->report.reachable_objects.trees.vec_largest_by_nr_entries); + survey_report_largest_vec(ctx, ctx->report.reachable_objects.trees.vec_largest_by_size_bytes); + survey_report_largest_vec(ctx, ctx->report.reachable_objects.blobs.vec_largest_by_size_bytes); } /* @@ -1053,6 +1054,10 @@ static int survey_load_config_cb(const char *var, const char *value, ctx->opts.show_progress = git_config_bool(var, value); return 0; } + if (!strcmp(var, "survey.namerev")) { + ctx->opts.show_name_rev = git_config_bool(var, value); + return 0; + } if (!strcmp(var, "survey.showcommitparents")) { ctx->opts.show_largest_commits_by_nr_parents = git_config_ulong(var, value, cctx->kvi); return 0; @@ -1188,6 +1193,13 @@ static void large_item_vec_lookup_name_rev(struct survey_context *ctx, static void do_lookup_name_rev(struct survey_context *ctx) { + /* + * `git name-rev` can be very expensive when there are lots of + * refs, so make it optional. + */ + if (!ctx->opts.show_name_rev) + return; + if (ctx->opts.show_progress) { ctx->progress_total = 0; ctx->progress = start_progress(_("Resolving name-revs..."), 0); @@ -1559,10 +1571,12 @@ static void survey_phase_objects(struct survey_context *ctx) release_revisions(&revs); trace2_region_leave("survey", "phase/objects", ctx->repo); - - trace2_region_enter("survey", "phase/namerev", the_repository); - do_lookup_name_rev(ctx); - trace2_region_enter("survey", "phase/namerev", the_repository);} + if (ctx->opts.show_name_rev) { + trace2_region_enter("survey", "phase/namerev", the_repository); + do_lookup_name_rev(ctx); + trace2_region_enter("survey", "phase/namerev", the_repository); + } +} int cmd_survey(int argc, const char **argv, const char *prefix, struct repository *repo) { @@ -1586,6 +1600,7 @@ int cmd_survey(int argc, const char **argv, const char *prefix, struct repositor static struct option survey_options[] = { OPT__VERBOSE(&ctx.opts.verbose, N_("verbose output")), OPT_BOOL(0, "progress", &ctx.opts.show_progress, N_("show progress")), + OPT_BOOL(0, "name-rev", &ctx.opts.show_name_rev, N_("run name-rev on each reported commit")), OPT_INTEGER('n', "top", &ctx.opts.top_nr, N_("number of entries to include in detail tables")), From 36931e9e3c5dbbe11496bcc739e6b14f8e4988ad Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 17 Jun 2024 15:20:05 -0400 Subject: [PATCH 016/196] survey: started TODO list at bottom of source file --- builtin/survey.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 46 insertions(+) diff --git a/builtin/survey.c b/builtin/survey.c index db6fd0956b5337..1488ec1883c319 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -1685,3 +1685,49 @@ int cmd_survey(int argc, const char **argv, const char *prefix, struct repositor clear_survey_context(&ctx); return 0; } + +/* + * NEEDSWORK: The following is a bit of a laundry list of things + * that I'd like to add. + * + * [] Dump stats on all of the packfiles. The number and size of each. + * Whether each is in the .git directory or in an alternate. The state + * of the IDX or MIDX files and etc. Delta chain stats. All of this + * data is relative to the "lived-in" state of the repository. Stuff + * that may change after a GC or repack. + * + * [] Dump stats on each remote. When we fetch from a remote the size + * of the response is related to the set of haves on the server. You + * can see this in `GIT_TRACE_CURL=1 git fetch`. We get a `ls-refs` + * payload that lists all of the branches and tags on the server, so + * at a minimum the RefName and SHA for each. But for annotated tags + * we also get the peeled SHA. The size of this overhead on every + * fetch is proporational to the size of the `git ls-remote` response + * (roughly, although the latter repeats the RefName of the peeled + * tag). If, for example, you have 500K refs on a remote, you're + * going to have a long "haves" message, so every fetch will be slow + * just because of that overhead (not counting new objects to be + * downloaded). + * + * Note that the local set of tags in "refs/tags/" is a union over all + * remotes. However, since most people only have one remote, we can + * probaly estimate the overhead value directly from the size of the + * set of "refs/tags/" that we visited while building the `ref_info` + * and `ref_array` and not need to ask the remote. + * + * [] Dump info on the complexity of the DAG. Criss-cross merges. + * The number of edges that must be touched to compute merge bases. + * Edge length. The number of parallel lanes in the history that must + * be navigated to get to the merge base. What affects the cost of + * the Ahead/Behind computation? How often do criss-crosses occur and + * do they cause various operations to slow down? + * + * [] If there are primary branches (like "main" or "master") are they + * always on the left side of merges? Does the graph have a clean + * left edge? Or are there normal and "backwards" merges? Do these + * cause problems at scale? + * + * [] If we have a hierarchy of FI/RI branches like "L1", "L2, ..., + * can we learn anything about the shape of the repo around these FI + * and RI integrations? + */ From cbc68151b521471eb3a7c709465e8d0f5f2448f7 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 28 Jun 2024 15:22:46 -0400 Subject: [PATCH 017/196] survey: expanded TODO list at the bottom of the source file Signed-off-by: Jeff Hostetler --- builtin/survey.c | 148 +++++++++++++++++++++++++++++++++++++---------- 1 file changed, 116 insertions(+), 32 deletions(-) diff --git a/builtin/survey.c b/builtin/survey.c index 1488ec1883c319..ee8500dd5fe177 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -1687,47 +1687,131 @@ int cmd_survey(int argc, const char **argv, const char *prefix, struct repositor } /* - * NEEDSWORK: The following is a bit of a laundry list of things - * that I'd like to add. + * NEEDSWORK: So far, I only have iteration on the requested set of + * refs and treewalk/reachable objects on that set of refs. The + * following is a bit of a laundry list of things that I'd like to + * add. * * [] Dump stats on all of the packfiles. The number and size of each. - * Whether each is in the .git directory or in an alternate. The state - * of the IDX or MIDX files and etc. Delta chain stats. All of this - * data is relative to the "lived-in" state of the repository. Stuff - * that may change after a GC or repack. + * Whether each is in the .git directory or in an alternate. The + * state of the IDX or MIDX files and etc. Delta chain stats. All + * of this data is relative to the "lived-in" state of the + * repository. Stuff that may change after a GC or repack. + * + * [] Clone and Index stats. partial, shallow, sparse-checkout, + * sparse-index, etc. Hydration stats. * * [] Dump stats on each remote. When we fetch from a remote the size - * of the response is related to the set of haves on the server. You - * can see this in `GIT_TRACE_CURL=1 git fetch`. We get a `ls-refs` - * payload that lists all of the branches and tags on the server, so - * at a minimum the RefName and SHA for each. But for annotated tags - * we also get the peeled SHA. The size of this overhead on every - * fetch is proporational to the size of the `git ls-remote` response - * (roughly, although the latter repeats the RefName of the peeled - * tag). If, for example, you have 500K refs on a remote, you're - * going to have a long "haves" message, so every fetch will be slow - * just because of that overhead (not counting new objects to be - * downloaded). + * of the response is related to the set of haves on the server. + * You can see this in `GIT_TRACE_CURL=1 git fetch`. We get a + * `ls-refs` payload that lists all of the branches and tags on the + * server, so at a minimum the RefName and SHA for each. But for + * annotated tags we also get the peeled SHA. The size of this + * overhead on every fetch is proporational to the size of the `git + * ls-remote` response (roughly, although the latter repeats the + * RefName of the peeled tag). If, for example, you have 500K refs + * on a remote, you're going to have a long "haves" message, so + * every fetch will be slow just because of that overhead (not + * counting new objects to be downloaded). * - * Note that the local set of tags in "refs/tags/" is a union over all - * remotes. However, since most people only have one remote, we can - * probaly estimate the overhead value directly from the size of the - * set of "refs/tags/" that we visited while building the `ref_info` - * and `ref_array` and not need to ask the remote. + * Note that the local set of tags in "refs/tags/" is a union over + * all remotes. However, since most people only have one remote, + * we can probaly estimate the overhead value directly from the + * size of the set of "refs/tags/" that we visited while building + * the `ref_info` and `ref_array` and not need to ask the remote. * * [] Dump info on the complexity of the DAG. Criss-cross merges. - * The number of edges that must be touched to compute merge bases. - * Edge length. The number of parallel lanes in the history that must - * be navigated to get to the merge base. What affects the cost of - * the Ahead/Behind computation? How often do criss-crosses occur and - * do they cause various operations to slow down? + * The number of edges that must be touched to compute merge bases. + * Edge length. The number of parallel lanes in the history that + * must be navigated to get to the merge base. What affects the + * cost of the Ahead/Behind computation? How often do + * criss-crosses occur and do they cause various operations to slow + * down? * * [] If there are primary branches (like "main" or "master") are they - * always on the left side of merges? Does the graph have a clean - * left edge? Or are there normal and "backwards" merges? Do these - * cause problems at scale? + * always on the left side of merges? Does the graph have a clean + * left edge? Or are there normal and "backwards" merges? Do + * these cause problems at scale? * * [] If we have a hierarchy of FI/RI branches like "L1", "L2, ..., - * can we learn anything about the shape of the repo around these FI - * and RI integrations? + * can we learn anything about the shape of the repo around these + * FI and RI integrations? + * + * [] Do we need a no-PII flag to omit pathnames or branch/tag names + * in the various histograms? (This would turn off --name-rev + * too.) + * + * [] I have so far avoided adding opinions about individual fields + * (such as the way `git-sizer` prints a row of stars or bangs in + * the last column). + * + * I'm wondering if that is a job of this executable or if it + * should be done in a post-processing step using the JSON output. + * + * My problem with the `git-sizer` approach is that it doesn't give + * the (casual) user any information on why it has stars or bangs. + * And there isn't a good way to print detailed information in the + * ASCII-art tables that would be easy to understand. + * + * [] For example, a large number of refs does not define a cliff. + * Performance will drop off (linearly, quadratically, ... ??). + * The tool should refer them to article(s) talking about the + * different problems that it could cause. So should `git + * survey` just print the number and (implicitly) refer them to + * the man page (chapter/verse) or to a tool that will interpret + * the number and explain it? + * + * [] Alternatively, should `git survey` do that analysis too and + * just print footnotes for each large number? + * + * [] The computation of the raw survey JSON data can take HOURS on + * a very large repo (like Windows), so I'm wondering if we + * want to keep the opinion portion separate. + * + * [] In addition to opinions based on the static data, I would like + * to dump the JSON results (or the Trace2 telemetry) into a DB and + * aggregate it with other users. + * + * Granted, they should all see the same DAG and the same set of + * reachable objects, but we could average across all datasets + * generated on a particular date and detect outlier users. + * + * [] Maybe someone cloned from the `_full` endpoint rather than + * the limited refs endpoint. + * + * [] Maybe that user is having problems with repacking / GC / + * maintenance without knowing it. + * + * [] I'd also like to dump use the DB to compare survey datasets over + * a time. How fast is their repository growing and in what ways? + * + * [] I'd rather have the delta analysis NOT be inside `git + * survey`, so it makes sense to consider having all of it in a + * post-process step. + * + * [] Another reason to put the opinion analysis in a post-process + * is that it would be easier to generate plots on the data tables. + * Granted, we can get plots from telemetry, but a stand-alone user + * could run the JSON thru python or jq or something and generate + * something nicer than ASCII-art and it could handle cross-referencing + * and hyperlinking to helpful information on each issue. + * + * [] I think there are several classes of data that we can report on: + * + * [] The "inherit repo properties", such as the shape and size of + * the DAG -- these should be universal in each enlistment. + * + * [] The "ODB lived in properties", such as the efficiency + * of the repack and things like partial and shallow clone. + * These will vary, but indicate health of the ODB. + * + * [] The "index related properties", such as sparse-checkout, + * sparse-index, cache-tree, untracked-cache, fsmonitor, and + * etc. These will also vary, but are more like knobs for + * the user to adjust. + * + * [] I want to compare these with Matt's "dimensions of scale" + * notes and see if there are other pieces of data that we + * could compute/consider. + * */ From 75357a35a16357219ad8f2fd8691a9051dab34a5 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 1 Jul 2024 12:07:01 -0400 Subject: [PATCH 018/196] survey: expanded TODO with more notes Signed-off-by: Jeff Hostetler --- builtin/survey.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/builtin/survey.c b/builtin/survey.c index ee8500dd5fe177..5bf1adb6668eed 100644 --- a/builtin/survey.c +++ b/builtin/survey.c @@ -1720,6 +1720,16 @@ int cmd_survey(int argc, const char **argv, const char *prefix, struct repositor * size of the set of "refs/tags/" that we visited while building * the `ref_info` and `ref_array` and not need to ask the remote. * + * [] Should the "string length of refnames / remote refs", for + * example, be sub-divided by remote so we can project the + * cost of the haves/wants overhead a fetch. + * + * [] Can we examine the merge commits and classify them as clean or + * dirty? (ie. ones with merge conflicts that needed to be + * addressed during the merge itself.) + * + * [] Do dirty merges affect performance of later operations? + * * [] Dump info on the complexity of the DAG. Criss-cross merges. * The number of edges that must be touched to compute merge bases. * Edge length. The number of parallel lanes in the history that From 590aed4fd5c8d801c688adcc75ab813885a4a56e Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Wed, 5 Apr 2017 10:58:09 -0600 Subject: [PATCH 019/196] reset --stdin: trim carriage return from the paths While using the reset --stdin feature on windows path added may have a \r at the end of the path that wasn't getting removed so didn't match the path in the index and wasn't reset. Signed-off-by: Kevin Willford --- t/t7108-reset-stdin.sh | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/t/t7108-reset-stdin.sh b/t/t7108-reset-stdin.sh index b7cbcbf869296c..db5483b8f10052 100755 --- a/t/t7108-reset-stdin.sh +++ b/t/t7108-reset-stdin.sh @@ -29,4 +29,13 @@ test_expect_success '--stdin requires --mixed' ' git reset --mixed --stdin list && + git reset --stdin Date: Tue, 23 Apr 2024 12:21:01 +0200 Subject: [PATCH 020/196] Identify microsoft/git via a distinct version suffix It has been a long-standing practice in Git for Windows to append `.windows.`, and in microsoft/git to append `.vfs.0.0`. Let's keep doing that. Signed-off-by: Johannes Schindelin --- GIT-VERSION-GEN | 3 +++ 1 file changed, 3 insertions(+) diff --git a/GIT-VERSION-GEN b/GIT-VERSION-GEN index b3265f7becb422..83eabe6850f7c0 100755 --- a/GIT-VERSION-GEN +++ b/GIT-VERSION-GEN @@ -3,6 +3,9 @@ GVF=GIT-VERSION-FILE DEF_VER=v2.47.0 +# Identify microsoft/git via a distinct version suffix +DEF_VER=$DEF_VER.vfs.0.0 + LF=' ' From a0ff4112b413935a326461e4486fa7120f4f1e24 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 4 Apr 2017 12:04:11 +0200 Subject: [PATCH 021/196] gvfs: ensure that the version is based on a GVFS tag Since we really want to be based on a `.vfs.*` tag, let's make sure that there was a new-enough one, i.e. one that agrees with the first three version numbers of the recorded default version. This prevents e.g. v2.22.0.vfs.0.. from being used when the current release train was not yet tagged. It is important to get the first three numbers of the version right because e.g. Scalar makes decisions depending on those (such as assuming that the `git maintenance` built-in is not available, even though it actually _is_ available). Signed-off-by: Johannes Schindelin --- GIT-VERSION-GEN | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/GIT-VERSION-GEN b/GIT-VERSION-GEN index 83eabe6850f7c0..9678f88decd3b7 100755 --- a/GIT-VERSION-GEN +++ b/GIT-VERSION-GEN @@ -15,10 +15,15 @@ if test -f version then VN=$(cat version) || VN="$DEF_VER" elif { test -d "${GIT_DIR:-.git}" || test -f .git; } && - VN=$(git describe --match "v[0-9]*" HEAD 2>/dev/null) && + VN=$(git describe --match "v[0-9]*vfs*" HEAD 2>/dev/null) && case "$VN" in *$LF*) (exit 1) ;; v[0-9]*) + if test "${VN%%.vfs.*}" != "${DEF_VER%%.vfs.*}" + then + echo "Found version $VN, which is not based on $DEF_VER" >&2 + exit 1 + fi git update-index -q --refresh test -z "$(git diff-index --name-only HEAD --)" || VN="$VN-dirty" ;; From 6ef654049facb02c5eb1f16d9364fe39fd2addc0 Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Tue, 24 Jan 2017 17:30:59 +0100 Subject: [PATCH 022/196] gvfs: add a GVFS-specific header file This header file will accumulate GVFS-specific definitions. Signed-off-by: Kevin Willford --- gvfs.h | 9 +++++++++ 1 file changed, 9 insertions(+) create mode 100644 gvfs.h diff --git a/gvfs.h b/gvfs.h new file mode 100644 index 00000000000000..b6dbe85eae4071 --- /dev/null +++ b/gvfs.h @@ -0,0 +1,9 @@ +#ifndef GVFS_H +#define GVFS_H + +/* + * This file is for the specific settings and methods + * used for GVFS functionality + */ + +#endif /* GVFS_H */ From 030b553b912f5055e6491c24f703452fde21ff9a Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 18 May 2021 22:48:24 +0200 Subject: [PATCH 023/196] git_config_set_multivar_in_file_gently(): add a lock timeout In particular when multiple processes want to write to the config simultaneously, it would come in handy to not fail immediately when another process locked the config, but to gently try again. This will help with Scalar's functional test suite which wants to register multiple repositories for maintenance semi-simultaneously. As not all code paths calling this function read the config (e.g. `git config`), we have to read the config setting via `git_config_get_ulong()`. Signed-off-by: Johannes Schindelin --- Documentation/config/core.txt | 9 +++++++++ config.c | 8 +++++++- 2 files changed, 16 insertions(+), 1 deletion(-) diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index 458a3f4f7f5d80..c81882b358ec8d 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -777,3 +777,12 @@ core.WSLCompat:: The default value is false. When set to true, Git will set the mode bits of the file in the way of wsl, so that the executable flag of files can be set or read correctly. + +core.configWriteLockTimeoutMS:: + When processes try to write to the config concurrently, it is likely + that one process "wins" and the other process(es) fail to lock the + config file. By configuring a timeout larger than zero, Git can be + told to try to lock the config again a couple times within the + specified timeout. If the timeout is configure to zero (which is the + default), Git will fail immediately when the config is already + locked. diff --git a/config.c b/config.c index bfca0ea5fd97f5..fe1992220934cc 100644 --- a/config.c +++ b/config.c @@ -3203,6 +3203,7 @@ int repo_config_set_multivar_in_file_gently(struct repository *r, const char *comment, unsigned flags) { + static unsigned long timeout_ms = ULONG_MAX; int fd = -1, in_fd = -1; int ret; struct lock_file lock = LOCK_INIT; @@ -3223,11 +3224,16 @@ int repo_config_set_multivar_in_file_gently(struct repository *r, if (!config_filename) config_filename = filename_buf = repo_git_path(r, "config"); + if ((long)timeout_ms < 0 && + git_config_get_ulong("core.configWriteLockTimeoutMS", &timeout_ms)) + timeout_ms = 0; + /* * The lock serves a purpose in addition to locking: the new * contents of .git/config will be written into it. */ - fd = hold_lock_file_for_update(&lock, config_filename, 0); + fd = hold_lock_file_for_update_timeout(&lock, config_filename, 0, + timeout_ms); if (fd < 0) { error_errno(_("could not lock config file %s"), config_filename); ret = CONFIG_NO_LOCK; From b73befa6386138a011e4f76552d559bb5fdc2fed Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 18 May 2021 23:22:56 +0200 Subject: [PATCH 024/196] scalar: set the config write-lock timeout to 150ms By default, Git fails immediately when locking a config file for writing fails due to an existing lock. With this change, Scalar-registered repositories will fall back to trying a couple times within a 150ms timeout. Signed-off-by: Johannes Schindelin --- scalar.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/scalar.c b/scalar.c index 7db9c5bbfe03df..18d3504c3fbee8 100644 --- a/scalar.c +++ b/scalar.c @@ -171,6 +171,7 @@ static int set_recommended_config(int reconfigure) { "core.safeCRLF", "false" }, { "fetch.showForcedUpdates", "false" }, { "push.usePathWalk", "true" }, + { "core.configWriteLockTimeoutMS", "150" }, { NULL, NULL }, }; int i; @@ -212,6 +213,11 @@ static int set_recommended_config(int reconfigure) static int toggle_maintenance(int enable) { + unsigned long ul; + + if (git_config_get_ulong("core.configWriteLockTimeoutMS", &ul)) + git_config_push_parameter("core.configWriteLockTimeoutMS=150"); + return run_git("maintenance", enable ? "start" : "unregister", enable ? NULL : "--force", @@ -221,10 +227,14 @@ static int toggle_maintenance(int enable) static int add_or_remove_enlistment(int add) { int res; + unsigned long ul; if (!the_repository->worktree) die(_("Scalar enlistments require a worktree")); + if (git_config_get_ulong("core.configWriteLockTimeoutMS", &ul)) + git_config_push_parameter("core.configWriteLockTimeoutMS=150"); + res = run_git("config", "--global", "--get", "--fixed-value", "scalar.repo", the_repository->worktree, NULL); From d08111155aae0669b6680154b35737bc09a5c204 Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Wed, 16 Jun 2021 10:01:37 -0400 Subject: [PATCH 025/196] scalar: add docs from microsoft/scalar These docs have been altered to fit the version implemented in C within microsoft/git. This means in particular that the advanced.md file no longer applied at all. Some other areas were removed or significantly edited. Signed-off-by: Derrick Stolee --- contrib/scalar/docs/faq.md | 51 ++++++++++++++ contrib/scalar/docs/getting-started.md | 93 ++++++++++++++++++++++++++ contrib/scalar/docs/index.md | 50 ++++++++++++++ contrib/scalar/docs/philosophy.md | 66 ++++++++++++++++++ contrib/scalar/docs/troubleshooting.md | 20 ++++++ 5 files changed, 280 insertions(+) create mode 100644 contrib/scalar/docs/faq.md create mode 100644 contrib/scalar/docs/getting-started.md create mode 100644 contrib/scalar/docs/index.md create mode 100644 contrib/scalar/docs/philosophy.md create mode 100644 contrib/scalar/docs/troubleshooting.md diff --git a/contrib/scalar/docs/faq.md b/contrib/scalar/docs/faq.md new file mode 100644 index 00000000000000..a14f78a996d5d5 --- /dev/null +++ b/contrib/scalar/docs/faq.md @@ -0,0 +1,51 @@ +Frequently Asked Questions +========================== + +Using Scalar +------------ + +### I don't want a sparse clone, I want every file after I clone! + +Run `scalar clone --full-clone ` to initialize your repo to include +every file. You can switch to a sparse-checkout later by running +`git sparse-checkout init --cone`. + +### I already cloned without `--full-clone`. How do I get everything? + +Run `git sparse-checkout disable`. + +Scalar Design Decisions +----------------------- + +There may be many design decisions within Scalar that are confusing at first +glance. Some of them may cause friction when you use Scalar with your existing +repos and existing habits. + +> Scalar has the most benefit when users design repositories +> with efficient patterns. + +For example: Scalar uses the sparse-checkout feature to limit the size of the +working directory within a large monorepo. It is designed to work efficiently +with monorepos that are highly componentized, allowing most developers to +need many fewer files in their daily work. + +### Why does `scalar clone` create a `/src` folder? + +Scalar uses a file system watcher to keep track of changes under this `src` folder. +Any activity in this folder is assumed to be important to Git operations. By +creating the `src` folder, we are making it easy for your build system to +create output folders outside the `src` directory. We commonly see systems +create folders for build outputs and package downloads. Scalar itself creates +these folders during its builds. + +Your build system may create build artifacts such as `.obj` or `.lib` files +next to your source code. These are commonly "hidden" from Git using +`.gitignore` files. Having such artifacts in your source tree creates +additional work for Git because it needs to look at these files and match them +against the `.gitignore` patterns. + +By following the `src` pattern Scalar tries to establish and placing your build +intermediates and outputs parallel with the `src` folder and not inside it, +you can help optimize Git command performance for developers in the repository +by limiting the number of files Git needs to consider for many common +operations. diff --git a/contrib/scalar/docs/getting-started.md b/contrib/scalar/docs/getting-started.md new file mode 100644 index 00000000000000..ef7ea07b0f948c --- /dev/null +++ b/contrib/scalar/docs/getting-started.md @@ -0,0 +1,93 @@ +Getting Started +=============== + +Registering existing Git repos +------------------------------ + +To add a repository to the list of registered repos, run `scalar register []`. +If `` is not provided, then the "current repository" is discovered from +the working directory by scanning the parent paths for a path containing a `.git` +folder, possibly inside a `src` folder. + +To see which repositories are currently tracked by the service, run +`scalar list`. + +Run `scalar unregister []` to remove the repo from this list. + +Creating a new Scalar clone +--------------------------------------------------- + +The `clone` verb creates a local enlistment of a remote repository using the +partial clone feature available e.g. on GitHub. + + +``` +scalar clone [options] [] +``` + +Create a local copy of the repository at ``. If specified, create the `` +directory and place the repository there. Otherwise, the last section of the `` +will be used for ``. + +At the end, the repo is located at `/src`. By default, the sparse-checkout +feature is enabled and the only files present are those in the root of your +Git repository. Use `git sparse-checkout set` to expand the set of directories +you want to see, or `git sparse-checkout disable` to expand to all files. You +can explore the subdirectories outside your sparse-checkout specification using +`git ls-tree HEAD`. + +### Sparse Repo Mode + +By default, Scalar reduces your working directory to only the files at the +root of the repository. You need to add the folders you care about to build up +to your working set. + +* `scalar clone ` + * Please choose the **Clone with HTTPS** option in the `Clone Repository` dialog in Azure Repos, not **Clone with SSH**. +* `cd \src` +* At this point, your `src` directory only contains files that appear in your root + tree. No folders are populated. +* Set the directory list for your sparse-checkout using: + 1. `git sparse-checkout set ...` + 2. `git sparse-checkout set --stdin < dir-list.txt` +* Run git commands as you normally would. +* To fully populate your working directory, run `git sparse-checkout disable`. + +If instead you want to start with all files on-disk, you can clone with the +`--full-clone` option. To enable sparse-checkout after the fact, run +`git sparse-checkout init --cone`. This will initialize your sparse-checkout +patterns to only match the files at root. + +If you are unfamiliar with what directories are available in the repository, +then you can run `git ls-tree -d --name-only HEAD` to discover the directories +at root, or `git ls-tree -d --name-only HEAD ` to discover the directories +in ``. + +### Options + +These options allow a user to customize their initial enlistment. + +* `--full-clone`: If specified, do not initialize the sparse-checkout feature. + All files will be present in your `src` directory. This uses a Git partial + clone: blobs are downloaded on demand. + +* `--branch=`: Specify the branch to checkout after clone. + +### Advanced Options + +The options below are not intended for use by a typical user. These are +usually used by build machines to create a temporary enlistment that +operates on a single commit. + +* `--single-branch`: Use this option to only download metadata for the branch + that will be checked out. This is helpful for build machines that target + a remote with many branches. Any `git fetch` commands after the clone will + still ask for all branches. + +Removing a Scalar Clone +----------------------- + +Since the `scalar clone` command sets up a file-system watcher (when available), +that watcher could prevent deleting the enlistment. Run `scalar delete ` +from outside of your enlistment to unregister the enlistment from the filesystem +watcher and delete the enlistment at ``. diff --git a/contrib/scalar/docs/index.md b/contrib/scalar/docs/index.md new file mode 100644 index 00000000000000..f9f5ab06e09253 --- /dev/null +++ b/contrib/scalar/docs/index.md @@ -0,0 +1,50 @@ +Scalar: Enabling Git at Scale +============================= + +Scalar is a tool that helps Git scale to some of the largest Git repositories. +It achieves this by enabling some advanced Git features, such as: + +* *Partial clone:* reduces time to get a working repository by not + downloading all Git objects right away. + +* *Background prefetch:* downloads Git object data from all remotes every + hour, reducing the amount of time for foreground `git fetch` calls. + +* *Sparse-checkout:* limits the size of your working directory. + +* *File system monitor:* tracks the recently modified files and eliminates + the need for Git to scan the entire worktree. + +* *Commit-graph:* accelerates commit walks and reachability calculations, + speeding up commands like `git log`. + +* *Multi-pack-index:* enables fast object lookups across many pack-files. + +* *Incremental repack:* Repacks the packed Git data into fewer pack-file + without disrupting concurrent commands by using the multi-pack-index. + +By running `scalar register` in any Git repo, Scalar will automatically enable +these features for that repo (except partial clone) and start running suggested +maintenance in the background using +[the `git maintenance` feature](https://git-scm.com/docs/git-maintenance). + +Repos cloned with the `scalar clone` command use partial clone to significantly +reduce the amount of data required to get started using a repository. By +delaying all blob downloads until they are required, Scalar allows you to work +with very large repositories quickly. + +Documentation +------------- + +* [Getting Started](getting-started.md): Get started with Scalar. + Includes `scalar register`, `scalar unregister`, `scalar clone`, and + `scalar delete`. + +* [Troubleshooting](troubleshooting.md): + Collect diagnostic information or update custom settings. Includes + `scalar diagnose`. + +* [The Philosophy of Scalar](philosophy.md): Why does Scalar work the way + it does, and how do we make decisions about its future? + +* [Frequently Asked Questions](faq.md) diff --git a/contrib/scalar/docs/philosophy.md b/contrib/scalar/docs/philosophy.md new file mode 100644 index 00000000000000..51486a75e41f0d --- /dev/null +++ b/contrib/scalar/docs/philosophy.md @@ -0,0 +1,66 @@ +The Philosophy of Scalar +======================== + +The team building Scalar has **opinions** about Git performance. Scalar +takes out the guesswork by automatically configuring your Git repositories +to take advantage of the latest and greatest features. It is difficult to +say that these are the absolute best settings for every repository, but +these settings do work for some of the largest repositories in the world. + +Scalar intends to do very little more than the standard Git client. We +actively implement new features into Git instead of Scalar, then update +Scalar only to configure those new settings. In particular, we ported +features like background maintenance to Git to make Scalar simpler and +make Git more powerful. + +Services such as GitHub support partial clone , a standard adopted by the Git +project to download only part of the Git objects when cloning, and fetching +further objects on demand. If your hosting service supports partial clone, then +we absolutely recommend it as a way to greatly speed up your clone and fetch +times and to reduce how much disk space your Git repository requires. Scalar +will help with this! + +Most of the value of Scalar can be found in the core Git client. However, most +of the advanced features that really optimize Git's performance are off by +default for compatibility reasons. To really take advantage of Git's latest and +greatest features, you either need to study the [`git config` +documentation](https://git-scm.com/docs/git-config) and regularly read [the Git +release notes](https://github.com/git/git/tree/master/Documentation/RelNotes). +Even if you do all that work and customize your Git settings on your machines, +you likely will want to share those settings with other team members. Or, you +can just use Scalar! + +Using `scalar register` on an existing Git repository will give you these +benefits: + +* Additional compression of your `.git/index` file. +* Hourly background `git fetch` operations, keeping you in-sync with your + remotes. +* Advanced data structures, such as the `commit-graph` and `multi-pack-index` + are updated automatically in the background. +* If using macOS or Windows, then Scalar configures Git's builtin File System + Monitor, providing faster commands such as `git status` or `git add`. + +Additionally, if you use `scalar clone` to create a new repository, then +you will automatically get these benefits: + +* Use Git's partial clone feature to only download the files you need for + your current checkout. +* Use Git's [sparse-checkout feature][sparse-checkout] to minimize the + number of files required in your working directory. + [Read more about sparse-checkout here.][sparse-checkout-blog] +* Create the Git repository inside `/src` to make it easy to + place build artifacts outside of the Git repository, such as in + `/bin` or `/packages`. + +We also admit that these **opinions** can always be improved! If you have +an idea of how to improve our setup, consider +[creating an issue](https://github.com/microsoft/scalar/issues/new) or +contributing a pull request! Some [existing](https://github.com/microsoft/scalar/issues/382) +[issues](https://github.com/microsoft/scalar/issues/388) have already +improved our configuration settings and roadmap! + +[gvfs-protocol]: https://github.com/microsoft/VFSForGit/blob/HEAD/Protocol.md +[microsoft-git]: https://github.com/microsoft/git +[sparse-checkout]: https://git-scm.com/docs/git-sparse-checkout +[sparse-checkout-blog]: https://github.blog/2020-01-17-bring-your-monorepo-down-to-size-with-sparse-checkout/ diff --git a/contrib/scalar/docs/troubleshooting.md b/contrib/scalar/docs/troubleshooting.md new file mode 100644 index 00000000000000..8ec56ad437ff09 --- /dev/null +++ b/contrib/scalar/docs/troubleshooting.md @@ -0,0 +1,20 @@ +Troubleshooting +=============== + +Diagnosing Issues +----------------- + +The `scalar diagnose` command collects logs and config details for the current +repository. The resulting zip file helps root-cause issues. + +When run inside your repository, creates a zip file containing several important +files for that repository. This includes: + +* Configuration files from your `.git` folder, such as the `config` file, + `index`, `hooks`, and `refs`. + +* A summary of your Git object database, including the number of loose objects + and the names and sizes of pack-files. + +As the `diagnose` command completes, it provides the path of the resulting +zip file. This zip can be attached to bug reports to make the analysis easier. From 7b212db5d3a144ce2820212ded54a1227686d98f Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 10 May 2022 13:43:05 +0200 Subject: [PATCH 026/196] scalar (Windows): use forward slashes as directory separators Git traditionally uses those, not backslashes, ever. Signed-off-by: Johannes Schindelin --- scalar.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/scalar.c b/scalar.c index 18d3504c3fbee8..f761741e3212c9 100644 --- a/scalar.c +++ b/scalar.c @@ -46,6 +46,9 @@ static void setup_enlistment_directory(int argc, const char **argv, die(_("need a working directory")); strbuf_trim_trailing_dir_sep(&path); +#ifdef GIT_WINDOWS_NATIVE + convert_slashes(path.buf); +#endif /* check if currently in enlistment root with src/ workdir */ len = path.len; From 8521b1ad58762c267556c29ee9574db4ad8a9833 Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Thu, 17 Jun 2021 11:40:09 -0400 Subject: [PATCH 027/196] scalar: add retry logic to run_git() Use a fixed 3 tries total to see how that increases our chances of success for subcommands such as 'git fetch'. We special-case the `diagnose` command here: When 672196a3073 (scalar-diagnose: use 'git diagnose --mode=all', 2022-08-12) updated 'scalar diagnose' to run 'git diagnose' as a subprocess, it was passed through the run_git() caller. We need to avoid repeating the call when the underlying 'git diagnose' command fails. Signed-off-by: Derrick Stolee --- scalar.c | 25 ++++++++++++++++++++----- 1 file changed, 20 insertions(+), 5 deletions(-) diff --git a/scalar.c b/scalar.c index f761741e3212c9..a249383d4aacdb 100644 --- a/scalar.c +++ b/scalar.c @@ -75,21 +75,34 @@ static void setup_enlistment_directory(int argc, const char **argv, strbuf_release(&path); } +static int git_retries = 3; + LAST_ARG_MUST_BE_NULL static int run_git(const char *arg, ...) { - struct child_process cmd = CHILD_PROCESS_INIT; va_list args; const char *p; + struct strvec argv = STRVEC_INIT; + int res = 0, attempts; va_start(args, arg); - strvec_push(&cmd.args, arg); + strvec_push(&argv, arg); while ((p = va_arg(args, const char *))) - strvec_push(&cmd.args, p); + strvec_push(&argv, p); va_end(args); - cmd.git_cmd = 1; - return run_command(&cmd); + for (attempts = 0, res = 1; + res && attempts < git_retries; + attempts++) { + struct child_process cmd = CHILD_PROCESS_INIT; + + cmd.git_cmd = 1; + strvec_pushv(&cmd.args, argv.v); + res = run_command(&cmd); + } + + strvec_clear(&argv); + return res; } struct scalar_config { @@ -590,6 +603,8 @@ static int cmd_diagnose(int argc, const char **argv) setup_enlistment_directory(argc, argv, usage, options, &diagnostics_root); strbuf_addstr(&diagnostics_root, "/.scalarDiagnostics"); + /* Here, a failure should not repeat itself. */ + git_retries = 1; res = run_git("diagnose", "--mode=all", "-s", "%Y%m%d_%H%M%S", "-o", diagnostics_root.buf, NULL); From d97b7b28a49263dc03749db9c7ef8f4bc157e8f0 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Thu, 27 May 2021 07:26:11 +0200 Subject: [PATCH 028/196] scalar: support the `config` command for backwards compatibility The .NET version supported running `scalar config` to reconfigure the current enlistment, and now the C port does, too. Signed-off-by: Johannes Schindelin --- scalar.