linux/fs/verity/verify.c
Nathan Huckleberry f959325e6a fsverity: Remove WQ_UNBOUND from fsverity read workqueue
WQ_UNBOUND causes significant scheduler latency on ARM64/Android.  This
is problematic for latency sensitive workloads, like I/O
post-processing.

Removing WQ_UNBOUND gives a 96% reduction in fsverity workqueue related
scheduler latency and improves app cold startup times by ~30ms.
WQ_UNBOUND was also removed from the dm-verity workqueue for the same
reason [1].

This code was tested by running Android app startup benchmarks and
measuring how long the fsverity workqueue spent in the runnable state.

Before
Total workqueue scheduler latency: 553800us
After
Total workqueue scheduler latency: 18962us

[1]: https://lore.kernel.org/all/20230202012348.885402-1-nhuck@google.com/

Signed-off-by: Nathan Huckleberry <nhuck@google.com>
Fixes: 8a1d0f9cac ("fs-verity: add data verification hooks for ->readpages()")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230310193325.620493-1-nhuck@google.com
Signed-off-by: Eric Biggers <ebiggers@google.com>
2023-03-14 16:16:36 -07:00

410 lines
13 KiB
C

// SPDX-License-Identifier: GPL-2.0
/*
* Data verification functions, i.e. hooks for ->readahead()
*
* Copyright 2019 Google LLC
*/
#include "fsverity_private.h"
#include <crypto/hash.h>
#include <linux/bio.h>
static struct workqueue_struct *fsverity_read_workqueue;
static inline int cmp_hashes(const struct fsverity_info *vi,
const u8 *want_hash, const u8 *real_hash,
u64 data_pos, int level)
{
const unsigned int hsize = vi->tree_params.digest_size;
if (memcmp(want_hash, real_hash, hsize) == 0)
return 0;
fsverity_err(vi->inode,
"FILE CORRUPTED! pos=%llu, level=%d, want_hash=%s:%*phN, real_hash=%s:%*phN",
data_pos, level,
vi->tree_params.hash_alg->name, hsize, want_hash,
vi->tree_params.hash_alg->name, hsize, real_hash);
return -EBADMSG;
}
static bool data_is_zeroed(struct inode *inode, struct page *page,
unsigned int len, unsigned int offset)
{
void *virt = kmap_local_page(page);
if (memchr_inv(virt + offset, 0, len)) {
kunmap_local(virt);
fsverity_err(inode,
"FILE CORRUPTED! Data past EOF is not zeroed");
return false;
}
kunmap_local(virt);
return true;
}
/*
* Returns true if the hash block with index @hblock_idx in the tree, located in
* @hpage, has already been verified.
*/
static bool is_hash_block_verified(struct fsverity_info *vi, struct page *hpage,
unsigned long hblock_idx)
{
bool verified;
unsigned int blocks_per_page;
unsigned int i;
/*
* When the Merkle tree block size and page size are the same, then the
* ->hash_block_verified bitmap isn't allocated, and we use PG_checked
* to directly indicate whether the page's block has been verified.
*
* Using PG_checked also guarantees that we re-verify hash pages that
* get evicted and re-instantiated from the backing storage, as new
* pages always start out with PG_checked cleared.
*/
if (!vi->hash_block_verified)
return PageChecked(hpage);
/*
* When the Merkle tree block size and page size differ, we use a bitmap
* to indicate whether each hash block has been verified.
*
* However, we still need to ensure that hash pages that get evicted and
* re-instantiated from the backing storage are re-verified. To do
* this, we use PG_checked again, but now it doesn't really mean
* "checked". Instead, now it just serves as an indicator for whether
* the hash page is newly instantiated or not.
*
* The first thread that sees PG_checked=0 must clear the corresponding
* bitmap bits, then set PG_checked=1. This requires a spinlock. To
* avoid having to take this spinlock in the common case of
* PG_checked=1, we start with an opportunistic lockless read.
*/
if (PageChecked(hpage)) {
/*
* A read memory barrier is needed here to give ACQUIRE
* semantics to the above PageChecked() test.
*/
smp_rmb();
return test_bit(hblock_idx, vi->hash_block_verified);
}
spin_lock(&vi->hash_page_init_lock);
if (PageChecked(hpage)) {
verified = test_bit(hblock_idx, vi->hash_block_verified);
} else {
blocks_per_page = vi->tree_params.blocks_per_page;
hblock_idx = round_down(hblock_idx, blocks_per_page);
for (i = 0; i < blocks_per_page; i++)
clear_bit(hblock_idx + i, vi->hash_block_verified);
/*
* A write memory barrier is needed here to give RELEASE
* semantics to the below SetPageChecked() operation.
*/
smp_wmb();
SetPageChecked(hpage);
verified = false;
}
spin_unlock(&vi->hash_page_init_lock);
return verified;
}
/*
* Verify a single data block against the file's Merkle tree.
*
* In principle, we need to verify the entire path to the root node. However,
