fsverity: use shash API instead of ahash API

The "ahash" API, like the other scatterlist-based crypto APIs such as
"skcipher", comes with some well-known limitations.  First, it can't
easily be used with vmalloc addresses.  Second, the request struct can't
be allocated on the stack.  This adds complexity and a possible failure
point that needs to be worked around, e.g. using a mempool.

The only benefit of ahash over "shash" is that ahash is needed to access
traditional memory-to-memory crypto accelerators, i.e. drivers/crypto/.
However, this style of crypto acceleration has largely fallen out of
favor and been superseded by CPU-based acceleration or inline crypto
engines.  Also, ahash needs to be used asynchronously to take full
advantage of such hardware, but fs/verity/ has never done this.

On all systems that aren't actually using one of these ahash-only crypto
accelerators, ahash just adds unnecessary overhead as it sits between
the user and the underlying shash algorithms.

Also, XFS is planned to cache fsverity Merkle tree blocks in the
existing XFS buffer cache.  As a result, it will be possible for a
single Merkle tree block to be split across discontiguous pages
(https://lore.kernel.org/r/20230405233753.GU3223426@dread.disaster.area).
This data will need to be hashed.  It is easiest to work with a vmapped
address in this case.  However, ahash is incompatible with this.

Therefore, let's convert fs/verity/ from ahash to shash.  This
simplifies the code, and it should also slightly improve performance for
everyone who wasn't actually using one of these ahash-only crypto
accelerators, i.e. almost everyone (or maybe even everyone)!

Link: https://lore.kernel.org/r/20230516052306.99600-1-ebiggers@kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>

Author: ebiggers

fs/verity/enable.c

@@ -7,6 +7,7 @@
 
 #include "fsverity_private.h"
 
+#include <crypto/hash.h>
 #include <linux/mount.h>
 #include <linux/sched/signal.h>
 #include <linux/uaccess.h>
@@ -20,7 +21,7 @@ struct block_buffer {
 /* Hash a block, writing the result to the next level's pending block buffer. */
 static int hash_one_block(struct inode *inode,
 			  const struct merkle_tree_params *params,
-			  struct ahash_request *req, struct block_buffer *cur)
+			  struct block_buffer *cur)
 {
 	struct block_buffer *next = cur + 1;
 	int err;
@@ -36,8 +37,7 @@ static int hash_one_block(struct inode *inode,
 	/* Zero-pad the block if it's shorter than the block size. */
 	memset(&cur->data[cur->filled], 0, params->block_size - cur->filled);
 
-	err = fsverity_hash_block(params, inode, req, virt_to_page(cur->data),
-				  offset_in_page(cur->data),
+	err = fsverity_hash_block(params, inode, cur->data,
 				  &next->data[next->filled]);
 	if (err)
 		return err;
@@ -76,7 +76,6 @@ static int build_merkle_tree(struct file *filp,
 	struct inode *inode = file_inode(filp);
 	const u64 data_size = inode->i_size;
 	const int num_levels = params->num_levels;
-	struct ahash_request *req;
 	struct block_buffer _buffers[1 + FS_VERITY_MAX_LEVELS + 1] = {};
 	struct block_buffer *buffers = &_buffers[1];
 	unsigned long level_offset[FS_VERITY_MAX_LEVELS];
@@ -90,9 +89,6 @@ static int build_merkle_tree(struct file *filp,
 		return 0;
 	}
 
-	/* This allocation never fails, since it's mempool-backed. */
-	req = fsverity_alloc_hash_request(params->hash_alg, GFP_KERNEL);
-
 	/*
 	 * Allocate the block buffers.  Buffer "-1" is for data blocks.
 	 * Buffers 0 <= level < num_levels are for the actual tree levels.
@@ -130,7 +126,7 @@ static int build_merkle_tree(struct file *filp,
 			fsverity_err(inode, "Short read of file data");
 			goto out;
 		}
-		err = hash_one_block(inode, params, req, &buffers[-1]);
+		err = hash_one_block(inode, params, &buffers[-1]);
 		if (err)
 			goto out;
 		for (level = 0; level < num_levels; level++) {
@@ -141,8 +137,7 @@ static int build_merkle_tree(struct file *filp,
 			}
 			/* Next block at @level is full */
 