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/scalar.c b/scalar.c index a249383d4aacdb..7918563fd1f72f 100644 --- a/scalar.c +++ b/scalar.c @@ -1015,6 +1015,9 @@ int cmd_main(int argc, const char **argv) argv++; argc--; + if (!strcmp(argv[0], "config")) + argv[0] = "reconfigure"; + for (i = 0; builtins[i].name; i++) if (!strcmp(builtins[i].name, argv[0])) return !!builtins[i].fn(argc, argv); From a9de9d085a3fd39d909491445b07b1dc3d3f8dc9 Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Thu, 23 Sep 2021 09:59:06 -0400 Subject: [PATCH 029/196] sequencer: avoid progress when stderr is redirected During a run of the Scalar functional tests, we hit a case where the inexact rename detection of a 'git cherry-pick' command slowed to the point of writing its delayed progress, failing the test because stderr differed from the control case. Showing progress like this when stderr is not a terminal is non-standard for Git, so inject an isatty(2) when initializing the progress option in sequencer.c. Unfortunately, there is no '--quiet' option in 'git cherry-pick' currently wired up. This could be considered in the future, and the isatty(2) could be moved to that position. This would also be needed for commands like 'git rebase', so we leave that for another time. Reported-by: Victoria Dye Signed-off-by: Derrick Stolee --- sequencer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sequencer.c b/sequencer.c index 8d01cd50ac943c..36b478750ee9d0 100644 --- a/sequencer.c +++ b/sequencer.c @@ -769,7 +769,7 @@ static int do_recursive_merge(struct repository *r, o.branch2 = next ? next_label : "(empty tree)"; if (is_rebase_i(opts)) o.buffer_output = 2; - o.show_rename_progress = 1; + o.show_rename_progress = isatty(2); head_tree = parse_tree_indirect(head); if (!head_tree) From e3d9c14861fe0c5aec6b381dbec16a3900e08cbe Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Tue, 24 Jan 2017 17:34:12 +0100 Subject: [PATCH 030/196] gvfs: add the core.gvfs config setting This does not do anything yet. The next patches will add various values for that config setting that correspond to the various features offered/required by GVFS. Signed-off-by: Kevin Willford gvfs: refactor loading the core.gvfs config value This code change makes sure that the config value for core_gvfs is always loaded before checking it. Signed-off-by: Kevin Willford --- Documentation/config/core.txt | 3 +++ Makefile | 1 + config.c | 6 +++++ environment.c | 1 + environment.h | 1 + gvfs.c | 45 +++++++++++++++++++++++++++++++++++ gvfs.h | 4 ++++ 7 files changed, 61 insertions(+) create mode 100644 gvfs.c diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index 458a3f4f7f5d80..3f3343c7cbd982 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -743,6 +743,9 @@ core.multiPackIndex:: single index. See linkgit:git-multi-pack-index[1] for more information. Defaults to true. +core.gvfs:: + Enable the features needed for GVFS. + core.sparseCheckout:: Enable "sparse checkout" feature. See linkgit:git-sparse-checkout[1] for more information. diff --git a/Makefile b/Makefile index aea9b14eb8b079..d6f0a9495f5972 100644 --- a/Makefile +++ b/Makefile @@ -1042,6 +1042,7 @@ LIB_OBJS += git-zlib.o LIB_OBJS += gpg-interface.o LIB_OBJS += graph.o LIB_OBJS += grep.o +LIB_OBJS += gvfs.o LIB_OBJS += hash-lookup.o LIB_OBJS += hashmap.o LIB_OBJS += help.o diff --git a/config.c b/config.c index bfca0ea5fd97f5..480264a398c407 100644 --- a/config.c +++ b/config.c @@ -12,6 +12,7 @@ #include "abspath.h" #include "advice.h" #include "date.h" +#include "gvfs.h" #include "branch.h" #include "config.h" #include "parse.h" @@ -1642,6 +1643,11 @@ int git_default_core_config(const char *var, const char *value, return 0; } + if (!strcmp(var, "core.gvfs")) { + gvfs_load_config_value(value); + return 0; + } + if (!strcmp(var, "core.sparsecheckout")) { core_apply_sparse_checkout = git_config_bool(var, value); return 0; diff --git a/environment.c b/environment.c index a2ce998081864d..6c821d4c5862dc 100644 --- a/environment.c +++ b/environment.c @@ -71,6 +71,7 @@ int grafts_keep_true_parents; int core_apply_sparse_checkout; int core_sparse_checkout_cone; int sparse_expect_files_outside_of_patterns; +int core_gvfs; int merge_log_config = -1; int precomposed_unicode = -1; /* see probe_utf8_pathname_composition() */ unsigned long pack_size_limit_cfg; diff --git a/environment.h b/environment.h index 923e12661e19e4..ebb5beae7b22dc 100644 --- a/environment.h +++ b/environment.h @@ -171,6 +171,7 @@ extern unsigned long pack_size_limit_cfg; extern int max_allowed_tree_depth; extern int core_preload_index; +extern int core_gvfs; extern int precomposed_unicode; extern int protect_hfs; extern int protect_ntfs; diff --git a/gvfs.c b/gvfs.c new file mode 100644 index 00000000000000..3cdd8a055d6021 --- /dev/null +++ b/gvfs.c @@ -0,0 +1,45 @@ +#define USE_THE_REPOSITORY_VARIABLE +#include "git-compat-util.h" +#include "environment.h" +#include "gvfs.h" +#include "setup.h" +#include "config.h" + +static int gvfs_config_loaded; +static int core_gvfs_is_bool; + +static int early_core_gvfs_config(const char *var, const char *value, + const struct config_context *ctx, void *cb UNUSED) +{ + if (!strcmp(var, "core.gvfs")) + core_gvfs = git_config_bool_or_int("core.gvfs", value, ctx->kvi, + &core_gvfs_is_bool); + return 0; +} + +void gvfs_load_config_value(const char *value) +{ + if (gvfs_config_loaded) + return; + + if (value) { + struct key_value_info default_kvi = KVI_INIT; + core_gvfs = git_config_bool_or_int("core.gvfs", value, &default_kvi, &core_gvfs_is_bool); + } else if (startup_info->have_repository == 0) + read_early_config(the_repository, early_core_gvfs_config, NULL); + else + repo_config_get_bool_or_int(the_repository, "core.gvfs", + &core_gvfs_is_bool, &core_gvfs); + + /* Turn on all bits if a bool was set in the settings */ + if (core_gvfs_is_bool && core_gvfs) + core_gvfs = -1; + + gvfs_config_loaded = 1; +} + +int gvfs_config_is_set(int mask) +{ + gvfs_load_config_value(NULL); + return (core_gvfs & mask) == mask; +} diff --git a/gvfs.h b/gvfs.h index b6dbe85eae4071..011185dea93734 100644 --- a/gvfs.h +++ b/gvfs.h @@ -1,9 +1,13 @@ #ifndef GVFS_H #define GVFS_H + /* * This file is for the specific settings and methods * used for GVFS functionality */ +void gvfs_load_config_value(const char *value); +int gvfs_config_is_set(int mask); + #endif /* GVFS_H */ From dd7004d1e25478b33284caf55c284e08e8708bbc Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Tue, 24 Jan 2017 17:38:59 +0100 Subject: [PATCH 031/196] gvfs: add the feature to skip writing the index' SHA-1 This takes a substantial amount of time, and if the user is reasonably sure that the files' integrity is not compromised, that time can be saved. Git no longer verifies the SHA-1 by default, anyway. Signed-off-by: Kevin Willford Update for 2023-02-27: This feature was upstreamed as the index.skipHash config option. This resulted in some changes to the struct and some of the setup code. In particular, the config reading was moved to prepare_repo_settings(), so the core.gvfs bit check was moved there, too. Signed-off-by: Derrick Stolee --- Documentation/config/core.txt | 10 +++++++++- gvfs.h | 6 ++++++ repo-settings.c | 8 ++++++++ t/t1017-read-tree-skip-sha-on-read.sh | 22 ++++++++++++++++++++++ 4 files changed, 45 insertions(+), 1 deletion(-) create mode 100755 t/t1017-read-tree-skip-sha-on-read.sh diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index 3f3343c7cbd982..925f164842ff00 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -744,7 +744,15 @@ core.multiPackIndex:: information. Defaults to true. core.gvfs:: - Enable the features needed for GVFS. + Enable the features needed for GVFS. This value can be set to true + to indicate all features should be turned on or the bit values listed + below can be used to turn on specific features. ++ +-- + GVFS_SKIP_SHA_ON_INDEX:: + Bit value 1 + Disables the calculation of the sha when writing the index +-- core.sparseCheckout:: Enable "sparse checkout" feature. See linkgit:git-sparse-checkout[1] diff --git a/gvfs.h b/gvfs.h index 011185dea93734..c75991530fa1fa 100644 --- a/gvfs.h +++ b/gvfs.h @@ -7,6 +7,12 @@ * used for GVFS functionality */ + +/* + * The list of bits in the core_gvfs setting + */ +#define GVFS_SKIP_SHA_ON_INDEX (1 << 0) + void gvfs_load_config_value(const char *value); int gvfs_config_is_set(int mask); diff --git a/repo-settings.c b/repo-settings.c index d8123b9323d23e..f9a355f498098e 100644 --- a/repo-settings.c +++ b/repo-settings.c @@ -3,6 +3,7 @@ #include "repo-settings.h" #include "repository.h" #include "midx.h" +#include "gvfs.h" static void repo_cfg_bool(struct repository *r, const char *key, int *dest, int def) @@ -76,6 +77,13 @@ void prepare_repo_settings(struct repository *r) r->settings.pack_use_bitmap_boundary_traversal); repo_cfg_bool(r, "core.usereplacerefs", &r->settings.read_replace_refs, 1); + /* + * For historical compatibility reasons, enable index.skipHash based + * on a bit in core.gvfs. + */ + if (gvfs_config_is_set(GVFS_SKIP_SHA_ON_INDEX)) + r->settings.index_skip_hash = 1; + /* * The GIT_TEST_MULTI_PACK_INDEX variable is special in that * either it *or* the config sets diff --git a/t/t1017-read-tree-skip-sha-on-read.sh b/t/t1017-read-tree-skip-sha-on-read.sh new file mode 100755 index 00000000000000..5b76a80a0020dc --- /dev/null +++ b/t/t1017-read-tree-skip-sha-on-read.sh @@ -0,0 +1,22 @@ +#!/bin/sh + +test_description='check that read-tree works with core.gvfs config value' + +. ./test-lib.sh +. "$TEST_DIRECTORY"/lib-read-tree.sh + +test_expect_success setup ' + echo one >a && + git add a && + git commit -m initial +' +test_expect_success 'read-tree without core.gvsf' ' + read_tree_u_must_succeed -m -u HEAD +' + +test_expect_success 'read-tree with core.gvfs set to 1' ' + git config core.gvfs 1 && + read_tree_u_must_succeed -m -u HEAD +' + +test_done From f9b8c1e11a50899fce953db483c543d72cea146e Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Tue, 24 Jan 2017 17:54:55 +0100 Subject: [PATCH 032/196] gvfs: add the feature that blobs may be missing Signed-off-by: Kevin Willford --- Documentation/config/core.txt | 4 ++++ cache-tree.c | 4 +++- commit.c | 9 +++++++-- gvfs.h | 1 + t/t0000-basic.sh | 5 +++++ 5 files changed, 20 insertions(+), 3 deletions(-) diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index 925f164842ff00..7c640bbf36bb68 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -752,6 +752,10 @@ core.gvfs:: GVFS_SKIP_SHA_ON_INDEX:: Bit value 1 Disables the calculation of the sha when writing the index + GVFS_MISSING_OK:: + Bit value 4 + Normally git write-tree ensures that the objects referenced by the + directory exist in the object database. This option disables this check. -- core.sparseCheckout:: diff --git a/cache-tree.c b/cache-tree.c index b482167a69ae6a..287658357a9dec 100644 --- a/cache-tree.c +++ b/cache-tree.c @@ -2,6 +2,7 @@ #include "git-compat-util.h" #include "hex.h" +#include "gvfs.h" #include "lockfile.h" #include "tree.h" #include "tree-walk.h" @@ -259,7 +260,8 @@ static int update_one(struct cache_tree *it, int flags) { struct strbuf buffer; - int missing_ok = flags & WRITE_TREE_MISSING_OK; + int missing_ok = gvfs_config_is_set(GVFS_MISSING_OK) ? + WRITE_TREE_MISSING_OK : (flags & WRITE_TREE_MISSING_OK); int dryrun = flags & WRITE_TREE_DRY_RUN; int repair = flags & WRITE_TREE_REPAIR; int to_invalidate = 0; diff --git a/commit.c b/commit.c index cc03a9303623ad..2498d3bd9409e3 100644 --- a/commit.c +++ b/commit.c @@ -1,6 +1,7 @@ #define USE_THE_REPOSITORY_VARIABLE #include "git-compat-util.h" +#include "gvfs.h" #include "tag.h" #include "commit.h" #include "commit-graph.h" @@ -556,13 +557,17 @@ int repo_parse_commit_internal(struct repository *r, .sizep = &size, .contentp = &buffer, }; + int ret; /* * Git does not support partial clones that exclude commits, so set * OBJECT_INFO_SKIP_FETCH_OBJECT to fail fast when an object is missing. */ int flags = OBJECT_INFO_LOOKUP_REPLACE | OBJECT_INFO_SKIP_FETCH_OBJECT | - OBJECT_INFO_DIE_IF_CORRUPT; - int ret; + OBJECT_INFO_DIE_IF_CORRUPT; + + /* But the GVFS Protocol _does_ support missing commits! */ + if (gvfs_config_is_set(GVFS_MISSING_OK)) + flags ^= OBJECT_INFO_SKIP_FETCH_OBJECT; if (!item) return -1; diff --git a/gvfs.h b/gvfs.h index c75991530fa1fa..7bedfaacf6d684 100644 --- a/gvfs.h +++ b/gvfs.h @@ -12,6 +12,7 @@ * The list of bits in the core_gvfs setting */ #define GVFS_SKIP_SHA_ON_INDEX (1 << 0) +#define GVFS_MISSING_OK (1 << 2) void gvfs_load_config_value(const char *value); int gvfs_config_is_set(int mask); diff --git a/t/t0000-basic.sh b/t/t0000-basic.sh index 98b81e4d63fa4c..7baa03212bf3ba 100755 --- a/t/t0000-basic.sh +++ b/t/t0000-basic.sh @@ -1106,6 +1106,11 @@ test_expect_success 'writing this tree with --missing-ok' ' git write-tree --missing-ok ' +test_expect_success 'writing this tree with missing ok config value' ' + git config core.gvfs 4 && + git write-tree +' + ################################################################ test_expect_success 'git read-tree followed by write-tree should be idempotent' ' From c7ee92640f2106873ffafe7b9f3291503df53e45 Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Wed, 18 May 2016 13:40:39 +0000 Subject: [PATCH 033/196] gvfs: prevent files to be deleted outside the sparse checkout Prevent the sparse checkout to delete files that were marked with skip-worktree bit and are not in the sparse-checkout file. This is because everything with the skip-worktree bit turned on is being virtualized and will be removed with the change of HEAD. There was only one failing test when running with these changes that was checking to make sure the worktree narrows on checkout which was expected since we would no longer be narrowing the worktree. Update 2022-04-05: temporarily set 'sparse.expectfilesoutsideofpatterns' in test (until we start disabling the "remove present-despite-SKIP_WORKTREE" behavior with 'core.virtualfilesystem' in a later commit). Signed-off-by: Kevin Willford --- Documentation/config/core.txt | 9 +++++++++ gvfs.h | 1 + t/t1090-sparse-checkout-scope.sh | 20 ++++++++++++++++++++ unpack-trees.c | 22 ++++++++++++++++++++++ 4 files changed, 52 insertions(+) diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index 7c640bbf36bb68..cb37bca6d7ac08 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -756,6 +756,15 @@ core.gvfs:: Bit value 4 Normally git write-tree ensures that the objects referenced by the directory exist in the object database. This option disables this check. + GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT:: + Bit value 8 + When marking entries to remove from the index and the working + directory this option will take into account what the + skip-worktree bit was set to so that if the entry has the + skip-worktree bit set it will not be removed from the working + directory. This will allow virtualized working directories to + detect the change to HEAD and use the new commit tree to show + the files that are in the working directory. -- core.sparseCheckout:: diff --git a/gvfs.h b/gvfs.h index 7bedfaacf6d684..44131625828cfa 100644 --- a/gvfs.h +++ b/gvfs.h @@ -13,6 +13,7 @@ */ #define GVFS_SKIP_SHA_ON_INDEX (1 << 0) #define GVFS_MISSING_OK (1 << 2) +#define GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT (1 << 3) void gvfs_load_config_value(const char *value); int gvfs_config_is_set(int mask); diff --git a/t/t1090-sparse-checkout-scope.sh b/t/t1090-sparse-checkout-scope.sh index b6298e7e7d8ab1..7af01dc00b75ee 100755 --- a/t/t1090-sparse-checkout-scope.sh +++ b/t/t1090-sparse-checkout-scope.sh @@ -107,6 +107,26 @@ test_expect_success 'in partial clone, sparse checkout only fetches needed blobs test_cmp expect actual ' +test_expect_success 'checkout does not delete items outside the sparse checkout file' ' + # The "sparse.expectfilesoutsideofpatterns" config will prevent the + # SKIP_WORKTREE flag from being dropped on files present on-disk. + test_config sparse.expectfilesoutsideofpatterns true && + + test_config core.gvfs 8 && + git checkout -b outside && + echo "new file1" >d && + git add --sparse d && + git commit -m "branch initial" && + echo "new file1" >e && + git add --sparse e && + git commit -m "skipped worktree" && + git update-index --skip-worktree e && + echo "/d" >.git/info/sparse-checkout && + git checkout HEAD^ && + test_path_is_file d && + test_path_is_file e +' + test_expect_success MINGW 'no unnecessary opendir() with fscache' ' git clone . fscache-test && ( diff --git a/unpack-trees.c b/unpack-trees.c index 701eec295f9fde..9d55cc94874b3f 100644 --- a/unpack-trees.c +++ b/unpack-trees.c @@ -2,6 +2,7 @@ #include "git-compat-util.h" #include "advice.h" +#include "gvfs.h" #include "strvec.h" #include "repository.h" #include "parse.h" @@ -2679,6 +2680,27 @@ static int deleted_entry(const struct cache_entry *ce, if (!(old->ce_flags & CE_CONFLICTED) && verify_uptodate(old, o)) return -1; + + /* + * When marking entries to remove from the index and the working + * directory this option will take into account what the + * skip-worktree bit was set to so that if the entry has the + * skip-worktree bit set it will not be removed from the working + * directory. This will allow virtualized working directories to + * detect the change to HEAD and use the new commit tree to show + * the files that are in the working directory. + * + * old is the cache_entry that will have the skip-worktree bit set + * which will need to be preserved when the CE_REMOVE entry is added + */ + if (gvfs_config_is_set(GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT) && + old && + old->ce_flags & CE_SKIP_WORKTREE) { + add_entry(o, old, CE_REMOVE, 0); + invalidate_ce_path(old, o); + return 1; + } + add_entry(o, ce, CE_REMOVE, 0); invalidate_ce_path(ce, o); return 1; From d25865c8f5718f0c8524a74a2903aaae37050c14 Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Mon, 30 May 2016 10:55:53 -0400 Subject: [PATCH 034/196] gvfs: optionally skip reachability checks/upload pack during fetch While performing a fetch with a virtual file system we know that there will be missing objects and we don't want to download them just because of the reachability of the commits. We also don't want to download a pack file with commits, trees, and blobs since these will be downloaded on demand. This flag will skip the first connectivity check and by returning zero will skip the upload pack. It will also skip the second connectivity check but continue to update the branches to the latest commit ids. Signed-off-by: Kevin Willford --- Documentation/config/core.txt | 9 +++++++++ connected.c | 19 +++++++++++++++++++ gvfs.h | 1 + t/t5584-vfs.sh | 24 ++++++++++++++++++++++++ 4 files changed, 53 insertions(+) create mode 100755 t/t5584-vfs.sh diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index cb37bca6d7ac08..c2063fcaf6e407 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -765,6 +765,15 @@ core.gvfs:: directory. This will allow virtualized working directories to detect the change to HEAD and use the new commit tree to show the files that are in the working directory. + GVFS_FETCH_SKIP_REACHABILITY_AND_UPLOADPACK:: + Bit value 16 + While performing a fetch with a virtual file system we know + that there will be missing objects and we don't want to download + them just because of the reachability of the commits. We also + don't want to download a pack file with commits, trees, and blobs + since these will be downloaded on demand. This flag will skip the + checks on the reachability of objects during a fetch as well as + the upload pack so that extraneous objects don't get downloaded. -- core.sparseCheckout:: diff --git a/connected.c b/connected.c index 87cc4b57a17ce5..44e5f75eb063b4 100644 --- a/connected.c +++ b/connected.c @@ -3,6 +3,7 @@ #include "git-compat-util.h" #include "gettext.h" #include "hex.h" +#include "gvfs.h" #include "object-store-ll.h" #include "run-command.h" #include "sigchain.h" @@ -34,6 +35,24 @@ int check_connected(oid_iterate_fn fn, void *cb_data, struct transport *transport; size_t base_len; + /* + * Running a virtual file system there will be objects that are + * missing locally and we don't want to download a bunch of + * commits, trees, and blobs just to make sure everything is + * reachable locally so this option will skip reachablility + * checks below that use rev-list. This will stop the check + * before uploadpack runs to determine if there is anything to + * fetch. Returning zero for the first check will also prevent the + * uploadpack from happening. It will also skip the check after + * the fetch is finished to make sure all the objects where + * downloaded in the pack file. This will allow the fetch to + * run and get all the latest tip commit ids for all the branches + * in the fetch but not pull down commits, trees, or blobs via + * upload pack. + */ + if (gvfs_config_is_set(GVFS_FETCH_SKIP_REACHABILITY_AND_UPLOADPACK)) + return 0; + if (!opt) opt = &defaults; transport = opt->transport; diff --git a/gvfs.h b/gvfs.h index 44131625828cfa..e69dd4b2c8f785 100644 --- a/gvfs.h +++ b/gvfs.h @@ -14,6 +14,7 @@ #define GVFS_SKIP_SHA_ON_INDEX (1 << 0) #define GVFS_MISSING_OK (1 << 2) #define GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT (1 << 3) +#define GVFS_FETCH_SKIP_REACHABILITY_AND_UPLOADPACK (1 << 4) void gvfs_load_config_value(const char *value); int gvfs_config_is_set(int mask); diff --git a/t/t5584-vfs.sh b/t/t5584-vfs.sh new file mode 100755 index 00000000000000..8a703cbb640387 --- /dev/null +++ b/t/t5584-vfs.sh @@ -0,0 +1,24 @@ +#!/bin/sh + +test_description='fetch using the flag to skip reachability and upload pack' + +. ./test-lib.sh + + +test_expect_success setup ' + echo inital >a && + git add a && + git commit -m initial && + git clone . one +' + +test_expect_success "fetch test" ' + cd one && + git config core.gvfs 16 && + rm -rf .git/objects/* && + git -C .. cat-file commit HEAD | git hash-object -w --stdin -t commit && + git fetch && + test_must_fail git rev-parse --verify HEAD^{tree} +' + +test_done \ No newline at end of file From 162137d9290dbf49dd21a7cabdb4b5eaade17bbc Mon Sep 17 00:00:00 2001 From: Ben Peart Date: Wed, 15 Jun 2016 14:59:16 +0000 Subject: [PATCH 035/196] gvfs: ensure all filters and EOL conversions are blocked Ensure all filters and EOL conversions are blocked when running under GVFS so that our projected file sizes will match the actual file size when it is hydrated on the local machine. Signed-off-by: Ben Peart --- Documentation/config/core.txt | 9 +++++++++ convert.c | 22 +++++++++++++++++++++ gvfs.h | 1 + t/t0021-conversion.sh | 37 +++++++++++++++++++++++++++++++++++ t/t0027-auto-crlf.sh | 12 ++++++++++++ 5 files changed, 81 insertions(+) diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index c2063fcaf6e407..2fd2a10f51ad2c 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -774,6 +774,15 @@ core.gvfs:: since these will be downloaded on demand. This flag will skip the checks on the reachability of objects during a fetch as well as the upload pack so that extraneous objects don't get downloaded. + GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS:: + Bit value 64 + With a virtual file system we only know the file size before any + CRLF or smudge/clean filters processing is done on the client. + To prevent file corruption due to truncation or expansion with + garbage at the end, these filters must not run when the file + is first accessed and brought down to the client. Git.exe can't + currently tell the first access vs subsequent accesses so this + flag just blocks them from occurring at all. -- core.sparseCheckout:: diff --git a/convert.c b/convert.c index c9a31eb4f03d6c..b59bb58f8a3d9b 100644 --- a/convert.c +++ b/convert.c @@ -2,6 +2,7 @@ #include "git-compat-util.h" #include "advice.h" +#include "gvfs.h" #include "config.h" #include "convert.h" #include "copy.h" @@ -562,6 +563,9 @@ static int crlf_to_git(struct index_state *istate, if (!buf) return 1; + if (gvfs_config_is_set(GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS)) + die("CRLF conversions not supported when running under GVFS"); + /* only grow if not in place */ if (strbuf_avail(buf) + buf->len < len) strbuf_grow(buf, len - buf->len); @@ -601,6 +605,9 @@ static int crlf_to_worktree(const char *src, size_t len, struct strbuf *buf, if (!will_convert_lf_to_crlf(&stats, crlf_action)) return 0; + if (gvfs_config_is_set(GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS)) + die("CRLF conversions not supported when running under GVFS"); + /* are we "faking" in place editing ? */ if (src == buf->buf) to_free = strbuf_detach(buf, NULL); @@ -710,6 +717,9 @@ static int apply_single_file_filter(const char *path, const char *src, size_t le struct async async; struct filter_params params; + if (gvfs_config_is_set(GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS)) + die("Filter \"%s\" not supported when running under GVFS", cmd); + memset(&async, 0, sizeof(async)); async.proc = filter_buffer_or_fd; async.data = ¶ms; @@ -1129,6 +1139,9 @@ static int ident_to_git(const char *src, size_t len, if (!buf) return 1; + if (gvfs_config_is_set(GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS)) + die("ident conversions not supported when running under GVFS"); + /* only grow if not in place */ if (strbuf_avail(buf) + buf->len < len) strbuf_grow(buf, len - buf->len); @@ -1176,6 +1189,9 @@ static int ident_to_worktree(const char *src, size_t len, if (!cnt) return 0; + if (gvfs_config_is_set(GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS)) + die("ident conversions not supported when running under GVFS"); + /* are we "faking" in place editing ? */ if (src == buf->buf) to_free = strbuf_detach(buf, NULL); @@ -1628,6 +1644,9 @@ static int lf_to_crlf_filter_fn(struct stream_filter *filter, size_t count, o = 0; struct lf_to_crlf_filter *lf_to_crlf = (struct lf_to_crlf_filter *)filter; + if (gvfs_config_is_set(GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS)) + die("CRLF conversions not supported when running under GVFS"); + /* * We may be holding onto the CR to see if it is followed by a * LF, in which case we would need to go to the main loop. @@ -1872,6 +1891,9 @@ static int ident_filter_fn(struct stream_filter *filter, struct ident_filter *ident = (struct ident_filter *)filter; static const char head[] = "$Id"; + if (gvfs_config_is_set(GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS)) + die("ident conversions not supported when running under GVFS"); + if (!input) { /* drain upon eof */ switch (ident->state) { diff --git a/gvfs.h b/gvfs.h index e69dd4b2c8f785..7c9367866f502a 100644 --- a/gvfs.h +++ b/gvfs.h @@ -15,6 +15,7 @@ #define GVFS_MISSING_OK (1 << 2) #define GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT (1 << 3) #define GVFS_FETCH_SKIP_REACHABILITY_AND_UPLOADPACK (1 << 4) +#define GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS (1 << 6) void gvfs_load_config_value(const char *value); int gvfs_config_is_set(int mask); diff --git a/t/t0021-conversion.sh b/t/t0021-conversion.sh index 38025608f6a951..3fd454b1e5c59a 100755 --- a/t/t0021-conversion.sh +++ b/t/t0021-conversion.sh @@ -336,6 +336,43 @@ test_expect_success "filter: smudge empty file" ' test_cmp expected filtered-empty-in-repo ' +test_expect_success "filter: clean filters blocked when under GVFS" ' + test_config filter.empty-in-repo.clean "cat >/dev/null" && + test_config filter.empty-in-repo.smudge "echo smudged && cat" && + test_config core.gvfs 64 && + + echo dead data walking >empty-in-repo && + test_must_fail git add empty-in-repo +' + +test_expect_success "filter: smudge filters blocked when under GVFS" ' + test_config filter.empty-in-repo.clean "cat >/dev/null" && + test_config filter.empty-in-repo.smudge "echo smudged && cat" && + test_config core.gvfs 64 && + + test_must_fail git checkout +' + +test_expect_success "ident blocked on add when under GVFS" ' + test_config core.gvfs 64 && + test_config core.autocrlf false && + + echo "*.i ident" >.gitattributes && + echo "\$Id\$" > ident.i && + + test_must_fail git add ident.i +' + +test_expect_success "ident blocked when under GVFS" ' + git add ident.i && + + git commit -m "added ident.i" && + test_config core.gvfs 64 && + rm ident.i && + + test_must_fail git checkout -- ident.i +' + test_expect_success 'disable filter with empty override' ' test_config_global filter.disable.smudge false && test_config_global filter.disable.clean false && diff --git a/t/t0027-auto-crlf.sh b/t/t0027-auto-crlf.sh index 2f57c8669cb5af..c42bf106e6f4ea 100755 --- a/t/t0027-auto-crlf.sh +++ b/t/t0027-auto-crlf.sh @@ -344,6 +344,18 @@ checkout_files () { " } +test_expect_success 'crlf conversions blocked when under GVFS' ' + git checkout -b gvfs && + test_commit initial && + rm initial.t && + test_config core.gvfs 64 && + test_config core.autocrlf true && + test_must_fail git read-tree --reset -u HEAD && + + git config core.autocrlf false && + git read-tree --reset -u HEAD +' + # Test control characters # NUL SOH CR EOF==^Z test_expect_success 'ls-files --eol -o Text/Binary' ' From 2516df27a52c2b4d86d648bb2a24fe3f504dd81c Mon Sep 17 00:00:00 2001 From: Ben Peart Date: Tue, 10 Jan 2017 18:47:14 +0000 Subject: [PATCH 036/196] gvfs: allow "virtualizing" objects The idea is to allow blob objects to be missing from the local repository, and to load them lazily on demand. After discussing this idea on the mailing list, we will rename the feature to "lazy clone" and work more on this. Signed-off-by: Ben Peart Signed-off-by: Johannes Schindelin --- config.c | 5 +++++ connected.c | 3 +++ environment.c | 1 + environment.h | 1 + object-file.c | 23 +++++++++++++++++++++++ 5 files changed, 33 insertions(+) diff --git a/config.c b/config.c index 480264a398c407..335459faee6b3c 100644 --- a/config.c +++ b/config.c @@ -1678,6 +1678,11 @@ int git_default_core_config(const char *var, const char *value, return 0; } + if (!strcmp(var, "core.virtualizeobjects")) { + core_virtualize_objects = git_config_bool(var, value); + return 0; + } + /* Add other config variables here and to Documentation/config.txt. */ return platform_core_config(var, value, ctx, cb); } diff --git a/connected.c b/connected.c index 44e5f75eb063b4..a8cf3dc351fc54 100644 --- a/connected.c +++ b/connected.c @@ -1,6 +1,7 @@ #define USE_THE_REPOSITORY_VARIABLE #include "git-compat-util.h" +#include "environment.h" #include "gettext.h" #include "hex.h" #include "gvfs.h" @@ -52,6 +53,8 @@ int check_connected(oid_iterate_fn fn, void *cb_data, */ if (gvfs_config_is_set(GVFS_FETCH_SKIP_REACHABILITY_AND_UPLOADPACK)) return 0; + if (core_virtualize_objects) + return 0; if (!opt) opt = &defaults; diff --git a/environment.c b/environment.c index 6c821d4c5862dc..91962ab8d0ebaf 100644 --- a/environment.c +++ b/environment.c @@ -75,6 +75,7 @@ int core_gvfs; int merge_log_config = -1; int precomposed_unicode = -1; /* see probe_utf8_pathname_composition() */ unsigned long pack_size_limit_cfg; +int core_virtualize_objects; int max_allowed_tree_depth = #ifdef _MSC_VER /* diff --git a/environment.h b/environment.h index ebb5beae7b22dc..d3d88598256237 100644 --- a/environment.h +++ b/environment.h @@ -226,5 +226,6 @@ extern const char *comment_line_str; extern char *comment_line_str_to_free; extern int auto_comment_line_char; +extern int core_virtualize_objects; # endif /* USE_THE_REPOSITORY_VARIABLE */ #endif /* ENVIRONMENT_H */ diff --git a/object-file.c b/object-file.c index a19be7222a19af..afdca5fe453c04 100644 --- a/object-file.c +++ b/object-file.c @@ -40,6 +40,8 @@ #include "fsck.h" #include "loose.h" #include "object-file-convert.h" +#include "trace.h" +#include "hook.h" /* The maximum size for an object header. */ #define MAX_HEADER_LEN 32 @@ -1620,6 +1622,20 @@ void disable_obj_read_lock(void) pthread_mutex_destroy(&obj_read_mutex); } +static int run_read_object_hook(struct repository *r, const struct object_id *oid) +{ + struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT; + int ret; + uint64_t start; + + start = getnanotime(); + strvec_push(&opt.args, oid_to_hex(oid)); + ret = run_hooks_opt(r, "read-object", &opt); + trace_performance_since(start, "run_read_object_hook"); + + return ret; +} + int fetch_if_missing = 1; static int do_oid_object_info_extended(struct repository *r, @@ -1632,6 +1648,7 @@ static int do_oid_object_info_extended(struct repository *r, int rtype; const struct object_id *real = oid; int already_retried = 0; + int tried_hook = 0; if (flags & OBJECT_INFO_LOOKUP_REPLACE) @@ -1643,6 +1660,7 @@ static int do_oid_object_info_extended(struct repository *r, if (!oi) oi = &blank_oi; +retry: co = find_cached_object(real); if (co) { if (oi->typep) @@ -1674,6 +1692,11 @@ static int do_oid_object_info_extended(struct repository *r, reprepare_packed_git(r); if (find_pack_entry(r, real, &e)) break; + if (core_virtualize_objects && !tried_hook) { + tried_hook = 1; + if (!run_read_object_hook(r, oid)) + goto retry; + } } /* From 06584ba482fe2488dd71cf1a510a41207cadc045 Mon Sep 17 00:00:00 2001 From: Ben Peart Date: Wed, 15 Mar 2017 18:43:05 +0000 Subject: [PATCH 037/196] Hydrate missing loose objects in check_and_freshen() Hydrate missing loose objects in check_and_freshen() when running virtualized. Add test cases to verify read-object hook works when running virtualized. This hook is called in check_and_freshen() rather than check_and_freshen_local() to make the hook work also with alternates. Helped-by: Kevin Willford Signed-off-by: Ben Peart --- .../technical/read-object-protocol.txt | 102 +++++++++++++ contrib/long-running-read-object/example.pl | 114 ++++++++++++++ object-file.c | 141 ++++++++++++++++-- t/t0410/read-object | 114 ++++++++++++++ t/t0499-read-object.sh | 30 ++++ 5 files changed, 485 insertions(+), 16 deletions(-) create mode 100644 Documentation/technical/read-object-protocol.txt create mode 100644 contrib/long-running-read-object/example.pl create mode 100755 t/t0410/read-object create mode 100755 t/t0499-read-object.sh diff --git a/Documentation/technical/read-object-protocol.txt b/Documentation/technical/read-object-protocol.txt new file mode 100644 index 00000000000000..a893b46e7c28a9 --- /dev/null +++ b/Documentation/technical/read-object-protocol.txt @@ -0,0 +1,102 @@ +Read Object Process +^^^^^^^^^^^^^^^^^^^^^^^^^^^ + +The read-object process enables Git to read all missing blobs with a +single process invocation for the entire life of a single Git command. +This is achieved by using a packet format (pkt-line, see technical/ +protocol-common.txt) based protocol over standard input and standard +output as follows. All packets, except for the "*CONTENT" packets and +the "0000" flush packet, are considered text and therefore are +terminated by a LF. + +Git starts the process when it encounters the first missing object that +needs to be retrieved. After the process is started, Git sends a welcome +message ("git-read-object-client"), a list of supported protocol version +numbers, and a flush packet. Git expects to read a welcome response +message ("git-read-object-server"), exactly one protocol version number +from the previously sent list, and a flush packet. All further +communication will be based on the selected version. + +The remaining protocol description below documents "version=1". Please +note that "version=42" in the example below does not exist and is only +there to illustrate how the protocol would look with more than one +version. + +After the version negotiation Git sends a list of all capabilities that +it supports and a flush packet. Git expects to read a list of desired +capabilities, which must be a subset of the supported capabilities list, +and a flush packet as response: +------------------------ +packet: git> git-read-object-client +packet: git> version=1 +packet: git> version=42 +packet: git> 0000 +packet: git< git-read-object-server +packet: git< version=1 +packet: git< 0000 +packet: git> capability=get +packet: git> capability=have +packet: git> capability=put +packet: git> capability=not-yet-invented +packet: git> 0000 +packet: git< capability=get +packet: git< 0000 +------------------------ +The only supported capability in version 1 is "get". + +Afterwards Git sends a list of "key=value" pairs terminated with a flush +packet. The list will contain at least the command (based on the +supported capabilities) and the sha1 of the object to retrieve. Please +note, that the process must not send any response before it received the +final flush packet. + +When the process receives the "get" command, it should make the requested +object available in the git object store and then return success. Git will +then check the object store again and this time find it and proceed. +------------------------ +packet: git> command=get +packet: git> sha1=0a214a649e1b3d5011e14a3dc227753f2bd2be05 +packet: git> 0000 +------------------------ + +The process is expected to respond with a list of "key=value" pairs +terminated with a flush packet. If the process does not experience +problems then the list must contain a "success" status. +------------------------ +packet: git< status=success +packet: git< 0000 +------------------------ + +In case the process cannot or does not want to process the content, it +is expected to respond with an "error" status. +------------------------ +packet: git< status=error +packet: git< 0000 +------------------------ + +In case the process cannot or does not want to process the content as +well as any future content for the lifetime of the Git process, then it +is expected to respond with an "abort" status at any point in the +protocol. +------------------------ +packet: git< status=abort +packet: git< 0000 +------------------------ + +Git neither stops nor restarts the process in case the "error"/"abort" +status is set. + +If the process dies during the communication or does not adhere to the +protocol then Git will stop the process and restart it with the next +object that needs to be processed. + +After the read-object process has processed an object it is expected to +wait for the next "key=value" list containing a command. Git will close +the command pipe on exit. The process is expected to detect EOF and exit +gracefully on its own. Git will wait until the process has stopped. + +A long running read-object process demo implementation can be found in +`contrib/long-running-read-object/example.pl` located in the Git core +repository. If you develop your own long running process then the +`GIT_TRACE_PACKET` environment variables can be very helpful for +debugging (see linkgit:git[1]). diff --git a/contrib/long-running-read-object/example.pl b/contrib/long-running-read-object/example.pl new file mode 100644 index 00000000000000..b8f37f836a813c --- /dev/null +++ b/contrib/long-running-read-object/example.pl @@ -0,0 +1,114 @@ +#!