* for efficiency the filesystem may cache the hash blocks. Therefore we need
* only ascend the tree until an already-verified hash block is seen, and then
* verify the path to that block.
*
* Return: %true if the data block is valid, else %false.
*/
static bool
verify_data_block(struct inode *inode, struct fsverity_info *vi,
struct ahash_request *req, struct page *data_page,
u64 data_pos, unsigned int dblock_offset_in_page,
unsigned long max_ra_pages)
{
const struct merkle_tree_params *params = &vi->tree_params;
const unsigned int hsize = params->digest_size;
int level;
u8 _want_hash[FS_VERITY_MAX_DIGEST_SIZE];
const u8 *want_hash;
u8 real_hash[FS_VERITY_MAX_DIGEST_SIZE];
/* The hash blocks that are traversed, indexed by level */
struct {
/* Page containing the hash block */
struct page *page;
/* Index of the hash block in the tree overall */
unsigned long index;
/* Byte offset of the hash block within @page */
unsigned int offset_in_page;
/* Byte offset of the wanted hash within @page */
unsigned int hoffset;
} hblocks[FS_VERITY_MAX_LEVELS];
/*
* The index of the previous level's block within that level; also the
* index of that block's hash within the current level.
*/
u64 hidx = data_pos >> params->log_blocksize;
int err;
if (unlikely(data_pos >= inode->i_size)) {
/*
* This can happen in the data page spanning EOF when the Merkle
* tree block size is less than the page size. The Merkle tree
* doesn't cover data blocks fully past EOF. But the entire
* page spanning EOF can be visible to userspace via a mmap, and
* any part past EOF should be all zeroes. Therefore, we need
* to verify that any data blocks fully past EOF are all zeroes.
*/
return data_is_zeroed(inode, data_page, params->block_size,
dblock_offset_in_page);
}
/*
* Starting at the leaf level, ascend the tree saving hash blocks along
* the way until we find a hash block that has already been verified, or
* until we reach the root.
*/
for (level = 0; level < params->num_levels; level++) {
unsigned long next_hidx;
unsigned long hblock_idx;
pgoff_t hpage_idx;
unsigned int hblock_offset_in_page;
unsigned int hoffset;
struct page *hpage;
/*
* The index of the block in the current level; also the index
* of that block's hash within the next level.
*/
next_hidx = hidx >> params->log_arity;
/* Index of the hash block in the tree overall */
hblock_idx = params->level_start[level] + next_hidx;
/* Index of the hash page in the tree overall */
hpage_idx = hblock_idx >> params->log_blocks_per_page;
/* Byte offset of the hash block within the page */
hblock_offset_in_page =
(hblock_idx << params->log_blocksize) & ~PAGE_MASK;
/* Byte offset of the hash within the page */
hoffset = hblock_offset_in_page +
((hidx << params->log_digestsize) &
(params->block_size - 1));
hpage = inode->i_sb->s_vop->read_merkle_tree_page(inode,
hpage_idx, level == 0 ? min(max_ra_pages,
params->tree_pages - hpage_idx) : 0);
if (IS_ERR(hpage)) {
err = PTR_ERR(hpage);
fsverity_err(inode,
"Error %d reading Merkle tree page %lu",
err, hpage_idx);
goto out;
}
if (is_hash_block_verified(vi, hpage, hblock_idx)) {
memcpy_from_page(_want_hash, hpage, hoffset, hsize);
want_hash = _want_hash;
put_page(hpage);
goto descend;
}
hblocks[level].page = hpage;
hblocks[level].index = hblock_idx;
hblocks[level].offset_in_page = hblock_offset_in_page;
hblocks[level].hoffset = hoffset;
hidx = next_hidx;
}
want_hash = vi->root_hash;
descend:
/* Descend the tree verifying hash blocks. */
for (; level > 0; level--) {
struct page *hpage = hblocks[level - 1].page;
unsigned long hblock_idx = hblocks[level - 1].index;
unsigned int hblock_offset_in_page =
hblocks[level - 1].offset_in_page;
unsigned int hoffset = hblocks[level - 1].hoffset;
err = fsverity_hash_block(params, inode, req, hpage,
hblock_offset_in_page, real_hash);
if (err)
goto out;
err = cmp_hashes(vi, want_hash, real_hash, data_pos, level - 1);
if (err)
goto out;
/*
* Mark the hash block as verified. This must be atomic and
* idempotent, as the same hash block might be verified by
* multiple threads concurrently.
*/
if (vi->hash_block_verified)
set_bit(hblock_idx, vi->hash_block_verified);
else
SetPageChecked(hpage);
memcpy_from_page(_want_hash, hpage, hoffset, hsize);
want_hash = _want_hash;
put_page(hpage);
}
/* Finally, verify the data block. */
err = fsverity_hash_block(params, inode, req, data_page,
dblock_offset_in_page, real_hash);
if (err)
goto out;
err = cmp_hashes(vi, want_hash, real_hash, data_pos, -1);
out:
for (; level > 0; level--)
put_page(hblocks[level - 1].page);
return err == 0;
}
static bool