-			err = hash_one_block(inode, params, req,
-					     &buffers[level]);
+			err = hash_one_block(inode, params, &buffers[level]);
 			if (err)
 				goto out;
 			err = write_merkle_tree_block(inode,
@@ -162,8 +157,7 @@ static int build_merkle_tree(struct file *filp,
 	/* Finish all nonempty pending tree blocks. */
 	for (level = 0; level < num_levels; level++) {
 		if (buffers[level].filled != 0) {
-			err = hash_one_block(inode, params, req,
-					     &buffers[level]);
+			err = hash_one_block(inode, params, &buffers[level]);
 			if (err)
 				goto out;
 			err = write_merkle_tree_block(inode,
@@ -183,7 +177,6 @@ static int build_merkle_tree(struct file *filp,
 out:
 	for (level = -1; level < num_levels; level++)
 		kfree(buffers[level].data);
-	fsverity_free_hash_request(params->hash_alg, req);
 	return err;
 }
 

fs/verity/fsverity_private.h

@@ -11,9 +11,6 @@
 #define pr_fmt(fmt) "fs-verity: " fmt
 
 #include <linux/fsverity.h>
-#include <linux/mempool.h>
-
-struct ahash_request;
 
 /*
  * Implementation limit: maximum depth of the Merkle tree.  For now 8 is plenty;
@@ -23,11 +20,10 @@ struct ahash_request;
 
 /* A hash algorithm supported by fs-verity */
 struct fsverity_hash_alg {
-	struct crypto_ahash *tfm; /* hash tfm, allocated on demand */
+	struct crypto_shash *tfm; /* hash tfm, allocated on demand */
 	const char *name;	  /* crypto API name, e.g. sha256 */
 	unsigned int digest_size; /* digest size in bytes, e.g. 32 for SHA-256 */
 	unsigned int block_size;  /* block size in bytes, e.g. 64 for SHA-256 */
-	mempool_t req_pool;	  /* mempool with a preallocated hash request */
 	/*
 	 * The HASH_ALGO_* constant for this algorithm.  This is different from
 	 * FS_VERITY_HASH_ALG_*, which uses a different numbering scheme.
@@ -85,15 +81,10 @@ extern struct fsverity_hash_alg fsverity_hash_algs[];
 
 struct fsverity_hash_alg *fsverity_get_hash_alg(const struct inode *inode,
 						unsigned int num);
-struct ahash_request *fsverity_alloc_hash_request(struct fsverity_hash_alg *alg,
-						  gfp_t gfp_flags);
-void fsverity_free_hash_request(struct fsverity_hash_alg *alg,
-				struct ahash_request *req);
 const u8 *fsverity_prepare_hash_state(struct fsverity_hash_alg *alg,
 				      const u8 *salt, size_t salt_size);
 int fsverity_hash_block(const struct merkle_tree_params *params,
-			const struct inode *inode, struct ahash_request *req,
-			struct page *page, unsigned int offset, u8 *out);
+			const struct inode *inode, const void *data, u8 *out);
 int fsverity_hash_buffer(struct fsverity_hash_alg *alg,
 			 const void *data, size_t size, u8 *out);
 void __init fsverity_check_hash_algs(void);

fs/verity/hash_algs.c

@@ -8,7 +8,6 @@
 #include "fsverity_private.h"
 
 #include <crypto/hash.h>
-#include <linux/scatterlist.h>
 
 /* The hash algorithms supported by fs-verity */
 struct fsverity_hash_alg fsverity_hash_algs[] = {
@@ -44,7 +43,7 @@ struct fsverity_hash_alg *fsverity_get_hash_alg(const struct inode *inode,
 						unsigned int num)
 {
 	struct fsverity_hash_alg *alg;
-	struct crypto_ahash *tfm;
+	struct crypto_shash *tfm;
 	int err;
 
 	if (num >= ARRAY_SIZE(fsverity_hash_algs) ||
@@ -63,11 +62,7 @@ struct fsverity_hash_alg *fsverity_get_hash_alg(const struct inode *inode,
 	if (alg->tfm != NULL)
 		goto out_unlock;
 