/usr/bin/perl +# +# Example implementation for the Git read-object protocol version 1 +# See Documentation/technical/read-object-protocol.txt +# +# Allows you to test the ability for blobs to be pulled from a host git repo +# "on demand." Called when git needs a blob it couldn't find locally due to +# a lazy clone that only cloned the commits and trees. +# +# A lazy clone can be simulated via the following commands from the host repo +# you wish to create a lazy clone of: +# +# cd /host_repo +# git rev-parse HEAD +# git init /guest_repo +# git cat-file --batch-check --batch-all-objects | grep -v 'blob' | +# cut -d' ' -f1 | git pack-objects /guest_repo/.git/objects/pack/noblobs +# cd /guest_repo +# git config core.virtualizeobjects true +# git reset --hard +# +# Please note, this sample is a minimal skeleton. No proper error handling +# was implemented. +# + +use strict; +use warnings; + +# +# Point $DIR to the folder where your host git repo is located so we can pull +# missing objects from it +# +my $DIR = "/host_repo/.git/"; + +sub packet_bin_read { + my $buffer; + my $bytes_read = read STDIN, $buffer, 4; + if ( $bytes_read == 0 ) { + + # EOF - Git stopped talking to us! + exit(); + } + elsif ( $bytes_read != 4 ) { + die "invalid packet: '$buffer'"; + } + my $pkt_size = hex($buffer); + if ( $pkt_size == 0 ) { + return ( 1, "" ); + } + elsif ( $pkt_size > 4 ) { + my $content_size = $pkt_size - 4; + $bytes_read = read STDIN, $buffer, $content_size; + if ( $bytes_read != $content_size ) { + die "invalid packet ($content_size bytes expected; $bytes_read bytes read)"; + } + return ( 0, $buffer ); + } + else { + die "invalid packet size: $pkt_size"; + } +} + +sub packet_txt_read { + my ( $res, $buf ) = packet_bin_read(); + unless ( $buf =~ s/\n$// ) { + die "A non-binary line MUST be terminated by an LF."; + } + return ( $res, $buf ); +} + +sub packet_bin_write { + my $buf = shift; + print STDOUT sprintf( "%04x", length($buf) + 4 ); + print STDOUT $buf; + STDOUT->flush(); +} + +sub packet_txt_write { + packet_bin_write( $_[0] . "\n" ); +} + +sub packet_flush { + print STDOUT sprintf( "%04x", 0 ); + STDOUT->flush(); +} + +( packet_txt_read() eq ( 0, "git-read-object-client" ) ) || die "bad initialize"; +( packet_txt_read() eq ( 0, "version=1" ) ) || die "bad version"; +( packet_bin_read() eq ( 1, "" ) ) || die "bad version end"; + +packet_txt_write("git-read-object-server"); +packet_txt_write("version=1"); +packet_flush(); + +( packet_txt_read() eq ( 0, "capability=get" ) ) || die "bad capability"; +( packet_bin_read() eq ( 1, "" ) ) || die "bad capability end"; + +packet_txt_write("capability=get"); +packet_flush(); + +while (1) { + my ($command) = packet_txt_read() =~ /^command=([^=]+)$/; + + if ( $command eq "get" ) { + my ($sha1) = packet_txt_read() =~ /^sha1=([0-9a-f]{40})$/; + packet_bin_read(); + + system ('git --git-dir="' . $DIR . '" cat-file blob ' . $sha1 . ' | git -c core.virtualizeobjects=false hash-object -w --stdin >/dev/null 2>&1'); + packet_txt_write(($?) ? "status=error" : "status=success"); + packet_flush(); + } else { + die "bad command '$command'"; + } +} diff --git a/object-file.c b/object-file.c index afdca5fe453c04..fb8cbbb1859ee4 100644 --- a/object-file.c +++ b/object-file.c @@ -42,6 +42,9 @@ #include "object-file-convert.h" #include "trace.h" #include "hook.h" +#include "sigchain.h" +#include "sub-process.h" +#include "pkt-line.h" /* The maximum size for an object header. */ #define MAX_HEADER_LEN 32 @@ -1027,6 +1030,115 @@ int has_alt_odb(struct repository *r) return !!r->objects->odb->next; } +#define CAP_GET (1u<<0) + +static int subprocess_map_initialized; +static struct hashmap subprocess_map; + +struct read_object_process { + struct subprocess_entry subprocess; + unsigned int supported_capabilities; +}; + +static int start_read_object_fn(struct subprocess_entry *subprocess) +{ + struct read_object_process *entry = (struct read_object_process *)subprocess; + static int versions[] = {1, 0}; + static struct subprocess_capability capabilities[] = { + { "get", CAP_GET }, + { NULL, 0 } + }; + + return subprocess_handshake(subprocess, "git-read-object", versions, + NULL, capabilities, + &entry->supported_capabilities); +} + +static int read_object_process(const struct object_id *oid) +{ + int err; + struct read_object_process *entry; + struct child_process *process; + struct strbuf status = STRBUF_INIT; + const char *cmd = find_hook(the_repository, "read-object"); + uint64_t start; + + start = getnanotime(); + + if (!subprocess_map_initialized) { + subprocess_map_initialized = 1; + hashmap_init(&subprocess_map, (hashmap_cmp_fn)cmd2process_cmp, + NULL, 0); + entry = NULL; + } else { + entry = (struct read_object_process *) subprocess_find_entry(&subprocess_map, cmd); + } + + if (!entry) { + entry = xmalloc(sizeof(*entry)); + entry->supported_capabilities = 0; + + if (subprocess_start(&subprocess_map, &entry->subprocess, cmd, + start_read_object_fn)) { + free(entry); + return -1; + } + } + process = &entry->subprocess.process; + + if (!(CAP_GET & entry->supported_capabilities)) + return -1; + + sigchain_push(SIGPIPE, SIG_IGN); + + err = packet_write_fmt_gently(process->in, "command=get\n"); + if (err) + goto done; + + err = packet_write_fmt_gently(process->in, "sha1=%s\n", oid_to_hex(oid)); + if (err) + goto done; + + err = packet_flush_gently(process->in); + if (err) + goto done; + + err = subprocess_read_status(process->out, &status); + err = err ? err : strcmp(status.buf, "success"); + +done: + sigchain_pop(SIGPIPE); + + if (err || errno == EPIPE) { + err = err ? err : errno; + if (!strcmp(status.buf, "error")) { + /* The process signaled a problem with the file. */ + } + else if (!strcmp(status.buf, "abort")) { + /* + * The process signaled a permanent problem. Don't try to read + * objects with the same command for the lifetime of the current + * Git process. + */ + entry->supported_capabilities &= ~CAP_GET; + } + else { + /* + * Something went wrong with the read-object process. + * Force shutdown and restart if needed. + */ + error("external process '%s' failed", cmd); + subprocess_stop(&subprocess_map, + (struct subprocess_entry *)entry); + free(entry); + } + } + + trace_performance_since(start, "read_object_process"); + + return err; +} + /* Returns 1 if we have successfully freshened the file, 0 otherwise. */ static int freshen_file(const char *fn) { @@ -1077,8 +1189,19 @@ static int check_and_freshen_nonlocal(const struct object_id *oid, int freshen) static int check_and_freshen(const struct object_id *oid, int freshen) { - return check_and_freshen_local(oid, freshen) || + int ret; + int tried_hook = 0; + +retry: + ret = check_and_freshen_local(oid, freshen) || check_and_freshen_nonlocal(oid, freshen); + if (!ret && core_virtualize_objects && !tried_hook) { + tried_hook = 1; + if (!read_object_process(oid)) + goto retry; + } + + return ret; } int has_loose_object_nonlocal(const struct object_id *oid) @@ -1622,20 +1745,6 @@ void disable_obj_read_lock(void) pthread_mutex_destroy(&obj_read_mutex); } -static int run_read_object_hook(struct repository *r, const struct object_id *oid) -{ - struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT; - int ret; - uint64_t start; - - start = getnanotime(); - strvec_push(&opt.args, oid_to_hex(oid)); - ret = run_hooks_opt(r, "read-object", &opt); - trace_performance_since(start, "run_read_object_hook"); - - return ret; -} - int fetch_if_missing = 1; static int do_oid_object_info_extended(struct repository *r, @@ -1694,7 +1803,7 @@ static int do_oid_object_info_extended(struct repository *r, break; if (core_virtualize_objects && !tried_hook) { tried_hook = 1; - if (!run_read_object_hook(r, oid)) + if (!read_object_process(oid)) goto retry; } } diff --git a/t/t0410/read-object b/t/t0410/read-object new file mode 100755 index 00000000000000..2b8feacc78577f --- /dev/null +++ b/t/t0410/read-object @@ -0,0 +1,114 @@ +#!/usr/bin/perl +# +# Example implementation for the Git read-object protocol version 1 +# See Documentation/technical/read-object-protocol.txt +# +# Allows you to test the ability for blobs to be pulled from a host git repo +# "on demand." Called when git needs a blob it couldn't find locally due to +# a lazy clone that only cloned the commits and trees. +# +# A lazy clone can be simulated via the following commands from the host repo +# you wish to create a lazy clone of: +# +# cd /host_repo +# git rev-parse HEAD +# git init /guest_repo +# git cat-file --batch-check --batch-all-objects | grep -v 'blob' | +# cut -d' ' -f1 | git pack-objects /guest_repo/.git/objects/pack/noblobs +# cd /guest_repo +# git config core.virtualizeobjects true +# git reset --hard +# +# Please note, this sample is a minimal skeleton. No proper error handling +# was implemented. +# + +use strict; +use warnings; + +# +# Point $DIR to the folder where your host git repo is located so we can pull +# missing objects from it +# +my $DIR = "../.git/"; + +sub packet_bin_read { + my $buffer; + my $bytes_read = read STDIN, $buffer, 4; + if ( $bytes_read == 0 ) { + + # EOF - Git stopped talking to us! + exit(); + } + elsif ( $bytes_read != 4 ) { + die "invalid packet: '$buffer'"; + } + my $pkt_size = hex($buffer); + if ( $pkt_size == 0 ) { + return ( 1, "" ); + } + elsif ( $pkt_size > 4 ) { + my $content_size = $pkt_size - 4; + $bytes_read = read STDIN, $buffer, $content_size; + if ( $bytes_read != $content_size ) { + die "invalid packet ($content_size bytes expected; $bytes_read bytes read)"; + } + return ( 0, $buffer ); + } + else { + die "invalid packet size: $pkt_size"; + } +} + +sub packet_txt_read { + my ( $res, $buf ) = packet_bin_read(); + unless ( $buf =~ s/\n$// ) { + die "A non-binary line MUST be terminated by an LF."; + } + return ( $res, $buf ); +} + +sub packet_bin_write { + my $buf = shift; + print STDOUT sprintf( "%04x", length($buf) + 4 ); + print STDOUT $buf; + STDOUT->flush(); +} + +sub packet_txt_write { + packet_bin_write( $_[0] . "\n" ); +} + +sub packet_flush { + print STDOUT sprintf( "%04x", 0 ); + STDOUT->flush(); +} + +( packet_txt_read() eq ( 0, "git-read-object-client" ) ) || die "bad initialize"; +( packet_txt_read() eq ( 0, "version=1" ) ) || die "bad version"; +( packet_bin_read() eq ( 1, "" ) ) || die "bad version end"; + +packet_txt_write("git-read-object-server"); +packet_txt_write("version=1"); +packet_flush(); + +( packet_txt_read() eq ( 0, "capability=get" ) ) || die "bad capability"; +( packet_bin_read() eq ( 1, "" ) ) || die "bad capability end"; + +packet_txt_write("capability=get"); +packet_flush(); + +while (1) { + my ($command) = packet_txt_read() =~ /^command=([^=]+)$/; + + if ( $command eq "get" ) { + my ($sha1) = packet_txt_read() =~ /^sha1=([0-9a-f]{40,64})$/; + packet_bin_read(); + + system ('git --git-dir="' . $DIR . '" cat-file blob ' . $sha1 . ' | git -c core.virtualizeobjects=false hash-object -w --stdin >/dev/null 2>&1'); + packet_txt_write(($?) ? "status=error" : "status=success"); + packet_flush(); + } else { + die "bad command '$command'"; + } +} diff --git a/t/t0499-read-object.sh b/t/t0499-read-object.sh new file mode 100755 index 00000000000000..2e208bdb46add5 --- /dev/null +++ b/t/t0499-read-object.sh @@ -0,0 +1,30 @@ +#!/bin/sh + +test_description='tests for long running read-object process' + +. ./test-lib.sh + +test_expect_success 'setup host repo with a root commit' ' + test_commit zero && + hash1=$(git ls-tree HEAD | grep zero.t | cut -f1 | cut -d\ -f3) +' + +test_expect_success 'blobs can be retrieved from the host repo' ' + git init guest-repo && + (cd guest-repo && + mkdir -p .git/hooks && + sed "1s|/usr/bin/perl|$PERL_PATH|" \ + <$TEST_DIRECTORY/t0410/read-object \ + >.git/hooks/read-object && + chmod +x .git/hooks/read-object && + git config core.virtualizeobjects true && + git cat-file blob "$hash1") +' + +test_expect_success 'invalid blobs generate errors' ' + (cd guest-repo && + test_must_fail git cat-file blob "invalid") +' + + +test_done From f3f8f889f5a11aaa4c15993ac58e58231ff15a92 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 8 Sep 2017 11:32:43 +0200 Subject: [PATCH 038/196] sha1_file: when writing objects, skip the read_object_hook If we are going to write an object there is no use in calling the read object hook to get an object from a potentially remote source. We would rather just write out the object and avoid the potential round trip for an object that doesn't exist. This change adds a flag to the check_and_freshen() and freshen_loose_object() functions' signatures so that the hook is bypassed when the functions are called before writing loose objects. The check for a local object is still performed so we don't overwrite something that has already been written to one of the objects directories. Based on a patch by Kevin Willford. Signed-off-by: Johannes Schindelin --- object-file.c | 19 +++++++++++-------- t/t0410/read-object | 4 ++++ t/t0499-read-object.sh | 7 +++++++ 3 files changed, 22 insertions(+), 8 deletions(-) diff --git a/object-file.c b/object-file.c index fb8cbbb1859ee4..7a9574c09e17be 100644 --- a/object-file.c +++ b/object-file.c @@ -1187,7 +1187,8 @@ static int check_and_freshen_nonlocal(const struct object_id *oid, int freshen) return 0; } -static int check_and_freshen(const struct object_id *oid, int freshen) +static int check_and_freshen(const struct object_id *oid, int freshen, + int skip_virtualized_objects) { int ret; int tried_hook = 0; @@ -1195,7 +1196,8 @@ static int check_and_freshen(const struct object_id *oid, int freshen) retry: ret = check_and_freshen_local(oid, freshen) || check_and_freshen_nonlocal(oid, freshen); - if (!ret && core_virtualize_objects && !tried_hook) { + if (!ret && core_virtualize_objects && !skip_virtualized_objects && + !tried_hook) { tried_hook = 1; if (!read_object_process(oid)) goto retry; @@ -1211,7 +1213,7 @@ int has_loose_object_nonlocal(const struct object_id *oid) int has_loose_object(const struct object_id *oid) { - return check_and_freshen(oid, 0); + return check_and_freshen(oid, 0, 0); } static void mmap_limit_check(size_t length) @@ -2465,9 +2467,10 @@ static int write_loose_object(const struct object_id *oid, char *hdr, FOF_SKIP_COLLISION_CHECK); } -static int freshen_loose_object(const struct object_id *oid) +static int freshen_loose_object(const struct object_id *oid, + int skip_virtualized_objects) { - return check_and_freshen(oid, 1); + return check_and_freshen(oid, 1, skip_virtualized_objects); } static int freshen_packed_object(const struct object_id *oid) @@ -2563,7 +2566,7 @@ int stream_loose_object(struct input_stream *in_stream, size_t len, die(_("deflateEnd on stream object failed (%d)"), ret); close_loose_object(fd, tmp_file.buf); - if (freshen_packed_object(oid) || freshen_loose_object(oid)) { + if (freshen_packed_object(oid) || freshen_loose_object(oid, 1)) { unlink_or_warn(tmp_file.buf); goto cleanup; } @@ -2625,7 +2628,7 @@ int write_object_file_flags(const void *buf, size_t len, * it out into .git/objects/??/?{38} file. */ write_object_file_prepare(algo, buf, len, type, oid, hdr, &hdrlen); - if (freshen_packed_object(oid) || freshen_loose_object(oid)) + if (freshen_packed_object(oid) || freshen_loose_object(oid, 1)) return 0; if (write_loose_object(oid, hdr, hdrlen, buf, len, 0, flags)) return -1; @@ -2669,7 +2672,7 @@ int write_object_file_literally(const void *buf, size_t len, if (!(flags & HASH_WRITE_OBJECT)) goto cleanup; - if (freshen_packed_object(oid) || freshen_loose_object(oid)) + if (freshen_packed_object(oid) || freshen_loose_object(oid, 1)) goto cleanup; status = write_loose_object(oid, header, hdrlen, buf, len, 0, 0); if (compat_type != -1) diff --git a/t/t0410/read-object b/t/t0410/read-object index 2b8feacc78577f..02c799837f4057 100755 --- a/t/t0410/read-object +++ b/t/t0410/read-object @@ -108,6 +108,10 @@ while (1) { system ('git --git-dir="' . $DIR . '" cat-file blob ' . $sha1 . ' | git -c core.virtualizeobjects=false hash-object -w --stdin >/dev/null 2>&1'); packet_txt_write(($?) ? "status=error" : "status=success"); packet_flush(); + + open my $log, '>>.git/read-object-hook.log'; + print $log "Read object $sha1, exit code $?\n"; + close $log; } else { die "bad command '$command'"; } diff --git a/t/t0499-read-object.sh b/t/t0499-read-object.sh index 2e208bdb46add5..0cee1963cf091e 100755 --- a/t/t0499-read-object.sh +++ b/t/t0499-read-object.sh @@ -26,5 +26,12 @@ test_expect_success 'invalid blobs generate errors' ' test_must_fail git cat-file blob "invalid") ' +test_expect_success 'read-object-hook is bypassed when writing objects' ' + (cd guest-repo && + echo hello >hello.txt && + git add hello.txt && + hash="$(git rev-parse --verify :hello.txt)" && + ! grep "$hash" .git/read-object-hook.log) +' test_done From ec080b04e3cd2449eed65bd23cde4e88992cecd8 Mon Sep 17 00:00:00 2001 From: Ben Peart Date: Tue, 24 May 2016 00:32:38 +0000 Subject: [PATCH 039/196] gvfs: add global command pre and post hook procs This adds hard-coded call to GVFS.hooks.exe before and after each Git command runs. To make sure that this is only called on repositories cloned with GVFS, we test for the tell-tale .gvfs. 2021-10-30: Recent movement of find_hook() to hook.c required moving these changes out of run-command.c to hook.c. Signed-off-by: Ben Peart --- git.c | 84 ++++++++++++++++++++++++++++++++++-- hook.c | 57 +++++++++++++++++++++++- t/t0400-pre-command-hook.sh | 34 +++++++++++++++ t/t0401-post-command-hook.sh | 32 ++++++++++++++ 4 files changed, 203 insertions(+), 4 deletions(-) create mode 100755 t/t0400-pre-command-hook.sh create mode 100755 t/t0401-post-command-hook.sh diff --git a/git.c b/git.c index 1825350267470d..5271fe20bc9177 100644 --- a/git.c +++ b/git.c @@ -17,6 +17,8 @@ #include "shallow.h" #include "trace.h" #include "trace2.h" +#include "dir.h" +#include "hook.h" #define RUN_SETUP (1<<0) #define RUN_SETUP_GENTLY (1<<1) @@ -441,6 +443,67 @@ static int handle_alias(int *argcp, const char ***argv) return ret; } +/* Runs pre/post-command hook */ +static struct strvec sargv = STRVEC_INIT; +static int run_post_hook = 0; +static int exit_code = -1; + +static int run_pre_command_hook(struct repository *r, const char **argv) +{ + char *lock; + int ret = 0; + struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT; + + /* + * Ensure the global pre/post command hook is only called for + * the outer command and not when git is called recursively + * or spawns multiple commands (like with the alias command) + */ + lock = getenv("COMMAND_HOOK_LOCK"); + if (lock && !strcmp(lock, "true")) + return 0; + setenv("COMMAND_HOOK_LOCK", "true", 1); + + /* call the hook proc */ + strvec_pushv(&sargv, argv); + strvec_pushv(&opt.args, sargv.v); + ret = run_hooks_opt(r, "pre-command", &opt); + + if (!ret) + run_post_hook = 1; + return ret; +} + +static int run_post_command_hook(struct repository *r) +{ + char *lock; + int ret = 0; + struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT; + + /* + * Only run post_command if pre_command succeeded in this process + */ + if (!run_post_hook) + return 0; + lock = getenv("COMMAND_HOOK_LOCK"); + if (!lock || strcmp(lock, "true")) + return 0; + + strvec_pushv(&opt.args, sargv.v); + strvec_pushf(&opt.args, "--exit_code=%u", exit_code); + ret = run_hooks_opt(r, "post-command", &opt); + + run_post_hook = 0; + strvec_clear(&sargv); + setenv("COMMAND_HOOK_LOCK", "false", 1); + return ret; +} + +static void post_command_hook_atexit(void) +{ + run_post_command_hook(the_repository); +} + static int run_builtin(struct cmd_struct *p, int argc, const char **argv, struct repository *repo) { int status, help; @@ -476,16 +539,21 @@ static int run_builtin(struct cmd_struct *p, int argc, const char **argv, struct if (!help && p->option & NEED_WORK_TREE) setup_work_tree(); + if (run_pre_command_hook(the_repository, argv)) + die("pre-command hook aborted command"); + trace_argv_printf(argv, "trace: built-in: git"); trace2_cmd_name(p->cmd); validate_cache_entries(repo->index); - status = p->fn(argc, argv, prefix, (p->option & RUN_SETUP)? repo : NULL); + exit_code = status = p->fn(argc, argv, prefix, (p->option & RUN_SETUP)? repo : NULL); validate_cache_entries(repo->index); if (status) return status; + run_post_command_hook(the_repository); + /* Somebody closed stdout? */ if (fstat(fileno(stdout), &st)) return 0; @@ -782,13 +850,16 @@ static void execv_dashed_external(const char **argv) */ trace_argv_printf(cmd.args.v, "trace: exec:"); + if (run_pre_command_hook(the_repository, cmd.args.v)) + die("pre-command hook aborted command"); + /* * If we fail because the command is not found, it is * OK to return. Otherwise, we just pass along the status code, * or our usual generic code if we were not even able to exec * the program. */ - status = run_command(&cmd); + exit_code = status = run_command(&cmd); /* * If the child process ran and we are now going to exit, emit a @@ -799,6 +870,8 @@ static void execv_dashed_external(const char **argv) exit(status); else if (errno != ENOENT) exit(128); + + run_post_command_hook(the_repository); } static int run_argv(int *argcp, const char ***argv) @@ -906,6 +979,7 @@ int cmd_main(int argc, const char **argv) } trace_command_performance(argv); + atexit(post_command_hook_atexit); /* * "git-xxxx" is the same as "git xxxx", but we obviously: @@ -931,10 +1005,14 @@ int cmd_main(int argc, const char **argv) if (!argc) { /* The user didn't specify a command; give them help */ commit_pager_choice(); + if (run_pre_command_hook(the_repository, argv)) + die("pre-command hook aborted command"); printf(_("usage: %s\n\n"), git_usage_string); list_common_cmds_help(); printf("\n%s\n", _(git_more_info_string)); - exit(1); + exit_code = 1; + run_post_command_hook(the_repository); + exit(exit_code); } if (!strcmp("--version", argv[0]) || !strcmp("-v", argv[0])) diff --git a/hook.c b/hook.c index a9320cb0ce95e5..9755c001900afa 100644 --- a/hook.c +++ b/hook.c @@ -1,5 +1,8 @@ +#define USE_THE_REPOSITORY_VARIABLE + #include "git-compat-util.h" #include "abspath.h" +#include "environment.h" #include "advice.h" #include "gettext.h" #include "hook.h" @@ -10,6 +13,54 @@ #include "environment.h" #include "setup.h" +static int early_hooks_path_config(const char *var, const char *value, + const struct config_context *ctx UNUSED, void *cb) +{ + if (!strcmp(var, "core.hookspath")) + return git_config_pathname((char **)cb, var, value); + + return 0; +} + +/* Discover the hook before setup_git_directory() was called */ +static const char *hook_path_early(const char *name, struct strbuf *result) +{ + static struct strbuf hooks_dir = STRBUF_INIT; + static int initialized; + + if (initialized < 0) + return NULL; + + if (!initialized) { + struct strbuf gitdir = STRBUF_INIT, commondir = STRBUF_INIT; + char *early_hooks_dir = NULL; + + if (discover_git_directory(&commondir, &gitdir) < 0) { + strbuf_release(&gitdir); + strbuf_release(&commondir); + initialized = -1; + return NULL; + } + + read_early_config(the_repository, early_hooks_path_config, &early_hooks_dir); + if (!early_hooks_dir) + strbuf_addf(&hooks_dir, "%s/hooks/", commondir.buf); + else { + strbuf_add_absolute_path(&hooks_dir, early_hooks_dir); + free(early_hooks_dir); + strbuf_addch(&hooks_dir, '/'); + } + + strbuf_release(&gitdir); + strbuf_release(&commondir); + + initialized = 1; + } + + strbuf_addf(result, "%s%s", hooks_dir.buf, name); + return result->buf; +} + const char *find_hook(struct repository *r, const char *name) { static struct strbuf path = STRBUF_INIT; @@ -17,7 +68,11 @@ const char *find_hook(struct repository *r, const char *name) int found_hook; strbuf_reset(&path); - strbuf_repo_git_path(&path, r, "hooks/%s", name); + if (have_git_dir()) + strbuf_repo_git_path(&path, r, "hooks/%s", name); + else if (!hook_path_early(name, &path)) + return NULL; + found_hook = access(path.buf, X_OK) >= 0; #ifdef STRIP_EXTENSION if (!found_hook) { diff --git a/t/t0400-pre-command-hook.sh b/t/t0400-pre-command-hook.sh new file mode 100755 index 00000000000000..4f4f610b52b0a0 --- /dev/null +++ b/t/t0400-pre-command-hook.sh @@ -0,0 +1,34 @@ +#!/bin/sh + +test_description='pre-command hook' + +. ./test-lib.sh + +test_expect_success 'with no hook' ' + echo "first" > file && + git add file && + git commit -m "first" +' + +test_expect_success 'with succeeding hook' ' + mkdir -p .git/hooks && + write_script .git/hooks/pre-command <<-EOF && + echo "\$*" >\$(git rev-parse --git-dir)/pre-command.out + EOF + echo "second" >> file && + git add file && + test "add file" = "$(cat .git/pre-command.out)" && + echo Hello | git hash-object --stdin && + test "hash-object --stdin" = "$(cat .git/pre-command.out)" +' + +test_expect_success 'with failing hook' ' + write_script .git/hooks/pre-command <<-EOF && + exit 1 + EOF + echo "third" >> file && + test_must_fail git add file && + test_path_is_missing "$(cat .git/pre-command.out)" +' + +test_done diff --git a/t/t0401-post-command-hook.sh b/t/t0401-post-command-hook.sh new file mode 100755 index 00000000000000..64646f7ad03b57 --- /dev/null +++ b/t/t0401-post-command-hook.sh @@ -0,0 +1,32 @@ +#!/bin/sh + +test_description='post-command hook' + +. ./test-lib.sh + +test_expect_success 'with no hook' ' + echo "first" > file && + git add file && + git commit -m "first" +' + +test_expect_success 'with succeeding hook' ' + mkdir -p .git/hooks && + write_script .git/hooks/post-command <<-EOF && + echo "\$*" >\$(git rev-parse --git-dir)/post-command.out + EOF + echo "second" >> file && + git add file && + test "add file --exit_code=0" = "$(cat .git/post-command.out)" +' + +test_expect_success 'with failing pre-command hook' ' + write_script .git/hooks/pre-command <<-EOF && + exit 1 + EOF + echo "third" >> file && + test_must_fail git add file && + test_path_is_missing "$(cat .git/post-command.out)" +' + +test_done From 348aec030787922fdd311b52878518a87eb88b99 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Thu, 16 Mar 2017 21:07:54 +0100 Subject: [PATCH 040/196] t0400: verify that the hook is called correctly from a subdirectory Suggested by Ben Peart. Signed-off-by: Johannes Schindelin --- t/t0400-pre-command-hook.sh | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) diff --git a/t/t0400-pre-command-hook.sh b/t/t0400-pre-command-hook.sh index 4f4f610b52b0a0..83c453c9643eae 100755 --- a/t/t0400-pre-command-hook.sh +++ b/t/t0400-pre-command-hook.sh @@ -31,4 +31,27 @@ test_expect_success 'with failing hook' ' test_path_is_missing "$(cat .git/pre-command.out)" ' +test_expect_success 'in a subdirectory' ' + echo touch i-was-here | write_script .git/hooks/pre-command && + mkdir sub && + ( + cd sub && + git version + ) && + test_path_is_file sub/i-was-here +' + +test_expect_success 'in a subdirectory, using an alias' ' + git reset --hard && + echo "echo \"\$@; \$(pwd)\" >>log" | + write_script .git/hooks/pre-command && + mkdir -p sub && + ( + cd sub && + git -c alias.v="version" v + ) && + test_path_is_missing log && + test_line_count = 2 sub/log +' + test_done From af565b2aed7da168e0608a6469f9fd53e862652b Mon Sep 17 00:00:00 2001 From: Alejandro Pauly Date: Mon, 10 Apr 2017 13:26:14 -0400 Subject: [PATCH 041/196] Pass PID of git process to hooks. Signed-off-by: Alejandro Pauly --- git.c | 1 + t/t0400-pre-command-hook.sh | 3 ++- t/t0401-post-command-hook.sh | 3 ++- 3 files changed, 5 insertions(+), 2 deletions(-) diff --git a/git.c b/git.c index 5271fe20bc9177..5c69f2e6cf1857 100644 --- a/git.c +++ b/git.c @@ -466,6 +466,7 @@ static int run_pre_command_hook(struct repository *r, const char **argv) /* call the hook proc */ strvec_pushv(&sargv, argv); + strvec_pushf(&sargv, "--git-pid=%"PRIuMAX, (uintmax_t)getpid()); strvec_pushv(&opt.args, sargv.v); ret = run_hooks_opt(r, "pre-command", &opt); diff --git a/t/t0400-pre-command-hook.sh b/t/t0400-pre-command-hook.sh index 83c453c9643eae..f04a55a695bc97 100755 --- a/t/t0400-pre-command-hook.sh +++ b/t/t0400-pre-command-hook.sh @@ -13,7 +13,8 @@ test_expect_success 'with no hook' ' test_expect_success 'with succeeding hook' ' mkdir -p .git/hooks && write_script .git/hooks/pre-command <<-EOF && - echo "\$*" >\$(git rev-parse --git-dir)/pre-command.out + echo "\$*" | sed "s/ --git-pid=[0-9]*//" \ + >\$(git rev-parse --git-dir)/pre-command.out EOF echo "second" >> file && git add file && diff --git a/t/t0401-post-command-hook.sh b/t/t0401-post-command-hook.sh index 64646f7ad03b57..fcbfc4a0c79c1e 100755 --- a/t/t0401-post-command-hook.sh +++ b/t/t0401-post-command-hook.sh @@ -13,7 +13,8 @@ test_expect_success 'with no hook' ' test_expect_success 'with succeeding hook' ' mkdir -p .git/hooks && write_script .git/hooks/post-command <<-EOF && - echo "\$*" >\$(git rev-parse --git-dir)/post-command.out + echo "\$*" | sed "s/ --git-pid=[0-9]*//" \ + >\$(git rev-parse --git-dir)/post-command.out EOF echo "second" >> file && git add file && From 0e0264ed9ab3378f2335b278ae359dc4990591ec Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 8 Aug 2017 00:27:50 +0200 Subject: [PATCH 042/196] pre-command: always respect core.hooksPath We need to respect that config setting even if we already know that we have a repository, but have not yet read the config. The regression test was written by Alejandro Pauly. 2021-10-30: Recent movement of find_hook() into hook.c required moving this change from run-command.c. Signed-off-by: Johannes Schindelin --- hook.c | 15 +++++++++++++-- t/t0400-pre-command-hook.sh | 11 +++++++++++ 2 files changed, 24 insertions(+), 2 deletions(-) diff --git a/hook.c b/hook.c index 9755c001900afa..180d4185284a50 100644 --- a/hook.c +++ b/hook.c @@ -68,9 +68,20 @@ const char *find_hook(struct repository *r, const char *name) int found_hook; strbuf_reset(&path); - if (have_git_dir()) + if (have_git_dir()) { + static int forced_config; + + if (!forced_config) { + if (!git_hooks_path) { + git_config_get_pathname("core.hookspath", + &git_hooks_path); + UNLEAK(git_hooks_path); + } + forced_config = 1; + } + strbuf_repo_git_path(&path, r, "hooks/%s", name); - else if (!hook_path_early(name, &path)) + } else if (!hook_path_early(name, &path)) return NULL; found_hook = access(path.buf, X_OK) >= 0; diff --git a/t/t0400-pre-command-hook.sh b/t/t0400-pre-command-hook.sh index f04a55a695bc97..f2a9115e299385 100755 --- a/t/t0400-pre-command-hook.sh +++ b/t/t0400-pre-command-hook.sh @@ -55,4 +55,15 @@ test_expect_success 'in a subdirectory, using an alias' ' test_line_count = 2 sub/log ' +test_expect_success 'with core.hooksPath' ' + mkdir -p .git/alternateHooks && + write_script .git/alternateHooks/pre-command <<-EOF && + echo "alternate" >\$(git rev-parse --git-dir)/pre-command.out + EOF + write_script .git/hooks/pre-command <<-EOF && + echo "original" >\$(git rev-parse --git-dir)/pre-command.out + EOF + git -c core.hooksPath=.git/alternateHooks status && + test "alternate" = "$(cat .git/pre-command.out)" +' test_done From 2b67c6a74a1e2876e2fc13782600561dbc1fe6ce Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Wed, 22 Feb 2017 12:50:43 -0700 Subject: [PATCH 043/196] sparse-checkout: update files with a modify/delete conflict When using the sparse-checkout feature, the file might not be on disk because the skip-worktree bit is on. Signed-off-by: Kevin Willford --- merge-recursive.c | 2 +- t/t7615-merge-sparse-checkout.sh | 31 +++++++++++++++++++++++++++++++ 2 files changed, 32 insertions(+), 1 deletion(-) create mode 100755 t/t7615-merge-sparse-checkout.sh diff --git a/merge-recursive.c b/merge-recursive.c index 46be3b03286b9f..1992c8c2b237d5 100644 --- a/merge-recursive.c +++ b/merge-recursive.c @@ -1593,7 +1593,7 @@ static int handle_change_delete(struct merge_options *opt, * path. We could call update_file_flags() with update_cache=0 * and update_wd=0, but that's a no-op. */ - if (change_branch != opt->branch1 || alt_path) + if (change_branch != opt->branch1 || alt_path || !file_exists(update_path)) ret = update_file(opt, 0, changed, update_path); } free(alt_path); diff --git a/t/t7615-merge-sparse-checkout.sh b/t/t7615-merge-sparse-checkout.sh new file mode 100755 index 00000000000000..5ce12431f62ad1 --- /dev/null +++ b/t/t7615-merge-sparse-checkout.sh @@ -0,0 +1,31 @@ +#!/bin/sh + +test_description='merge can handle sparse-checkout' + +. ./test-lib.sh + +# merges with conflicts + +test_expect_success 'setup' ' + git branch -M main && + test_commit a && + test_commit file && + git checkout -b delete-file && + git rm file.t && + test_tick && + git commit -m "remove file" && + git checkout main && + test_commit modify file.t changed +' + +test_expect_success 'merge conflict deleted file and modified' ' + echo "/a.t" >.git/info/sparse-checkout && + test_config core.sparsecheckout true && + git checkout -f && + test_path_is_missing file.t && + test_must_fail git merge delete-file && + test_path_is_file file.t && + test "changed" = "$(cat file.t)" +' + +test_done From 3191f6e48959f2adb2aba5b5cd3ab200199dafc9 Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Wed, 1 Mar 2017 15:17:12 -0800 Subject: [PATCH 044/196] sparse-checkout: avoid writing entries with the skip-worktree bit When using the sparse-checkout feature git should not write to the working directory for files with the skip-worktree bit on. With the skip-worktree bit on the file may or may not be in the working directory and if it is not we don't want or need to create it by calling checkout_entry. There are two callers of checkout_target. Both of which check that the file does not exist before calling checkout_target. load_current which make a call to lstat right before calling checkout_target and check_preimage which will only run checkout_taret it stat_ret is less than zero. It sets stat_ret to zero and only if !stat->cached will it lstat the file and set stat_ret to something other than zero. This patch checks if skip-worktree bit is on in checkout_target and just returns so that the entry doesn't not end up in the working directory. This is so that apply will not create a file in the working directory, then update the index but not keep the working directory up to date with the changes that happened in the index. Signed-off-by: Kevin Willford --- apply.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/apply.c b/apply.c index 2088f7516393ae..15a17be9735e08 100644 --- a/apply.c +++ b/apply.c @@ -3369,6 +3369,24 @@ static int checkout_target(struct index_state *istate, { struct checkout costate = CHECKOUT_INIT; + /* + * Do not checkout the entry if the skipworktree bit is set + * + * Both callers of this method (check_preimage and load_current) + * check for the existance of the file before calling this + * method so we know that the file doesn't exist at this point + * and we don't need to perform that check again here. + * We just need to check the skip-worktree and return. + * + * This is to prevent git from creating a file in the + * working directory that has the skip-worktree bit on, + * then updating the index from the patch and not keeping + * the working directory version up to date with what it + * changed the index version to be. + */ + if (ce_skip_worktree(ce)) + return 0; + costate.refresh_cache = 1; costate.istate = istate; if (checkout_entry(ce, &costate, NULL, NULL) || From 481bff49749f631b36549a48bc38628050f1feeb Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Wed, 5 Apr 2017 10:55:32 -0600 Subject: [PATCH 045/196] Do not remove files outside the sparse-checkout Signed-off-by: Kevin Willford --- unpack-trees.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/unpack-trees.c b/unpack-trees.c index 9d55cc94874b3f..611b114a8e3def 100644 --- a/unpack-trees.c +++ b/unpack-trees.c @@ -572,7 +572,9 @@ static int apply_sparse_checkout(struct index_state *istate, ce->ce_flags &= ~CE_SKIP_WORKTREE; return -1; } - ce->ce_flags |= CE_WT_REMOVE; + if (!gvfs_config_is_set(GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT)) + ce->ce_flags |= CE_WT_REMOVE; + ce->ce_flags &= ~CE_UPDATE; } if (was_skip_worktree && !ce_skip_worktree(ce)) { From 31d7cc7855e21036f6f575d492702681700b8f2e Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Fri, 16 Nov 2018 11:28:59 -0700 Subject: [PATCH 046/196] send-pack: do not check for sha1 file when GVFS_MISSING_OK set --- send-pack.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/send-pack.c b/send-pack.c index 3481ecc8cfbc17..60ab129b68a218 100644 --- a/send-pack.c +++ b/send-pack.c @@ -6,6 +6,7 @@ #include "date.h" #include "gettext.h" #include "hex.h" +#include "gvfs.h" #include "object-store-ll.h" #include "pkt-line.h" #include "sideband.h" @@ -46,7 +47,7 @@ int option_parse_push_signed(const struct option *opt, static void feed_object(const struct object_id *oid, FILE *fh, int negative) { - if (negative && + if (negative && !gvfs_config_is_set(GVFS_MISSING_OK) && !repo_has_object_file_with_flags(the_repository, oid, OBJECT_INFO_SKIP_FETCH_OBJECT | OBJECT_INFO_QUICK)) From d24c5ab140e9f9d62a11f8ae1934fa95a51e7811 Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Mon, 3 Jul 2017 13:39:45 -0600 Subject: [PATCH 047/196] cache-tree: remove use of strbuf_addf in update_one String formatting can be a performance issue when there are hundreds of thousands of trees. Change to stop using the strbuf_addf and just add the strings or characters individually. There are a limited number of modes so added a switch for the known ones and a default case if something comes through that are not a known one for git. In one scenario regarding a huge worktree, this reduces the time required for a `git checkout ` from 44 seconds to 38 seconds, i.e. it is a non-negligible performance improvement. Signed-off-by: Kevin Willford --- cache-tree.c | 24 +++++++++++++++++++++++- 1 file changed, 23 insertions(+), 1 deletion(-) diff --git a/cache-tree.c b/cache-tree.c index 287658357a9dec..5db6587cc34a4e 100644 --- a/cache-tree.c +++ b/cache-tree.c @@ -430,7 +430,29 @@ static int update_one(struct cache_tree *it, continue; strbuf_grow(&buffer, entlen + 100); - strbuf_addf(&buffer, "%o %.*s%c", mode, entlen, path + baselen, '\0'); + + switch (mode) { + case 0100644: + strbuf_add(&buffer, "100644 ", 7); + break; + case 0100664: + strbuf_add(&buffer, "100664 ", 7); + break; + case 0100755: + strbuf_add(&buffer, "100755 ", 7); + break; + case 0120000: + strbuf_add(&buffer, "120000 ", 7); + break; + case 0160000: + strbuf_add(&buffer, "160000 ", 7); + break; + default: + strbuf_addf(&buffer, "%o ", mode); + break; + } + strbuf_add(&buffer, path + baselen, entlen); + strbuf_addch(&buffer, '\0'); strbuf_add(&buffer, oid->hash, the_hash_algo->rawsz); #if DEBUG_CACHE_TREE From f681e1a6814a91b43dd0b86055ff8f2aa26d7127 Mon Sep 17 00:00:00 2001 From: Ben Peart Date: Thu, 6 Dec 2018 11:09:19 -0500 Subject: [PATCH 048/196] gvfs: block unsupported commands when running in a GVFS repo The following commands and options are not currently supported when working in a GVFS repo. Add code to detect and block these commands from executing. 