verify_data_blocks(struct inode *inode, struct fsverity_info *vi,
struct ahash_request *req, struct folio *data_folio,
size_t len, size_t offset, unsigned long max_ra_pages)
{
const unsigned int block_size = vi->tree_params.block_size;
u64 pos = (u64)data_folio->index << PAGE_SHIFT;
if (WARN_ON_ONCE(len <= 0 || !IS_ALIGNED(len | offset, block_size)))
return false;
if (WARN_ON_ONCE(!folio_test_locked(data_folio) ||
folio_test_uptodate(data_folio)))
return false;
do {
struct page *data_page =
folio_page(data_folio, offset >> PAGE_SHIFT);
if (!verify_data_block(inode, vi, req, data_page, pos + offset,
offset & ~PAGE_MASK, max_ra_pages))
return false;
offset += block_size;
len -= block_size;
} while (len);
return true;
}
/**
* fsverity_verify_blocks() - verify data in a folio
* @folio: the folio containing the data to verify
* @len: the length of the data to verify in the folio
* @offset: the offset of the data to verify in the folio
*
* Verify data that has just been read from a verity file. The data must be
* located in a pagecache folio that is still locked and not yet uptodate. The
* length and offset of the data must be Merkle tree block size aligned.
*
* Return: %true if the data is valid, else %false.
*/
bool fsverity_verify_blocks(struct folio *folio, size_t len, size_t offset)
{
struct inode *inode = folio->mapping->host;
struct fsverity_info *vi = inode->i_verity_info;
struct ahash_request *req;
bool valid;
/* This allocation never fails, since it's mempool-backed. */
req = fsverity_alloc_hash_request(vi->tree_params.hash_alg, GFP_NOFS);
valid = verify_data_blocks(inode, vi, req, folio, len, offset, 0);
fsverity_free_hash_request(vi->tree_params.hash_alg, req);
return valid;
}
EXPORT_SYMBOL_GPL(fsverity_verify_blocks);
#ifdef CONFIG_BLOCK
/**
* fsverity_verify_bio() - verify a 'read' bio that has just completed
* @bio: the bio to verify
*
* Verify the bio's data against the file's Merkle tree. All bio data segments
* must be aligned to the file's Merkle tree block size. If any data fails
* verification, then bio->bi_status is set to an error status.
*
* This is a helper function for use by the ->readahead() method of filesystems
* that issue bios to read data directly into the page cache. Filesystems that
* populate the page cache without issuing bios (e.g. non block-based
* filesystems) must instead call fsverity_verify_page() directly on each page.
* All filesystems must also call fsverity_verify_page() on holes.
*/
void fsverity_verify_bio(struct bio *bio)
{
struct inode *inode = bio_first_page_all(bio)->mapping->host;
struct fsverity_info *vi = inode->i_verity_info;
struct ahash_request *req;
struct folio_iter fi;
unsigned long max_ra_pages = 0;
/* This allocation never fails, since it's mempool-backed. */
req = fsverity_alloc_hash_request(vi->tree_params.hash_alg, GFP_NOFS);
if (bio->bi_opf & REQ_RAHEAD) {
/*
* If this bio is for data readahead, then we also do readahead
* of the first (largest) level of the Merkle tree. Namely,
* when a Merkle tree page is read, we also try to piggy-back on
* some additional pages -- up to 1/4 the number of data pages.
*
* This improves sequential read performance, as it greatly
* reduces the number of I/O requests made to the Merkle tree.
*/
max_ra_pages = bio->bi_iter.bi_size >> (PAGE_SHIFT + 2);
}
bio_for_each_folio_all(fi, bio) {
if (!verify_data_blocks(inode, vi, req, fi.folio, fi.length,
fi.offset, max_ra_pages)) {
bio->bi_status = BLK_STS_IOERR;
break;
}
}
fsverity_free_hash_request(vi->tree_params.hash_alg, req);
}
EXPORT_SYMBOL_GPL(fsverity_verify_bio);
#endif /* CONFIG_BLOCK */
/**
* fsverity_enqueue_verify_work() - enqueue work on the fs-verity workqueue
* @work: the work to enqueue
*
* Enqueue verification work for asynchronous processing.
*/
void fsverity_enqueue_verify_work(struct work_struct *work)
{
queue_work(fsverity_read_workqueue, work);
}
EXPORT_SYMBOL_GPL(fsverity_enqueue_verify_work);
int __init fsverity_init_workqueue(void)
{
/*
* Use a high-priority workqueue to prioritize verification work, which
* blocks reads from completing, over regular application tasks.
*
* For performance reasons, don't use an unbound workqueue. Using an
* unbound workqueue for crypto operations causes excessive scheduler
* latency on ARM64.
*/
fsverity_read_workqueue = alloc_workqueue("fsverity_read_queue",
WQ_HIGHPRI,
num_online_cpus());
if (!fsverity_read_workqueue)
return -ENOMEM;
return 0;
}
void __init fsverity_exit_workqueue(void)
{
destroy_workqueue(fsverity_read_workqueue);
fsverity_read_workqueue = NULL;
}