-	/*
-	 * Using the shash API would make things a bit simpler, but the ahash
-	 * API is preferable as it allows the use of crypto accelerators.
-	 */
-	tfm = crypto_alloc_ahash(alg->name, 0, 0);
+	tfm = crypto_alloc_shash(alg->name, 0, 0);
 	if (IS_ERR(tfm)) {
 		if (PTR_ERR(tfm) == -ENOENT) {
 			fsverity_warn(inode,
@@ -84,68 +79,26 @@ struct fsverity_hash_alg *fsverity_get_hash_alg(const struct inode *inode,
 	}
 
 	err = -EINVAL;
-	if (WARN_ON_ONCE(alg->digest_size != crypto_ahash_digestsize(tfm)))
+	if (WARN_ON_ONCE(alg->digest_size != crypto_shash_digestsize(tfm)))
 		goto err_free_tfm;
-	if (WARN_ON_ONCE(alg->block_size != crypto_ahash_blocksize(tfm)))
-		goto err_free_tfm;
-
-	err = mempool_init_kmalloc_pool(&alg->req_pool, 1,
-					sizeof(struct ahash_request) +
-					crypto_ahash_reqsize(tfm));
-	if (err)
+	if (WARN_ON_ONCE(alg->block_size != crypto_shash_blocksize(tfm)))
 		goto err_free_tfm;
 
 	pr_info("%s using implementation \"%s\"\n",
-		alg->name, crypto_ahash_driver_name(tfm));
+		alg->name, crypto_shash_driver_name(tfm));
 
 	/* pairs with smp_load_acquire() above */
 	smp_store_release(&alg->tfm, tfm);
 	goto out_unlock;
 
 err_free_tfm:
-	crypto_free_ahash(tfm);
+	crypto_free_shash(tfm);
 	alg = ERR_PTR(err);
 out_unlock:
 	mutex_unlock(&fsverity_hash_alg_init_mutex);
 	return alg;
 }
 
-/**
- * fsverity_alloc_hash_request() - allocate a hash request object
- * @alg: the hash algorithm for which to allocate the request
- * @gfp_flags: memory allocation flags
- *
- * This is mempool-backed, so this never fails if __GFP_DIRECT_RECLAIM is set in
- * @gfp_flags.  However, in that case this might need to wait for all
- * previously-allocated requests to be freed.  So to avoid deadlocks, callers
- * must never need multiple requests at a time to make forward progress.
- *
- * Return: the request object on success; NULL on failure (but see above)
- */
-struct ahash_request *fsverity_alloc_hash_request(struct fsverity_hash_alg *alg,
-						  gfp_t gfp_flags)
-{
-	struct ahash_request *req = mempool_alloc(&alg->req_pool, gfp_flags);
-
-	if (req)
-		ahash_request_set_tfm(req, alg->tfm);
-	return req;
-}
-
-/**
- * fsverity_free_hash_request() - free a hash request object
- * @alg: the hash algorithm
- * @req: the hash request object to free
- */
-void fsverity_free_hash_request(struct fsverity_hash_alg *alg,
-				struct ahash_request *req)
-{
-	if (req) {
-		ahash_request_zero(req);
-		mempool_free(req, &alg->req_pool);
-	}
-}
-
 /**
  * fsverity_prepare_hash_state() - precompute the initial hash state
  * @alg: hash algorithm
@@ -159,23 +112,20 @@ const u8 *fsverity_prepare_hash_state(struct fsverity_hash_alg *alg,
 				      const u8 *salt, size_t salt_size)
 {
 	u8 *hashstate = NULL;
-	struct ahash_request *req = NULL;
+	SHASH_DESC_ON_STACK(desc, alg->tfm);
 	u8 *padded_salt = NULL;
 	size_t padded_salt_size;
-	struct scatterlist sg;
-	DECLARE_CRYPTO_WAIT(wait);
 	int err;
 
+	desc->tfm = alg->tfm;
+
 	if (salt_size == 0)
 		return NULL;
 
-	hashstate = kmalloc(crypto_ahash_statesize(alg->tfm), GFP_KERNEL);
+	hashstate = kmalloc(crypto_shash_statesize(alg->tfm), GFP_KERNEL);
 	if (!hashstate)
 		return ERR_PTR(-ENOMEM);
 
-	/* This allocation never fails, since it's mempool-backed. */
-	req = fsverity_alloc_hash_request(alg, GFP_KERNEL);
-
 	/*
 	 * Zero-pad the salt to the next multiple of the input size of the hash
 	 * algorithm's compression function, e.g. 64 bytes for SHA-256 or 128
@@ -190,26 +140,18 @@ const u8 *fsverity_prepare_hash_state(struct fsverity_hash_alg *alg,
 		goto err_free;
 	}
 	memcpy(padded_salt, salt, salt_size);
-
-	sg_init_one(&sg, padded_salt, padded_salt_size);
-	ahash_request_set_callback(req, CRYPTO_TFM_REQ_MAY_SLEEP |
-					CRYPTO_TFM_REQ_MAY_BACKLOG,
-				   crypto_req_done, &wait);
-	ahash_request_set_crypt(req, &sg, NULL, padded_salt_size);
-
-	err = crypto_wait_req(crypto_ahash_init(req), &wait);
+	err = crypto_shash_init(desc);
 	if (err)
 		goto err_free;
 