1) fsck 2) gc 4) prune 5) repack 6) submodule 8) update-index --split-index 9) update-index --index-version (other than 4) 10) update-index --[no-]skip-worktree 11) worktree Signed-off-by: Ben Peart Signed-off-by: Johannes Schindelin --- builtin/gc.c | 4 ++++ builtin/update-index.c | 10 ++++++++ git.c | 15 ++++++++---- gvfs.h | 1 + t/t0402-block-command-on-gvfs.sh | 39 ++++++++++++++++++++++++++++++++ 5 files changed, 64 insertions(+), 5 deletions(-) create mode 100755 t/t0402-block-command-on-gvfs.sh diff --git a/builtin/gc.c b/builtin/gc.c index 6a7a2da006eeee..7966eeb933a091 100644 --- a/builtin/gc.c +++ b/builtin/gc.c @@ -15,6 +15,7 @@ #include "date.h" #include "environment.h" #include "hex.h" +#include "gvfs.h" #include "config.h" #include "tempfile.h" #include "lockfile.h" @@ -736,6 +737,9 @@ struct repository *repo UNUSED) if (quiet) strvec_push(&repack, "-q"); + if ((!opts.auto_flag || (opts.auto_flag && cfg.gc_auto_threshold > 0)) && gvfs_config_is_set(GVFS_BLOCK_COMMANDS)) + die(_("'git gc' is not supported on a GVFS repo")); + if (opts.auto_flag) { if (cfg.detach_auto && opts.detach < 0) opts.detach = 1; diff --git a/builtin/update-index.c b/builtin/update-index.c index 45b4a8b5556e08..c838c8e6a30073 100644 --- a/builtin/update-index.c +++ b/builtin/update-index.c @@ -5,6 +5,7 @@ */ #define USE_THE_REPOSITORY_VARIABLE #include "builtin.h" +#include "gvfs.h" #include "bulk-checkin.h" #include "config.h" #include "environment.h" @@ -1112,7 +1113,13 @@ int cmd_update_index(int argc, argc = parse_options_end(&ctx); getline_fn = nul_term_line ? strbuf_getline_nul : strbuf_getline_lf; + if (mark_skip_worktree_only && gvfs_config_is_set(GVFS_BLOCK_COMMANDS)) + die(_("modifying the skip worktree bit is not supported on a GVFS repo")); + if (preferred_index_format) { + if (preferred_index_format != 4 && gvfs_config_is_set(GVFS_BLOCK_COMMANDS)) + die(_("changing the index version is not supported on a GVFS repo")); + if (preferred_index_format < 0) { printf(_("%d\n"), the_repository->index->version); } else if (preferred_index_format < INDEX_FORMAT_LB || @@ -1158,6 +1165,9 @@ int cmd_update_index(int argc, end_odb_transaction(); if (split_index > 0) { + if (gvfs_config_is_set(GVFS_BLOCK_COMMANDS)) + die(_("split index is not supported on a GVFS repo")); + if (repo_config_get_split_index(the_repository) == 0) warning(_("core.splitIndex is set to false; " "remove or change it, if you really want to " diff --git a/git.c b/git.c index 5c69f2e6cf1857..fbc1838905ad9e 100644 --- a/git.c +++ b/git.c @@ -1,6 +1,7 @@ #define USE_THE_REPOSITORY_VARIABLE #include "builtin.h" +#include "gvfs.h" #include "config.h" #include "environment.h" #include "exec-cmd.h" @@ -30,6 +31,7 @@ #define NEED_WORK_TREE (1<<3) #define DELAY_PAGER_CONFIG (1<<4) #define NO_PARSEOPT (1<<5) /* parse-options is not used */ +#define BLOCK_ON_GVFS_REPO (1<<6) /* command not allowed in GVFS repos */ struct cmd_struct { const char *cmd; @@ -540,6 +542,9 @@ static int run_builtin(struct cmd_struct *p, int argc, const char **argv, struct if (!help && p->option & NEED_WORK_TREE) setup_work_tree(); + if (!help && p->option & BLOCK_ON_GVFS_REPO && gvfs_config_is_set(GVFS_BLOCK_COMMANDS)) + die("'git %s' is not supported on a GVFS repo", p->cmd); + if (run_pre_command_hook(the_repository, argv)) die("pre-command hook aborted command"); @@ -623,7 +628,7 @@ static struct cmd_struct commands[] = { { "for-each-ref", cmd_for_each_ref, RUN_SETUP }, { "for-each-repo", cmd_for_each_repo, RUN_SETUP_GENTLY }, { "format-patch", cmd_format_patch, RUN_SETUP }, - { "fsck", cmd_fsck, RUN_SETUP }, + { "fsck", cmd_fsck, RUN_SETUP | BLOCK_ON_GVFS_REPO}, { "fsck-objects", cmd_fsck, RUN_SETUP }, { "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP }, { "gc", cmd_gc, RUN_SETUP }, @@ -664,7 +669,7 @@ static struct cmd_struct commands[] = { { "pack-refs", cmd_pack_refs, RUN_SETUP }, { "patch-id", cmd_patch_id, RUN_SETUP_GENTLY | NO_PARSEOPT }, { "pickaxe", cmd_blame, RUN_SETUP }, - { "prune", cmd_prune, RUN_SETUP }, + { "prune", cmd_prune, RUN_SETUP | BLOCK_ON_GVFS_REPO}, { "prune-packed", cmd_prune_packed, RUN_SETUP }, { "pull", cmd_pull, RUN_SETUP | NEED_WORK_TREE }, { "push", cmd_push, RUN_SETUP }, @@ -677,7 +682,7 @@ static struct cmd_struct commands[] = { { "remote", cmd_remote, RUN_SETUP }, { "remote-ext", cmd_remote_ext, NO_PARSEOPT }, { "remote-fd", cmd_remote_fd, NO_PARSEOPT }, - { "repack", cmd_repack, RUN_SETUP }, + { "repack", cmd_repack, RUN_SETUP | BLOCK_ON_GVFS_REPO }, { "replace", cmd_replace, RUN_SETUP }, { "replay", cmd_replay, RUN_SETUP }, { "rerere", cmd_rerere, RUN_SETUP }, @@ -698,7 +703,7 @@ static struct cmd_struct commands[] = { { "stash", cmd_stash, RUN_SETUP | NEED_WORK_TREE }, { "status", cmd_status, RUN_SETUP | NEED_WORK_TREE }, { "stripspace", cmd_stripspace }, - { "submodule--helper", cmd_submodule__helper, RUN_SETUP }, + { "submodule--helper", cmd_submodule__helper, RUN_SETUP | BLOCK_ON_GVFS_REPO }, { "survey", cmd_survey, RUN_SETUP }, { "switch", cmd_switch, RUN_SETUP | NEED_WORK_TREE }, { "symbolic-ref", cmd_symbolic_ref, RUN_SETUP }, @@ -717,7 +722,7 @@ static struct cmd_struct commands[] = { { "verify-tag", cmd_verify_tag, RUN_SETUP }, { "version", cmd_version }, { "whatchanged", cmd_whatchanged, RUN_SETUP }, - { "worktree", cmd_worktree, RUN_SETUP }, + { "worktree", cmd_worktree, RUN_SETUP | BLOCK_ON_GVFS_REPO }, { "write-tree", cmd_write_tree, RUN_SETUP }, }; diff --git a/gvfs.h b/gvfs.h index 7c9367866f502a..e193502151467a 100644 --- a/gvfs.h +++ b/gvfs.h @@ -12,6 +12,7 @@ * The list of bits in the core_gvfs setting */ #define GVFS_SKIP_SHA_ON_INDEX (1 << 0) +#define GVFS_BLOCK_COMMANDS (1 << 1) #define GVFS_MISSING_OK (1 << 2) #define GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT (1 << 3) #define GVFS_FETCH_SKIP_REACHABILITY_AND_UPLOADPACK (1 << 4) diff --git a/t/t0402-block-command-on-gvfs.sh b/t/t0402-block-command-on-gvfs.sh new file mode 100755 index 00000000000000..3ec7620ce6194d --- /dev/null +++ b/t/t0402-block-command-on-gvfs.sh @@ -0,0 +1,39 @@ +#!/bin/sh + +test_description='block commands in GVFS repo' + +. ./test-lib.sh + +not_with_gvfs () { + command=$1 && + shift && + test_expect_success "test $command $*" " + test_config alias.g4rbled $command && + test_config core.gvfs true && + test_must_fail git $command $* && + test_must_fail git g4rbled $* && + test_unconfig core.gvfs && + test_must_fail git -c core.gvfs=true $command $* && + test_must_fail git -c core.gvfs=true g4rbled $* + " +} + +not_with_gvfs fsck +not_with_gvfs gc +not_with_gvfs gc --auto +not_with_gvfs prune +not_with_gvfs repack +not_with_gvfs submodule status +not_with_gvfs update-index --index-version 2 +not_with_gvfs update-index --skip-worktree +not_with_gvfs update-index --no-skip-worktree +not_with_gvfs update-index --split-index +not_with_gvfs worktree list + +test_expect_success 'test gc --auto succeeds when disabled via config' ' + test_config core.gvfs true && + test_config gc.auto 0 && + git gc --auto +' + +test_done From 46eca79c7e8e82a50b8397269c505ec62ca7dc95 Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Fri, 30 Sep 2022 12:59:40 -0400 Subject: [PATCH 049/196] worktree: allow in Scalar repositories The 'git worktree' command was marked as BLOCK_ON_GVFS_REPO because it does not interact well with the virtual filesystem of VFS for Git. When a Scalar clone uses the GVFS protocol, it enables the GVFS_BLOCK_COMMANDS flag, since commands like 'git gc' do not work well with the GVFS protocol. However, 'git worktree' works just fine with the GVFS protocol since it isn't doing anything special. It copies the sparse-checkout from the current worktree, so it does not have performance issues. This is a highly requested option. The solution is to stop using the BLOCK_ON_GVFS_REPO option and instead add a special-case check in cmd_worktree() specifically for a particular bit of the 'core_gvfs' global variable (loaded by very early config reading) that corresponds to the virtual filesystem. The bit that most closely resembled this behavior was non-obviously named, but does provide a signal that we are in a Scalar clone and not a VFS for Git clone. The error message is copied from git.c, so it will have the same output as before if a user runs this in a VFS for Git clone. Signed-off-by: Derrick Stolee --- builtin/worktree.c | 8 ++++++++ git.c | 2 +- gvfs.h | 11 +++++++++++ 3 files changed, 20 insertions(+), 1 deletion(-) diff --git a/builtin/worktree.c b/builtin/worktree.c index fc31d072a620d7..8d1b3fd854979f 100644 --- a/builtin/worktree.c +++ b/builtin/worktree.c @@ -2,6 +2,7 @@ #include "builtin.h" #include "abspath.h" #include "advice.h" +#include "gvfs.h" #include "checkout.h" #include "config.h" #include "copy.h" @@ -1412,6 +1413,13 @@ int cmd_worktree(int ac, git_config(git_worktree_config, NULL); + /* + * git-worktree is special-cased to work in Scalar repositories + * even when they use the GVFS Protocol. + */ + if (core_gvfs & GVFS_USE_VIRTUAL_FILESYSTEM) + die("'git %s' is not supported on a GVFS repo", "worktree"); + if (!prefix) prefix = ""; diff --git a/git.c b/git.c index fbc1838905ad9e..01261d76d6dc59 100644 --- a/git.c +++ b/git.c @@ -722,7 +722,7 @@ static struct cmd_struct commands[] = { { "verify-tag", cmd_verify_tag, RUN_SETUP }, { "version", cmd_version }, { "whatchanged", cmd_whatchanged, RUN_SETUP }, - { "worktree", cmd_worktree, RUN_SETUP | BLOCK_ON_GVFS_REPO }, + { "worktree", cmd_worktree, RUN_SETUP }, { "write-tree", cmd_write_tree, RUN_SETUP }, }; diff --git a/gvfs.h b/gvfs.h index e193502151467a..a8e58a6ebc88b8 100644 --- a/gvfs.h +++ b/gvfs.h @@ -14,7 +14,18 @@ #define GVFS_SKIP_SHA_ON_INDEX (1 << 0) #define GVFS_BLOCK_COMMANDS (1 << 1) #define GVFS_MISSING_OK (1 << 2) + +/* + * This behavior of not deleting outside of the sparse-checkout + * is specific to the virtual filesystem support. It is only + * enabled by VFS for Git, and so can be used as an indicator + * that we are in a virtualized filesystem environment and not + * in a Scalar environment. This bit has two names to reflect + * that. + */ #define GVFS_NO_DELETE_OUTSIDE_SPARSECHECKOUT (1 << 3) +#define GVFS_USE_VIRTUAL_FILESYSTEM (1 << 3) + #define GVFS_FETCH_SKIP_REACHABILITY_AND_UPLOADPACK (1 << 4) #define GVFS_BLOCK_FILTERS_AND_EOL_CONVERSIONS (1 << 6) From 04634cf2eab54aab2b2df2988a194f61416f50d6 Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Wed, 15 Apr 2020 16:19:31 +0000 Subject: [PATCH 050/196] gvfs: allow overriding core.gvfs We found a user who had set "core.gvfs = false" in their global config. This should not have been necessary, but it also should not have caused a problem. However, it did. The reason is that gvfs_load_config_value() is called from config.c when reading config key/value pairs from all the config files. The local config should override the global config, and this is done by config.c reading the global config first then reading the local config. However, our logic only allowed writing the core_gvfs variable once. Put the guards against multiple assignments of core_gvfs into gvfs_config_is_set() instead, because that will fix the problem _and_ keep multiple calls to gvfs_config_is_set() from slowing down. Signed-off-by: Derrick Stolee --- gvfs.c | 10 ++++------ t/t0021-conversion.sh | 4 ++++ 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/gvfs.c b/gvfs.c index 3cdd8a055d6021..11635237893968 100644 --- a/gvfs.c +++ b/gvfs.c @@ -19,9 +19,6 @@ static int early_core_gvfs_config(const char *var, const char *value, void gvfs_load_config_value(const char *value) { - if (gvfs_config_loaded) - return; - if (value) { struct key_value_info default_kvi = KVI_INIT; core_gvfs = git_config_bool_or_int("core.gvfs", value, &default_kvi, &core_gvfs_is_bool); @@ -34,12 +31,13 @@ void gvfs_load_config_value(const char *value) /* Turn on all bits if a bool was set in the settings */ if (core_gvfs_is_bool && core_gvfs) core_gvfs = -1; - - gvfs_config_loaded = 1; } int gvfs_config_is_set(int mask) { - gvfs_load_config_value(NULL); + if (!gvfs_config_loaded) + gvfs_load_config_value(NULL); + + gvfs_config_loaded = 1; return (core_gvfs & mask) == mask; } diff --git a/t/t0021-conversion.sh b/t/t0021-conversion.sh index 3fd454b1e5c59a..5351fe9f0362bb 100755 --- a/t/t0021-conversion.sh +++ b/t/t0021-conversion.sh @@ -350,6 +350,10 @@ test_expect_success "filter: smudge filters blocked when under GVFS" ' test_config filter.empty-in-repo.smudge "echo smudged && cat" && test_config core.gvfs 64 && + test_must_fail git checkout && + + # ensure the local core.gvfs setting overwrites the global setting + git config --global core.gvfs false && test_must_fail git checkout ' From b8f54a17882ef3eae2ba4910d37c92b4b4ed513a Mon Sep 17 00:00:00 2001 From: Ben Peart Date: Thu, 11 Jan 2018 16:25:08 -0500 Subject: [PATCH 051/196] Add virtual file system settings and hook proc On index load, clear/set the skip worktree bits based on the virtual file system data. Use virtual file system data to update skip-worktree bit in unpack-trees. Use virtual file system data to exclude files and folders not explicitly requested. Update 2022-04-05: disable the "present-despite-SKIP_WORKTREE" file removal behavior when 'core.virtualfilesystem' is enabled. Signed-off-by: Ben Peart --- Documentation/config/core.txt | 8 + Documentation/githooks.txt | 20 ++ Makefile | 1 + config.c | 30 ++- config.h | 2 + dir.c | 36 ++- environment.c | 1 + environment.h | 1 + read-cache.c | 2 + sparse-index.c | 5 +- t/t1090-sparse-checkout-scope.sh | 4 +- t/t1093-virtualfilesystem.sh | 369 +++++++++++++++++++++++++++++++ unpack-trees.c | 14 +- virtualfilesystem.c | 314 ++++++++++++++++++++++++++ virtualfilesystem.h | 25 +++ wt-status.c | 2 + 16 files changed, 826 insertions(+), 8 deletions(-) create mode 100755 t/t1093-virtualfilesystem.sh create mode 100644 virtualfilesystem.c create mode 100644 virtualfilesystem.h diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt index 2fd2a10f51ad2c..50e4146af1d24e 100644 --- a/Documentation/config/core.txt +++ b/Documentation/config/core.txt @@ -111,6 +111,14 @@ Version 2 uses an opaque string so that the monitor can return something that can be used to determine what files have changed without race conditions. +core.virtualFilesystem:: + If set, the value of this variable is used as a command which + will identify all files and directories that are present in + the working directory. Git will only track and update files + listed in the virtual file system. Using the virtual file system + will supersede the sparse-checkout settings which will be ignored. + See the "virtual file system" section of linkgit:githooks[5]. + core.trustctime:: If false, the ctime differences between the index and the working tree are ignored; useful when the inode change time diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt index 0397dec64d7315..86c78b3b4825e2 100644 --- a/Documentation/githooks.txt +++ b/Documentation/githooks.txt @@ -758,6 +758,26 @@ and "0" meaning they were not. Only one parameter should be set to "1" when the hook runs. The hook running passing "1", "1" should not be possible. +virtualFilesystem +~~~~~~~~~~~~~~~~~~ + +"Virtual File System" allows populating the working directory sparsely. +The projection data is typically automatically generated by an external +process. Git will limit what files it checks for changes as well as which +directories are checked for untracked files based on the path names given. +Git will also only update those files listed in the projection. + +The hook is invoked when the configuration option core.virtualFilesystem +is set. It takes one argument, a version (currently 1). + +The hook should output to stdout the list of all files in the working +directory that git should track. The paths are relative to the root +of the working directory and are separated by a single NUL. Full paths +('dir1/a.txt') as well as directories are supported (ie 'dir1/'). + +The exit status determines whether git will use the data from the +hook. On error, git will abort the command with an error message. + SEE ALSO -------- linkgit:git-hook[1] diff --git a/Makefile b/Makefile index d6f0a9495f5972..9b81082b03a6f5 100644 --- a/Makefile +++ b/Makefile @@ -1201,6 +1201,7 @@ LIB_OBJS += utf8.o LIB_OBJS += varint.o LIB_OBJS += version.o LIB_OBJS += versioncmp.o +LIB_OBJS += virtualfilesystem.o LIB_OBJS += walker.o LIB_OBJS += wildmatch.o LIB_OBJS += worktree.o diff --git a/config.c b/config.c index 335459faee6b3c..b1da0e4e0d9aae 100644 --- a/config.c +++ b/config.c @@ -1649,7 +1649,11 @@ int git_default_core_config(const char *var, const char *value, } if (!strcmp(var, "core.sparsecheckout")) { - core_apply_sparse_checkout = git_config_bool(var, value); + /* virtual file system relies on the sparse checkout logic so force it on */ + if (core_virtualfilesystem) + core_apply_sparse_checkout = 1; + else + core_apply_sparse_checkout = git_config_bool(var, value); return 0; } @@ -2740,6 +2744,30 @@ int repo_config_get_max_percent_split_change(struct repository *r) return -1; /* default value */ } +int repo_config_get_virtualfilesystem(struct repository *r) +{ + /* Run only once. */ + static int virtual_filesystem_result = -1; + if (virtual_filesystem_result >= 0) + return virtual_filesystem_result; + + if (repo_config_get_pathname(r, "core.virtualfilesystem", &core_virtualfilesystem)) + core_virtualfilesystem = getenv("GIT_VIRTUALFILESYSTEM_TEST"); + + if (core_virtualfilesystem && !*core_virtualfilesystem) + core_virtualfilesystem = NULL; + + /* virtual file system relies on the sparse checkout logic so force it on */ + if (core_virtualfilesystem) { + core_apply_sparse_checkout = 1; + virtual_filesystem_result = 1; + return 1; + } + + virtual_filesystem_result = 0; + return 0; +} + int repo_config_get_index_threads(struct repository *r, int *dest) { int is_bool, val; diff --git a/config.h b/config.h index e4199bbdc07685..b640ded10f822f 100644 --- a/config.h +++ b/config.h @@ -684,6 +684,8 @@ int repo_config_get_index_threads(struct repository *r, int *dest); int repo_config_get_split_index(struct repository *r); int repo_config_get_max_percent_split_change(struct repository *r); +int repo_config_get_virtualfilesystem(struct repository *r); + /* This dies if the configured or default date is in the future */ int repo_config_get_expiry(struct repository *r, const char *key, char **output); diff --git a/dir.c b/dir.c index 90fabfa4ce00fd..1c3c7ef18ebb58 100644 --- a/dir.c +++ b/dir.c @@ -10,6 +10,7 @@ #include "git-compat-util.h" #include "abspath.h" +#include "virtualfilesystem.h" #include "config.h" #include "convert.h" #include "dir.h" @@ -1475,6 +1476,19 @@ enum pattern_match_result path_matches_pattern_list( int result = NOT_MATCHED; size_t slash_pos; + if (core_virtualfilesystem) { + /* + * The virtual file system data is used to prevent git from traversing + * any part of the tree that is not in the virtual file system. Return + * 1 to exclude the entry if it is not found in the virtual file system, + * else fall through to the regular excludes logic as it may further exclude. + */ + if (*dtype == DT_UNKNOWN) + *dtype = resolve_dtype(DT_UNKNOWN, istate, pathname, pathlen); + if (is_excluded_from_virtualfilesystem(pathname, pathlen, *dtype) > 0) + return 1; + } + if (!pl->use_cone_patterns) { pattern = last_matching_pattern_from_list(pathname, pathlen, basename, dtype, pl, istate); @@ -1819,8 +1833,22 @@ struct path_pattern *last_matching_pattern(struct dir_struct *dir, int is_excluded(struct dir_struct *dir, struct index_state *istate, const char *pathname, int *dtype_p) { - struct path_pattern *pattern = - last_matching_pattern(dir, istate, pathname, dtype_p); + struct path_pattern *pattern; + + if (core_virtualfilesystem) { + /* + * The virtual file system data is used to prevent git from traversing + * any part of the tree that is not in the virtual file system. Return + * 1 to exclude the entry if it is not found in the virtual file system, + * else fall through to the regular excludes logic as it may further exclude. + */ + if (*dtype_p == DT_UNKNOWN) + *dtype_p = resolve_dtype(DT_UNKNOWN, istate, pathname, strlen(pathname)); + if (is_excluded_from_virtualfilesystem(pathname, strlen(pathname), *dtype_p) > 0) + return 1; + } + + pattern = last_matching_pattern(dir, istate, pathname, dtype_p); if (pattern) return pattern->flags & PATTERN_FLAG_NEGATIVE ? 0 : 1; return 0; @@ -2438,6 +2466,8 @@ static enum path_treatment treat_path(struct dir_struct *dir, ignore_case); if (dtype != DT_DIR && has_path_in_index) return path_none; + if (is_excluded_from_virtualfilesystem(path->buf, path->len, dtype) > 0) + return path_excluded; /* * When we are looking at a directory P in the working tree, @@ -2642,6 +2672,8 @@ static void add_path_to_appropriate_result_list(struct dir_struct *dir, /* add the path to the appropriate result list */ switch (state) { case path_excluded: + if (is_excluded_from_virtualfilesystem(path->buf, path->len, DT_DIR) > 0) + break; if (dir->flags & DIR_SHOW_IGNORED) dir_add_name(dir, istate, path->buf, path->len); else if ((dir->flags & DIR_SHOW_IGNORED_TOO) || diff --git a/environment.c b/environment.c index 91962ab8d0ebaf..19700245e10471 100644 --- a/environment.c +++ b/environment.c @@ -72,6 +72,7 @@ int core_apply_sparse_checkout; int core_sparse_checkout_cone; int sparse_expect_files_outside_of_patterns; int core_gvfs; +char *core_virtualfilesystem; int merge_log_config = -1; int precomposed_unicode = -1; /* see probe_utf8_pathname_composition() */ unsigned long pack_size_limit_cfg; diff --git a/environment.h b/environment.h index d3d88598256237..e34e9b61ca8b5a 100644 --- a/environment.h +++ b/environment.h @@ -171,6 +171,7 @@ extern unsigned long pack_size_limit_cfg; extern int max_allowed_tree_depth; extern int core_preload_index; +extern char *core_virtualfilesystem; extern int core_gvfs; extern int precomposed_unicode; extern int protect_hfs; diff --git a/read-cache.c b/read-cache.c index 3624f8c1d56371..6c46ce61d2a50d 100644 --- a/read-cache.c +++ b/read-cache.c @@ -8,6 +8,7 @@ #include "git-compat-util.h" #include "bulk-checkin.h" +#include "virtualfilesystem.h" #include "config.h" #include "date.h" #include "diff.h" @@ -1979,6 +1980,7 @@ static void post_read_index_from(struct index_state *istate) tweak_untracked_cache(istate); tweak_split_index(istate); tweak_fsmonitor(istate); + apply_virtualfilesystem(istate); } static size_t estimate_cache_size_from_compressed(unsigned int entries) diff --git a/sparse-index.c b/sparse-index.c index 62b2393f5fb117..9a80c34b09de91 100644 --- a/sparse-index.c +++ b/sparse-index.c @@ -267,7 +267,7 @@ static int add_path_to_index(const struct object_id *oid, size_t len = base->len; if (S_ISDIR(mode)) { - int dtype; + int dtype = DT_DIR; size_t baselen = base->len; if (!ctx->pl) return READ_TREE_RECURSIVE; @@ -391,7 +391,7 @@ void expand_index(struct index_state *istate, struct pattern_list *pl) struct cache_entry *ce = istate->cache[i]; struct tree *tree; struct pathspec ps; - int dtype; + int dtype = DT_UNKNOWN; if (!S_ISSPARSEDIR(ce->ce_mode)) { set_index_entry(full, full->cache_nr++, ce); @@ -666,6 +666,7 @@ static void clear_skip_worktree_from_present_files_full(struct index_state *ista void clear_skip_worktree_from_present_files(struct index_state *istate) { if (!core_apply_sparse_checkout || + core_virtualfilesystem || sparse_expect_files_outside_of_patterns) return; diff --git a/t/t1090-sparse-checkout-scope.sh b/t/t1090-sparse-checkout-scope.sh index 7af01dc00b75ee..1433e7e1fcbb58 100755 --- a/t/t1090-sparse-checkout-scope.sh +++ b/t/t1090-sparse-checkout-scope.sh @@ -108,9 +108,9 @@ test_expect_success 'in partial clone, sparse checkout only fetches needed blobs ' test_expect_success 'checkout does not delete items outside the sparse checkout file' ' - # The "sparse.expectfilesoutsideofpatterns" config will prevent the + # The "core.virtualfilesystem" config will prevent the # SKIP_WORKTREE flag from being dropped on files present on-disk. - test_config sparse.expectfilesoutsideofpatterns true && + test_config core.virtualfilesystem true && test_config core.gvfs 8 && git checkout -b outside && diff --git a/t/t1093-virtualfilesystem.sh b/t/t1093-virtualfilesystem.sh new file mode 100755 index 00000000000000..bd0c9f72ba3c4a --- /dev/null +++ b/t/t1093-virtualfilesystem.sh @@ -0,0 +1,369 @@ +#!/bin/sh + +test_description='virtual file system tests' + +. ./test-lib.sh + +clean_repo () { + rm .git/index && + git -c core.virtualfilesystem= reset --hard HEAD && + git -c core.virtualfilesystem= clean -fd && + touch untracked.txt && + touch dir1/untracked.txt && + touch dir2/untracked.txt +} + +test_expect_success 'setup' ' + git branch -M main && + mkdir -p .git/hooks/ && + cat > .gitignore <<-\EOF && + .gitignore + expect* + actual* + EOF + mkdir -p dir1 && + touch dir1/file1.txt && + touch dir1/file2.txt && + mkdir -p dir2 && + touch dir2/file1.txt && + touch dir2/file2.txt && + git add . && + git commit -m "initial" && + git config --local core.virtualfilesystem .git/hooks/virtualfilesystem +' + +test_expect_success 'test hook parameters and version' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + if test "$#" -ne 1 + then + echo "$0: Exactly 1 argument expected" >&2 + exit 2 + fi + + if test "$1" != 1 + then + echo "$0: Unsupported hook version." >&2 + exit 1 + fi + EOF + git status && + write_script .git/hooks/virtualfilesystem <<-\EOF && + exit 3 + EOF + test_must_fail git status +' + +test_expect_success 'verify status is clean' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir2/file1.txt\0" + EOF + rm -f .git/index && + git checkout -f && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir2/file1.txt\0" + printf "dir1/file1.txt\0" + printf "dir1/file2.txt\0" + EOF + git status > actual && + cat > expected <<-\EOF && + On branch main + nothing to commit, working tree clean + EOF + test_cmp expected actual +' + +test_expect_success 'verify skip-worktree bit is set for absolute path' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file1.txt\0" + EOF + git ls-files -v > actual && + cat > expected <<-\EOF && + H dir1/file1.txt + S dir1/file2.txt + S dir2/file1.txt + S dir2/file2.txt + EOF + test_cmp expected actual +' + +test_expect_success 'verify skip-worktree bit is cleared for absolute path' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file2.txt\0" + EOF + git ls-files -v > actual && + cat > expected <<-\EOF && + S dir1/file1.txt + H dir1/file2.txt + S dir2/file1.txt + S dir2/file2.txt + EOF + test_cmp expected actual +' + +test_expect_success 'verify folder wild cards' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/\0" + EOF + git ls-files -v > actual && + cat > expected <<-\EOF && + H dir1/file1.txt + H dir1/file2.txt + S dir2/file1.txt + S dir2/file2.txt + EOF + test_cmp expected actual +' + +test_expect_success 'verify folders not included are ignored' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file1.txt\0" + printf "dir1/file2.txt\0" + EOF + mkdir -p dir1/dir2 && + touch dir1/a && + touch dir1/b && + touch dir1/dir2/a && + touch dir1/dir2/b && + git add . && + git ls-files -v > actual && + cat > expected <<-\EOF && + H dir1/file1.txt + H dir1/file2.txt + S dir2/file1.txt + S dir2/file2.txt + EOF + test_cmp expected actual +' + +test_expect_success 'verify including one file doesnt include the rest' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file1.txt\0" + printf "dir1/file2.txt\0" + printf "dir1/dir2/a\0" + EOF + mkdir -p dir1/dir2 && + touch dir1/a && + touch dir1/b && + touch dir1/dir2/a && + touch dir1/dir2/b && + git add . && + git ls-files -v > actual && + cat > expected <<-\EOF && + H dir1/dir2/a + H dir1/file1.txt + H dir1/file2.txt + S dir2/file1.txt + S dir2/file2.txt + EOF + test_cmp expected actual +' + +test_expect_success 'verify files not listed are ignored by git clean -f -x' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "untracked.txt\0" + printf "dir1/\0" + EOF + mkdir -p dir3 && + touch dir3/untracked.txt && + git clean -f -x && + test ! -f untracked.txt && + test -d dir1 && + test -f dir1/file1.txt && + test -f dir1/file2.txt && + test ! -f dir1/untracked.txt && + test -f dir2/file1.txt && + test -f dir2/file2.txt && + test -f dir2/untracked.txt && + test -d dir3 && + test -f dir3/untracked.txt +' + +test_expect_success 'verify files not listed are ignored by git clean -f -d -x' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "untracked.txt\0" + printf "dir1/\0" + printf "dir3/\0" + EOF + mkdir -p dir3 && + touch dir3/untracked.txt && + git clean -f -d -x && + test ! -f untracked.txt && + test -d dir1 && + test -f dir1/file1.txt && + test -f dir1/file2.txt && + test ! -f dir1/untracked.txt && + test -f dir2/file1.txt && + test -f dir2/file2.txt && + test -f dir2/untracked.txt && + test ! -d dir3 && + test ! -f dir3/untracked.txt +' + +test_expect_success 'verify folder entries include all files' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/\0" + EOF + mkdir -p dir1/dir2 && + touch dir1/a && + touch dir1/b && + touch dir1/dir2/a && + touch dir1/dir2/b && + git status -su > actual && + cat > expected <<-\EOF && + ?? dir1/a + ?? dir1/b + ?? dir1/untracked.txt + EOF + test_cmp expected actual +' + +test_expect_success 'verify case insensitivity of virtual file system entries' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/a\0" + printf "Dir1/Dir2/a\0" + printf "DIR2/\0" + EOF + mkdir -p dir1/dir2 && + touch dir1/a && + touch dir1/b && + touch dir1/dir2/a && + touch dir1/dir2/b && + git -c core.ignorecase=false status -su > actual && + cat > expected <<-\EOF && + ?? dir1/a + EOF + test_cmp expected actual && + git -c core.ignorecase=true status -su > actual && + cat > expected <<-\EOF && + ?? dir1/a + ?? dir1/dir2/a + ?? dir2/untracked.txt + EOF + test_cmp expected actual +' + +test_expect_success 'on file created' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file3.txt\0" + EOF + touch dir1/file3.txt && + git add . && + git ls-files -v > actual && + cat > expected <<-\EOF && + S dir1/file1.txt + S dir1/file2.txt + H dir1/file3.txt + S dir2/file1.txt + S dir2/file2.txt + EOF + test_cmp expected actual +' + +test_expect_success 'on file renamed' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file1.txt\0" + printf "dir1/file3.txt\0" + EOF + mv dir1/file1.txt dir1/file3.txt && + git status -su > actual && + cat > expected <<-\EOF && + D dir1/file1.txt + ?? dir1/file3.txt + EOF + test_cmp expected actual +' + +test_expect_success 'on file deleted' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file1.txt\0" + EOF + rm dir1/file1.txt && + git status -su > actual && + cat > expected <<-\EOF && + D dir1/file1.txt + EOF + test_cmp expected actual +' + +test_expect_success 'on file overwritten' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/file1.txt\0" + EOF + echo "overwritten" > dir1/file1.txt && + git status -su > actual && + cat > expected <<-\EOF && + M dir1/file1.txt + EOF + test_cmp expected actual +' + +test_expect_success 'on folder created' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/dir1/\0" + EOF + mkdir -p dir1/dir1 && + git status -su > actual && + cat > expected <<-\EOF && + EOF + test_cmp expected actual && + git clean -fd && + test ! -d "/dir1/dir1" +' + +test_expect_success 'on folder renamed' ' + clean_repo && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir3/\0" + printf "dir1/file1.txt\0" + printf "dir1/file2.txt\0" + printf "dir3/file1.txt\0" + printf "dir3/file2.txt\0" + EOF + mv dir1 dir3 && + git status -su > actual && + cat > expected <<-\EOF && + D dir1/file1.txt + D dir1/file2.txt + ?? dir3/file1.txt + ?? dir3/file2.txt + ?? dir3/untracked.txt + EOF + test_cmp expected actual +' + +test_expect_success 'folder with same prefix as file' ' + clean_repo && + touch dir1.sln && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/\0" + printf "dir1.sln\0" + EOF + git add dir1.sln && + git ls-files -v > actual && + cat > expected <<-\EOF && + H dir1.sln + H dir1/file1.txt + H dir1/file2.txt + S dir2/file1.txt + S dir2/file2.txt + EOF + test_cmp expected actual +' + +test_done diff --git a/unpack-trees.c b/unpack-trees.c index 611b114a8e3def..47affd038d5dff 100644 --- a/unpack-trees.c +++ b/unpack-trees.c @@ -3,6 +3,7 @@ #include "git-compat-util.h" #include "advice.h" #include "gvfs.h" +#include "virtualfilesystem.h" #include "strvec.h" #include "repository.h" #include "parse.h" @@ -1703,6 +1704,14 @@ static int clear_ce_flags_1(struct index_state *istate, continue; } + /* if it's not in the virtual file system, exit early */ + if (core_virtualfilesystem) { + if (is_included_in_virtualfilesystem(ce->name, ce->ce_namelen) > 0) + ce->ce_flags &= ~clear_mask; + cache++; + continue; + } + if (prefix->len && strncmp(ce->name, prefix->buf, prefix->len)) break; @@ -1929,7 +1938,10 @@ int unpack_trees(unsigned len, struct tree_desc *t, struct unpack_trees_options if (!o->skip_sparse_checkout) { memset(&pl, 0, sizeof(pl)); free_pattern_list = 1; - populate_from_existing_patterns(o, &pl); + if (core_virtualfilesystem) + o->internal.pl = &pl; + else + populate_from_existing_patterns(o, &pl); } index_state_init(&o->internal.result, o->src_index->repo); diff --git a/virtualfilesystem.c b/virtualfilesystem.c new file mode 100644 index 00000000000000..6eedbcb3d4edc7 --- /dev/null +++ b/virtualfilesystem.c @@ -0,0 +1,314 @@ +#define USE_THE_REPOSITORY_VARIABLE + +#include "git-compat-util.h" +#include "environment.h" +#include "gettext.h" +#include "config.h" +#include "dir.h" +#include "hashmap.h" +#include "run-command.h" +#include "name-hash.h" +#include "read-cache-ll.h" +#include "virtualfilesystem.h" + +#define HOOK_INTERFACE_VERSION (1) + +static struct strbuf virtual_filesystem_data = STRBUF_INIT; +static struct hashmap virtual_filesystem_hashmap; +static struct hashmap parent_directory_hashmap; + +struct virtualfilesystem { + struct hashmap_entry ent; /* must be the first member! */ + const char *pattern; + int patternlen; +}; + +static unsigned int(*vfshash)(const void *buf, size_t len); +static int(*vfscmp)(const char *a, const char *b, size_t len); + +static int vfs_hashmap_cmp(const void *cmp_data UNUSED, + const struct hashmap_entry *he1, + const struct hashmap_entry *he2, + const void *key UNUSED) +{ + const struct virtualfilesystem *vfs1 = + container_of(he1, const struct virtualfilesystem, ent); + const struct virtualfilesystem *vfs2 = + container_of(he2, const struct virtualfilesystem, ent); + + return vfscmp(vfs1->pattern, vfs2->pattern, vfs1->patternlen); +} + +static void get_virtual_filesystem_data(struct repository *r, struct strbuf *vfs_data) +{ + struct child_process cp = CHILD_PROCESS_INIT; + int err; + + strbuf_init(vfs_data, 0); + + strvec_push(&cp.args, core_virtualfilesystem); + strvec_pushf(&cp.args, "%d", HOOK_INTERFACE_VERSION); + cp.use_shell = 1; + cp.dir = repo_get_work_tree(r); + + err = capture_command(&cp, vfs_data, 1024); + if (err) + die("unable to load virtual file system"); +} + +static int check_includes_hashmap(struct hashmap *map, const char *pattern, int patternlen) +{ + struct strbuf sb = STRBUF_INIT; + struct virtualfilesystem vfs; + char *slash; + + /* Check straight mapping */ + strbuf_reset(&sb); + strbuf_add(&sb, pattern, patternlen); + vfs.pattern = sb.buf; + vfs.patternlen = sb.len; + hashmap_entry_init(&vfs.ent, vfshash(vfs.pattern, vfs.patternlen)); + if (hashmap_get_entry(map, &vfs, ent, NULL)) { + strbuf_release(&sb); + return 1; + } + + /* + * Check to see if it matches a directory or any path + * underneath it. In other words, 'a/b/foo.txt' will match + * '/', 'a/', and 'a/b/'. + */ + slash = strchr(sb.buf, '/'); + while (slash) { + vfs.pattern = sb.buf; + vfs.patternlen = slash - sb.buf + 1; + hashmap_entry_init(&vfs.ent, vfshash(vfs.pattern, vfs.patternlen)); + if (hashmap_get_entry(map, &vfs, ent, NULL)) { + strbuf_release(&sb); + return 1; + } + slash = strchr(slash + 1, '/'); + } + + strbuf_release(&sb); + return 0; +} + +static void includes_hashmap_add(struct hashmap *map, const char *pattern, const int patternlen) +{ + struct virtualfilesystem *vfs; + + vfs = xmalloc(sizeof(struct virtualfilesystem)); + vfs->pattern = pattern; + vfs->patternlen = patternlen; + hashmap_entry_init(&vfs->ent, vfshash(vfs->pattern, vfs->patternlen)); + hashmap_add(map, &vfs->ent); +} + +static void initialize_includes_hashmap(struct hashmap *map, struct strbuf *vfs_data) +{ + char *buf, *entry; + size_t len; + int i; + + /* + * Build a hashmap of the virtual file system data we can use to look + * for cache entry matches quickly + */ + vfshash = ignore_case ? memihash : memhash; + vfscmp = ignore_case ? strncasecmp : strncmp; + hashmap_init(map, vfs_hashmap_cmp, NULL, 0); + + entry = buf = vfs_data->buf; + len = vfs_data->len; + for (i = 0; i < len; i++) { + if (buf[i] == '\0') { + includes_hashmap_add(map, entry, buf + i - entry); + entry = buf + i + 1; + } + } +} + +/* + * Return 1 if the requested item is found in the virtual file system, + * 0 for not found and -1 for undecided. + */ +int is_included_in_virtualfilesystem(const char *pathname, int pathlen) +{ + if (!core_virtualfilesystem) + return -1; + + if (!virtual_filesystem_hashmap.tablesize && virtual_filesystem_data.len) + initialize_includes_hashmap(&virtual_filesystem_hashmap, &virtual_filesystem_data); + if (!virtual_filesystem_hashmap.tablesize) + return -1; + + return check_includes_hashmap(&virtual_filesystem_hashmap, pathname, pathlen); +} + +static void parent_directory_hashmap_add(struct hashmap *map, const char *pattern, const int patternlen) +{ + char *slash; + struct virtualfilesystem *vfs; + + /* + * Add any directories leading up to the file as the excludes logic + * needs to match directories leading up to the files as well. Detect + * and prevent unnecessary duplicate entries which will be common. + */ + if (patternlen > 1) { + slash = strchr(pattern + 1, '/'); + while (slash) { + vfs = xmalloc(sizeof(struct virtualfilesystem)); + vfs->pattern = pattern; + vfs->patternlen = slash - pattern + 1; + hashmap_entry_init(&vfs->ent, vfshash(vfs->pattern, vfs->patternlen)); + if (hashmap_get_entry(map, vfs, ent, NULL)) + free(vfs); + else + hashmap_add(map, &vfs->ent); + slash = strchr(slash + 1, '/'); + } + } +} + +static void initialize_parent_directory_hashmap(struct hashmap *map, struct strbuf *vfs_data) +{ + char *buf, *entry; + size_t len; + int i; + + /* + * Build a hashmap of the parent directories contained in the virtual + * file system data we can use to look for matches quickly + */ + vfshash = ignore_case ? memihash : memhash; + vfscmp = ignore_case ? strncasecmp : strncmp; + hashmap_init(map, vfs_hashmap_cmp, NULL, 0); + + entry = buf = vfs_data->buf; + len = vfs_data->len; + for (i = 0; i < len; i++) { + if (buf[i] == '\0') { + parent_directory_hashmap_add(map, entry, buf + i - entry); + entry = buf + i + 1; + } + } +} + +static int check_directory_hashmap(struct hashmap *map, const char *pathname, int pathlen) +{ + struct strbuf sb = STRBUF_INIT; + struct virtualfilesystem vfs; + + /* Check for directory */ + strbuf_reset(&sb); + strbuf_add(&sb, pathname, pathlen); + strbuf_addch(&sb, '/'); + vfs.pattern = sb.buf; + vfs.patternlen = sb.len; + hashmap_entry_init(&vfs.ent, vfshash(vfs.pattern, vfs.patternlen)); + if (hashmap_get_entry(map, &vfs, ent, NULL)) { + strbuf_release(&sb); + return 0; + } + + strbuf_release(&sb); + return 1; +} + +/* + * Return 1 for exclude, 0 for include and -1 for undecided. + */ +int is_excluded_from_virtualfilesystem(const char *pathname, int pathlen, int dtype) +{ + if (!core_virtualfilesystem) + return -1; + + if (dtype != DT_REG && dtype != DT_DIR && dtype != DT_LNK) + die(_("is_excluded_from_virtualfilesystem passed unhandled dtype")); + + if (dtype == DT_REG || dtype == DT_LNK) { + int ret = is_included_in_virtualfilesystem(pathname, pathlen); + if (ret > 0) + return 0; + if (ret == 0) + return 1; + return ret; + } + + if (dtype == DT_DIR) { + if (!parent_directory_hashmap.tablesize && virtual_filesystem_data.len) + initialize_parent_directory_hashmap(&parent_directory_hashmap, &virtual_filesystem_data); + if (!parent_directory_hashmap.tablesize) + return -1; + + return check_directory_hashmap(&parent_directory_hashmap, pathname, pathlen); + } + + return -1; +} + +/* + * Update the CE_SKIP_WORKTREE bits based on the virtual file system. + */ +void apply_virtualfilesystem(struct index_state *istate) +{ + char *buf, *entry; + int i; + + if (!repo_config_get_virtualfilesystem(istate->repo)) + return; + + if (!virtual_filesystem_data.len) + get_virtual_filesystem_data(istate->repo, &virtual_filesystem_data); + + /* set CE_SKIP_WORKTREE bit on all entries */ + for (i = 0; i < istate->cache_nr; i++) + istate->cache[i]->ce_flags |= CE_SKIP_WORKTREE; + + /* clear CE_SKIP_WORKTREE bit for everything in the virtual file system */ + entry = buf = virtual_filesystem_data.buf; + for (i = 0; i < virtual_filesystem_data.len; i++) { + if (buf[i] == '\0') { + int pos, len; + + len = buf + i - entry; + + /* look for a directory wild card (ie "dir1/") */ + if (buf[i - 1] == '/') { + if (ignore_case) + adjust_dirname_case(istate, entry); + pos = index_name_pos(istate, entry, len); + if (pos < 0) { + pos = -pos - 1; + while (pos < istate->cache_nr && !fspathncmp(istate->cache[pos]->name, entry, len)) { + istate->cache[pos]->ce_flags &= ~CE_SKIP_WORKTREE; + pos++; + } + } + } else { + if (ignore_case) { + struct cache_entry *ce = index_file_exists(istate, entry, len, ignore_case); + if (ce) + ce->ce_flags &= ~CE_SKIP_WORKTREE; + } else { + int pos = index_name_pos(istate, entry, len); + if (pos >= 0) + istate->cache[pos]->ce_flags &= ~CE_SKIP_WORKTREE; + } + } + + entry += len + 1; + } + } +} + +/* + * Free the virtual file system data structures. + */ +void free_virtualfilesystem(void) { + hashmap_clear_and_free(&virtual_filesystem_hashmap, struct virtualfilesystem, ent); + hashmap_clear_and_free(&parent_directory_hashmap, struct virtualfilesystem, ent); + strbuf_release(&virtual_filesystem_data); +} diff --git a/virtualfilesystem.h b/virtualfilesystem.h new file mode 100644 index 00000000000000..5e8c5b096df09a --- /dev/null +++ b/virtualfilesystem.h @@ -0,0 +1,25 @@ +#ifndef VIRTUALFILESYSTEM_H +#define VIRTUALFILESYSTEM_H + +/* + * Update the CE_SKIP_WORKTREE bits based on the virtual file system. + */ +void apply_virtualfilesystem(struct index_state *istate); + +/* + * Return 1 if the requested item is found in the virtual file system, + * 0 for not found and -1 for undecided. + */ +int is_included_in_virtualfilesystem(const char *pathname, int pathlen); + +/* + * Return 1 for exclude, 0 for include and -1 for undecided. + */ +int is_excluded_from_virtualfilesystem(const char *pathname, int pathlen, int dtype); + +/* + * Free the virtual file system data structures. + */ +void free_virtualfilesystem(void); + +#endif diff --git a/wt-status.c b/wt-status.c index 43280115b8351e..40961c070c4780 100644 --- a/wt-status.c +++ b/wt-status.c @@ -1607,6 +1607,8 @@ static void show_sparse_checkout_in_use(struct wt_status *s, { if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_DISABLED) return; + if (core_virtualfilesystem) + return; if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_SPARSE_INDEX) status_printf_ln(s, color, _("You are in a sparse checkout.")); From 8289b81d14e550c17dc70603f5054e94dcba5e14 Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Fri, 27 Jul 2018 12:00:44 -0600 Subject: [PATCH 052/196] BRANCHES.md: Add explanation of branches and using forks --- BRANCHES.md | 59 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 59 insertions(+) create mode 100644 BRANCHES.md diff --git a/BRANCHES.md b/BRANCHES.md new file mode 100644 index 00000000000000..364158375e7d55 --- /dev/null +++ b/BRANCHES.md @@ -0,0 +1,59 @@ +Branches used in this repo +========================== + +The document explains the branching structure that we are using in the VFSForGit repository as well as the forking strategy that we have adopted for contributing. + +Repo Branches +------------- + +1. `vfs-#` + + These branches are used to track the specific version that match Git for Windows with the VFSForGit specific patches on top. When a new version of Git for Windows is released, the VFSForGit patches will be rebased on that windows version and a new gvfs-# branch created to create pull requests against. + + #### Examples + + ``` + vfs-2.27.0 + vfs-2.30.0 + ``` + + The versions of git for VFSForGit are based on the Git for Windows versions. v2.20.0.vfs.1 will correspond with the v2.20.0.windows.1 with the VFSForGit specific patches applied to the windows version. + +2. `vfs-#-exp` + + These branches are for releasing experimental features to early adopters. They + should contain everything within the corresponding `vfs-#` branch; if the base + branch updates, then merge into the `vfs-#-exp` branch as well. + +Tags +---- + +We are using annotated tags to build the version number for git. The build will look back through the commit history to find the first tag matching `v[0-9]*vfs*` and build the git version number using that tag. + +Full releases are of the form `v2.XX.Y.vfs.Z.W` where `v2.XX.Y` comes from the +upstream version and `Z.W` are custom updates within our fork. Specifically, +the `.Z` value represents the "compatibility level" with VFS for Git. Only +increase this version when making a breaking change with a released version +of VFS for Git. The `.W` version is used for minor updates between major +versions. + +Experimental releases are of the form `v2.XX.Y.vfs.Z.W.exp`. The `.exp` +suffix indicates that experimental features are available. The rest of the +version string comes from the full release tag. These versions will only +be made available as pre-releases on the releases page, never a full release. + +Forking +------- + +A personal fork of this repository and a branch in that repository should be used for development. + +These branches should be based on the latest vfs-# branch. If there are work in progress pull requests that you have based on a previous version branch when a new version branch is created, you will need to move your patches to the new branch to get them in that latest version. + +#### Example + +``` +git clone +git remote add ms https://github.com/Microsoft/git.git +git checkout -b my-changes ms/vfs-2.20.0 --no-track +git push -fu origin HEAD +``` From b09c255d3050d620aa0eeb40aa249f9483c6481d Mon Sep 17 00:00:00 2001 From: Ben Peart Date: Wed, 1 Aug 2018 13:26:22 -0400 Subject: [PATCH 053/196] virtualfilesystem: don't run the virtual file system hook if the index has been redirected Fixes #13 Some git commands spawn helpers and redirect the index to a different location. These include "difftool -d" and the sequencer (i.e. `git rebase -i`, `git cherry-pick` and `git revert`) and others. In those instances we don't want to update their temporary index with our virtualization data. Helped-by: Johannes Schindelin Signed-off-by: Ben Peart --- config.c | 22 ++++++++++++++++++---- 1 file changed, 18 insertions(+), 4 deletions(-) diff --git a/config.c b/config.c index b1da0e4e0d9aae..7c927b8c91dc7f 100644 --- a/config.c +++ b/config.c @@ -2757,11 +2757,25 @@ int repo_config_get_virtualfilesystem(struct repository *r) if (core_virtualfilesystem && !