Skip to contentSkip to conversation

squashfs: fixups for caching

PATCH Superseded Vincent Whitchurch 2 participants 5 messages May 26, 2023 SquashFS file system

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover squashfs: fixups for caching

These are a couple of fixups for squashfs-cache-partial-compressed-blocks.patch which is currently in mm-nonmm-unstable.

---
Vincent Whitchurch (2):
squashfs: fix page update race
squashfs: fix page indices

 fs/squashfs/block.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

---
base-commit: 84cc8b966a3d4cde585761d05cc448dc1da0824f
change-id: 20230526-squashfs-cache-fixup-194431de42eb

Best regards,

Signature · 1 line
Vincent Whitchurch <[email protected]>
#1 lore raw

PATCH 1/2 squashfs: fix page update race

We only put the page into the cache after we've read it, so the PageUptodate() check should not be necessary. In fact, it's actively harmful since the check could fail (since we used find_get_page() and not find_lock_page()) and we could end up submitting a page for I/O after it has been read and while it's actively being used, which could lead to corruption depending on what the block driver does with it.

---
 fs/squashfs/block.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/squashfs/block.c +1 -1
index 6285f5afb6c6..f2412e5fc84b 100644
@@ -92,7 +92,7 @@ static int squashfs_bio_read_cached(struct bio *fullbio,
 	bio_for_each_segment_all(bv, fullbio, iter_all) {
 		struct page *page = bv->bv_page;
 
-		if (page->mapping == cache_mapping && PageUptodate(page)) {
+		if (page->mapping == cache_mapping) {
 			idx++;
 			continue;
 		}
Signature · 1 line
2.34.1
#3 lore raw

PATCH 2/2 squashfs: fix page indices

In reply to Vincent Whitchurch

The page cache functions want the page index as an argument but we're currently passing in the byte address.

---
 fs/squashfs/block.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
fs/squashfs/block.c +3 -3
index f2412e5fc84b..447fb04f2b61 100644
@@ -142,7 +142,7 @@ static int squashfs_bio_read_cached(struct bio *fullbio,
 
 	if (head_to_cache) {
 		int ret = add_to_page_cache_lru(head_to_cache, cache_mapping,
-						read_start, GFP_NOIO);
+						read_start >> PAGE_SHIFT, GFP_NOIO);
 
 		if (!ret) {
 			SetPageUptodate(head_to_cache);
@@ -153,7 +153,7 @@ static int squashfs_bio_read_cached(struct bio *fullbio,
 
 	if (tail_to_cache) {
 		int ret = add_to_page_cache_lru(tail_to_cache, cache_mapping,
-						read_end - PAGE_SIZE, GFP_NOIO);
+						(read_end >> PAGE_SHIFT) - 1, GFP_NOIO);
 
 		if (!ret) {
 			SetPageUptodate(tail_to_cache);
@@ -192,7 +192,7 @@ static int squashfs_bio_read(struct super_block *sb, u64 index, int length,
 
 		if (cache_mapping)
 			page = find_get_page(cache_mapping,
-					     read_start + i * PAGE_SIZE);
+					     (read_start >> PAGE_SHIFT) + i);
 		if (!page)
 			page = alloc_page(GFP_NOIO);
 
Signature · 1 line
2.34.1
#4 lore raw

Re: [PATCH mm-nonmm-unstable 2/2] squashfs: fix page indices

Quoted 2 lines · Vincent Whitchurch

The page cache functions want the page index as an argument but we're currently passing in the byte address.

Can you avoid the overly long lines by movning the GFP_NOFS parameters to the following lines?

The rest look good.