-	err = crypto_wait_req(crypto_ahash_update(req), &wait);
+	err = crypto_shash_update(desc, padded_salt, padded_salt_size);
 	if (err)
 		goto err_free;
 
-	err = crypto_ahash_export(req, hashstate);
+	err = crypto_shash_export(desc, hashstate);
 	if (err)
 		goto err_free;
 out:
-	fsverity_free_hash_request(alg, req);
 	kfree(padded_salt);
 	return hashstate;
 
@@ -223,9 +165,7 @@ const u8 *fsverity_prepare_hash_state(struct fsverity_hash_alg *alg,
  * fsverity_hash_block() - hash a single data or hash block
  * @params: the Merkle tree's parameters
  * @inode: inode for which the hashing is being done
- * @req: preallocated hash request
- * @page: the page containing the block to hash
- * @offset: the offset of the block within @page
+ * @data: virtual address of a buffer containing the block to hash
  * @out: output digest, size 'params->digest_size' bytes
  *
  * Hash a single data or hash block.  The hash is salted if a salt is specified
@@ -234,33 +174,24 @@ const u8 *fsverity_prepare_hash_state(struct fsverity_hash_alg *alg,
  * Return: 0 on success, -errno on failure
  */
 int fsverity_hash_block(const struct merkle_tree_params *params,
-			const struct inode *inode, struct ahash_request *req,
-			struct page *page, unsigned int offset, u8 *out)
+			const struct inode *inode, const void *data, u8 *out)
 {
-	struct scatterlist sg;
-	DECLARE_CRYPTO_WAIT(wait);
+	SHASH_DESC_ON_STACK(desc, params->hash_alg->tfm);
 	int err;
 
-	sg_init_table(&sg, 1);
-	sg_set_page(&sg, page, params->block_size, offset);
-	ahash_request_set_callback(req, CRYPTO_TFM_REQ_MAY_SLEEP |
-					CRYPTO_TFM_REQ_MAY_BACKLOG,
-				   crypto_req_done, &wait);
-	ahash_request_set_crypt(req, &sg, out, params->block_size);
+	desc->tfm = params->hash_alg->tfm;
 
 	if (params->hashstate) {
-		err = crypto_ahash_import(req, params->hashstate);
+		err = crypto_shash_import(desc, params->hashstate);
 		if (err) {
 			fsverity_err(inode,
 				     "Error %d importing hash state", err);
 			return err;
 		}
-		err = crypto_ahash_finup(req);
+		err = crypto_shash_finup(desc, data, params->block_size, out);
 	} else {
-		err = crypto_ahash_digest(req);
+		err = crypto_shash_digest(desc, data, params->block_size, out);
 	}
-
-	err = crypto_wait_req(err, &wait);
 	if (err)
 		fsverity_err(inode, "Error %d computing block hash", err);
 	return err;
@@ -273,32 +204,12 @@ int fsverity_hash_block(const struct merkle_tree_params *params,
  * @size: size of data to hash, in bytes
  * @out: output digest, size 'alg->digest_size' bytes
  *
- * Hash some data which is located in physically contiguous memory (i.e. memory
- * allocated by kmalloc(), not by vmalloc()).  No salt is used.
- *
  * Return: 0 on success, -errno on failure
  */
 int fsverity_hash_buffer(struct fsverity_hash_alg *alg,
 			 const void *data, size_t size, u8 *out)
 {
-	struct ahash_request *req;
-	struct scatterlist sg;
-	DECLARE_CRYPTO_WAIT(wait);
-	int err;
-
-	/* This allocation never fails, since it's mempool-backed. */
-	req = fsverity_alloc_hash_request(alg, GFP_KERNEL);
-
-	sg_init_one(&sg, data, size);
-	ahash_request_set_callback(req, CRYPTO_TFM_REQ_MAY_SLEEP |
-					CRYPTO_TFM_REQ_MAY_BACKLOG,
-				   crypto_req_done, &wait);
-	ahash_request_set_crypt(req, &sg, out, size);
-
-	err = crypto_wait_req(crypto_ahash_digest(req), &wait);
-
-	fsverity_free_hash_request(alg, req);
-	return err;
+	return crypto_shash_tfm_digest(alg->tfm, data, size, out);
 }
 
 void __init fsverity_check_hash_algs(void)