*core_virtualfilesystem) core_virtualfilesystem = NULL; - /* virtual file system relies on the sparse checkout logic so force it on */ if (core_virtualfilesystem) { - core_apply_sparse_checkout = 1; - virtual_filesystem_result = 1; - return 1; + /* + * Some git commands spawn helpers and redirect the index to a different + * location. These include "difftool -d" and the sequencer + * (i.e. `git rebase -i`, `git cherry-pick` and `git revert`) and others. + * In those instances we don't want to update their temporary index with + * our virtualization data. + */ + char *default_index_file = xstrfmt("%s/%s", the_repository->gitdir, "index"); + int should_run_hook = !strcmp(default_index_file, the_repository->index_file); + + free(default_index_file); + if (should_run_hook) { + /* virtual file system relies on the sparse checkout logic so force it on */ + core_apply_sparse_checkout = 1; + virtual_filesystem_result = 1; + return 1; + } + core_virtualfilesystem = NULL; } virtual_filesystem_result = 0; From dcb6eeed77edada65f1f6e46b8c29a77445c535b Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Tue, 9 Oct 2018 10:19:14 -0600 Subject: [PATCH 054/196] virtualfilesystem: check if directory is included Add check to see if a directory is included in the virtualfilesystem before checking the directory hashmap. This allows a directory entry like foo/ to find all untracked files in subdirectories. --- t/t1093-virtualfilesystem.sh | 2 ++ virtualfilesystem.c | 4 ++++ 2 files changed, 6 insertions(+) diff --git a/t/t1093-virtualfilesystem.sh b/t/t1093-virtualfilesystem.sh index bd0c9f72ba3c4a..8ba9a2a75e093a 100755 --- a/t/t1093-virtualfilesystem.sh +++ b/t/t1093-virtualfilesystem.sh @@ -222,6 +222,8 @@ test_expect_success 'verify folder entries include all files' ' cat > expected <<-\EOF && ?? dir1/a ?? dir1/b + ?? dir1/dir2/a + ?? dir1/dir2/b ?? dir1/untracked.txt EOF test_cmp expected actual diff --git a/virtualfilesystem.c b/virtualfilesystem.c index 6eedbcb3d4edc7..ae191ad05adf4c 100644 --- a/virtualfilesystem.c +++ b/virtualfilesystem.c @@ -238,6 +238,10 @@ int is_excluded_from_virtualfilesystem(const char *pathname, int pathlen, int dt } if (dtype == DT_DIR) { + int ret = is_included_in_virtualfilesystem(pathname, pathlen); + if (ret > 0) + return 0; + if (!parent_directory_hashmap.tablesize && virtual_filesystem_data.len) initialize_parent_directory_hashmap(&parent_directory_hashmap, &virtual_filesystem_data); if (!parent_directory_hashmap.tablesize) From f2621e5cccebc74a4ae4f2658c0d3a63cea6fc29 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 28 May 2019 21:48:08 +0200 Subject: [PATCH 055/196] backwards-compatibility: support the post-indexchanged hook When our patches to support that hook were upstreamed, the hook's name was eliciting some reviewer suggestions, and it was renamed to `post-index-change`. These patches (with the new name) made it into v2.22.0. However, VFSforGit users may very well have checkouts with that hook installed under the original name. To support this, let's just introduce a hack where we look a bit more closely when we just failed to find the `post-index-change` hook, and allow any `post-indexchanged` hook to run instead (if it exists). --- hook.c | 14 +++++++++++++- t/t7113-post-index-change-hook.sh | 30 ++++++++++++++++++++++++++++++ 2 files changed, 43 insertions(+), 1 deletion(-) diff --git a/hook.c b/hook.c index 180d4185284a50..38207def6037b9 100644 --- a/hook.c +++ b/hook.c @@ -196,7 +196,7 @@ int run_hooks_opt(struct repository *r, const char *hook_name, .hook_name = hook_name, .options = options, }; - const char *const hook_path = find_hook(r, hook_name); + const char *hook_path = find_hook(r, hook_name); int ret = 0; const struct run_process_parallel_opts opts = { .tr2_category = "hook", @@ -212,6 +212,18 @@ int run_hooks_opt(struct repository *r, const char *hook_name, .data = &cb_data, }; + /* + * Backwards compatibility hack in VFS for Git: when originally + * introduced (and used!), it was called `post-indexchanged`, but this + * name was changed during the review on the Git mailing list. + * + * Therefore, when the `post-index-change` hook is not found, let's + * look for a hook with the old name (which would be found in case of + * already-existing checkouts). + */ + if (!hook_path && !strcmp(hook_name, "post-index-change")) + hook_path = find_hook(r, "post-indexchanged"); + if (!options) BUG("a struct run_hooks_opt must be provided to run_hooks"); diff --git a/t/t7113-post-index-change-hook.sh b/t/t7113-post-index-change-hook.sh index 58e55a7c779161..455c61a92542ae 100755 --- a/t/t7113-post-index-change-hook.sh +++ b/t/t7113-post-index-change-hook.sh @@ -16,6 +16,36 @@ test_expect_success 'setup' ' git commit -m "initial" ' +test_expect_success 'post-indexchanged' ' + mkdir -p .git/hooks && + test_when_finished "rm -f .git/hooks/post-indexchanged marker" && + write_script .git/hooks/post-indexchanged <<-\EOF && + : >marker + EOF + + : make sure -changed is called if -change does not exist && + test_when_finished "echo testing >dir1/file2.txt && git status" && + echo changed >dir1/file2.txt && + : force index to be dirty && + test-tool chmtime -60 .git/index && + git status && + test_path_is_file marker && + + test_when_finished "rm -f .git/hooks/post-index-change marker2" && + write_script .git/hooks/post-index-change <<-\EOF && + : >marker2 + EOF + + : make sure -changed is not called if -change exists && + rm -f marker marker2 && + echo testing >dir1/file2.txt && + : force index to be dirty && + test-tool chmtime -60 .git/index && + git status && + test_path_is_missing marker && + test_path_is_file marker2 +' + test_expect_success 'test status, add, commit, others trigger hook without flags set' ' test_hook post-index-change <<-\EOF && if test "$1" -eq 1; then From 5e58e1f1199d64c6b7087a3061d87873ff6a910e Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 18 Jun 2021 14:45:20 +0200 Subject: [PATCH 056/196] gvfs: verify that the built-in FSMonitor is disabled When using a virtual file system layer, the FSMonitor does not make sense. Signed-off-by: Johannes Schindelin --- t/t1093-virtualfilesystem.sh | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/t/t1093-virtualfilesystem.sh b/t/t1093-virtualfilesystem.sh index 8ba9a2a75e093a..cad13d680cb199 100755 --- a/t/t1093-virtualfilesystem.sh +++ b/t/t1093-virtualfilesystem.sh @@ -368,4 +368,15 @@ test_expect_success 'folder with same prefix as file' ' test_cmp expected actual ' +test_expect_success MINGW,FSMONITOR_DAEMON 'virtualfilesystem hook disables built-in FSMonitor' ' + clean_repo && + test_config core.usebuiltinfsmonitor true && + write_script .git/hooks/virtualfilesystem <<-\EOF && + printf "dir1/\0" + EOF + git config core.virtualfilesystem .git/hooks/virtualfilesystem && + git status && + test_must_fail git fsmonitor--daemon status +' + test_done From ce2abce41c0fde6f69b80ef06b711bef986cc16c Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 21 Jun 2024 17:18:59 -0400 Subject: [PATCH 057/196] wt-status: add trace2 data for sparse-checkout percentage When sparse-checkout is enabled, add the sparse-checkout percentage to the Trace2 data stream. This number was already computed and printed on the console in the "You are in a sparse checkout..." message. It would be helpful to log it too for performance monitoring. Signed-off-by: Jeff Hostetler --- wt-status.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+) diff --git a/wt-status.c b/wt-status.c index 40961c070c4780..070936638f41d1 100644 --- a/wt-status.c +++ b/wt-status.c @@ -2565,6 +2565,36 @@ void wt_status_print(struct wt_status *s) s->untracked.nr); trace2_data_intmax("status", s->repo, "count/ignored", s->ignored.nr); + switch (s->state.sparse_checkout_percentage) { + case SPARSE_CHECKOUT_DISABLED: + break; + case SPARSE_CHECKOUT_SPARSE_INDEX: + /* + * Log just the observed size of the sparse-index. + * + * When sparse-index is enabled we can have + * sparse-directory entries in addition to individual + * sparse-file entries, so we don't know the complete + * size of the index. And we do not want to force + * expand it just to emit some telemetry data. So we + * cannot report a percentage for the space savings. + * + * It is possible that if the telemetry data is + * aggregated, someone will have a good estimate for + * the size of a fully populated index and can compute + * a percentage after the fact. + */ + trace2_data_intmax("status", s->repo, + "sparse-index/size", + s->repo->index->cache_nr); + break; + default: + trace2_data_intmax("status", s->repo, + "sparse-checkout/percentage", + s->state.sparse_checkout_percentage); + break; + } + trace2_region_enter("status", "print", s->repo); switch (s->status_format) { From 7821224ee6a8b317a4111107e739bcdb46120cc9 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 24 Jun 2024 11:24:20 -0400 Subject: [PATCH 058/196] wt-status: add VFS hydration percentage to normal `git status` output Add VFS checkout hydration percentage information to the default `git status` output. When VFS is enable, users will now see a "You are in a partially-hydrated checkout with of tracked files present." message. Upstream `git status` normally prints a "You are in a sparse checkout with of tracked files present." This message was hidden in `microsoft/git` when `core_virtualfilesystem` is set (because GVFS users are always (and secretly) in a sparse checkout) and it was thought that it would annoy users. However, we now believe that it may be helpful for users to always see the percentage and know when they are over-hyrdated, since over-hyrdation can occur by accident and may greatly impact their Git performance. Knowing this value may help with GVFS support. Helped-by: Johannes Schindelin Signed-off-by: Jeff Hostetler --- t/t1093-virtualfilesystem.sh | 2 ++ wt-status.c | 13 +++++++++---- 2 files changed, 11 insertions(+), 4 deletions(-) diff --git a/t/t1093-virtualfilesystem.sh b/t/t1093-virtualfilesystem.sh index cad13d680cb199..7786735dffec06 100755 --- a/t/t1093-virtualfilesystem.sh +++ b/t/t1093-virtualfilesystem.sh @@ -69,6 +69,8 @@ test_expect_success 'verify status is clean' ' git status > actual && cat > expected <<-\EOF && On branch main + You are in a partially-hydrated checkout with 75% of tracked files present. + nothing to commit, working tree clean EOF test_cmp expected actual diff --git a/wt-status.c b/wt-status.c index 070936638f41d1..061cf9024fdba8 100644 --- a/wt-status.c +++ b/wt-status.c @@ -1607,10 +1607,15 @@ static void show_sparse_checkout_in_use(struct wt_status *s, { if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_DISABLED) return; - if (core_virtualfilesystem) - return; - - if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_SPARSE_INDEX) + if (core_virtualfilesystem) { + if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_SPARSE_INDEX) + status_printf_ln(s, color, + _("You are in a partially-hydrated checkout with a sparse index.")); + else + status_printf_ln(s, color, + _("You are in a partially-hydrated checkout with %d%% of tracked files present."), + s->state.sparse_checkout_percentage); + } else if (s->state.sparse_checkout_percentage == SPARSE_CHECKOUT_SPARSE_INDEX) status_printf_ln(s, color, _("You are in a sparse checkout.")); else status_printf_ln(s, color, From 326915f7a9563bd28372cd90d9794b3b7589aada Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 22 Aug 2017 11:54:23 -0400 Subject: [PATCH 059/196] status: add status serialization mechanism Teach STATUS to optionally serialize the results of a status computation to a file. Teach STATUS to optionally read an existing serialization file and simply print the results, rather than actually scanning. This is intended for immediate status results on extremely large repos and assumes the use of a service/daemon to maintain a fresh current status snapshot. 2021-10-30: packet_read() changed its prototype in ec9a37d (pkt-line.[ch]: remove unused packet_read_line_buf(), 2021-10-14). 2021-10-30: sscanf() now does an extra check that "%d" goes into an "int" and complains about "uint32_t". Replacing with "%u" fixes the compile-time error. 2021-10-30: string_list_init() was removed by abf897b (string-list.[ch]: remove string_list_init() compatibility function, 2021-09-28), so we need to initialize manually. Signed-off-by: Jeff Hostetler Signed-off-by: Derrick Stolee --- Documentation/config/status.txt | 6 + Documentation/git-status.txt | 33 + .../technical/status-serialization-format.txt | 107 +++ Makefile | 2 + builtin/commit.c | 123 +++- contrib/completion/git-completion.bash | 2 +- pkt-line.c | 2 +- pkt-line.h | 1 + t/t7522-serialized-status.sh | 141 ++++ t/t7523-status-complete-untracked.sh | 39 ++ wt-status-deserialize.c | 608 ++++++++++++++++++ wt-status-serialize.c | 219 +++++++ wt-status.c | 6 + wt-status.h | 53 +- 14 files changed, 1338 insertions(+), 4 deletions(-) create mode 100644 Documentation/technical/status-serialization-format.txt create mode 100755 t/t7522-serialized-status.sh create mode 100755 t/t7523-status-complete-untracked.sh create mode 100644 wt-status-deserialize.c create mode 100644 wt-status-serialize.c diff --git a/Documentation/config/status.txt b/Documentation/config/status.txt index 8caf90f51c19a3..7302b066644e73 100644 --- a/Documentation/config/status.txt +++ b/Documentation/config/status.txt @@ -77,3 +77,9 @@ status.submoduleSummary:: the --ignore-submodules=dirty command-line option or the 'git submodule summary' command, which shows a similar output but does not honor these settings. + +status.deserializePath:: + EXPERIMENTAL, Pathname to a file containing cached status results + generated by `--serialize`. This will be overridden by + `--deserialize=` on the command line. If the cache file is + invalid or stale, git will fall-back and compute status normally. diff --git a/Documentation/git-status.txt b/Documentation/git-status.txt index 9a376886a5867a..fedf86d32718eb 100644 --- a/Documentation/git-status.txt +++ b/Documentation/git-status.txt @@ -151,6 +151,19 @@ ignored, then the directory is not shown, but all contents are shown. threshold. See also linkgit:git-diff[1] `--find-renames`. +--serialize[=]:: + (EXPERIMENTAL) Serialize raw status results to stdout in a + format suitable for use by `--deserialize`. Valid values for + `` are "1" and "v1". + +--deserialize[=]:: + (EXPERIMENTAL) Deserialize raw status results from a file or + stdin rather than scanning the worktree. If `` is omitted + and `status.deserializePath` is unset, input is read from stdin. +--no-deserialize:: + (EXPERIMENTAL) Disable implicit deserialization of status results + from the value of `status.deserializePath`. + ...:: See the 'pathspec' entry in linkgit:gitglossary[7]. @@ -424,6 +437,26 @@ quoted as explained for the configuration variable `core.quotePath` (see linkgit:git-config[1]). +SERIALIZATION and DESERIALIZATION (EXPERIMENTAL) +------------------------------------------------ + +The `--serialize` option allows git to cache the result of a +possibly time-consuming status scan to a binary file. A local +service/daemon watching file system events could use this to +periodically pre-compute a fresh status result. + +Interactive users could then use `--deserialize` to simply +(and immediately) print the last-known-good result without +waiting for the status scan. + +The binary serialization file format includes some worktree state +information allowing `--deserialize` to reject the cached data +and force a normal status scan if, for example, the commit, branch, +or status modes/options change. The format cannot, however, indicate +when the cached data is otherwise stale -- that coordination belongs +to the task driving the serializations. + + CONFIGURATION ------------- diff --git a/Documentation/technical/status-serialization-format.txt b/Documentation/technical/status-serialization-format.txt new file mode 100644 index 00000000000000..475ae814495581 --- /dev/null +++ b/Documentation/technical/status-serialization-format.txt @@ -0,0 +1,107 @@ +Git status serialization format +=============================== + +Git status serialization enables git to dump the results of a status scan +to a binary file. This file can then be loaded by later status invocations +to print the cached status results. + +The file contains the essential fields from: +() the index +() the "struct wt_status" for the overall results +() the contents of "struct wt_status_change_data" for tracked changed files +() the list of untracked and ignored files + +Version 1 Format: +================= + +The V1 file begins with a required header section followed by optional +sections for each type of item (changed, untracked, ignored). Individual +item sections are only present if necessary. Each item section begins +with an item-type header with the number of items in the section. + +Each "line" in the format is encoded using pkt-line with a final LF. +Flush packets are used to terminate sections. + +----------------- +PKT-LINE("version" SP "1") + +[] +[] +[] +----------------- + + +V1 Header +--------- + +The v1-header-section fields are taken directly from "struct wt_status". +Each field is printed on a separate pkt-line. Lines for NULL string +values are omitted. All integers are printed with "%d". OIDs are +printed in hex. + +v1-header-section = + + PKT-LINE() + +v1-index-headers = PKT-LINE("index_mtime" SP SP LF) + +v1-wt-status-headers = PKT-LINE("is_initial" SP LF) + [ PKT-LINE("branch" SP LF) ] + [ PKT-LINE("reference" SP LF) ] + PKT-LINE("show_ignored_files" SP LF) + PKT-LINE("show_untracked_files" SP LF) + PKT-LINE("show_ignored_directory" SP LF) + [ PKT-LINE("ignore_submodule_arg" SP LF) ] + PKT-LINE("detect_rename" SP LF) + PKT-LINE("rename_score" SP LF) + PKT-LINE("rename_limit" SP LF) + PKT-LINE("detect_break" SP LF) + PKT-LINE("sha1_commit" SP LF) + PKT-LINE("committable" SP LF) + PKT-LINE("workdir_dirty" SP LF) + + +V1 Changed Items +---------------- + +The v1-changed-item-section lists all of the changed items with one +item per pkt-line. Each pkt-line contains: a binary block of data +from "struct wt_status_serialize_data_fixed" in a fixed header where +integers are in network byte order and OIDs are in raw (non-hex) form. +This is followed by one or two raw pathnames (not c-quoted) with NUL +terminators (both NULs are always present even if there is no rename). + +v1-changed-item-section = PKT-LINE("changed" SP LF) + [ PKT-LINE( LF) ]+ + PKT-LINE() + +changed_item = + + + + + + + + + + + + NUL + [ ] + NUL + + +V1 Untracked and Ignored Items +------------------------------ + +These sections are simple lists of pathnames. They ARE NOT +c-quoted. + +v1-untracked-item-section = PKT-LINE("untracked" SP LF) + [ PKT-LINE( LF) ]+ + PKT-LINE() + +v1-ignored-item-section = PKT-LINE("ignored" SP LF) + [ PKT-LINE( LF) ]+ + PKT-LINE() diff --git a/Makefile b/Makefile index 9b81082b03a6f5..a199ef5dead752 100644 --- a/Makefile +++ b/Makefile @@ -1209,6 +1209,8 @@ LIB_OBJS += wrapper.o LIB_OBJS += write-or-die.o LIB_OBJS += ws.o LIB_OBJS += wt-status.o +LIB_OBJS += wt-status-deserialize.o +LIB_OBJS += wt-status-serialize.o LIB_OBJS += xdiff-interface.o BUILTIN_OBJS += builtin/add.o diff --git a/builtin/commit.c b/builtin/commit.c index ba76c213792736..dc513e698dc0b2 100644 --- a/builtin/commit.c +++ b/builtin/commit.c @@ -161,6 +161,70 @@ static int opt_parse_porcelain(const struct option *opt, const char *arg, int un return 0; } +static int do_serialize = 0; +static int do_implicit_deserialize = 0; +static int do_explicit_deserialize = 0; +static char *deserialize_path = NULL; + +/* + * --serialize | --serialize=1 | --serialize=v1 + * + * Request that we serialize our output rather than printing in + * any of the established formats. Optionally specify serialization + * version. + */ +static int opt_parse_serialize(const struct option *opt, const char *arg, int unset) +{ + enum wt_status_format *value = (enum wt_status_format *)opt->value; + if (unset || !arg) + *value = STATUS_FORMAT_SERIALIZE_V1; + else if (!strcmp(arg, "v1") || !strcmp(arg, "1")) + *value = STATUS_FORMAT_SERIALIZE_V1; + else + die("unsupported serialize version '%s'", arg); + + if (do_explicit_deserialize) + die("cannot mix --serialize and --deserialize"); + do_implicit_deserialize = 0; + + do_serialize = 1; + return 0; +} + +/* + * --deserialize | --deserialize= | + * --no-deserialize + * + * Request that we deserialize status data from some existing resource + * rather than performing a status scan. + * + * The input source can come from stdin or a path given here -- or be + * inherited from the config settings. + */ +static int opt_parse_deserialize(const struct option *opt UNUSED, const char *arg, int unset) +{ + if (unset) { + do_implicit_deserialize = 0; + do_explicit_deserialize = 0; + } else { + if (do_serialize) + die("cannot mix --serialize and --deserialize"); + if (arg) { + /* override config or stdin */ + free(deserialize_path); + deserialize_path = xstrdup(arg); + } + if (deserialize_path && *deserialize_path + && (access(deserialize_path, R_OK) != 0)) + die("cannot find serialization file '%s'", + deserialize_path); + + do_explicit_deserialize = 1; + } + + return 0; +} + static int opt_parse_m(const struct option *opt, const char *arg, int unset) { struct strbuf *buf = opt->value; @@ -1176,6 +1240,8 @@ static enum untracked_status_type parse_untracked_setting_name(const char *u) return SHOW_NORMAL_UNTRACKED_FILES; else if (!strcmp(u, "all")) return SHOW_ALL_UNTRACKED_FILES; + else if (!strcmp(u,"complete")) + return SHOW_COMPLETE_UNTRACKED_FILES; else return SHOW_UNTRACKED_FILES_ERROR; } @@ -1473,6 +1539,19 @@ static int git_status_config(const char *k, const char *v, s->relative_paths = git_config_bool(k, v); return 0; } + if (!strcmp(k, "status.deserializepath")) { + /* + * Automatically assume deserialization if this is + * set in the config and the file exists. Do not + * complain if the file does not exist, because we + * silently fall back to normal mode. + */ + if (v && *v && access(v, R_OK) == 0) { + do_implicit_deserialize = 1; + deserialize_path = xstrdup(v); + } + return 0; + } if (!strcmp(k, "status.showuntrackedfiles")) { enum untracked_status_type u; @@ -1512,7 +1591,8 @@ struct repository *repo UNUSED) static const char *rename_score_arg = (const char *)-1; static struct wt_status s; unsigned int progress_flag = 0; - int fd; + int try_deserialize; + int fd = -1; struct object_id oid; static struct option builtin_status_options[] = { OPT__VERBOSE(&verbose, N_("be verbose")), @@ -1527,6 +1607,12 @@ struct repository *repo UNUSED) OPT_CALLBACK_F(0, "porcelain", &status_format, N_("version"), N_("machine-readable output"), PARSE_OPT_OPTARG, opt_parse_porcelain), + { OPTION_CALLBACK, 0, "serialize", &status_format, + N_("version"), N_("serialize raw status data to stdout"), + PARSE_OPT_OPTARG | PARSE_OPT_NONEG, opt_parse_serialize }, + { OPTION_CALLBACK, 0, "deserialize", NULL, + N_("path"), N_("deserialize raw status data from file"), + PARSE_OPT_OPTARG, opt_parse_deserialize }, OPT_SET_INT(0, "long", &status_format, N_("show status in long format (default)"), STATUS_FORMAT_LONG), @@ -1571,10 +1657,26 @@ struct repository *repo UNUSED) s.show_untracked_files == SHOW_NO_UNTRACKED_FILES) die(_("Unsupported combination of ignored and untracked-files arguments")); + if (s.show_untracked_files == SHOW_COMPLETE_UNTRACKED_FILES && + s.show_ignored_mode == SHOW_NO_IGNORED) + die(_("Complete Untracked only supported with ignored files")); + parse_pathspec(&s.pathspec, 0, PATHSPEC_PREFER_FULL, prefix, argv); + /* + * If we want to try to deserialize status data from a cache file, + * we need to re-order the initialization code. The problem is that + * this makes for a very nasty diff and causes merge conflicts as we + * carry it forward. And it easy to mess up the merge, so we + * duplicate some code here to hopefully reduce conflicts. + */ + try_deserialize = (!do_serialize && + (do_implicit_deserialize || do_explicit_deserialize)); + if (try_deserialize) + goto skip_init; + enable_fscache(0); if (status_format != STATUS_FORMAT_PORCELAIN && status_format != STATUS_FORMAT_PORCELAIN_V2) @@ -1589,6 +1691,7 @@ struct repository *repo UNUSED) else fd = -1; +skip_init: s.is_initial = repo_get_oid(the_repository, s.reference, &oid) ? 1 : 0; if (!s.is_initial) oidcpy(&s.oid_commit, &oid); @@ -1605,6 +1708,24 @@ struct repository *repo UNUSED) s.rename_score = parse_rename_score(&rename_score_arg); } + if (try_deserialize) { + if (s.relative_paths) + s.prefix = prefix; + + if (wt_status_deserialize(&s, deserialize_path) == DESERIALIZE_OK) + return 0; + + /* deserialize failed, so force the initialization we skipped above. */ + enable_fscache(1); + repo_read_index_preload(the_repository, &s.pathspec, 0); + refresh_index(the_repository->index, REFRESH_QUIET|REFRESH_UNMERGED, &s.pathspec, NULL, NULL); + + if (use_optional_locks()) + fd = repo_hold_locked_index(the_repository, &index_lock, 0); + else + fd = -1; + } + wt_status_collect(&s); if (0 <= fd) diff --git a/contrib/completion/git-completion.bash b/contrib/completion/git-completion.bash index 60a22d619a85cc..94ff4f9363e1d8 100644 --- a/contrib/completion/git-completion.bash +++ b/contrib/completion/git-completion.bash @@ -1802,7 +1802,7 @@ _git_clone () esac } -__git_untracked_file_modes="all no normal" +__git_untracked_file_modes="all no normal complete" __git_trailer_tokens () { diff --git a/pkt-line.c b/pkt-line.c index 24479eae4dbe2a..b3ec9e5f0d636a 100644 --- a/pkt-line.c +++ b/pkt-line.c @@ -231,7 +231,7 @@ static int do_packet_write(const int fd_out, const char *buf, size_t size, return 0; } -static int packet_write_gently(const int fd_out, const char *buf, size_t size) +int packet_write_gently(const int fd_out, const char *buf, size_t size) { struct strbuf err = STRBUF_INIT; if (do_packet_write(fd_out, buf, size, &err)) { diff --git a/pkt-line.h b/pkt-line.h index 3b33cc64f34dcc..10fd9a812e1935 100644 --- a/pkt-line.h +++ b/pkt-line.h @@ -29,6 +29,7 @@ void packet_write(int fd_out, const char *buf, size_t size); void packet_buf_write(struct strbuf *buf, const char *fmt, ...) __attribute__((format (printf, 2, 3))); int packet_flush_gently(int fd); int packet_write_fmt_gently(int fd, const char *fmt, ...) __attribute__((format (printf, 2, 3))); +int packet_write_gently(const int fd_out, const char *buf, size_t size); int write_packetized_from_fd_no_flush(int fd_in, int fd_out); int write_packetized_from_buf_no_flush_count(const char *src_in, size_t len, int fd_out, int *packet_counter); diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh new file mode 100755 index 00000000000000..283a98bdf750e6 --- /dev/null +++ b/t/t7522-serialized-status.sh @@ -0,0 +1,141 @@ +#!/bin/sh + +test_description='git serialized status tests' + +. ./test-lib.sh + +# This file includes tests for serializing / deserializing +# status data. These tests cover two basic features: +# +# [1] Because users can request different types of untracked-file +# and ignored file reporting, the cache data generated by +# serialize must use either the same untracked and ignored +# parameters as the later deserialize invocation; otherwise, +# the deserialize invocation must disregard the cached data +# and run a full scan itself. +# +# To increase the number of cases where the cached status can +# be used, we have added a "--untracked-file=complete" option +# that reports a superset or union of the results from the +# "-u normal" and "-u all". We combine this with a filter in +# deserialize to filter the results. +# +# Ignored file reporting is simpler in that is an all or +# nothing; there are no subsets. +# +# The tests here (in addition to confirming that a cache +# file can be generated and used by a subsequent status +# command) need to test this untracked-file filtering. +# +# [2] ensuring the status calls are using data from the status +# cache as expected. This includes verifying cached data +# is used when appropriate as well as falling back to +# performing a new status scan when the data in the cache +# is insufficient/known stale. + +test_expect_success 'setup' ' + git branch -M main && + cat >.gitignore <<-\EOF && + *.ign + ignored_dir/ + EOF + + mkdir tracked ignored_dir && + touch tracked_1.txt tracked/tracked_1.txt && + git add . && + test_tick && + git commit -m"Adding original file." && + mkdir untracked && + touch ignored.ign ignored_dir/ignored_2.txt \ + untracked_1.txt untracked/untracked_2.txt untracked/untracked_3.txt +' + +test_expect_success 'verify untracked-files=complete with no conversion' ' + test_when_finished "rm serialized_status.dat new_change.txt output" && + cat >expect <<-\EOF && + ? expect + ? serialized_status.dat + ? untracked/ + ? untracked/untracked_2.txt + ? untracked/untracked_3.txt + ? untracked_1.txt + ! ignored.ign + ! ignored_dir/ + EOF + + git status --untracked-files=complete --ignored=matching --serialize >serialized_status.dat && + touch new_change.txt && + + git status --porcelain=v2 --untracked-files=complete --ignored=matching --deserialize=serialized_status.dat >output && + test_cmp expect output +' + +test_expect_success 'verify untracked-files=complete to untracked-files=normal conversion' ' + test_when_finished "rm serialized_status.dat new_change.txt output" && + cat >expect <<-\EOF && + ? expect + ? serialized_status.dat + ? untracked/ + ? untracked_1.txt + EOF + + git status --untracked-files=complete --ignored=matching --serialize >serialized_status.dat && + touch new_change.txt && + + git status --porcelain=v2 --deserialize=serialized_status.dat >output && + test_cmp expect output +' + +test_expect_success 'verify untracked-files=complete to untracked-files=all conversion' ' + test_when_finished "rm serialized_status.dat new_change.txt output" && + cat >expect <<-\EOF && + ? expect + ? serialized_status.dat + ? untracked/untracked_2.txt + ? untracked/untracked_3.txt + ? untracked_1.txt + ! ignored.ign + ! ignored_dir/ + EOF + + git status --untracked-files=complete --ignored=matching --serialize >serialized_status.dat && + touch new_change.txt && + + git status --porcelain=v2 --untracked-files=all --ignored=matching --deserialize=serialized_status.dat >output && + test_cmp expect output +' + +test_expect_success 'verify serialized status with non-convertible ignore mode does new scan' ' + test_when_finished "rm serialized_status.dat new_change.txt output" && + cat >expect <<-\EOF && + ? expect + ? new_change.txt + ? output + ? serialized_status.dat + ? untracked/ + ? untracked_1.txt + ! ignored.ign + ! ignored_dir/ + EOF + + git status --untracked-files=complete --ignored=matching --serialize >serialized_status.dat && + touch new_change.txt && + + git status --porcelain=v2 --ignored --deserialize=serialized_status.dat >output && + test_cmp expect output +' + +test_expect_success 'verify serialized status handles path scopes' ' + test_when_finished "rm serialized_status.dat new_change.txt output" && + cat >expect <<-\EOF && + ? untracked/ + EOF + + git status --untracked-files=complete --ignored=matching --serialize >serialized_status.dat && + touch new_change.txt && + + git status --porcelain=v2 --deserialize=serialized_status.dat untracked >output && + test_cmp expect output +' + +test_done diff --git a/t/t7523-status-complete-untracked.sh b/t/t7523-status-complete-untracked.sh new file mode 100755 index 00000000000000..f79611fc024f48 --- /dev/null +++ b/t/t7523-status-complete-untracked.sh @@ -0,0 +1,39 @@ +#!/bin/sh + +test_description='git status untracked complete tests' + +. ./test-lib.sh + +test_expect_success 'setup' ' + cat >.gitignore <<-\EOF && + *.ign + ignored_dir/ + EOF + + mkdir tracked ignored_dir && + touch tracked_1.txt tracked/tracked_1.txt && + git add . && + test_tick && + git commit -m"Adding original file." && + mkdir untracked && + touch ignored.ign ignored_dir/ignored_2.txt \ + untracked_1.txt untracked/untracked_2.txt untracked/untracked_3.txt +' + +test_expect_success 'verify untracked-files=complete' ' + cat >expect <<-\EOF && + ? expect + ? output + ? untracked/ + ? untracked/untracked_2.txt + ? untracked/untracked_3.txt + ? untracked_1.txt + ! ignored.ign + ! ignored_dir/ + EOF + + git status --porcelain=v2 --untracked-files=complete --ignored >output && + test_cmp expect output +' + +test_done diff --git a/wt-status-deserialize.c b/wt-status-deserialize.c new file mode 100644 index 00000000000000..95cb7324a50697 --- /dev/null +++ b/wt-status-deserialize.c @@ -0,0 +1,608 @@ +#define USE_THE_REPOSITORY_VARIABLE +#include "git-compat-util.h" +#include "environment.h" +#include "hex.h" +#include "hash.h" +#include "wt-status.h" +#include "pkt-line.h" +#include "trace.h" +#include "statinfo.h" +#include "hex.h" + +static struct trace_key trace_deserialize = TRACE_KEY_INIT(DESERIALIZE); + +enum deserialize_parse_strategy { + DESERIALIZE_STRATEGY_AS_IS, + DESERIALIZE_STRATEGY_SKIP, + DESERIALIZE_STRATEGY_NORMAL, + DESERIALIZE_STRATEGY_ALL +}; + +static int check_path_contains(const char *out, int out_len, const char *in, int in_len) +{ + return (out_len > 0 && + out_len < in_len && + (out[out_len - 1] == '/') && + !memcmp(out, in, out_len)); +} + +static const char *my_packet_read_line(int fd, int *line_len) +{ + static char buf[LARGE_PACKET_MAX]; + + *line_len = packet_read(fd, buf, sizeof(buf), + PACKET_READ_CHOMP_NEWLINE | + PACKET_READ_GENTLE_ON_EOF); + return (*line_len > 0) ? buf : NULL; +} + +/* + * mtime_reported contains the mtime of the index when the + * serialization snapshot was computed. + * + * mtime_observed_on_disk contains the mtime of the index now. + * + * If these 2 times are different, then the .git/index has + * changed since the serialization cache was created and we + * must reject the cache because anything could have changed. + * + * If they are the same, we continue trying to use the cache. + */ +static int my_validate_index(const char *path, const struct cache_time *mtime_reported) +{ + struct stat st; + struct cache_time mtime_observed_on_disk; + + if (lstat(path, &st)) { + trace_printf_key(&trace_deserialize, "could not stat index"); + return DESERIALIZE_ERR; + } + mtime_observed_on_disk.sec = st.st_mtime; + mtime_observed_on_disk.nsec = ST_MTIME_NSEC(st); + if ((mtime_observed_on_disk.sec != mtime_reported->sec) || + (mtime_observed_on_disk.nsec != mtime_reported->nsec)) { + trace_printf_key(&trace_deserialize, "index mtime changed [des %d.%d][obs %d.