Exclude threads test from rr tracing

(cherry picked from commit 7a3c153)

Author: Keno

base/options.jl

@@ -45,6 +45,7 @@ struct JLOptions
     image_file_specified::Int8
     warn_scope::Int8
     image_codegen::Int8
+    rr_detach::Int8
 end
 
 # This runs early in the sysimage != is not defined yet

src/jlapi.c

@@ -651,6 +651,17 @@ static void lock_low32(void)
 // Actual definition in `ast.c`
 void jl_lisp_prompt(void);
 
+static void rr_detach_teleport(void) {
+#ifdef _OS_LINUX_
+#define RR_CALL_BASE 1000
+#define SYS_rrcall_detach_teleport (RR_CALL_BASE + 9)
+    int err = syscall(SYS_rrcall_detach_teleport, 0, 0, 0, 0, 0, 0);
+    if (err < 0 || jl_running_under_rr(1)) {
+        jl_error("Failed to detach from rr session");
+    }
+#endif
+}
+
 JL_DLLEXPORT int repl_entrypoint(int argc, char *argv[])
 {
     // no-op on Windows, note that the caller must have already converted
@@ -666,7 +677,19 @@ JL_DLLEXPORT int repl_entrypoint(int argc, char *argv[])
         memmove(&argv[1], &argv[2], (argc-2)*sizeof(void*));
         argc--;
     }
+    char **orig_argv = argv;
     jl_parse_opts(&argc, (char***)&argv);
+
+    // The parent process requested that we detach from the rr session.
+    // N.B.: In a perfect world, we would only do this for the portion of
+    // the execution where we actually need to exclude rr (e.g. because we're
+    // testing for the absence of a memory-model-dependent bug).
+    if (jl_options.rr_detach && jl_running_under_rr(0)) {
+        rr_detach_teleport();
+        execv("/proc/self/exe", orig_argv);
+        jl_error("Failed to self-execute");
+    }
+
     julia_init(jl_options.image_file_specified ? JL_IMAGE_CWD : JL_IMAGE_JULIA_HOME);
     if (lisp_prompt) {
         jl_get_ptls_states()->world_age = jl_get_world_counter();

src/jloptions.c

@@ -75,6 +75,7 @@ jl_options_t jl_options = { 0,    // quiet
                             0, // image_file_specified
                             JL_OPTIONS_WARN_SCOPE_ON,  // ambiguous scope warning
                             0, // image-codegen
+                            0, // rr-detach
 };
 
 static const char usage[] = "julia [switches] -- [programfile] [args...]\n";
@@ -203,6 +204,7 @@ JL_DLLEXPORT void jl_parse_opts(int *argcp, char ***argvp)
            opt_project,
            opt_bug_report,
            opt_image_codegen,
+           opt_rr_detach,
     };
     static const char* const shortopts = "+vhqH:e:E:L:J:C:it:p:O:g:";
     static const struct option longopts[] = {
@@ -254,6 +256,7 @@ JL_DLLEXPORT void jl_parse_opts(int *argcp, char ***argvp)
         { "bind-to",         required_argument, 0, opt_bind_to },
         { "lisp",            no_argument,       0, 1 },
         { "image-codegen",   no_argument,       0, opt_image_codegen },
+        { "rr-detach",       no_argument,       0, opt_rr_detach },
         { 0, 0, 0, 0 }
     };
 
@@ -655,6 +658,9 @@ JL_DLLEXPORT void jl_parse_opts(int *argcp, char ***argvp)
         case opt_image_codegen:
             jl_options.image_codegen = 1;
             break;
+        case opt_rr_detach:
+            jl_options.rr_detach = 1;
+            break;
         default:
             jl_errorf("julia: unhandled option -- %c\n"
                       "This is a bug, please report it.", c);

src/julia.h

@@ -1944,6 +1944,7 @@ typedef struct {
     int8_t image_file_specified;
     int8_t warn_scope;
     int8_t image_codegen;
+    int8_t rr_detach;
 } jl_options_t;
 
 extern JL_DLLEXPORT jl_options_t jl_options;

src/julia_internal.h

@@ -83,6 +83,8 @@ void JL_UV_LOCK(void);
 extern "C" {
 #endif
 
+int jl_running_under_rr(int recheck);
+
 //--------------------------------------------------
 // timers
 // Returns time in nanosec

src/partr.c

@@ -274,14 +274,14 @@ JL_DLLEXPORT int jl_enqueue_task(jl_task_t *task)
 }
 
 
-static int running_under_rr(void)
+int jl_running_under_rr(int recheck)
 {
 #ifdef _OS_LINUX_
 #define RR_CALL_BASE 1000
 #define SYS_rrcall_check_presence (RR_CALL_BASE + 8)
     static int checked_running_under_rr = 0;
     static int is_running_under_rr = 0;
-    if (!checked_running_under_rr) {
+    if (!checked_running_under_rr || recheck) {
         int ret = syscall(SYS_rrcall_check_presence, 0, 0, 0, 0, 0, 0);
         if (ret == -1) {
             // Should always be ENOSYS, but who knows what people do for
@@ -311,7 +311,7 @@ static int sleep_check_after_threshold(uint64_t *start_cycles)
      * scheduling logic from switching to other threads. Just don't bother
      * trying to wait here
      */
-    if (running_under_rr())
+    if (jl_running_under_rr(0))
         return 1;
     if (!(*start_cycles)) {
         *start_cycles = jl_hrtime();

test/threads.jl

@@ -1,18 +1,20 @@
 # This file is a part of Julia. License is MIT: https://julialang.org/license
 
-let cmd = `$(Base.julia_cmd()) --depwarn=error --startup-file=no threads_exec.jl`
+using Test
+
+let cmd = `$(Base.julia_cmd()) --depwarn=error --rr-detach --startup-file=no threads_exec.jl`
     for test_nthreads in (1, 2, 4, 4) # run once to try single-threaded mode, then try a couple times to trigger bad races
         run(pipeline(setenv(cmd, "JULIA_NUM_THREADS" => test_nthreads), stdout = stdout, stderr = stderr))
     end
 end
 
 # issue #34415 - make sure external affinity settings work
+const SYS_rrcall_check_presence = 1008
+running_under_rr() = 0 == ccall(:syscall, Int,
+    (Int, Int, Int, Int, Int, Int, Int),
+    SYS_rrcall_check_presence, 0, 0, 0, 0, 0, 0)
 
 if Sys.islinux()
-    const SYS_rrcall_check_presence = 1008
-    running_under_rr() = 0 == ccall(:syscall, Int,
-        (Int, Int, Int, Int, Int, Int, Int),
-        SYS_rrcall_check_presence, 0, 0, 0, 0, 0, 0)
     if Sys.CPU_THREADS > 1 && Sys.which("taskset") !== nothing && !running_under_rr()
         run_with_affinity(spec) = readchomp(`taskset -c $spec $(Base.julia_cmd()) -e "run(\`taskset -p \$(getpid())\`)"`)
         @test endswith(run_with_affinity("1"), "2")