%d]", + mtime_reported->sec, mtime_reported->nsec, + mtime_observed_on_disk.sec, mtime_observed_on_disk.nsec); + return DESERIALIZE_ERR; + } + + return DESERIALIZE_OK; +} + +static int wt_deserialize_v1_header(struct wt_status *s, int fd) +{ + struct cache_time index_mtime; + int line_len, nr_fields; + const char *line; + const char *arg; + + /* + * parse header lines up to the first flush packet. + */ + while ((line = my_packet_read_line(fd, &line_len))) { + + if (skip_prefix(line, "index_mtime ", &arg)) { + nr_fields = sscanf(arg, "%u %u", + &index_mtime.sec, + &index_mtime.nsec); + if (nr_fields != 2) { + trace_printf_key(&trace_deserialize, "invalid index_mtime (%d) '%s'", + nr_fields, line); + return DESERIALIZE_ERR; + } + continue; + } + + if (skip_prefix(line, "is_initial ", &arg)) { + s->is_initial = (int)strtol(arg, NULL, 10); + continue; + } + if (skip_prefix(line, "branch ", &arg)) { + s->branch = xstrdup(arg); + continue; + } + if (skip_prefix(line, "reference ", &arg)) { + s->reference = xstrdup(arg); + continue; + } + /* pathspec */ + /* verbose */ + /* amend */ + if (skip_prefix(line, "whence ", &arg)) { + s->whence = (int)strtol(arg, NULL, 10); + continue; + } + /* nowarn */ + /* use_color */ + /* no_gettext */ + /* display_comment_prefix */ + /* relative_paths */ + /* submodule_summary */ + if (skip_prefix(line, "show_ignored_mode ", &arg)) { + s->show_ignored_mode = (int)strtol(arg, NULL, 10); + continue; + } + if (skip_prefix(line, "show_untracked_files ", &arg)) { + s->show_untracked_files = (int)strtol(arg, NULL, 10); + continue; + } + if (skip_prefix(line, "ignore_submodule_arg ", &arg)) { + s->ignore_submodule_arg = xstrdup(arg); + continue; + } + /* color_palette */ + /* colopts */ + /* null_termination */ + /* commit_template */ + /* show_branch */ + /* show_stash */ + if (skip_prefix(line, "hints ", &arg)) { + s->hints = (int)strtol(arg, NULL, 10); + continue; + } + if (skip_prefix(line, "detect_rename ", &arg)) { + s->detect_rename = (int)strtol(arg, NULL, 10); + continue; + } + if (skip_prefix(line, "rename_score ", &arg)) { + s->rename_score = (int)strtol(arg, NULL, 10); + continue; + } + if (skip_prefix(line, "rename_limit ", &arg)) { + s->rename_limit = (int)strtol(arg, NULL, 10); + continue; + } + /* status_format */ + if (skip_prefix(line, "sha1_commit ", &arg)) { + if (get_oid_hex(arg, &s->oid_commit)) { + trace_printf_key(&trace_deserialize, "invalid sha1_commit"); + return DESERIALIZE_ERR; + } + continue; + } + if (skip_prefix(line, "committable ", &arg)) { + s->committable = (int)strtol(arg, NULL, 10); + continue; + } + if (skip_prefix(line, "workdir_dirty ", &arg)) { + s->workdir_dirty = (int)strtol(arg, NULL, 10); + continue; + } + /* prefix */ + + trace_printf_key(&trace_deserialize, "unexpected line '%s'", line); + return DESERIALIZE_ERR; + } + + return my_validate_index(s->index_file, &index_mtime); +} + +/* + * Build a string-list of (count) lines from the input. + */ +static int wt_deserialize_v1_changed_items(struct wt_status *s, int fd, int count UNUSED) +{ + struct wt_status_serialize_data *sd; + char *p; + int line_len; + const char *line; + struct string_list_item *item; + + memset(&s->change, 0, sizeof(s->change)); + s->change.strdup_strings = 1; + + /* + * + + * + * + * NUL [] NUL + */ + while ((line = my_packet_read_line(fd, &line_len))) { + struct wt_status_change_data *d = xcalloc(1, sizeof(*d)); + sd = (struct wt_status_serialize_data *)line; + + d->worktree_status = ntohl(sd->fixed.worktree_status); + d->index_status = ntohl(sd->fixed.index_status); + d->stagemask = ntohl(sd->fixed.stagemask); + d->rename_status = ntohl(sd->fixed.rename_status); + d->rename_score = ntohl(sd->fixed.rename_score); + d->mode_head = ntohl(sd->fixed.mode_head); + d->mode_index = ntohl(sd->fixed.mode_index); + d->mode_worktree = ntohl(sd->fixed.mode_worktree); + d->dirty_submodule = ntohl(sd->fixed.dirty_submodule); + d->new_submodule_commits = ntohl(sd->fixed.new_submodule_commits); + oidcpy(&d->oid_head, &sd->fixed.oid_head); + oidcpy(&d->oid_index, &sd->fixed.oid_index); + + p = sd->variant; + item = string_list_append(&s->change, p); + p += strlen(p) + 1; + if (*p) + d->rename_source = xstrdup(p); + item->util = d; + + trace_printf_key( + &trace_deserialize, + "change: %d %d %d %d %d %o %o %o %d %d %s %s '%s' '%s'", + d->worktree_status, + d->index_status, + d->stagemask, + d->rename_status, + d->rename_score, + d->mode_head, + d->mode_index, + d->mode_worktree, + d->dirty_submodule, + d->new_submodule_commits, + oid_to_hex(&d->oid_head), + oid_to_hex(&d->oid_index), + item->string, + (d->rename_source ? d->rename_source : "")); + } + + return DESERIALIZE_OK; +} + +static int wt_deserialize_v1_untracked_items(struct wt_status *s, + int fd, + int count UNUSED, + enum deserialize_parse_strategy strategy) +{ + int line_len; + const char *line; + char *out = NULL; + int out_len = 0; + + memset(&s->untracked, 0, sizeof(s->untracked)); + s->untracked.strdup_strings = 1; + + /* + * + + * + */ + while ((line = my_packet_read_line(fd, &line_len))) { + if (strategy == DESERIALIZE_STRATEGY_AS_IS) + string_list_append(&s->untracked, line); + if (strategy == DESERIALIZE_STRATEGY_SKIP) + continue; + if (strategy == DESERIALIZE_STRATEGY_NORMAL) { + + /* Only add "normal" entries to list */ + if (out && + check_path_contains(out, out_len, line, line_len)) { + continue; + } + else { + out = string_list_append(&s->untracked, line)->string; + out_len = line_len; + } + } + if (strategy == DESERIALIZE_STRATEGY_ALL) { + /* Only add "all" entries to list */ + if (line[line_len - 1] != '/') + string_list_append(&s->untracked, line); + } + } + + return DESERIALIZE_OK; +} + +static int wt_deserialize_v1_ignored_items(struct wt_status *s, + int fd, + int count UNUSED, + enum deserialize_parse_strategy strategy) +{ + int line_len; + const char *line; + + memset(&s->ignored, 0, sizeof(s->ignored)); + s->ignored.strdup_strings = 1; + + /* + * + + * + */ + while ((line = my_packet_read_line(fd, &line_len))) { + if (strategy == DESERIALIZE_STRATEGY_AS_IS) + string_list_append(&s->ignored, line); + else + continue; + } + + return DESERIALIZE_OK; +} + +static int validate_untracked_files_arg(enum untracked_status_type cmd, + enum untracked_status_type des, + enum deserialize_parse_strategy *strategy) +{ + *strategy = DESERIALIZE_STRATEGY_AS_IS; + + if (cmd == des) { + *strategy = DESERIALIZE_STRATEGY_AS_IS; + } else if (cmd == SHOW_NO_UNTRACKED_FILES) { + *strategy = DESERIALIZE_STRATEGY_SKIP; + } else if (des == SHOW_COMPLETE_UNTRACKED_FILES) { + if (cmd == SHOW_ALL_UNTRACKED_FILES) + *strategy = DESERIALIZE_STRATEGY_ALL; + else if (cmd == SHOW_NORMAL_UNTRACKED_FILES) + *strategy = DESERIALIZE_STRATEGY_NORMAL; + } else { + return DESERIALIZE_ERR; + } + + return DESERIALIZE_OK; +} + +static int validate_ignored_files_arg(enum show_ignored_type cmd, + enum show_ignored_type des, + enum deserialize_parse_strategy *strategy) +{ + *strategy = DESERIALIZE_STRATEGY_AS_IS; + + if (cmd == SHOW_NO_IGNORED) { + *strategy = DESERIALIZE_STRATEGY_SKIP; + } + else if (cmd != des) { + return DESERIALIZE_ERR; + } + + return DESERIALIZE_OK; +} + +static int wt_deserialize_v1(const struct wt_status *cmd_s, struct wt_status *s, int fd) +{ + int line_len; + const char *line; + const char *arg; + int nr_changed = 0; + int nr_untracked = 0; + int nr_ignored = 0; + + enum deserialize_parse_strategy ignored_strategy = DESERIALIZE_STRATEGY_AS_IS, untracked_strategy = DESERIALIZE_STRATEGY_AS_IS; + + if (wt_deserialize_v1_header(s, fd) == DESERIALIZE_ERR) + return DESERIALIZE_ERR; + + /* + * We now have the header parsed. Look at the command args (as passed in), and see how to parse + * the serialized data + */ + if (validate_untracked_files_arg(cmd_s->show_untracked_files, s->show_untracked_files, &untracked_strategy)) { + trace_printf_key(&trace_deserialize, "reject: show_untracked_file: command: %d, serialized : %d", + cmd_s->show_untracked_files, + s->show_untracked_files); + return DESERIALIZE_ERR; + } + + if (validate_ignored_files_arg(cmd_s->show_ignored_mode, s->show_ignored_mode, &ignored_strategy)) { + trace_printf_key(&trace_deserialize, "reject: show_ignored_mode: command: %d, serialized: %d", + cmd_s->show_ignored_mode, + s->show_ignored_mode); + return DESERIALIZE_ERR; + } + + /* + * [ [+] ] + * [ [+] ] + * [ [+] ] + */ + while ((line = my_packet_read_line(fd, &line_len))) { + if (skip_prefix(line, "changed ", &arg)) { + nr_changed = (int)strtol(arg, NULL, 10); + if (wt_deserialize_v1_changed_items(s, fd, nr_changed) + == DESERIALIZE_ERR) + return DESERIALIZE_ERR; + continue; + } + if (skip_prefix(line, "untracked ", &arg)) { + nr_untracked = (int)strtol(arg, NULL, 10); + if (wt_deserialize_v1_untracked_items(s, fd, nr_untracked, untracked_strategy) + == DESERIALIZE_ERR) + return DESERIALIZE_ERR; + continue; + } + if (skip_prefix(line, "ignored ", &arg)) { + nr_ignored = (int)strtol(arg, NULL, 10); + if (wt_deserialize_v1_ignored_items(s, fd, nr_ignored, ignored_strategy) + == DESERIALIZE_ERR) + return DESERIALIZE_ERR; + continue; + } + trace_printf_key(&trace_deserialize, "unexpected line '%s'", line); + return DESERIALIZE_ERR; + } + + return DESERIALIZE_OK; +} + +static int wt_deserialize_parse(const struct wt_status *cmd_s, struct wt_status *s, int fd) +{ + int line_len; + const char *line; + const char *arg; + + memset(s, 0, sizeof(*s)); + + if ((line = my_packet_read_line(fd, &line_len)) && + (skip_prefix(line, "version ", &arg))) { + int version = (int)strtol(arg, NULL, 10); + if (version == 1) + return wt_deserialize_v1(cmd_s, s, fd); + } + trace_printf_key(&trace_deserialize, "missing/unsupported version"); + return DESERIALIZE_ERR; +} + +static inline int my_strcmp_null(const char *a, const char *b) +{ + const char *alt_a = (a) ? a : ""; + const char *alt_b = (b) ? b : ""; + + return strcmp(alt_a, alt_b); +} + +static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *des_s, int fd) +{ + /* + * Check the path spec on the current command + */ + if (cmd_s->pathspec.nr > 1) { + trace_printf_key(&trace_deserialize, "reject: multiple pathspecs"); + return DESERIALIZE_ERR; + } + + /* + * If we have a pathspec, but it maches the root (e.g. no filtering) + * then this is OK. + */ + if (cmd_s->pathspec.nr == 1 && + my_strcmp_null(cmd_s->pathspec.items[0].match, "")) { + trace_printf_key(&trace_deserialize, "reject: pathspec"); + return DESERIALIZE_ERR; + } + + /* + * Deserialize cached status + */ + if (wt_deserialize_parse(cmd_s, des_s, fd) == DESERIALIZE_ERR) + return DESERIALIZE_ERR; + + /* + * Compare fields in cmd_s with those observed in des_s and + * complain if they are incompatible (such as different "-u" + * or "--ignored" settings). + */ + if (cmd_s->is_initial != des_s->is_initial) { + trace_printf_key(&trace_deserialize, "reject: is_initial"); + return DESERIALIZE_ERR; + } + if (my_strcmp_null(cmd_s->branch, des_s->branch)) { + trace_printf_key(&trace_deserialize, "reject: branch"); + return DESERIALIZE_ERR; + } + if (my_strcmp_null(cmd_s->reference, des_s->reference)) { + trace_printf_key(&trace_deserialize, "reject: reference"); + return DESERIALIZE_ERR; + } + /* verbose */ + /* amend */ + if (cmd_s->whence != des_s->whence) { + trace_printf_key(&trace_deserialize, "reject: whence"); + return DESERIALIZE_ERR; + } + /* nowarn */ + /* use_color */ + /* no_gettext */ + /* display_comment_prefix */ + /* relative_paths */ + /* submodule_summary */ + + /* show_ignored_files - already validated */ + /* show_untrackes_files - already validated */ + + /* + * Submodules are not supported by status serialization. + * The status will not be serialized if it contains submodules, + * and so this check is not needed. + * + * if (my_strcmp_null(cmd_s->ignore_submodule_arg, des_s->ignore_submodule_arg)) { + * trace_printf_key(&trace_deserialize, "reject: ignore_submodule_arg"); + * return DESERIALIZE_ERR; + * } + */ + + /* color_palette */ + /* colopts */ + /* null_termination */ + /* commit_template */ + /* show_branch */ + /* show_stash */ + /* hints */ + if (cmd_s->detect_rename != des_s->detect_rename) { + trace_printf_key(&trace_deserialize, "reject: detect_rename"); + return DESERIALIZE_ERR; + } + if (cmd_s->rename_score != des_s->rename_score) { + trace_printf_key(&trace_deserialize, "reject: rename_score"); + return DESERIALIZE_ERR; + } + if (cmd_s->rename_limit != des_s->rename_limit) { + trace_printf_key(&trace_deserialize, "reject: rename_limit"); + return DESERIALIZE_ERR; + } + /* status_format */ + if (!oideq(&cmd_s->oid_commit, &des_s->oid_commit)) { + trace_printf_key(&trace_deserialize, "reject: sha1_commit"); + return DESERIALIZE_ERR; + } + + /* + * Copy over display-related fields from the current command. + */ + des_s->repo = cmd_s->repo; + des_s->verbose = cmd_s->verbose; + /* amend */ + /* whence */ + des_s->nowarn = cmd_s->nowarn; + des_s->use_color = cmd_s->use_color; + des_s->no_gettext = cmd_s->no_gettext; + des_s->display_comment_prefix = cmd_s->display_comment_prefix; + des_s->relative_paths = cmd_s->relative_paths; + des_s->submodule_summary = cmd_s->submodule_summary; + memcpy(des_s->color_palette, cmd_s->color_palette, + sizeof(char)*WT_STATUS_MAXSLOT*COLOR_MAXLEN); + des_s->colopts = cmd_s->colopts; + des_s->null_termination = cmd_s->null_termination; + /* commit_template */ + des_s->show_branch = cmd_s->show_branch; + des_s->show_stash = cmd_s->show_stash; + /* hints */ + des_s->status_format = cmd_s->status_format; + des_s->fp = cmd_s->fp; + if (cmd_s->prefix && *cmd_s->prefix) + des_s->prefix = xstrdup(cmd_s->prefix); + + return DESERIALIZE_OK; +} + + +/* + * Read raw serialized status data from the given file + * + * Verify that the args specified in the current command + * are compatible with the deserialized data (such as "-uno"). + * + * Copy display-related fields from the current command + * into the deserialized data (so that the user can request + * long or short as they please). + */ +int wt_status_deserialize(const struct wt_status *cmd_s, + const char *path) +{ + struct wt_status des_s; + int result; + + if (path && *path && strcmp(path, "0")) { + int fd = xopen(path, O_RDONLY); + if (fd == -1) { + trace_printf_key(&trace_deserialize, "could not read '%s'", path); + return DESERIALIZE_ERR; + } + trace_printf_key(&trace_deserialize, "reading serialization file '%s'", path); + result = wt_deserialize_fd(cmd_s, &des_s, fd); + close(fd); + } else { + trace_printf_key(&trace_deserialize, "reading stdin"); + result = wt_deserialize_fd(cmd_s, &des_s, 0); + } + + if (result == DESERIALIZE_OK) { + wt_status_get_state(cmd_s->repo, &des_s.state, des_s.branch && + !strcmp(des_s.branch, "HEAD")); + wt_status_print(&des_s); + } + + return result; +} diff --git a/wt-status-serialize.c b/wt-status-serialize.c new file mode 100644 index 00000000000000..1f0e8a2c1e2eed --- /dev/null +++ b/wt-status-serialize.c @@ -0,0 +1,219 @@ +#include "git-compat-util.h" +#include "hex.h" +#include "repository.h" +#include "wt-status.h" +#include "pkt-line.h" +#include "trace.h" +#include "read-cache-ll.h" + +static struct trace_key trace_serialize = TRACE_KEY_INIT(SERIALIZE); + +/* + * Write V1 header fields. + */ +static void wt_serialize_v1_header(struct wt_status *s, int fd) +{ + /* + * Write select fields from the current index to help + * the deserializer recognize a stale data set. + */ + packet_write_fmt(fd, "index_mtime %d %d\n", + s->repo->index->timestamp.sec, + s->repo->index->timestamp.nsec); + + /* + * Write data from wt_status to qualify this status report. + * That is, if this run specified "-uno", the consumer of + * our serialization should know that. + */ + packet_write_fmt(fd, "is_initial %d\n", s->is_initial); + if (s->branch) + packet_write_fmt(fd, "branch %s\n", s->branch); + if (s->reference) + packet_write_fmt(fd, "reference %s\n", s->reference); + /* pathspec */ + /* verbose */ + /* amend */ + packet_write_fmt(fd, "whence %d\n", s->whence); + /* nowarn */ + /* use_color */ + /* no_gettext */ + /* display_comment_prefix */ + /* relative_paths */ + /* submodule_summary */ + packet_write_fmt(fd, "show_ignored_mode %d\n", s->show_ignored_mode); + packet_write_fmt(fd, "show_untracked_files %d\n", s->show_untracked_files); + if (s->ignore_submodule_arg) + packet_write_fmt(fd, "ignore_submodule_arg %s\n", s->ignore_submodule_arg); + /* color_palette */ + /* colopts */ + /* null_termination */ + /* commit_template */ + /* show_branch */ + /* show_stash */ + packet_write_fmt(fd, "hints %d\n", s->hints); + packet_write_fmt(fd, "detect_rename %d\n", s->detect_rename); + packet_write_fmt(fd, "rename_score %d\n", s->rename_score); + packet_write_fmt(fd, "rename_limit %d\n", s->rename_limit); + /* status_format */ + packet_write_fmt(fd, "sha1_commit %s\n", oid_to_hex(&s->oid_commit)); + packet_write_fmt(fd, "committable %d\n", s->committable); + packet_write_fmt(fd, "workdir_dirty %d\n", s->workdir_dirty); + /* prefix */ + packet_flush(fd); +} + +/* + * Print changed/unmerged items. + * We write raw (not c-quoted) pathname(s). The rename_source is only + * set when status computed a rename/copy. + * + * We ALWAYS write a final LF to the packet-line (for debugging) + * even though Linux pathnames allow LFs. + */ +static inline void wt_serialize_v1_changed(struct wt_status *s UNUSED, int fd, + struct string_list_item *item) +{ + struct wt_status_change_data *d = item->util; + struct wt_status_serialize_data sd; + char *begin; + char *end; + char *p; + int len_path, len_rename_source; + + trace_printf_key(&trace_serialize, + "change: %d %d %d %d %d %o %o %o %d %d %s %s '%s' '%s'", + d->worktree_status, + d->index_status, + d->stagemask, + d->rename_status, + d->rename_score, + d->mode_head, + d->mode_index, + d->mode_worktree, + d->dirty_submodule, + d->new_submodule_commits, + oid_to_hex(&d->oid_head), + oid_to_hex(&d->oid_index), + item->string, + (d->rename_source ? d->rename_source : "")); + + sd.fixed.worktree_status = htonl(d->worktree_status); + sd.fixed.index_status = htonl(d->index_status); + sd.fixed.stagemask = htonl(d->stagemask); + sd.fixed.rename_status = htonl(d->rename_status); + sd.fixed.rename_score = htonl(d->rename_score); + sd.fixed.mode_head = htonl(d->mode_head); + sd.fixed.mode_index = htonl(d->mode_index); + sd.fixed.mode_worktree = htonl(d->mode_worktree); + sd.fixed.dirty_submodule = htonl(d->dirty_submodule); + sd.fixed.new_submodule_commits = htonl(d->new_submodule_commits); + oidcpy(&sd.fixed.oid_head, &d->oid_head); + oidcpy(&sd.fixed.oid_index, &d->oid_index); + + begin = (char *)&sd; + end = begin + sizeof(sd); + + p = sd.variant; + + /* + * Write NUL [] NUL LF at the end of the buffer. + */ + len_path = strlen(item->string); + len_rename_source = d->rename_source ? strlen(d->rename_source) : 0; + + /* + * This is a bit of a hack, but I don't want to split the + * status detail record across multiple pkt-lines. + */ + if (p + len_path + 1 + len_rename_source + 1 + 1 >= end) + BUG("path to long to serialize '%s'", item->string); + + memcpy(p, item->string, len_path); + p += len_path; + *p++ = '\0'; + + if (len_rename_source) { + memcpy(p, d->rename_source, len_rename_source); + p += len_rename_source; + } + *p++ = '\0'; + *p++ = '\n'; + + if (packet_write_gently(fd, begin, (p - begin))) + BUG("cannot serialize '%s'", item->string); +} + +/* + * Write raw (not c-quoted) pathname for an untracked item. + * We ALWAYS write a final LF to the packet-line (for debugging) + * even though Linux pathnames allows LFs. That is, deserialization + * should use the packet-line length and omit the final LF. + */ +static inline void wt_serialize_v1_untracked(struct wt_status *s UNUSED, int fd, + struct string_list_item *item) +{ + packet_write_fmt(fd, "%s\n", item->string); +} + +/* + * Write raw (not c-quoted) pathname for an ignored item. + * We ALWAYS write a final LF to the packet-line (for debugging) + * even though Linux pathnames allows LFs. + */ +static inline void wt_serialize_v1_ignored(struct wt_status *s UNUSED, int fd, + struct string_list_item *item) +{ + packet_write_fmt(fd, "%s\n", item->string); +} + +/* + * Serialize the list of changes to stdout. The goal of this + * is to just serialize the key fields in wt_status so that a + * later command can rebuilt it and do the printing. + * + * We DO NOT include the contents of wt_status_state NOR + * current branch info. This info easily gets stale and + * is relatively quick for the status consumer to compute + * as necessary. + */ +void wt_status_serialize_v1(struct wt_status *s) +{ + int fd = 1; /* we always write to stdout */ + struct string_list_item *iter; + int k; + + /* + * version header must be first line. + */ + packet_write_fmt(fd, "version 1\n"); + + wt_serialize_v1_header(s, fd); + + if (s->change.nr > 0) { + packet_write_fmt(fd, "changed %"PRIuMAX"\n", (uintmax_t)s->change.nr); + for (k = 0; k < s->change.nr; k++) { + iter = &(s->change.items[k]); + wt_serialize_v1_changed(s, fd, iter); + } + packet_flush(fd); + } + + if (s->untracked.nr > 0) { + packet_write_fmt(fd, "untracked %"PRIuMAX"\n", (uintmax_t)s->untracked.nr); + for (k = 0; k < s->untracked.nr; k++) { + iter = &(s->untracked.items[k]); + wt_serialize_v1_untracked(s, fd, iter); + } + packet_flush(fd); + } + + if (s->ignored.nr > 0) { + packet_write_fmt(fd, "ignored %"PRIuMAX"\n", (uintmax_t)s->ignored.nr); + for (k = 0; k < s->ignored.nr; k++) { + iter = &(s->ignored.items[k]); + wt_serialize_v1_ignored(s, fd, iter); + } + packet_flush(fd); + } +} diff --git a/wt-status.c b/wt-status.c index 061cf9024fdba8..179df73197addd 100644 --- a/wt-status.c +++ b/wt-status.c @@ -787,6 +787,9 @@ static void wt_status_collect_untracked(struct wt_status *s) if (s->show_untracked_files != SHOW_ALL_UNTRACKED_FILES) dir.flags |= DIR_SHOW_OTHER_DIRECTORIES | DIR_HIDE_EMPTY_DIRECTORIES; + if (s->show_untracked_files == SHOW_COMPLETE_UNTRACKED_FILES) + dir.flags |= DIR_KEEP_UNTRACKED_CONTENTS; + if (s->show_ignored_mode) { dir.flags |= DIR_SHOW_IGNORED_TOO; @@ -2619,6 +2622,9 @@ void wt_status_print(struct wt_status *s) case STATUS_FORMAT_LONG: wt_longstatus_print(s); break; + case STATUS_FORMAT_SERIALIZE_V1: + wt_status_serialize_v1(s); + break; } trace2_region_leave("status", "print", s->repo); diff --git a/wt-status.h b/wt-status.h index 4e377ce62b8b28..afeaf5753a0494 100644 --- a/wt-status.h +++ b/wt-status.h @@ -4,6 +4,7 @@ #include "string-list.h" #include "color.h" #include "pathspec.h" +#include "pkt-line.h" #include "remote.h" struct repository; @@ -26,7 +27,8 @@ enum untracked_status_type { SHOW_UNTRACKED_FILES_ERROR = -1, SHOW_NO_UNTRACKED_FILES = 0, SHOW_NORMAL_UNTRACKED_FILES, - SHOW_ALL_UNTRACKED_FILES + SHOW_ALL_UNTRACKED_FILES, + SHOW_COMPLETE_UNTRACKED_FILES, }; enum show_ignored_type { @@ -74,6 +76,7 @@ enum wt_status_format { STATUS_FORMAT_SHORT, STATUS_FORMAT_PORCELAIN, STATUS_FORMAT_PORCELAIN_V2, + STATUS_FORMAT_SERIALIZE_V1, STATUS_FORMAT_UNSPECIFIED }; @@ -185,4 +188,52 @@ int require_clean_work_tree(struct repository *repo, int ignore_submodules, int gently); +#define DESERIALIZE_OK 0 +#define DESERIALIZE_ERR 1 + +struct wt_status_serialize_data_fixed +{ + uint32_t worktree_status; + uint32_t index_status; + uint32_t stagemask; + uint32_t rename_status; + uint32_t rename_score; + uint32_t mode_head; + uint32_t mode_index; + uint32_t mode_worktree; + uint32_t dirty_submodule; + uint32_t new_submodule_commits; + struct object_id oid_head; + struct object_id oid_index; +}; + +/* + * Consume the maximum amount of data possible in a + * packet-line record. This is overkill because we + * have at most 2 relative pathnames, but means we + * don't need to allocate a variable length structure. + */ +struct wt_status_serialize_data +{ + struct wt_status_serialize_data_fixed fixed; + char variant[LARGE_PACKET_DATA_MAX + - sizeof(struct wt_status_serialize_data_fixed)]; +}; + +/* + * Serialize computed status scan results using "version 1" format + * to the given file. + */ +void wt_status_serialize_v1(struct wt_status *s); + +/* + * Deserialize existing status results from the given file and + * populate a (new) "struct wt_status". Use the contents of "cmd_s" + * (computed from the command line arguments) to verify that the + * cached data is compatible and overlay various display-related + * fields. + */ +int wt_status_deserialize(const struct wt_status *cmd_s, + const char *path); + #endif /* STATUS_H */ From 315f2e18831b7d64b3374a1c5107fd5d8fac391c Mon Sep 17 00:00:00 2001 From: Jameson Miller Date: Wed, 10 Jan 2018 11:56:26 -0500 Subject: [PATCH 060/196] Teach ahead-behind and serialized status to play nicely together --- t/t7522-serialized-status.sh | 34 +++++++++++++++++++++++++++++++++- wt-status-deserialize.c | 3 ++- wt-status-serialize.c | 1 + 3 files changed, 36 insertions(+), 2 deletions(-) diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh index 283a98bdf750e6..0f5a33e2a23442 100755 --- a/t/t7522-serialized-status.sh +++ b/t/t7522-serialized-status.sh @@ -47,7 +47,13 @@ test_expect_success 'setup' ' git commit -m"Adding original file." && mkdir untracked && touch ignored.ign ignored_dir/ignored_2.txt \ - untracked_1.txt untracked/untracked_2.txt untracked/untracked_3.txt + untracked_1.txt untracked/untracked_2.txt untracked/untracked_3.txt && + + test_oid_cache <<-EOF + branch_oid sha1:68d4a437ea4c2de65800f48c053d4d543b55c410 + + branch_oid sha256:6b95e4b1ea911dad213f2020840f5e92d3066cf9e38cf35f79412ec58d409ce4 + EOF ' test_expect_success 'verify untracked-files=complete with no conversion' ' @@ -138,4 +144,30 @@ test_expect_success 'verify serialized status handles path scopes' ' test_cmp expect output ' +test_expect_success 'verify no-ahead-behind and serialized status integration' ' + test_when_finished "rm serialized_status.dat new_change.txt output" && + cat >expect <<-EOF && + # branch.oid $(test_oid branch_oid) + # branch.head alt_branch + # branch.upstream main + # branch.ab +1 -0 + ? expect + ? serialized_status.dat + ? untracked/ + ? untracked_1.txt + EOF + + git checkout -b alt_branch main --track >/dev/null && + touch alt_branch_changes.txt && + git add alt_branch_changes.txt && + test_tick && + git commit -m"New commit on alt branch" && + + git status --untracked-files=complete --ignored=matching --serialize >serialized_status.dat && + touch new_change.txt && + + git -c status.aheadBehind=false status --porcelain=v2 --branch --ahead-behind --deserialize=serialized_status.dat >output && + test_cmp expect output +' + test_done diff --git a/wt-status-deserialize.c b/wt-status-deserialize.c index 95cb7324a50697..120a5d34377581 100644 --- a/wt-status-deserialize.c +++ b/wt-status-deserialize.c @@ -520,6 +520,7 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de /* show_branch */ /* show_stash */ /* hints */ + /* ahead_behind_flags */ if (cmd_s->detect_rename != des_s->detect_rename) { trace_printf_key(&trace_deserialize, "reject: detect_rename"); return DESERIALIZE_ERR; @@ -559,6 +560,7 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de des_s->show_branch = cmd_s->show_branch; des_s->show_stash = cmd_s->show_stash; /* hints */ + des_s->ahead_behind_flags = cmd_s->ahead_behind_flags; des_s->status_format = cmd_s->status_format; des_s->fp = cmd_s->fp; if (cmd_s->prefix && *cmd_s->prefix) @@ -567,7 +569,6 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de return DESERIALIZE_OK; } - /* * Read raw serialized status data from the given file * diff --git a/wt-status-serialize.c b/wt-status-serialize.c index 1f0e8a2c1e2eed..0198366d49a310 100644 --- a/wt-status-serialize.c +++ b/wt-status-serialize.c @@ -52,6 +52,7 @@ static void wt_serialize_v1_header(struct wt_status *s, int fd) /* show_branch */ /* show_stash */ packet_write_fmt(fd, "hints %d\n", s->hints); + /* ahead_behind_flags */ packet_write_fmt(fd, "detect_rename %d\n", s->detect_rename); packet_write_fmt(fd, "rename_score %d\n", s->rename_score); packet_write_fmt(fd, "rename_limit %d\n", s->rename_limit); From 0ff0247948d10c3a191194a369e3bc43862cd73e Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 2 Feb 2018 14:17:05 -0500 Subject: [PATCH 061/196] status: serialize to path Teach status serialization to take an optional pathname on the command line to direct that cache data be written there rather than to stdout. When used this way, normal status results will still be written to stdout. When no path is given, only binary serialization data is written to stdout. Usage: git status --serialize[=] Signed-off-by: Jeff Hostetler --- Documentation/git-status.txt | 10 ++++++---- builtin/commit.c | 36 +++++++++++++++++++++++++++--------- t/t7522-serialized-status.sh | 35 +++++++++++++++++++++++++++++++++++ wt-status-serialize.c | 5 ++--- wt-status.c | 2 +- wt-status.h | 2 +- 6 files changed, 72 insertions(+), 18 deletions(-) diff --git a/Documentation/git-status.txt b/Documentation/git-status.txt index fedf86d32718eb..62254617dd7fc1 100644 --- a/Documentation/git-status.txt +++ b/Documentation/git-status.txt @@ -151,10 +151,12 @@ ignored, then the directory is not shown, but all contents are shown. threshold. See also linkgit:git-diff[1] `--find-renames`. ---serialize[=]:: - (EXPERIMENTAL) Serialize raw status results to stdout in a - format suitable for use by `--deserialize`. Valid values for - `` are "1" and "v1". +--serialize[=]:: + (EXPERIMENTAL) Serialize raw status results to a file or stdout + in a format suitable for use by `--deserialize`. If a path is + given, serialize data will be written to that path *and* normal + status output will be written to stdout. If path is omitted, + only binary serialization data will be written to stdout. --deserialize[=]:: (EXPERIMENTAL) Deserialize raw status results from a file or diff --git a/builtin/commit.c b/builtin/commit.c index dc513e698dc0b2..d7ad42ba6d153a 100644 --- a/builtin/commit.c +++ b/builtin/commit.c @@ -162,26 +162,34 @@ static int opt_parse_porcelain(const struct option *opt, const char *arg, int un } static int do_serialize = 0; +static char *serialize_path = NULL; + static int do_implicit_deserialize = 0; static int do_explicit_deserialize = 0; static char *deserialize_path = NULL; /* - * --serialize | --serialize=1 | --serialize=v1 + * --serialize | --serialize= + * + * Request that we serialize status output rather than or in addition to + * printing in any of the established formats. + * + * Without a path, we write binary serialization data to stdout (and omit + * the normal status output). * - * Request that we serialize our output rather than printing in - * any of the established formats. Optionally specify serialization - * version. + * With a path, we write binary serialization data to the and then + * write normal status output. */ static int opt_parse_serialize(const struct option *opt, const char *arg, int unset) { enum wt_status_format *value = (enum wt_status_format *)opt->value; if (unset || !arg) *value = STATUS_FORMAT_SERIALIZE_V1; - else if (!strcmp(arg, "v1") || !strcmp(arg, "1")) - *value = STATUS_FORMAT_SERIALIZE_V1; - else - die("unsupported serialize version '%s'", arg); + + if (arg) { + free(serialize_path); + serialize_path = xstrdup(arg); + } if (do_explicit_deserialize) die("cannot mix --serialize and --deserialize"); @@ -1608,7 +1616,7 @@ struct repository *repo UNUSED) N_("version"), N_("machine-readable output"), PARSE_OPT_OPTARG, opt_parse_porcelain), { OPTION_CALLBACK, 0, "serialize", &status_format, - N_("version"), N_("serialize raw status data to stdout"), + N_("path"), N_("serialize raw status data to path or stdout"), PARSE_OPT_OPTARG | PARSE_OPT_NONEG, opt_parse_serialize }, { OPTION_CALLBACK, 0, "deserialize", NULL, N_("path"), N_("deserialize raw status data from file"), @@ -1734,6 +1742,16 @@ struct repository *repo UNUSED) if (s.relative_paths) s.prefix = prefix; + if (serialize_path) { + int fd_serialize = xopen(serialize_path, + O_WRONLY | O_CREAT | O_TRUNC, 0666); + if (fd_serialize < 0) + die_errno(_("could not serialize to '%s'"), + serialize_path); + wt_status_serialize_v1(fd_serialize, &s); + close(fd_serialize); + } + wt_status_print(&s); wt_status_collect_free_buffers(&s); diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh index 0f5a33e2a23442..2a81b4e625ee59 100755 --- a/t/t7522-serialized-status.sh +++ b/t/t7522-serialized-status.sh @@ -170,4 +170,39 @@ test_expect_success 'verify no-ahead-behind and serialized status integration' ' test_cmp expect output ' +test_expect_success 'verify new --serialize=path mode' ' + #test_when_finished "rm serialized_status.dat expect new_change.txt output.1 output.2" && + cat >expect <<-\EOF && + ? expect + ? output.1 + ? untracked/ + ? untracked_1.txt + EOF + + git checkout -b serialize_path_branch main --track >/dev/null && + touch alt_branch_changes.txt && + git add alt_branch_changes.txt && + test_tick && + git commit -m"New commit on serialize_path_branch" && + + git status --porcelain=v2 --serialize=serialized_status.dat >output.1 && + touch new_change.txt && + + git status --porcelain=v2 --deserialize=serialized_status.dat >output.2 && + test_cmp expect output.1 && + test_cmp expect output.2 +' + +test_expect_success 'renames' ' + git init -b main rename_test && + echo OLDNAME >rename_test/OLDNAME && + git -C rename_test add OLDNAME && + git -C rename_test commit -m OLDNAME && + git -C rename_test mv OLDNAME NEWNAME && + git -C rename_test status --serialize=renamed.dat >output.1 && + echo DIRT >rename_test/DIRT && + git -C rename_test status --deserialize=renamed.dat >output.2 && + test_cmp output.1 output.2 +' + test_done diff --git a/wt-status-serialize.c b/wt-status-serialize.c index 0198366d49a310..15646ef4174e81 100644 --- a/wt-status-serialize.c +++ b/wt-status-serialize.c @@ -169,7 +169,7 @@ static inline void wt_serialize_v1_ignored(struct wt_status *s UNUSED, int fd, } /* - * Serialize the list of changes to stdout. The goal of this + * Serialize the list of changes to the given file. The goal of this * is to just serialize the key fields in wt_status so that a * later command can rebuilt it and do the printing. * @@ -178,9 +178,8 @@ static inline void wt_serialize_v1_ignored(struct wt_status *s UNUSED, int fd, * is relatively quick for the status consumer to compute * as necessary. */ -void wt_status_serialize_v1(struct wt_status *s) +void wt_status_serialize_v1(int fd, struct wt_status *s) { - int fd = 1; /* we always write to stdout */ struct string_list_item *iter; int k; diff --git a/wt-status.c b/wt-status.c index 179df73197addd..e50e1b2d945e16 100644 --- a/wt-status.c +++ b/wt-status.c @@ -2623,7 +2623,7 @@ void wt_status_print(struct wt_status *s) wt_longstatus_print(s); break; case STATUS_FORMAT_SERIALIZE_V1: - wt_status_serialize_v1(s); + wt_status_serialize_v1(1, s); break; } diff --git a/wt-status.h b/wt-status.h index afeaf5753a0494..314185cc1ae8f3 100644 --- a/wt-status.h +++ b/wt-status.h @@ -224,7 +224,7 @@ struct wt_status_serialize_data * Serialize computed status scan results using "version 1" format * to the given file. */ -void wt_status_serialize_v1(struct wt_status *s); +void wt_status_serialize_v1(int fd, struct wt_status *s); /* * Deserialize existing status results from the given file and From 5fe3d3ddd61fa74da764bc17746e48122e8ab73d Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 7 Feb 2018 10:59:03 -0500 Subject: [PATCH 062/196] status: reject deserialize in V2 and conflicts Teach status deserialize code to reject status cache when printing in porcelain V2 and there are unresolved conflicts in the cache file. A follow-on task might extend the cache format to include this additiona data. See code for longer explanation. Signed-off-by: Jeff Hostetler --- t/t7522-serialized-status.sh | 90 +++++++++++++++++++++++++++++++++++- wt-status-deserialize.c | 28 ++++++++++- 2 files changed, 115 insertions(+), 3 deletions(-) diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh index 2a81b4e625ee59..361afca94835e0 100755 --- a/t/t7522-serialized-status.sh +++ b/t/t7522-serialized-status.sh @@ -51,8 +51,14 @@ test_expect_success 'setup' ' test_oid_cache <<-EOF branch_oid sha1:68d4a437ea4c2de65800f48c053d4d543b55c410 + x_base sha1:587be6b4c3f93f93c489c0111bba5596147a26cb + x_ours sha1:b68025345d5301abad4d9ec9166f455243a0d746 + x_theirs sha1:975fbec8256d3e8a3797e7a3611380f27c49f4ac branch_oid sha256:6b95e4b1ea911dad213f2020840f5e92d3066cf9e38cf35f79412ec58d409ce4 + x_base sha256:14f5162e2fe3d240d0d37aaab0f90e4af9a7cfa79639f3bab005b5bfb4174d9f + x_ours sha256:3a404ba030a4afa912155c476a48a253d4b3a43d0098431b6d6ca6e554bd78fb + x_theirs sha256:44dc634218adec09e34f37839b3840bad8c6103693e9216626b32d00e093fa35 EOF ' @@ -171,7 +177,7 @@ test_expect_success 'verify no-ahead-behind and serialized status integration' ' ' test_expect_success 'verify new --serialize=path mode' ' - #test_when_finished "rm serialized_status.dat expect new_change.txt output.1 output.2" && + test_when_finished "rm serialized_status.dat expect new_change.txt output.1 output.2" && cat >expect <<-\EOF && ? expect ? output.1 @@ -193,6 +199,88 @@ test_expect_success 'verify new --serialize=path mode' ' test_cmp expect output.2 ' +test_expect_success 'merge conflicts' ' + + # create a merge conflict. + + git init -b main conflicts && + echo x >conflicts/x.txt && + git -C conflicts add x.txt && + git -C conflicts commit -m x && + git -C conflicts branch a && + git -C conflicts branch b && + git -C conflicts checkout a && + echo y >conflicts/x.txt && + git -C conflicts add x.txt && + git -C conflicts commit -m a && + git -C conflicts checkout b && + echo z >conflicts/x.txt && + git -C conflicts add x.txt && + git -C conflicts commit -m b && + test_must_fail git -C conflicts merge --no-commit a && + + # verify that regular status correctly identifies it + # in each format. + + cat >expect.v2 <observed.v2 && + test_cmp expect.v2 observed.v2 && + + cat >expect.long <..." to mark resolution) + both modified: x.txt + +no changes added to commit (use "git add" and/or "git commit -a") +EOF + git -C conflicts status --long >observed.long && + test_cmp expect.long observed.long && + + cat >expect.short <observed.short && + test_cmp expect.short observed.short && + + # save status data in serialized cache. + + git -C conflicts status --serialize >serialized && + + # make some dirt in the worktree so we can tell whether subsequent + # status commands used the cached data or did a fresh status. + + echo dirt >conflicts/dirt.txt && + + # run status using the cached data. + + git -C conflicts status --long --deserialize=../serialized >observed.long && + test_cmp expect.long observed.long && + + git -C conflicts status --short --deserialize=../serialized >observed.short && + test_cmp expect.short observed.short && + + # currently, the cached data does not have enough information about + # merge conflicts for porcelain V2 format. (And V2 format looks at + # the index to get that data, but the whole point of the serialization + # is to avoid reading the index unnecessarily.) So V2 always rejects + # the cached data when there is an unresolved conflict. + + cat >expect.v2.dirty <observed.v2 && + test_cmp expect.v2.dirty observed.v2 + +' + test_expect_success 'renames' ' git init -b main rename_test && echo OLDNAME >rename_test/OLDNAME && diff --git a/wt-status-deserialize.c b/wt-status-deserialize.c index 120a5d34377581..7fa0c481c85bb6 100644 --- a/wt-status-deserialize.c +++ b/wt-status-deserialize.c @@ -181,7 +181,8 @@ static int wt_deserialize_v1_header(struct wt_status *s, int fd) /* * Build a string-list of (count) lines from the input. */ -static int wt_deserialize_v1_changed_items(struct wt_status *s, int fd, int count UNUSED) +static int wt_deserialize_v1_changed_items(const struct wt_status *cmd_s, + struct wt_status *s, int fd, int count UNUSED) { struct wt_status_serialize_data *sd; char *p; @@ -239,6 +240,29 @@ static int wt_deserialize_v1_changed_items(struct wt_status *s, int fd, int coun oid_to_hex(&d->oid_index), item->string, (d->rename_source ? d->rename_source : "")); + + if (d->stagemask && + cmd_s->status_format == STATUS_FORMAT_PORCELAIN_V2) { + /* + * We have an unresolved conflict and the user wants + * to see porcelain V2 output. The cached status data + * does not contain enough information for V2 (because + * the main status computation does not capture it). + * We only get a single change record for the file with + * a single SHA -- we don't get the stage [123] mode + * and SHA data. The V2 detail-line print code looks + * up this information directly from the index. The + * whole point of this serialization cache is to avoid + * reading the index, so the V2 print code gets zeros. + * So we reject the status cache and let the fallback + * code run. + */ + trace_printf_key( + &trace_deserialize, + "reject: V2 format and umerged file: %s", + item->string); + return DESERIALIZE_ERR; + } } return DESERIALIZE_OK; @@ -391,7 +415,7 @@ static int wt_deserialize_v1(const struct wt_status *cmd_s, struct wt_status *s, while ((line = my_packet_read_line(fd, &line_len))) { if (skip_prefix(line, "changed ", &arg)) { nr_changed = (int)strtol(arg, NULL, 10); - if (wt_deserialize_v1_changed_items(s, fd, nr_changed) + if (wt_deserialize_v1_changed_items(cmd_s, s, fd, nr_changed) == DESERIALIZE_ERR) return DESERIALIZE_ERR; continue; From fe9f2d8233fb47eac67557d6f9b3ad3b1b4e1136 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 20 Jul 2018 12:08:50 -0400 Subject: [PATCH 063/196] serialize-status: serialize global and repo-local exclude file metadata Changes to the global or repo-local excludes files can change the results returned by "git status" for untracked files. Therefore, it is important that the exclude-file values used during serialization are still current at the time of deserialization. Teach "git status --serialize" to report metadata on the user's global exclude file (which defaults to "$XDG_HOME/git/ignore") and for the repo-local excludes file (which is in ".git/info/excludes"). Serialize will record the pathnames and mtimes for these files in the serialization header (next to the mtime data for the .git/index file). Teach "git status --deserialize" to validate this new metadata. If either exclude file has changed since the serialization-cache-file was written, then deserialize will reject the cache file and force a full/normal status run. Signed-off-by: Jeff Hostetler --- wt-status-deserialize.c | 85 ++++++++++++++++++++++++++++ wt-status-serialize.c | 122 ++++++++++++++++++++++++++++++++++++++++ wt-status.h | 8 +++ 3 files changed, 215 insertions(+) diff --git a/wt-status-deserialize.c b/wt-status-deserialize.c index 7fa0c481c85bb6..1f4c27868f306d 100644 --- a/wt-status-deserialize.c +++ b/wt-status-deserialize.c @@ -8,6 +8,7 @@ #include "trace.h" #include "statinfo.h" #include "hex.h" +#include "path.h" static struct trace_key trace_deserialize = TRACE_KEY_INIT(DESERIALIZE); @@ -70,12 +71,69 @@ static int my_validate_index(const char *path, const struct cache_time *mtime_re return DESERIALIZE_OK; } +/* + * Use the given key and exclude pathname to compute a serialization header + * reflecting the current contents on disk. See if that matches the value + * computed for this key when the cache was written. Reject the cache if + * anything has changed. + */ +static int my_validate_excludes(const char *path, const char *key, const char *line) +{ + struct strbuf sb = STRBUF_INIT; + int r; + + wt_serialize_compute_exclude_header(&sb, key, path); + + r = (strcmp(line, sb.buf) ? DESERIALIZE_ERR : DESERIALIZE_OK); + + if (r == DESERIALIZE_ERR) + trace_printf_key(&trace_deserialize, + "%s changed [cached '%s'][observed '%s']", + key, line, sb.buf); + + strbuf_release(&sb); + return r; +} + +static int my_parse_core_excludes(const char *line) +{ + /* + * In dir.c:setup_standard_excludes() they use either the value of + * the "core.excludefile" variable (stored in the global "excludes_file" + * variable) -or- the default value "$XDG_HOME/git/ignore". This is done + * during wt_status_collect_untracked() which we are hoping to not call. + * + * Fake the setup here. + */ + + if (excludes_file) { + return my_validate_excludes(excludes_file, "core_excludes", line); + } else { + char *path = xdg_config_home("ignore"); + int r = my_validate_excludes(path, "core_excludes", line); + free(path); + return r; + } +} + +static int my_parse_repo_excludes(const char *line) +{ + char *path = git_pathdup("info/exclude"); + int r = my_validate_excludes(path, "repo_excludes", line); + free(path); + + return r; +} + static int wt_deserialize_v1_header(struct wt_status *s, int fd) { struct cache_time index_mtime; int line_len, nr_fields; const char *line; const char *arg; + int have_required_index_mtime = 0; + int have_required_core_excludes = 0; + int have_required_repo_excludes = 0; /* * parse header lines up to the first flush packet. @@ -91,6 +149,20 @@ static int wt_deserialize_v1_header(struct wt_status *s, int fd) nr_fields, line); return DESERIALIZE_ERR; } + have_required_index_mtime = 1; + continue; + } + + if (skip_prefix(line, "core_excludes ", &arg)) { + if (my_parse_core_excludes(line) != DESERIALIZE_OK) + return DESERIALIZE_ERR; + have_required_core_excludes = 1; + continue; + } + if (skip_prefix(line, "repo_excludes ", &arg)) { + if (my_parse_repo_excludes(line) != DESERIALIZE_OK) + return DESERIALIZE_ERR; + have_required_repo_excludes = 1; continue; } @@ -175,6 +247,19 @@ static int wt_deserialize_v1_header(struct wt_status *s, int fd) return DESERIALIZE_ERR; } + if (!have_required_index_mtime) { + trace_printf_key(&trace_deserialize, "missing '%s'", "index_mtime"); + return DESERIALIZE_ERR; + } + if (!have_required_core_excludes) { + trace_printf_key(&trace_deserialize, "missing '%s'", "core_excludes"); + return DESERIALIZE_ERR; + } + if (!have_required_repo_excludes) { + trace_printf_key(&trace_deserialize, "missing '%s'", "repo_excludes"); + return DESERIALIZE_ERR; + } + return my_validate_index(s->index_file, &index_mtime); } diff --git a/wt-status-serialize.c b/wt-status-serialize.c index 15646ef4174e81..beaa7bd565cf2c 100644 --- a/wt-status-serialize.c +++ b/wt-status-serialize.c @@ -1,13 +1,133 @@ +#define USE_THE_REPOSITORY_VARIABLE + #include "git-compat-util.h" +#include "environment.h" #include "hex.h" #include "repository.h" #include "wt-status.h" #include "pkt-line.h" #include "trace.h" #include "read-cache-ll.h" +#include "path.h" static struct trace_key trace_serialize = TRACE_KEY_INIT(SERIALIZE); +/* + * Compute header record for exclude file using format: + * SP SP LF + */ +void wt_serialize_compute_exclude_header(struct strbuf *sb, + const char *key, + const char *path) +{ + struct stat st; + struct stat_data sd; + + memset(&sd, 0, sizeof(sd)); + + strbuf_setlen(sb, 0); + + if (!path || !*path) { + strbuf_addf(sb, "%s U (unset)", key); + } else if (lstat(path, &st) == -1) { + if (is_missing_file_error(errno)) + strbuf_addf(sb, "%s E (not-found) %s", key, path); + else + strbuf_addf(sb, "%s E (other) %s", key, path); + } else { + fill_stat_data(&sd, &st); + strbuf_addf(sb, "%s F %d %d %s", + key, sd.sd_mtime.sec, sd.sd_mtime.nsec, path); + } +} + +static void append_exclude_info(int fd, const char *path, const char *key) +{ + struct strbuf sb = STRBUF_INIT; + + wt_serialize_compute_exclude_header(&sb, key, path); + + packet_write_fmt(fd, "%s\n", sb.buf); + + strbuf_release(&sb); +} + +static void append_core_excludes_file_info(int fd) +{ + /* + * Write pathname and mtime of the core/global excludes file to + * the status cache header. Since a change in the global excludes + * will/may change the results reported by status, the deserialize + * code should be able to reject the status cache if the excludes + * file changes since when the cache was written. + * + * The "core.excludefile" setting defaults to $XDG_HOME/git/ignore + * and uses a global variable which should have been set during + * wt_status_collect_untracked(). + * + * See dir.c:setup_standard_excludes() + */ + append_exclude_info(fd, excludes_file, "core_excludes"); +} + +static void append_repo_excludes_file_info(int fd) +{ + /* + * Likewise, there is a per-repo excludes file in .git/info/excludes + * that can change the results reported by status. And the deserialize + * code needs to be able to reject the status cache if this file + * changes. + * + * See dir.c:setup_standard_excludes() and git_path_info_excludes(). + * We replicate the pathname construction here because of the static + * variables/functions used in dir.c. + */ + char *path = git_pathdup("info/exclude"); + + append_exclude_info(fd, path, "repo_excludes"); + + free(path); +} + +/* + * WARNING: The status cache attempts to preserve the essential in-memory + * status data after a status scan into a "serialization" (aka "status cache") + * file. It allows later "git status --deserialize=" instances to + * just print the cached status results without scanning the workdir (and + * without reading the index). + * + * The status cache file is valid as long as: + * [1] the set of functional command line options are the same (think "-u"). + * [2] repo-local and user-global configuration settings are compatible. + * [3] nothing in the workdir has changed. + * + * We rely on: + * [1.a] We remember the relevant (functional, non-display) command line + * arguments in the status cache header. + * [2.a] We use the mtime of the .git/index to detect staging changes. + * [2.b] We use the mtimes of the excludes files to detect changes that + * might affect untracked file reporting. + * + * But we need external help to verify [3]. + * [] This includes changes to tracked files. + * [] This includes changes to tracked .gitignore files that might change + * untracked file reporting. + * [] This includes the creation of new, untracked per-directory .gitignore + * files that might change untracked file reporting. + * + * [3.a] On GVFS repos, we rely on the GVFS service (mount) daemon to + * watch the filesystem and invalidate (delete) the status cache + * when anything changes inside the workdir. + * + * [3.b] TODO This problem is not solved for non-GVFS repos. + * [] It is possible that the untracked-cache index extension + * could help with this but that requires status to read the + * index to load the extension. + * [] It is possible that the new fsmonitor facility could also + * provide this information, but that to requires reading the + * index. + */ + /* * Write V1 header fields. */ @@ -20,6 +140,8 @@ static void wt_serialize_v1_header(struct wt_status *s, int fd) packet_write_fmt(fd, "index_mtime %d %d\n", s->repo->index->timestamp.sec, s->repo->index->timestamp.nsec); + append_core_excludes_file_info(fd); + append_repo_excludes_file_info(fd); /* * Write data from wt_status to qualify this status report. diff --git a/wt-status.h b/wt-status.h index 314185cc1ae8f3..9728117f0e9217 100644 --- a/wt-status.h +++ b/wt-status.h @@ -236,4 +236,12 @@ void wt_status_serialize_v1(int fd, struct wt_status *s); int wt_status_deserialize(const struct wt_status *cmd_s, const char *path); +/* + * A helper routine for serialize and deserialize to compute + * metadata for the user-global and repo-local excludes files. + */ +void wt_serialize_compute_exclude_header(struct strbuf *sb, + const char *key, + const char *path); + #endif /* STATUS_H */ From 377714581fff4b04dbdfb1f3867d8fec8177d072 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 25 Jul 2018 14:49:37 -0400 Subject: [PATCH 064/196] status: deserialization wait Teach `git status --deserialize` to either wait indefintely or immediately fail if the status serialization cache file is stale. Signed-off-by: Jeff Hostetler --- Documentation/config/status.txt | 16 +++++ builtin/commit.c | 59 +++++++++++++++- t/t7522-serialized-status.sh | 52 ++++++++++++++ wt-status-deserialize.c | 119 +++++++++++++++++++++++++++++--- wt-status.h | 12 +++- 5 files changed, 245 insertions(+), 13 deletions(-) diff --git a/Documentation/config/status.txt b/Documentation/config/status.txt index 7302b066644e73..4d863fdaaec2eb 100644 --- a/Documentation/config/status.txt +++ b/Documentation/config/status.txt @@ -83,3 +83,19 @@ status.deserializePath:: generated by `--serialize`. This will be overridden by `--deserialize=` on the command line. If the cache file is invalid or stale, git will fall-back and compute status normally. + +status.deserializeWait:: + EXPERIMENTAL, Specifies what `git status --deserialize` should do + if the serialization cache file is stale and whether it should + fall-back and compute status normally. This will be overridden by + `--deserialize-wait=` on the command line. ++ +-- +* `fail` - cause git to exit with an error when the status cache file +is stale; this is intended for testing and debugging. +* `block` - cause git to spin and periodically retry the cache file +every 100 ms; this is intended to help coordinate with another git +instance concurrently computing the cache file. +* `no` - to immediately fall-back if cache file is stale. This is the default. +* `` - time (in tenths of a second) to spin and retry. +-- diff --git a/builtin/commit.c b/builtin/commit.c index d7ad42ba6d153a..f2eaad52afb70f 100644 --- a/builtin/commit.c +++ b/builtin/commit.c @@ -168,6 +168,9 @@ static int do_implicit_deserialize = 0; static int do_explicit_deserialize = 0; static char *deserialize_path = NULL; +static enum wt_status_deserialize_wait implicit_deserialize_wait = DESERIALIZE_WAIT__UNSET; +static enum wt_status_deserialize_wait explicit_deserialize_wait = DESERIALIZE_WAIT__UNSET; + /* * --serialize | --serialize= * @@ -233,6 +236,40 @@ static int opt_parse_deserialize(const struct option *opt UNUSED, const char *ar return 0; } +static enum wt_status_deserialize_wait parse_dw(const char *arg) +{ + int tenths; + + if (!strcmp(arg, "fail")) + return DESERIALIZE_WAIT__FAIL; + else if (!strcmp(arg, "block")) + return DESERIALIZE_WAIT__BLOCK; + else if (!strcmp(arg, "no")) + return DESERIALIZE_WAIT__NO; + + /* + * Otherwise, assume it is a timeout in tenths of a second. + * If it contains a bogus value, atol() will return zero + * which is OK. + */ + tenths = atol(arg); + if (tenths < 0) + tenths = DESERIALIZE_WAIT__NO; + return tenths; +} + +static int opt_parse_deserialize_wait(const struct option *opt UNUSED, + const char *arg, + int unset) +{ + if (unset) + explicit_deserialize_wait = DESERIALIZE_WAIT__UNSET; + else + explicit_deserialize_wait = parse_dw(arg); + + return 0; +} + static int opt_parse_m(const struct option *opt, const char *arg, int unset) { struct strbuf *buf = opt->value; @@ -1560,6 +1597,13 @@ static int git_status_config(const char *k, const char *v, } return 0; } + if (!strcmp(k, "status.deserializewait")) { + if (!v || !*v) + implicit_deserialize_wait = DESERIALIZE_WAIT__UNSET; + else + implicit_deserialize_wait = parse_dw(v); + return 0; + } if (!strcmp(k, "status.showuntrackedfiles")) { enum untracked_status_type u; @@ -1621,6 +1665,9 @@ struct repository *repo UNUSED) { OPTION_CALLBACK, 0, "deserialize", NULL, N_("path"), N_("deserialize raw status data from file"), PARSE_OPT_OPTARG, opt_parse_deserialize }, + { OPTION_CALLBACK, 0, "deserialize-wait", NULL, + N_("fail|block|no"), N_("how to wait if status cache file is invalid"), + PARSE_OPT_OPTARG, opt_parse_deserialize_wait }, OPT_SET_INT(0, "long", &status_format, N_("show status in long format (default)"), STATUS_FORMAT_LONG), @@ -1717,11 +1764,21 @@ struct repository *repo UNUSED) } if (try_deserialize) { + int result; + enum wt_status_deserialize_wait dw = implicit_deserialize_wait; + if (explicit_deserialize_wait != DESERIALIZE_WAIT__UNSET) + dw = explicit_deserialize_wait; + if (dw == DESERIALIZE_WAIT__UNSET) + dw = DESERIALIZE_WAIT__NO; + if (s.relative_paths) s.prefix = prefix; - if (wt_status_deserialize(&s, deserialize_path) == DESERIALIZE_OK) + result = wt_status_deserialize(&s, deserialize_path, dw); + if (result == DESERIALIZE_OK) return 0; + if (dw == DESERIALIZE_WAIT__FAIL) + die(_("Rejected status serialization cache")); /* deserialize failed, so force the initialization we skipped above. */ enable_fscache(1); diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh index 361afca94835e0..edf15d7af45489 100755 --- a/t/t7522-serialized-status.sh +++ b/t/t7522-serialized-status.sh @@ -199,6 +199,58 @@ test_expect_success 'verify new --serialize=path mode' ' test_cmp expect output.2 ' +test_expect_success 'try deserialize-wait feature' ' + test_when_finished "rm -f serialized_status.dat dirt expect.* output.* trace.*" && + + git status --serialize=serialized_status.dat >output.1 && + + # make status cache stale by updating the mtime on the index. confirm that + # deserialize fails when requested. + sleep 1 && + touch .git/index && + test_must_fail git status --deserialize=serialized_status.dat --deserialize-wait=fail && + test_must_fail git -c status.deserializeWait=fail status --deserialize=serialized_status.dat && + + cat >expect.1 <<-\EOF && + ? expect.1 + ? output.1 + ? serialized_status.dat + ? untracked/ + ? untracked_1.txt + EOF + + # refresh the status cache. + git status --porcelain=v2 --serialize=serialized_status.dat >output.1 && + test_cmp expect.1 output.1 && + + # create some dirt. confirm deserialize used the existing status cache. + echo x >dirt && + git status --porcelain=v2 --deserialize=serialized_status.dat >output.2 && + test_cmp output.1 output.2 && + + # make the cache stale and try the timeout feature and wait upto + # 2 tenths of a second. confirm deserialize timed out and rejected + # the status cache and did a normal scan. + + cat >expect.2 <<-\EOF && + ? dirt + ? expect.1 + ? expect.2 + ? output.1 + ? output.2 + ? serialized_status.dat + ? trace.2 + ? untracked/ + ? untracked_1.txt + EOF + + sleep 1 && + touch .git/index && + GIT_TRACE_DESERIALIZE=1 git status --porcelain=v2 --deserialize=serialized_status.dat --deserialize-wait=2 >output.2 2>trace.2 && + test_cmp expect.2 output.2 && + grep "wait polled=2 result=1" trace.2 >trace.2g +' + test_expect_success 'merge conflicts' ' # create a merge conflict. diff --git a/wt-status-deserialize.c b/wt-status-deserialize.c index 1f4c27868f306d..5e77e4aaa82705 100644 --- a/wt-status-deserialize.c +++ b/wt-status-deserialize.c @@ -62,7 +62,8 @@ static int my_validate_index(const char *path, const struct cache_time *mtime_re mtime_observed_on_disk.nsec = ST_MTIME_NSEC(st); if ((mtime_observed_on_disk.sec != mtime_reported->sec) || (mtime_observed_on_disk.nsec != mtime_reported->nsec)) { - trace_printf_key(&trace_deserialize, "index mtime changed [des %d.%d][obs %d.%d]", + trace_printf_key(&trace_deserialize, + "index mtime changed [des %d %d][obs %d %d]", mtime_reported->sec, mtime_reported->nsec, mtime_observed_on_disk.sec, mtime_observed_on_disk.nsec); return DESERIALIZE_ERR; @@ -554,6 +555,8 @@ static inline int my_strcmp_null(const char *a, const char *b) static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *des_s, int fd) { + memset(des_s, 0, sizeof(*des_s)); + /* * Check the path spec on the current command */ @@ -678,8 +681,101 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de return DESERIALIZE_OK; } +static struct cache_time deserialize_prev_mtime = { 0, 0 }; + +static int try_deserialize_read_from_file_1(const struct wt_status *cmd_s, + const char *path, + struct wt_status *des_s) +{ + struct stat st; + int result; + int fd; + + /* + * If we are spinning waiting for the status cache to become + * valid, skip re-reading it if the mtime has not changed + * since the last time we read it. + */ + if (lstat(path, &st)) { + trace_printf_key(&trace_deserialize, + "could not lstat '%s'", path); + return DESERIALIZE_ERR; + } + if (st.st_mtime == deserialize_prev_mtime.sec && + ST_MTIME_NSEC(st) == deserialize_prev_mtime.nsec) { + trace_printf_key(&trace_deserialize, + "mtime has not changed '%s'", path); + return DESERIALIZE_ERR; + } + + fd = xopen(path, O_RDONLY); + if (fd == -1) { + trace_printf_key(&trace_deserialize, + "could not read '%s'", path); + return DESERIALIZE_ERR; + } + + deserialize_prev_mtime.sec = st.st_mtime; + deserialize_prev_mtime.nsec = ST_MTIME_NSEC(st); + + trace_printf_key(&trace_deserialize, + "reading serialization file (%d %d) '%s'", + deserialize_prev_mtime.sec, + deserialize_prev_mtime.nsec, + path); + + result = wt_deserialize_fd(cmd_s, des_s, fd); + close(fd); + + return result; +} + +static int try_deserialize_read_from_file(const struct wt_status *cmd_s, + const char *path, + enum wt_status_deserialize_wait dw, + struct wt_status *des_s) +{ + int k, limit; + int result = DESERIALIZE_ERR; + + /* + * For "fail" or "no", try exactly once to read the status cache. + * Return an error if the file is stale. + */ + if (dw == DESERIALIZE_WAIT__FAIL || dw == DESERIALIZE_WAIT__NO) + return try_deserialize_read_from_file_1(cmd_s, path, des_s); + + /* + * Wait for the status cache file to refresh. Wait duration can + * be in tenths of a second or unlimited. Poll every 100ms. + */ + if (dw == DESERIALIZE_WAIT__BLOCK) { + /* + * Convert "unlimited" to 1 day. + */ + limit = 10 * 60 * 60 * 24; + } else { + /* spin for dw tenths of a second */ + limit = dw; + } + for (k = 0; k < limit; k++) { + result = try_deserialize_read_from_file_1( + cmd_s, path, des_s); + + if (result == DESERIALIZE_OK) + break; + + sleep_millisec(100); + } + + trace_printf_key(&trace_deserialize, + "wait polled=%d result=%d '%s'", + k, result, path); + return result; +} + /* - * Read raw serialized status data from the given file + * Read raw serialized status data from the given file (or STDIN). * * Verify that the args specified in the current command * are compatible with the deserialized data (such as "-uno"). @@ -687,24 +783,25 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de * Copy display-related fields from the current command * into the deserialized data (so that the user can request * long or short as they please). + * + * Print status report using cached data. */ int wt_status_deserialize(const struct wt_status *cmd_s, - const char *path) + const char *path, + enum wt_status_deserialize_wait dw) { struct wt_status des_s; int result; if (path && *path && strcmp(path, "0")) { - int fd = xopen(path, O_RDONLY); - if (fd == -1) { - trace_printf_key(&trace_deserialize, "could not read '%s'", path); - return DESERIALIZE_ERR; - } - trace_printf_key(&trace_deserialize, "reading serialization file '%s'", path); - result = wt_deserialize_fd(cmd_s, &des_s, fd); - close(fd); + result = try_deserialize_read_from_file(cmd_s, path, dw, &des_s); } else { trace_printf_key(&trace_deserialize, "reading stdin"); + + /* + * Read status cache data from stdin. Ignore the deserialize-wait + * term, since we cannot read stdin multiple times. + */ result = wt_deserialize_fd(cmd_s, &des_s, 0); } diff --git a/wt-status.h b/wt-status.h index 9728117f0e9217..b6cf4531fe56a4 100644 --- a/wt-status.h +++ b/wt-status.h @@ -220,6 +220,15 @@ struct wt_status_serialize_data - sizeof(struct wt_status_serialize_data_fixed)]; }; +enum wt_status_deserialize_wait +{ + DESERIALIZE_WAIT__UNSET = -3, + DESERIALIZE_WAIT__FAIL = -2, /* return error, do not fallback */ + DESERIALIZE_WAIT__BLOCK = -1, /* unlimited timeout */ + DESERIALIZE_WAIT__NO = 0, /* immediately fallback */ + /* any positive value is a timeout in tenths of a second */ +}; + /* * Serialize computed status scan results using "version 1" format * to the given file. @@ -234,7 +243,8 @@ void wt_status_serialize_v1(int fd, struct wt_status *s); * fields. */ int wt_status_deserialize(const struct wt_status *cmd_s, - const char *path); + const char *path, + enum wt_status_deserialize_wait dw); /* * A helper routine for serialize and deserialize to compute From 2f09a38fe2307ae0898ab232ec6e44d7e1dc8c99 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 21 May 2019 23:14:48 +0200 Subject: [PATCH 065/196] merge-recursive: avoid confusing logic in was_dirty() It took this developer more than a moment to verify that was_dirty() really returns 0 (i.e. "false") if the file was not even tracked. In other words, the `dirty` variable that was initialized to 1 (i.e. "true") and then negated to be returned was not helping readability. The same holds for the final return: rather than assigning the value to return to `dirty` and then *immediately* returning that, we can simplify it to a single statement. --- merge-recursive.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/merge-recursive.c b/merge-recursive.c index 1992c8c2b237d5..838949e2f429bb 100644 --- a/merge-recursive.c +++ b/merge-recursive.c @@ -874,15 +874,13 @@ static int would_lose_untracked(struct merge_options *opt, const char *path) static int was_dirty(struct merge_options *opt, const char *path) { struct cache_entry *ce; - int dirty = 1; if (opt->priv->call_depth || !was_tracked(opt, path)) - return !dirty; + return 0; ce = index_file_exists(opt->priv->unpack_opts.src_index, path, strlen(path), ignore_case); - dirty = verify_uptodate(ce, &opt->priv->unpack_opts) != 0; - return dirty; + return verify_uptodate(ce, &opt->priv->unpack_opts) != 0; } static int make_room_for_path(struct merge_options *opt, const char *path) From de4dcad3a316bc58011cbebc3cf1afcee198663b Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 21 May 2019 23:17:46 +0200 Subject: [PATCH 066/196] merge-recursive: add some defensive coding to was_dirty() It took this developer quite a good while to understand why the current code cannot get a `NULL` returned by `index_file_exists()`. To un-confuse readers (and future-proof the code), let's just be safe and check before we dereference the returned pointer. --- merge-recursive.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/merge-recursive.c b/merge-recursive.c index 838949e2f429bb..830620af016262 100644 --- a/merge-recursive.c +++ b/merge-recursive.c @@ -880,7 +880,7 @@ static int was_dirty(struct merge_options *opt, const char *path) ce = index_file_exists(opt->priv->unpack_opts.src_index, path, strlen(path), ignore_case); - return verify_uptodate(ce, &opt->priv->unpack_opts) != 0; + return !ce || verify_uptodate(ce, &opt->priv->unpack_opts) != 0; } static int make_room_for_path(struct merge_options *opt, const char *path) From eccf80179b3228e2cc6eae705fc2e46069e9feac Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 21 May 2019 23:20:16 +0200 Subject: [PATCH 067/196] merge-recursive: teach was_dirty() about the virtualfilesystem The idea of the virtual file system really is to tell Git to avoid accessing certain paths. This fixes the case where a given path is not yet included in the virtual file system and we are about to write a conflicted version of it. --- merge-recursive.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/merge-recursive.c b/merge-recursive.c index 830620af016262..8a8561446a5842 100644 --- a/merge-recursive.c +++ b/merge-recursive.c @@ -8,6 +8,7 @@ #include "git-compat-util.h" #include "merge-recursive.h" +#include "virtualfilesystem.h" #include "alloc.h" #include "cache-tree.h" @@ -875,7 +876,8 @@ static int was_dirty(struct merge_options *opt, const char *path) { struct cache_entry *ce; - if (opt->priv->call_depth || !was_tracked(opt, path)) + if (opt->priv->call_depth || !was_tracked(opt, path) || + is_excluded_from_virtualfilesystem(path, strlen(path), DT_REG) == 1) return 0; ce = index_file_exists(opt->priv->unpack_opts.src_index, From 1e410a610d2f29d1d36f5d621e0118cbe1dcfdcd Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 25 Jun 2019 16:38:50 -0400 Subject: [PATCH 068/196] status: deserialize with -uno does not print correct hint With the "--untracked-files=complete" option status computes a superset of the untracked files. We use this when writing the status cache. If subsequent deserialize commands ask for either the complete set or one of the "no", "normal", or "all" subsets, it can still use the cache file because of filtering in the deserialize parser. When running status with the "-uno" option, the long format status would print a "(use -u to show untracked files)" hint. When deserializing with the "-uno" option and using a cache computed with "-ucomplete", the "nothing to commit, working tree clean" message would be printed instead of the hint. It was easy to miss because the correct hint message was printed if the cache was rejected for any reason (and status did the full fallback). The "struct wt_status des" structure was initialized with the content of the status cache (and thus defaulted to "complete"). This change sets "des.show_untracked_files" to the requested subset from the command-line or config. This allows the long format to print the hint. Signed-off-by: Jeff Hostetler --- t/t7522-serialized-status.sh | 55 ++++++++++++++++++++++++++++++++++++ wt-status-deserialize.c | 16 +++++++---- 2 files changed, 65 insertions(+), 6 deletions(-) diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh index edf15d7af45489..b52a9b7fa2f520 100755 --- a/t/t7522-serialized-status.sh +++ b/t/t7522-serialized-status.sh @@ -345,4 +345,59 @@ test_expect_success 'renames' ' test_cmp output.1 output.2 ' +test_expect_success 'hint message when cached with u=complete' ' + git init -b main hint && + echo xxx >hint/xxx && + git -C hint add xxx && + git -C hint commit -m xxx && + + cat >expect.clean <expect.use_u <hint.output_normal && + test_cmp expect.clean hint.output_normal && + + git -C hint status --untracked-files=all >hint.output_all && + test_cmp expect.clean hint.output_all && + + git -C hint status --untracked-files=no >hint.output_no && + test_cmp expect.use_u hint.output_no && + + # Create long format output for "complete" and create status cache. + + git -C hint status --untracked-files=complete --ignored=matching --serialize=../hint.dat >hint.output_complete && + test_cmp expect.clean hint.output_complete && + + # Capture long format output using the status cache and verify + # that the output matches the non-cached version. There are 2 + # ways to specify untracked-files, so do them both. + + git -C hint status --deserialize=../hint.dat -unormal >hint.d1_normal && + test_cmp expect.clean hint.d1_normal && + git -C hint -c status.showuntrackedfiles=normal status --deserialize=../hint.dat >hint.d2_normal && + test_cmp expect.clean hint.d2_normal && + + git -C hint status --deserialize=../hint.dat -uall >hint.d1_all && + test_cmp expect.clean hint.d1_all && + git -C hint -c status.showuntrackedfiles=all status --deserialize=../hint.dat >hint.d2_all && + test_cmp expect.clean hint.d2_all && + + git -C hint status --deserialize=../hint.dat -uno >hint.d1_no && + test_cmp expect.use_u hint.d1_no && + git -C hint -c status.showuntrackedfiles=no status --deserialize=../hint.dat >hint.d2_no && + test_cmp expect.use_u hint.d2_no + +' + test_done diff --git a/wt-status-deserialize.c b/wt-status-deserialize.c index 5e77e4aaa82705..73cbe00902903f 100644 --- a/wt-status-deserialize.c +++ b/wt-status-deserialize.c @@ -424,20 +424,24 @@ static int wt_deserialize_v1_ignored_items(struct wt_status *s, } static int validate_untracked_files_arg(enum untracked_status_type cmd, - enum untracked_status_type des, + enum untracked_status_type *des, enum deserialize_parse_strategy *strategy) { *strategy = DESERIALIZE_STRATEGY_AS_IS; - if (cmd == des) { + if (cmd == *des) { *strategy = DESERIALIZE_STRATEGY_AS_IS; } else if (cmd == SHOW_NO_UNTRACKED_FILES) { *strategy = DESERIALIZE_STRATEGY_SKIP; - } else if (des == SHOW_COMPLETE_UNTRACKED_FILES) { - if (cmd == SHOW_ALL_UNTRACKED_FILES) + *des = cmd; + } else if (*des == SHOW_COMPLETE_UNTRACKED_FILES) { + if (cmd == SHOW_ALL_UNTRACKED_FILES) { *strategy = DESERIALIZE_STRATEGY_ALL; - else if (cmd == SHOW_NORMAL_UNTRACKED_FILES) + *des = cmd; + } else if (cmd == SHOW_NORMAL_UNTRACKED_FILES) { *strategy = DESERIALIZE_STRATEGY_NORMAL; + *des = cmd; + } } else { return DESERIALIZE_ERR; } @@ -479,7 +483,7 @@ static int wt_deserialize_v1(const struct wt_status *cmd_s, struct wt_status *s, * We now have the header parsed. Look at the command args (as passed in), and see how to parse * the serialized data */ - if (validate_untracked_files_arg(cmd_s->show_untracked_files, s->show_untracked_files, &untracked_strategy)) { + if (validate_untracked_files_arg(cmd_s->show_untracked_files, &s->show_untracked_files, &untracked_strategy)) { trace_printf_key(&trace_deserialize, "reject: show_untracked_file: command: %d, serialized : %d", cmd_s->show_untracked_files, s->show_untracked_files); From a1eee567b97510dba0d1ae84d0b1bb35742ac76e Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Thu, 21 Nov 2019 12:01:04 -0700 Subject: [PATCH 069/196] fsmonitor: check CE_FSMONITOR_VALID in ce_uptodate When using fsmonitor the CE_FSMONITOR_VALID flag should be checked when wanting to know if the entry has been updated. If the flag is set the entry should be considered up to date and the same as if the CE_UPTODATE is set. In order to trust the CE_FSMONITOR_VALID flag, the fsmonitor data needs to be refreshed when the fsmonitor bitmap is applied to the index in tweak_fsmonitor. Since the fsmonitor data is kept up to date for every command, some tests needed to be updated to take that into account. istate->untracked->use_fsmonitor was set in tweak_fsmonitor when the fsmonitor bitmap data was loaded and is now in refresh_fsmonitor since that is being called in tweak_fsmonitor. refresh_fsmonitor will only be called once and any other callers should be setting it when refreshing the fsmonitor data so that code can use the fsmonitor data when checking untracked files. When writing the index, fsmonitor_last_update is used to determine if the fsmonitor bitmap should be created and the extension data written to the index. When running through unpack-trees this is not copied to the result index. This makes the next time a git command is ran do all the work of lstating all files to determine what is clean since all entries in the index are marked as dirty since there wasn't any fsmonitor data saved in the index extension. Copying the fsmonitor_last_update to the result index will cause the extension data for fsmonitor to be in the index for the next git command to use. Signed-off-by: Kevin Willford --- read-cache-ll.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/read-cache-ll.h b/read-cache-ll.h index b5d11d07a86962..e87599c942b413 100644 --- a/read-cache-ll.h +++ b/read-cache-ll.h @@ -118,7 +118,7 @@ static inline unsigned create_ce_flags(unsigned stage) #define ce_namelen(ce) ((ce)->ce_namelen) #define ce_size(ce) cache_entry_size(ce_namelen(ce)) #define ce_stage(ce) ((CE_STAGEMASK & (ce)->ce_flags) >> CE_STAGESHIFT) -#define ce_uptodate(ce) ((ce)->ce_flags & CE_UPTODATE) +#define ce_uptodate(ce) (((ce)->ce_flags & CE_UPTODATE) || ((ce)->ce_flags & CE_FSMONITOR_VALID)) #define ce_skip_worktree(ce) ((ce)->ce_flags & CE_SKIP_WORKTREE) #define ce_mark_uptodate(ce) ((ce)->ce_flags |= CE_UPTODATE) #define ce_intent_to_add(ce) ((ce)->ce_flags & CE_INTENT_TO_ADD) From b0d651c5b27daee7c25b07249a38ca3dcedfb63e Mon Sep 17 00:00:00 2001 From: Kevin Willford Date: Thu, 21 Nov 2019 09:24:36 -0700 Subject: [PATCH 070/196] fsmonitor: add script for debugging and update script for tests The fsmonitor script that can be used for running all the git tests using watchman was causing some of the tests to fail because it wrote to stderr and created some files for debugging purposes. Add a new debug script to use with debugging and modify the other script to remove the code that would cause tests to fail. Signed-off-by: Kevin Willford --- t/t7519/fsmonitor-watchman | 22 +----- t/t7519/fsmonitor-watchman-debug | 128 +++++++++++++++++++++++++++++++ 2 files changed, 129 insertions(+), 21 deletions(-) create mode 100755 t/t7519/fsmonitor-watchman-debug diff --git a/t/t7519/fsmonitor-watchman b/t/t7519/fsmonitor-watchman index 264b9daf834ec8..6461f625f64181 100755 --- a/t/t7519/fsmonitor-watchman +++ b/t/t7519/fsmonitor-watchman @@ -17,7 +17,6 @@ use IPC::Open2; # 'git config core.fsmonitor .git/hooks/query-watchman' # my ($version, $time) = @ARGV; -#print STDERR "$0 $version $time\n"; # Check the hook interface version @@ -44,7 +43,7 @@ launch_watchman(); sub launch_watchman { - my $pid = open2(\*CHLD_OUT, \*CHLD_IN, 'watchman -j') + my $pid = open2(\*CHLD_OUT, \*CHLD_IN, 'watchman -j --no-pretty') or die "open2() failed: $!\n" . "Falling back to scanning...\n"; @@ -62,19 +61,11 @@ sub launch_watchman { "fields": ["name"] }] END - - open (my $fh, ">", ".git/watchman-query.json"); - print $fh $query; - close $fh; print CHLD_IN $query; close CHLD_IN; my $response = do {local $/; }; - open ($fh, ">", ".git/watchman-response.json"); - print $fh $response; - close $fh; - die "Watchman: command returned no output.\n" . "Falling back to scanning...\n" if $response eq ""; die "Watchman: command returned invalid output: $response\n" . @@ -93,7 +84,6 @@ sub launch_watchman { my $o = $json_pkg->new->utf8->decode($response); if ($retry > 0 and $o->{error} and $o->{error} =~ m/unable to resolve root .* directory (.*) is not watched/) { - print STDERR "Adding '$git_work_tree' to watchman's watch list.\n"; $retry--; qx/watchman watch "$git_work_tree"/; die "Failed to make watchman watch '$git_work_tree'.\n" . @@ -103,11 +93,6 @@ sub launch_watchman { # return the fast "everything is dirty" flag to git and do the # Watchman query just to get it over with now so we won't pay # the cost in git to look up each individual file. - - open ($fh, ">", ".git/watchman-output.out"); - print "/\0"; - close $fh; - print "/\0"; eval { launch_watchman() }; exit 0; @@ -116,11 +101,6 @@ sub launch_watchman { die "Watchman: $o->{error}.\n" . "Falling back to scanning...\n" if $o->{error}; - open ($fh, ">", ".git/watchman-output.out"); - binmode $fh, ":utf8"; - print $fh @{$o->{files}}; - close $fh; - binmode STDOUT, ":utf8"; local $, = "\0"; print @{$o->{files}}; diff --git a/t/t7519/fsmonitor-watchman-debug b/t/t7519/fsmonitor-watchman-debug new file mode 100755 index 00000000000000..d8e7a1e5ba85c0 --- /dev/null +++ b/t/t7519/fsmonitor-watchman-debug @@ -0,0 +1,128 @@ +#!/usr/bin/perl + +use strict; +use warnings; +use IPC::Open2; + +# An example hook script to integrate Watchman +# (https://facebook.github.io/watchman/) with git to speed up detecting +# new and modified files. +# +# The hook is passed a version (currently 1) and a time in nanoseconds +# formatted as a string and outputs to stdout all files that have been +# modified since the given time. Paths must be relative to the root of +# the working tree and separated by a single NUL. +# +# To enable this hook, rename this file to "query-watchman" and set +# 'git config core.fsmonitor .git/hooks/query-watchman' +# +my ($version, $time) = @ARGV; +#print STDERR "$0 $version $time\n"; + +# Check the hook interface version + +if ($version == 1) { + # convert nanoseconds to seconds + # subtract one second to make sure watchman will return all changes + $time = int ($time / 1000000000) - 1; +} else { + die "Unsupported query-fsmonitor hook version '$version'.\n" . + "Falling back to scanning...\n"; +} + +my $git_work_tree; +if ($^O =~ 'msys' || $^O =~ 'cygwin') { + $git_work_tree = Win32::GetCwd(); + $git_work_tree =~ tr/\\/\//; +} else { + require Cwd; + $git_work_tree = Cwd::cwd(); +} + +my $retry = 1; + +launch_watchman(); + +sub launch_watchman { + + my $pid = open2(\*CHLD_OUT, \*CHLD_IN, 'watchman -j') + or die "open2() failed: $!\n" . + "Falling back to scanning...\n"; + + # In the query expression below we're asking for names of files that + # changed since $time but were not transient (ie created after + # $time but no longer exist). + # + # To accomplish this, we're using the "since" generator to use the + # recency index to select candidate nodes and "fields" to limit the + # output to file names only. + + my $query = <<" END"; + ["query", "$git_work_tree", { + "since": $time, + "fields": ["name"] + }] + END + + open (my $fh, ">", ".git/watchman-query.json"); + print $fh $query; + close $fh; + + print CHLD_IN $query; + close CHLD_IN; + my $response = do {local $/; }; + + open ($fh, ">", ".git/watchman-response.json"); + print $fh $response; + close $fh; + + die "Watchman: command returned no output.\n" . + "Falling back to scanning...\n" if $response eq ""; + die "Watchman: command returned invalid output: $response\n" . + "Falling back to scanning...\n" unless $response =~ /^\{/; + + my $json_pkg; + eval { + require JSON::XS; + $json_pkg = "JSON::XS"; + 1; + } or do { + require JSON::PP; + $json_pkg = "JSON::PP"; + }; + + my $o = $json_pkg->new->utf8->decode($response); + + if ($retry > 0 and $o->{error} and $o->{error} =~ m/unable to resolve root .* directory (.*) is not watched/) { + print STDERR "Adding '$git_work_tree' to watchman's watch list.\n"; + $retry--; + qx/watchman watch "$git_work_tree"/; + die "Failed to make watchman watch '$git_work_tree'.\n" . + "Falling back to scanning...\n" if $? != 0; + + # Watchman will always return all files on the first query so + # return the fast "everything is dirty" flag to git and do the + # Watchman query just to get it over with now so we won't pay + # the cost in git to look up each individual file. + + open ($fh, ">", ".git/watchman-output.out"); + print "/\0"; + close $fh; + + print "/\0"; + eval { launch_watchman() }; + exit 0; + } + + die "Watchman: $o->{error}.\n" . + "Falling back to scanning...\n" if $o->{error}; + + open ($fh, ">", ".git/watchman-output.out"); + binmode $fh, ":utf8"; + print $fh @{$o->{files}}; + close $fh; + + binmode STDOUT, ":utf8"; + local $, = "\0"; + print @{$o->{files}}; +} From d9b0c4de9d7922133130f5ae1de743caa2f37076 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 10 Apr 2020 21:14:44 -0400 Subject: [PATCH 071/196] status: disable deserialize when verbose output requested. Disable deserialization when verbose output requested. Verbose mode causes Git to print diffs for modified files. This requires the index to be loaded to have the currently staged OID values. Without loading the index, verbose output make it look like everything was deleted. Signed-off-by: Jeff Hostetler --- builtin/commit.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/builtin/commit.c b/builtin/commit.c index f2eaad52afb70f..c1f5e7d3f4a886 100644 --- a/builtin/commit.c +++ b/builtin/commit.c @@ -39,6 +39,7 @@ #include "commit-graph.h" #include "pretty.h" #include "trailer.h" +#include "trace2.h" static const char * const builtin_commit_usage[] = { N_("git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" @@ -1729,6 +1730,22 @@ struct repository *repo UNUSED) */ try_deserialize = (!do_serialize && (do_implicit_deserialize || do_explicit_deserialize)); + + /* + * Disable deserialize when verbose is set because it causes us to + * print diffs for each modified file, but that requires us to have + * the index loaded and we don't want to do that (at least not now for + * this seldom used feature). My fear is that would further tangle + * the merge conflict with upstream. + * + * TODO Reconsider this in the future. + */ + if (try_deserialize && verbose) { + trace2_data_string("status", the_repository, "deserialize/reject", + "args/verbose"); + try_deserialize = 0; + } + if (try_deserialize) goto skip_init; From 7ebc9cfe25793e6ead42ea18d4b886588b477e55 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 10 Apr 2020 21:18:41 -0400 Subject: [PATCH 072/196] t7524: add test for verbose status deserialzation Verify that `git status --deserialize=x -v` does not crash and generates the same output as a normal (scanning) status command. These issues are described in the previous 2 commits. Signed-off-by: Jeff Hostetler --- t/t7522-serialized-status.sh | 39 ++++++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+) diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh index b52a9b7fa2f520..6010fcd31635cd 100755 --- a/t/t7522-serialized-status.sh +++ b/t/t7522-serialized-status.sh @@ -400,4 +400,43 @@ EOF ' +test_expect_success 'ensure deserialize -v does not crash' ' + + git init -b main verbose_test && + touch verbose_test/a && + touch verbose_test/b && + touch verbose_test/c && + git -C verbose_test add a b c && + git -C verbose_test commit -m abc && + + echo green >>verbose_test/a && + git -C verbose_test add a && + echo red_1 >>verbose_test/b && + echo red_2 >verbose_test/dirt && + + git -C verbose_test status >output.ref && + git -C verbose_test status -v >output.ref_v && + + git -C verbose_test --no-optional-locks status --serialize=../verbose_test.dat >output.ser.long && + git -C verbose_test --no-optional-locks status --serialize=../verbose_test.dat_v -v >output.ser.long_v && + + # Verify that serialization does not affect the status output itself. + test_cmp output.ref output.ser.long && + test_cmp output.ref_v output.ser.long_v && + + GIT_TRACE2_PERF="$(pwd)"/verbose_test.log \ + git -C verbose_test status --deserialize=../verbose_test.dat >output.des.long && + + # Verify that normal deserialize was actually used and produces the same result. + test_cmp output.ser.long output.des.long && + grep -q "deserialize/result:ok" verbose_test.log && + + GIT_TRACE2_PERF="$(pwd)"/verbose_test.log_v \ + git -C verbose_test status --deserialize=../verbose_test.dat_v -v >output.des.long_v && + + # Verify that vebose mode produces the same result because verbose was rejected. + test_cmp output.ser.long_v output.des.long_v && + grep -q "deserialize/reject:args/verbose" verbose_test.log_v +' + test_done From 1518949888eb663861b79810efa7a0e362adf558 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 26 Sep 2018 12:29:26 -0400 Subject: [PATCH 073/196] gvfs:trace2:data: add trace2 tracing around read_object_process Add trace2 region around read_object_process to collect time spent waiting for missing objects to be dynamically fetched. Signed-off-by: Jeff Hostetler --- object-file.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/object-file.c b/object-file.c index 7a9574c09e17be..5144a339a8fd2b 100644 --- a/object-file.c +++ b/object-file.c @@ -41,6 +41,7 @@ #include "loose.h" #include "object-file-convert.h" #include "trace.h" +#include "trace2.h" #include "hook.h" #include "sigchain.h" #include "sub-process.h" @@ -1065,6 +1066,8 @@ static int read_object_process(const struct object_id *oid) start = getnanotime(); + trace2_region_enter("subprocess", "read_object", the_repository); + if (!subprocess_map_initialized) { subprocess_map_initialized = 1; hashmap_init(&subprocess_map, (hashmap_cmp_fn)cmd2process_cmp, @@ -1081,13 +1084,16 @@ static int read_object_process(const struct object_id *oid) if (subprocess_start(&subprocess_map, &entry->subprocess, cmd, start_read_object_fn)) { free(entry); - return -1; + err = -1; + goto leave_region; } } process = &entry->subprocess.process; - if (!(CAP_GET & entry->supported_capabilities)) - return -1; + if (!(CAP_GET & entry->supported_capabilities)) { + err = -1; + goto leave_region; + } sigchain_push(SIGPIPE, SIG_IGN); @@ -1136,6 +1142,10 @@ static int read_object_process(const struct object_id *oid) trace_performance_since(start, "read_object_process"); +leave_region: + trace2_region_leave_printf("subprocess", "read_object", the_repository, + "result %d", err); + return err; } From 894c6fa4dd87f0bae4836391feba284fef1ffbd1 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 13 May 2020 17:38:50 -0400 Subject: [PATCH 074/196] deserialize-status: silently fallback if we cannot read cache file Teach Git to not throw a fatal error when an explicitly-specified status-cache file (`git status --deserialize=`) could not be found or opened for reading and silently fallback to a traditional scan. This matches the behavior when the status-cache file is implicitly given via a config setting. Note: the current version causes a test to start failing. Mark this as an expected result for now. Signed-off-by: Jeff Hostetler Signed-off-by: Derrick Stolee --- builtin/commit.c | 18 ++++++++++++------ t/t7522-serialized-status.sh | 16 ++++++++++++++++ 2 files changed, 28 insertions(+), 6 deletions(-) diff --git a/builtin/commit.c b/builtin/commit.c index c1f5e7d3f4a886..b66eafac515736 100644 --- a/builtin/commit.c +++ b/builtin/commit.c @@ -226,12 +226,18 @@ static int opt_parse_deserialize(const struct option *opt UNUSED, const char *ar free(deserialize_path); deserialize_path = xstrdup(arg); } - if (deserialize_path && *deserialize_path - && (access(deserialize_path, R_OK) != 0)) - die("cannot find serialization file '%s'", - deserialize_path); - - do_explicit_deserialize = 1; + if (!deserialize_path || !*deserialize_path) + do_explicit_deserialize = 1; /* read stdin */ + else if (access(deserialize_path, R_OK) == 0) + do_explicit_deserialize = 1; /* can read from this file */ + else { + /* + * otherwise, silently fallback to the normal + * collection scan + */ + do_implicit_deserialize = 0; + do_explicit_deserialize = 0; + } } return 0; diff --git a/t/t7522-serialized-status.sh b/t/t7522-serialized-status.sh index 6010fcd31635cd..230e1e24cfc1c4 100755 --- a/t/t7522-serialized-status.sh +++ b/t/t7522-serialized-status.sh @@ -439,4 +439,20 @@ test_expect_success 'ensure deserialize -v does not crash' ' grep -q "deserialize/reject:args/verbose" verbose_test.log_v ' +test_expect_success 'fallback when implicit' ' + git init -b main implicit_fallback_test && + git -C implicit_fallback_test -c status.deserializepath=foobar status +' + +test_expect_success 'fallback when explicit' ' + git init -b main explicit_fallback_test && + git -C explicit_fallback_test status --deserialize=foobar +' + +test_expect_success 'deserialize from stdin' ' + git init -b main stdin_test && + git -C stdin_test status --serialize >serialized_status.dat && + cat serialize_status.dat | git -C stdin_test status --deserialize +' + test_done From 059147588e8fc3eca9b0237aa85b982ea56f2ba2 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Wed, 26 Sep 2018 11:21:22 -0400 Subject: [PATCH 075/196] gvfs:trace2:data: status deserialization information Add trace2 region and data events describing attempts to deserialize status data using a status cache. A category:status, label:deserialize region is pushed around the deserialize code. Deserialization results when reading from a file are: category:status, path = category:status, polled = category:status, result = "ok" | "reject" When reading from STDIN are: category:status, path = "STDIN" category:status, result = "ok" | "reject" Status will fallback and run a normal status scan when a "reject" is reported (unless "--deserialize-wait=fail"). If "ok" is reported, status was able to use the status cache and avoid scanning the workdir. Additionally, a cmd_mode is emitted for each step: collection, deserialization, and serialization. For example, if deserialization is attempted and fails and status falls back to actually computing the status, a cmd_mode message containing "deserialize" is issued and then a cmd_mode for "collect" is issued. Also, if deserialization fails, a data message containing the rejection reason is emitted. Signed-off-by: Jeff Hostetler --- builtin/commit.c | 19 +++++++++++- wt-status-deserialize.c | 67 ++++++++++++++++++++++++++++++++++++++--- wt-status.h | 2 ++ 3 files changed, 83 insertions(+), 5 deletions(-) diff --git a/builtin/commit.c b/builtin/commit.c index b66eafac515736..b552c940ee2ee7 100644 --- a/builtin/commit.c +++ b/builtin/commit.c @@ -165,6 +165,7 @@ static int opt_parse_porcelain(const struct option *opt, const char *arg, int un static int do_serialize = 0; static char *serialize_path = NULL; +static int reject_implicit = 0; static int do_implicit_deserialize = 0; static int do_explicit_deserialize = 0; static char *deserialize_path = NULL; @@ -228,7 +229,7 @@ static int opt_parse_deserialize(const struct option *opt UNUSED, const char *ar } if (!deserialize_path || !*deserialize_path) do_explicit_deserialize = 1; /* read stdin */ - else if (access(deserialize_path, R_OK) == 0) + else if (wt_status_deserialize_access(deserialize_path, R_OK) == 0) do_explicit_deserialize = 1; /* can read from this file */ else { /* @@ -1601,6 +1602,8 @@ static int git_status_config(const char *k, const char *v, if (v && *v && access(v, R_OK) == 0) { do_implicit_deserialize = 1; deserialize_path = xstrdup(v); + } else { + reject_implicit = 1; } return 0; } @@ -1754,6 +1757,17 @@ struct repository *repo UNUSED) if (try_deserialize) goto skip_init; + /* + * If we implicitly received a status cache pathname from the config + * and the file does not exist, we silently reject it and do the normal + * status "collect". Fake up some trace2 messages to reflect this and + * assist post-processors know this case is different. + */ + if (!do_serialize && reject_implicit) { + trace2_cmd_mode("implicit-deserialize"); + trace2_data_string("status", the_repository, "deserialize/reject", + "status-cache/access"); + } enable_fscache(0); if (status_format != STATUS_FORMAT_PORCELAIN && @@ -1797,6 +1811,7 @@ struct repository *repo UNUSED) if (s.relative_paths) s.prefix = prefix; + trace2_cmd_mode("deserialize"); result = wt_status_deserialize(&s, deserialize_path, dw); if (result == DESERIALIZE_OK) return 0; @@ -1814,6 +1829,7 @@ struct repository *repo UNUSED) fd = -1; } + trace2_cmd_mode("collect"); wt_status_collect(&s); if (0 <= fd) @@ -1828,6 +1844,7 @@ struct repository *repo UNUSED) if (fd_serialize < 0) die_errno(_("could not serialize to '%s'"), serialize_path); + trace2_cmd_mode("serialize"); wt_status_serialize_v1(fd_serialize, &s); close(fd_serialize); } diff --git a/wt-status-deserialize.c b/wt-status-deserialize.c index 73cbe00902903f..39c2673358b218 100644 --- a/wt-status-deserialize.c +++ b/wt-status-deserialize.c @@ -9,6 +9,23 @@ #include "statinfo.h" #include "hex.h" #include "path.h" +#include "trace2.h" + +static void set_deserialize_reject_reason(const char *reason) +{ + trace2_data_string("status", the_repository, "deserialize/reject", + reason); +} + +int wt_status_deserialize_access(const char *path, int mode) +{ + int a = access(path, mode); + + if (a != 0) + set_deserialize_reject_reason("status-cache/access"); + + return a; +} static struct trace_key trace_deserialize = TRACE_KEY_INIT(DESERIALIZE); @@ -55,6 +72,7 @@ static int my_validate_index(const char *path, const struct cache_time *mtime_re struct cache_time mtime_observed_on_disk; if (lstat(path, &st)) { + set_deserialize_reject_reason("index/not-found"); trace_printf_key(&trace_deserialize, "could not stat index"); return DESERIALIZE_ERR; } @@ -62,6 +80,7 @@ static int my_validate_index(const char *path, const struct cache_time *mtime_re mtime_observed_on_disk.nsec = ST_MTIME_NSEC(st); if ((mtime_observed_on_disk.sec != mtime_reported->sec) || (mtime_observed_on_disk.nsec != mtime_reported->nsec)) { + set_deserialize_reject_reason("index/mtime-changed"); trace_printf_key(&trace_deserialize, "index mtime changed [des %d %d][obs %d %d]", mtime_reported->sec, mtime_reported->nsec, @@ -87,10 +106,12 @@ static int my_validate_excludes(const char *path, const char *key, const char *l r = (strcmp(line, sb.buf) ? DESERIALIZE_ERR : DESERIALIZE_OK); - if (r == DESERIALIZE_ERR) + if (r == DESERIALIZE_ERR) { + set_deserialize_reject_reason("excludes/changed"); trace_printf_key(&trace_deserialize, "%s changed [cached '%s'][observed '%s']", key, line, sb.buf); + } strbuf_release(&sb); return r; @@ -146,6 +167,7 @@ static int wt_deserialize_v1_header(struct wt_status *s, int fd) &index_mtime.sec, &index_mtime.nsec); if (nr_fields != 2) { + set_deserialize_reject_reason("v1-header/invalid-index-mtime"); trace_printf_key(&trace_deserialize, "invalid index_mtime (%d) '%s'", nr_fields, line); return DESERIALIZE_ERR; @@ -229,6 +251,7 @@ static int wt_deserialize_v1_header(struct wt_status *s, int fd) /* status_format */ if (skip_prefix(line, "sha1_commit ", &arg)) { if (get_oid_hex(arg, &s->oid_commit)) { + set_deserialize_reject_reason("v1-header/invalid-commit-sha"); trace_printf_key(&trace_deserialize, "invalid sha1_commit"); return DESERIALIZE_ERR; } @@ -244,19 +267,23 @@ static int wt_deserialize_v1_header(struct wt_status *s, int fd) } /* prefix */ + set_deserialize_reject_reason("v1-header/unexpected-line"); trace_printf_key(&trace_deserialize, "unexpected line '%s'", line); return DESERIALIZE_ERR; } if (!have_required_index_mtime) { + set_deserialize_reject_reason("v1-header/missing-index-mtime"); trace_printf_key(&trace_deserialize, "missing '%s'", "index_mtime"); return DESERIALIZE_ERR; } if (!have_required_core_excludes) { + set_deserialize_reject_reason("v1-header/missing-core-excludes"); trace_printf_key(&trace_deserialize, "missing '%s'", "core_excludes"); return DESERIALIZE_ERR; } if (!have_required_repo_excludes) { + set_deserialize_reject_reason("v1-header/missing-repo-excludes"); trace_printf_key(&trace_deserialize, "missing '%s'", "repo_excludes"); return DESERIALIZE_ERR; } @@ -343,6 +370,7 @@ static int wt_deserialize_v1_changed_items(const struct wt_status *cmd_s, * So we reject the status cache and let the fallback * code run. */ + set_deserialize_reject_reason("v1-data/unmerged"); trace_printf_key( &trace_deserialize, "reject: V2 format and umerged file: %s", @@ -484,6 +512,7 @@ static int wt_deserialize_v1(const struct wt_status *cmd_s, struct wt_status *s, * the serialized data */ if (validate_untracked_files_arg(cmd_s->show_untracked_files, &s->show_untracked_files, &untracked_strategy)) { + set_deserialize_reject_reason("args/untracked-files"); trace_printf_key(&trace_deserialize, "reject: show_untracked_file: command: %d, serialized : %d", cmd_s->show_untracked_files, s->show_untracked_files); @@ -491,6 +520,7 @@ static int wt_deserialize_v1(const struct wt_status *cmd_s, struct wt_status *s, } if (validate_ignored_files_arg(cmd_s->show_ignored_mode, s->show_ignored_mode, &ignored_strategy)) { + set_deserialize_reject_reason("args/ignored-mode"); trace_printf_key(&trace_deserialize, "reject: show_ignored_mode: command: %d, serialized: %d", cmd_s->show_ignored_mode, s->show_ignored_mode); @@ -524,6 +554,7 @@ static int wt_deserialize_v1(const struct wt_status *cmd_s, struct wt_status *s, return DESERIALIZE_ERR; continue; } + set_deserialize_reject_reason("v1-data/unexpected-line"); trace_printf_key(&trace_deserialize, "unexpected line '%s'", line); return DESERIALIZE_ERR; } @@ -545,6 +576,7 @@ static int wt_deserialize_parse(const struct wt_status *cmd_s, struct wt_status if (version == 1) return wt_deserialize_v1(cmd_s, s, fd); } + set_deserialize_reject_reason("status-cache/unsupported-version"); trace_printf_key(&trace_deserialize, "missing/unsupported version"); return DESERIALIZE_ERR; } @@ -565,6 +597,7 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de * Check the path spec on the current command */ if (cmd_s->pathspec.nr > 1) { + set_deserialize_reject_reason("args/multiple-pathspecs"); trace_printf_key(&trace_deserialize, "reject: multiple pathspecs"); return DESERIALIZE_ERR; } @@ -575,6 +608,7 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de */ if (cmd_s->pathspec.nr == 1 && my_strcmp_null(cmd_s->pathspec.items[0].match, "")) { + set_deserialize_reject_reason("args/root-pathspec"); trace_printf_key(&trace_deserialize, "reject: pathspec"); return DESERIALIZE_ERR; } @@ -591,20 +625,24 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de * or "--ignored" settings). */ if (cmd_s->is_initial != des_s->is_initial) { + set_deserialize_reject_reason("args/is-initial-changed"); trace_printf_key(&trace_deserialize, "reject: is_initial"); return DESERIALIZE_ERR; } if (my_strcmp_null(cmd_s->branch, des_s->branch)) { + set_deserialize_reject_reason("args/branch-changed"); trace_printf_key(&trace_deserialize, "reject: branch"); return DESERIALIZE_ERR; } if (my_strcmp_null(cmd_s->reference, des_s->reference)) { + set_deserialize_reject_reason("args/reference-changed"); trace_printf_key(&trace_deserialize, "reject: reference"); return DESERIALIZE_ERR; } /* verbose */ /* amend */ if (cmd_s->whence != des_s->whence) { + set_deserialize_reject_reason("args/whence-changed"); trace_printf_key(&trace_deserialize, "reject: whence"); return DESERIALIZE_ERR; } @@ -638,19 +676,23 @@ static int wt_deserialize_fd(const struct wt_status *cmd_s, struct wt_status *de /* hints */ /* ahead_behind_flags */ if (cmd_s->detect_rename != des_s->detect_rename) { + set_deserialize_reject_reason("args/detect-rename-changed"); trace_printf_key(&trace_deserialize, "reject: detect_rename"); return DESERIALIZE_ERR; } if (cmd_s->rename_score != des_s->rename_score) { + set_deserialize_reject_reason("args/rename-score-changed"); trace_printf_key(&trace_deserialize, "reject: rename_score"); return DESERIALIZE_ERR; } if (cmd_s->rename_limit != des_s->rename_limit) { + set_deserialize_reject_reason("args/rename-limit-changed"); trace_printf_key(&trace_deserialize, "reject: rename_limit"); return DESERIALIZE_ERR; } /* status_format */ if (!oideq(&cmd_s->oid_commit, &des_s->oid_commit)) { + set_deserialize_reject_reason("args/commit-changed"); trace_printf_key(&trace_deserialize, "reject: sha1_commit"); return DESERIALIZE_ERR; } @@ -739,15 +781,18 @@ static int try_deserialize_read_from_file(const struct wt_status *cmd_s, enum wt_status_deserialize_wait dw, struct wt_status *des_s) { - int k, limit; + int k = 0; + int limit; int result = DESERIALIZE_ERR; /* * For "fail" or "no", try exactly once to read the status cache. * Return an error if the file is stale. */ - if (dw == DESERIALIZE_WAIT__FAIL || dw == DESERIALIZE_WAIT__NO) - return try_deserialize_read_from_file_1(cmd_s, path, des_s); + if (dw == DESERIALIZE_WAIT__FAIL || dw == DESERIALIZE_WAIT__NO) { + result = try_deserialize_read_from_file_1(cmd_s, path, des_s); + goto done; + } /* * Wait for the status cache file to refresh. Wait duration can @@ -772,6 +817,12 @@ static int try_deserialize_read_from_file(const struct wt_status *cmd_s, sleep_millisec(100); } +done: + trace2_data_string("status", the_repository, "deserialize/path", path); + trace2_data_intmax("status", the_repository, "deserialize/polled", k); + trace2_data_string("status", the_repository, "deserialize/result", + ((result == DESERIALIZE_OK) ? "ok" : "reject")); + trace_printf_key(&trace_deserialize, "wait polled=%d result=%d '%s'", k, result, path); @@ -797,6 +848,8 @@ int wt_status_deserialize(const struct wt_status *cmd_s, struct wt_status des_s; int result; + trace2_region_enter("status", "deserialize", the_repository); + if (path && *path && strcmp(path, "0")) { result = try_deserialize_read_from_file(cmd_s, path, dw, &des_s); } else { @@ -807,8 +860,14 @@ int wt_status_deserialize(const struct wt_status *cmd_s, * term, since we cannot read stdin multiple times. */ result = wt_deserialize_fd(cmd_s, &des_s, 0); + + trace2_data_string("status", the_repository, "deserialize/path", "STDIN"); + trace2_data_string("status", the_repository, "deserialize/result", + ((result == DESERIALIZE_OK) ? "ok" : "reject")); } + trace2_region_leave("status", "deserialize", the_repository); + if (result == DESERIALIZE_OK) { wt_status_get_state(cmd_s->repo, &des_s.state, des_s.branch && !strcmp(des_s.branch, "HEAD")); diff --git a/wt-status.h b/wt-status.h index b6cf4531fe56a4..03c25186491990 100644 --- a/wt-status.h +++ b/wt-status.h @@ -246,6 +246,8 @@ int wt_status_deserialize(const struct wt_status *cmd_s, const char *path, enum wt_status_deserialize_wait dw); +int wt_status_deserialize_access(const char *path, int mode); + /* * A helper routine for serialize and deserialize to compute * metadata for the user-global and repo-local excludes files. From d7543276acb73d84461cc36032da7b98870a1fcd Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 7 Jan 2019 12:45:48 -0500 Subject: [PATCH 076/196] gvfs:trace2:data: status serialization Add trace information around status serialization. Signed-off-by: Jeff Hostetler --- wt-status-serialize.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/wt-status-serialize.c b/wt-status-serialize.c index beaa7bd565cf2c..d5f8baa63efd66 100644 --- a/wt-status-serialize.c +++ b/wt-status-serialize.c @@ -9,6 +9,7 @@ #include "trace.h" #include "read-cache-ll.h" #include "path.h" +#include "trace2.h" static struct trace_key trace_serialize = TRACE_KEY_INIT(SERIALIZE); @@ -305,6 +306,8 @@ void wt_status_serialize_v1(int fd, struct wt_status *s) struct string_list_item *iter; int k; + trace2_region_enter("status", "serialize", the_repository); + /* * version header must be first line. */ @@ -338,4 +341,6 @@ void wt_status_serialize_v1(int fd, struct wt_status *s) } packet_flush(fd); } + + trace2_region_leave("status", "serialize", the_repository); } From b05adb373d08b1fa0718ccb5931a3cd794669317 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 19 Nov 2018 16:26:37 -0500 Subject: [PATCH 077/196] gvfs:trace2:data: add vfs stats Report virtual filesystem summary data. Signed-off-by: Jeff Hostetler --- virtualfilesystem.c | 38 ++++++++++++++++++++++++++++++++++++-- 1 file changed, 36 insertions(+), 2 deletions(-) diff --git a/virtualfilesystem.c b/virtualfilesystem.c index ae191ad05adf4c..20810c541558ce 100644 --- a/virtualfilesystem.c +++ b/virtualfilesystem.c @@ -3,6 +3,7 @@ #include "git-compat-util.h" #include "environment.h" #include "gettext.h" +#include "trace2.h" #include "config.h" #include "dir.h" #include "hashmap.h" @@ -260,6 +261,11 @@ void apply_virtualfilesystem(struct index_state *istate) { char *buf, *entry; int i; + int nr_unknown = 0; + int nr_vfs_dirs = 0; + int nr_vfs_rows = 0; + int nr_bulk_skip = 0; + int nr_explicit_skip = 0; if (!repo_config_get_virtualfilesystem(istate->repo)) return; @@ -277,16 +283,21 @@ void apply_virtualfilesystem(struct index_state *istate) if (buf[i] == '\0') { int pos, len; + nr_vfs_rows++; + len = buf + i - entry; /* look for a directory wild card (ie "dir1/") */ if (buf[i - 1] == '/') { + nr_vfs_dirs++; if (ignore_case) adjust_dirname_case(istate, entry); pos = index_name_pos(istate, entry, len); if (pos < 0) { pos = -pos - 1; while (pos < istate->cache_nr && !fspathncmp(istate->cache[pos]->name, entry, len)) { + if (istate->cache[pos]->ce_flags & CE_SKIP_WORKTREE) + nr_bulk_skip++; istate->cache[pos]->ce_flags &= ~CE_SKIP_WORKTREE; pos++; } @@ -294,18 +305,41 @@ void apply_virtualfilesystem(struct index_state *istate) } else { if (ignore_case) { struct cache_entry *ce = index_file_exists(istate, entry, len, ignore_case); - if (ce) + if (ce) { + if (ce->ce_flags & CE_SKIP_WORKTREE) + nr_explicit_skip++; ce->ce_flags &= ~CE_SKIP_WORKTREE; + } + else { + nr_unknown++; + } } else { int pos = index_name_pos(istate, entry, len); - if (pos >= 0) + if (pos >= 0) { + if (istate->cache[pos]->ce_flags & CE_SKIP_WORKTREE) + nr_explicit_skip++; istate->cache[pos]->ce_flags &= ~CE_SKIP_WORKTREE; + } + else { + nr_unknown++; + } } } entry += len + 1; } } + + if (nr_vfs_rows > 0) { + trace2_data_intmax("vfs", the_repository, "apply/tracked", nr_bulk_skip + nr_explicit_skip); + + trace2_data_intmax("vfs", the_repository, "apply/vfs_rows", nr_vfs_rows); + trace2_data_intmax("vfs", the_repository, "apply/vfs_dirs", nr_vfs_dirs); + + trace2_data_intmax("vfs", the_repository, "apply/nr_unknown", nr_unknown); + trace2_data_intmax("vfs", the_repository, "apply/nr_bulk_skip", nr_bulk_skip); + trace2_data_intmax("vfs", the_repository, "apply/nr_explicit_skip", nr_explicit_skip); + } } /* From 9f20d58a465b5b5949635744a54e7b6e41a7924e Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 15 Apr 2019 13:39:43 -0700 Subject: [PATCH 078/196] trace2: refactor setting process starting time Create trace2_initialize_clock() and call from main() to capture process start time in isolation and before other sub-systems are ready. Signed-off-by: Jeff Hostetler Signed-off-by: Junio C Hamano --- compat/mingw.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/compat/mingw.c b/compat/mingw.c index c1030e40f227d2..a76d9ef207867c 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -4090,6 +4090,8 @@ int wmain(int argc, const wchar_t **wargv) SetConsoleCtrlHandler(handle_ctrl_c, TRUE); + trace2_initialize_clock(); + maybe_redirect_std_handles(); adjust_symlink_flags(); fsync_object_files = 1; From 4728e1652dc6273ec8e686655c221ed7492a68d1 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 30 Apr 2019 14:12:51 -0400 Subject: [PATCH 079/196] trace2:gvfs:experiment: clear_ce_flags_1 Signed-off-by: Jeff Hostetler --- unpack-trees.c | 1 + 1 file changed, 1 insertion(+) diff --git a/unpack-trees.c b/unpack-trees.c index 47affd038d5dff..c881eb6a29f5c9 100644 --- a/unpack-trees.c +++ b/unpack-trees.c @@ -1788,6 +1788,7 @@ static int clear_ce_flags(struct index_state *istate, xsnprintf(label, sizeof(label), "clear_ce_flags(0x%08lx,0x%08lx)", (unsigned long)select_mask, (unsigned long)clear_mask); trace2_region_enter("unpack_trees", label, the_repository); + rval = clear_ce_flags_1(istate, istate->cache, istate->cache_nr, From e5c9e72e268624ae93d3f9e9dc63d5ab70319ceb Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 30 Apr 2019 16:02:39 -0400 Subject: [PATCH 080/196] trace2:gvfs:experiment: report_tracking Signed-off-by: Jeff Hostetler --- builtin/checkout.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/builtin/checkout.c b/builtin/checkout.c index eb811de04c1792..9e26f1cf15fb6d 100644 --- a/builtin/checkout.c +++ b/builtin/checkout.c @@ -1047,8 +1047,11 @@ static void update_refs_for_switch(const struct checkout_opts *opts, strbuf_release(&msg); if (!opts->quiet && !opts->force_detach && - (new_branch_info->path || !strcmp(new_branch_info->name, "HEAD"))) + (new_branch_info->path || !strcmp(new_branch_info->name, "HEAD"))) { + trace2_region_enter("exp", "report_tracking", the_repository); report_tracking(new_branch_info); + trace2_region_leave("exp", "report_tracking", the_repository); + } } static int add_pending_uninteresting_ref(const char *refname, const char *referent UNUSED, From 310ca5b76c9b617e4006cdb0c17a269baa40379b Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 14 Jun 2019 12:38:31 -0400 Subject: [PATCH 081/196] trace2:gvfs:experiment: read_cache: annotate thread usage in read-cache Add trace2_thread_start() and trace2_thread_exit() events to the worker threads used to read the index. This gives per-thread perf data. These workers were introduced in: abb4bb83845 read-cache: load cache extensions on a worker thread 77ff1127a4c read-cache: load cache entries on worker threads Signed-off-by: Jeff Hostetler --- read-cache.c | 18 +++++++++++++++++- 1 file changed, 17 insertions(+), 1 deletion(-) diff --git a/read-cache.c b/read-cache.c index 6c46ce61d2a50d..2d5a83157223a6 100644 --- a/read-cache.c +++ b/read-cache.c @@ -2053,6 +2053,17 @@ static void *load_index_extensions(void *_data) return NULL; } +static void *load_index_extensions_threadproc(void *_data) +{ + void *result; + + trace2_thread_start("load_index_extensions"); + result = load_index_extensions(_data); + trace2_thread_exit(); + + return result; +} + /* * A helper function that will load the specified range of cache entries * from the memory mapped file and add them to the given index. @@ -2129,12 +2140,17 @@ static void *load_cache_entries_thread(void *_data) struct load_cache_entries_thread_data *p = _data; int i; + trace2_thread_start("load_cache_entries"); + /* iterate across all ieot blocks assigned to this thread */ for (i = p->ieot_start; i < p->ieot_start + p->ieot_blocks; i++) { p->consumed += load_cache_entry_block(p->istate, p->ce_mem_pool, p->offset, p->ieot->entries[i].nr, p->mmap, p->ieot->entries[i].offset, NULL); p->offset += p->ieot->entries[i].nr; } + + trace2_thread_exit(); + return NULL; } @@ -2304,7 +2320,7 @@ int do_read_index(struct index_state *istate, const char *path, int must_exist) int err; p.src_offset = extension_offset; - err = pthread_create(&p.pthread, NULL, load_index_extensions, &p); + err = pthread_create(&p.pthread, NULL, load_index_extensions_threadproc, &p); if (err) die(_("unable to create load_index_extensions thread: %s"), strerror(err)); From dc05480ae7b8e1c9f52f3e68117213d7aaf5525d Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 9 Jul 2019 14:43:47 -0400 Subject: [PATCH 082/196] trace2:gvfs:experiment: read-cache: time read/write of cache-tree extension Add regions around code to read and write the cache-tree extension when the index is read or written. This is an experiment and may be dropped in future releases if we don't need it anymore. This experiment demonstrates that it takes more time to parse and deserialize the cache-tree extension than it does to read the cache-entries. Commits [1] and [2] spreads cache-entry reading across N-1 cores and dedicates a single core to simultaneously read the index extensions. Local testing (on my machine) shows that reading the cache-tree extension takes ~0.28 seconds. The 11 cache-entry threads take ~0.08 seconds. The main thread is blocked for 0.15 to 0.20 seconds waiting for the extension thread to finish. Let's use this commit to gather some telemetry and confirm this. My point is that improvements, such as index V5 which makes the cache entries smaller, may improve performance, but the gains may be limited because of this extension. And that we may need to look inside the cache-tree extension to truly improve do_read_index() performance. [1] abb4bb83845 read-cache: load cache extensions on a worker thread [2] 77ff1127a4c read-cache: load cache entries on worker threads Signed-off-by: Jeff Hostetler --- read-cache.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/read-cache.c b/read-cache.c index 2d5a83157223a6..e0c2fd6cc51d5e 100644 --- a/read-cache.c +++ b/read-cache.c @@ -1763,7 +1763,10 @@ static int read_index_extension(struct index_state *istate, { switch (CACHE_EXT(ext)) { case CACHE_EXT_TREE: + trace2_region_enter("index", "read/extension/cache_tree", NULL); istate->cache_tree = cache_tree_read(data, sz); + trace2_data_intmax("index", NULL, "read/extension/cache_tree/bytes", (intmax_t)sz); + trace2_region_leave("index", "read/extension/cache_tree", NULL); break; case CACHE_EXT_RESOLVE_UNDO: istate->resolve_undo = resolve_undo_read(data, sz); @@ -3050,9 +3053,13 @@ static int do_write_index(struct index_state *istate, struct tempfile *tempfile, !drop_cache_tree && istate->cache_tree) { strbuf_reset(&sb); + trace2_region_enter("index", "write/extension/cache_tree", NULL); cache_tree_write(&sb, istate->cache_tree); err = write_index_ext_header(f, eoie_c, CACHE_EXT_TREE, sb.len) < 0; hashwrite(f, sb.buf, sb.len); + trace2_data_intmax("index", NULL, "write/extension/cache_tree/bytes", (intmax_t)sb.len); + trace2_region_leave("index", "write/extension/cache_tree", NULL); + if (err) { ret = -1; goto out; From 41289c22576570fd6f5211a3184ad082fa9f8849 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 16 Jul 2019 09:09:53 -0400 Subject: [PATCH 083/196] trace2:gvfs:experiment: add region to apply_virtualfilesystem() Signed-off-by: Jeff Hostetler --- virtualfilesystem.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/virtualfilesystem.c b/virtualfilesystem.c index 20810c541558ce..bb7c2549481a74 100644 --- a/virtualfilesystem.c +++ b/virtualfilesystem.c @@ -270,6 +270,8 @@ void apply_virtualfilesystem(struct index_state *istate) if (!repo_config_get_virtualfilesystem(istate->repo)) return; + trace2_region_enter("vfs", "apply", the_repository); + if (!virtual_filesystem_data.len) get_virtual_filesystem_data(istate->repo, &virtual_filesystem_data); @@ -340,6 +342,8 @@ void apply_virtualfilesystem(struct index_state *istate) trace2_data_intmax("vfs", the_repository, "apply/nr_bulk_skip", nr_bulk_skip); trace2_data_intmax("vfs", the_repository, "apply/nr_explicit_skip", nr_explicit_skip); } + + trace2_region_leave("vfs", "apply", the_repository); } /* From a0f0ff0958b9beb93929d56ede75cfc1e5cca80a Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 16 Jul 2019 10:08:08 -0400 Subject: [PATCH 084/196] trace2:gvfs:experiment: add region around unpack_trees() Signed-off-by: Jeff Hostetler --- unpack-trees.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/unpack-trees.c b/unpack-trees.c index c881eb6a29f5c9..5ef04ed929b0de 100644 --- a/unpack-trees.c +++ b/unpack-trees.c @@ -1911,6 +1911,8 @@ int unpack_trees(unsigned len, struct tree_desc *t, struct unpack_trees_options if (o->df_conflict_entry) BUG("o->df_conflict_entry is an output only field"); + trace2_region_enter("exp", "unpack_trees", NULL); + trace_performance_enter(); trace2_region_enter("unpack_trees", "unpack_trees", the_repository); @@ -2116,6 +2118,7 @@ int unpack_trees(unsigned len, struct tree_desc *t, struct unpack_trees_options } trace2_region_leave("unpack_trees", "unpack_trees", the_repository); trace_performance_leave("unpack_trees"); + trace2_region_leave("exp", "unpack_trees", NULL); return ret; return_failed: From 98907cad07b6d89d5abc25dd18e92ba7180b761f Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 16 Jul 2019 10:16:37 -0400 Subject: [PATCH 085/196] trace2:gvfs:experiment: add region to cache_tree_fully_valid() Signed-off-by: Jeff Hostetler --- cache-tree.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/cache-tree.c b/cache-tree.c index 5db6587cc34a4e..c99bbfb256884b 100644 --- a/cache-tree.c +++ b/cache-tree.c @@ -232,7 +232,7 @@ static void discard_unused_subtrees(struct cache_tree *it) } } -int cache_tree_fully_valid(struct cache_tree *it) +static int cache_tree_fully_valid_1(struct cache_tree *it) { int i; if (!it) @@ -240,7 +240,7 @@ int cache_tree_fully_valid(struct cache_tree *it) if (it->entry_count < 0 || !repo_has_object_file(the_repository, &it->oid)) return 0; for (i = 0; i < it->subtree_nr; i++) { - if (!cache_tree_fully_valid(it->down[i]->cache_tree)) + if (!cache_tree_fully_valid_1(it->down[i]->cache_tree)) return 0; } return 1; @@ -251,6 +251,17 @@ static int must_check_existence(const struct cache_entry *ce) return !(repo_has_promisor_remote(the_repository) && ce_skip_worktree(ce)); } +int cache_tree_fully_valid(struct cache_tree *it) +{ + int result; + + trace2_region_enter("cache_tree", "fully_valid", NULL); + result = cache_tree_fully_valid_1(it); + trace2_region_leave("cache_tree", "fully_valid", NULL); + + return result; +} + static int update_one(struct cache_tree *it, struct cache_entry **cache, int entries, From a3ae6c2ef5bec688ebfacccb103843f70820c852 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 16 Jul 2019 10:40:56 -0400 Subject: [PATCH 086/196] trace2:gvfs:experiment: add unpack_entry() counter to unpack_trees() and report_tracking() Signed-off-by: Jeff Hostetler --- builtin/checkout.c | 6 ++++++ packfile.c | 9 +++++++++ packfile.h | 5 +++++ unpack-trees.c | 5 +++++ 4 files changed, 25 insertions(+) diff --git a/builtin/checkout.c b/builtin/checkout.c index 9e26f1cf15fb6d..dd8d6fbde81b29 100644 --- a/builtin/checkout.c +++ b/builtin/checkout.c @@ -18,6 +18,7 @@ #include "merge-recursive.h" #include "object-name.h" #include "object-store-ll.h" +#include "packfile.h" #include "parse-options.h" #include "path.h" #include "preload-index.h" @@ -1048,8 +1049,13 @@ static void update_refs_for_switch(const struct checkout_opts *opts, if (!opts->quiet && !opts->force_detach && (new_branch_info->path || !strcmp(new_branch_info->name, "HEAD"))) { + unsigned long nr_unpack_entry_at_start; + trace2_region_enter("exp", "report_tracking", the_repository); + nr_unpack_entry_at_start = get_nr_unpack_entry(); report_tracking(new_branch_info); + trace2_data_intmax("exp", NULL, "report_tracking/nr_unpack_entries", + (intmax_t)(get_nr_unpack_entry() - nr_unpack_entry_at_start)); trace2_region_leave("exp", "report_tracking", the_repository); } } diff --git a/packfile.c b/packfile.c index df4ba677197193..c3346197a08a2c 100644 --- a/packfile.c +++ b/packfile.c @@ -1678,6 +1678,13 @@ struct unpack_entry_stack_ent { unsigned long size; }; +static unsigned long g_nr_unpack_entry; + +unsigned long get_nr_unpack_entry(void) +{ + return g_nr_unpack_entry; +} + void *unpack_entry(struct repository *r, struct packed_git *p, off_t obj_offset, enum object_type *final_type, unsigned long *final_size) { @@ -1691,6 +1698,8 @@ void *unpack_entry(struct repository *r, struct packed_git *p, off_t obj_offset, int delta_stack_nr = 0, delta_stack_alloc = UNPACK_ENTRY_STACK_PREALLOC; int base_from_cache = 0; + g_nr_unpack_entry++; + write_pack_access_log(p, obj_offset); /* PHASE 1: drill down to the innermost base object */ diff --git a/packfile.h b/packfile.h index 0f7865822916ff..33efd0d4c0bfe1 100644 --- a/packfile.h +++ b/packfile.h @@ -214,4 +214,9 @@ int is_promisor_object(const struct object_id *oid); int load_idx(const char *path, const unsigned int hashsz, void *idx_map, size_t idx_size, struct packed_git *p); +/* + * Return the number of objects fetched from a packfile. + */ +unsigned long get_nr_unpack_entry(void); + #endif diff --git a/unpack-trees.c b/unpack-trees.c index 5ef04ed929b0de..7f898e61582538 100644 --- a/unpack-trees.c +++ b/unpack-trees.c @@ -16,6 +16,7 @@ #include "tree-walk.h" #include "cache-tree.h" #include "unpack-trees.h" +#include "packfile.h" #include "progress.h" #include "refs.h" #include "attr.h" @@ -1898,6 +1899,7 @@ int unpack_trees(unsigned len, struct tree_desc *t, struct unpack_trees_options struct pattern_list pl; int free_pattern_list = 0; struct dir_struct dir = DIR_INIT; + unsigned long nr_unpack_entry_at_start; if (o->reset == UNPACK_RESET_INVALID) BUG("o->reset had a value of 1; should be UNPACK_TREES_*_UNTRACKED"); @@ -1912,6 +1914,7 @@ int unpack_trees(unsigned len, struct tree_desc *t, struct unpack_trees_options BUG("o->df_conflict_entry is an output only field"); trace2_region_enter("exp", "unpack_trees", NULL); + nr_unpack_entry_at_start = get_nr_unpack_entry(); trace_performance_enter(); trace2_region_enter("unpack_trees", "unpack_trees", the_repository); @@ -2118,6 +2121,8 @@ int unpack_trees(unsigned len, struct tree_desc *t, struct unpack_trees_options } trace2_region_leave("unpack_trees", "unpack_trees", the_repository); trace_performance_leave("unpack_trees"); + trace2_data_intmax("unpack_trees", NULL, "unpack_trees/nr_unpack_entries", + (intmax_t)(get_nr_unpack_entry() - nr_unpack_entry_at_start)); trace2_region_leave("exp", "unpack_trees", NULL); return ret; From aeaecb8496c9eb83b73119b781792d52ca8c86b2 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Thu, 25 Jul 2019 15:43:50 -0400 Subject: [PATCH 087/196] trace2:gvfs:experiment: increase default event depth for unpack-tree data --- trace2/tr2_tgt_event.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/trace2/tr2_tgt_event.c b/trace2/tr2_tgt_event.c index 45b0850a5ec5c8..04f2e5390bc749 100644 --- a/trace2/tr2_tgt_event.c +++ b/trace2/tr2_tgt_event.c @@ -37,7 +37,7 @@ static struct tr2_dst tr2dst_event = { * event target. Use the TR2_SYSENV_EVENT_NESTING setting to increase * region details in the event target. */ -static int tr2env_event_max_nesting_levels = 2; +static int tr2env_event_max_nesting_levels = 4; /* * Use the TR2_SYSENV_EVENT_BRIEF to omit the