f2fs: fix to zero post-eof page
Chao Yu
maintainerPATCH v2 1/1 f2fs: fix to zero post-eof page
fstest reports a f2fs bug:
generic/363 42s ... [failed, exit status 1]- output mismatch (see /share/git/fstests/results//generic/363.out.bad)
--- tests/generic/363.out 2025-01-12 21:57:40.271440542 +0800
+++ /share/git/fstests/results//generic/363.out.bad 2025-05-19 19:55:58.000000000 +0800
@@ -1,2 +1,78 @@
QA output created by 363
fsx -q -S 0 -e 1 -N 100000
+READ BAD DATA: offset = 0xd6fb, size = 0xf044, fname = /mnt/f2fs/junk
+OFFSET GOOD BAD RANGE
+0x1540d 0x0000 0x2a25 0x0
+operation# (mod 256) for the bad data may be 37
+0x1540e 0x0000 0x2527 0x1
...
(Run 'diff -u /share/git/fstests/tests/generic/363.out /share/git/fstests/results//generic/363.out.bad' to see the entire diff)Ran: generic/363
Failures: generic/363
Failed 1 of 1 tests
The root cause is user can update post-eof page via mmap, however, f2fs missed to zero post-eof page in below operations, so, once it expands i_size, then it will include dummy data locates previous post-eof page, so during below operations, we need to zero post-eof page.
Operations which can include dummy data after previous i_size after expanding i_size:
- write
- mapwrite
- truncate
- fallocate
* preallocate
* zero_range
* insert_range
* collapse_range
- clone_range (doesn’t support in f2fs)
- copy_range (doesn’t support in f2fs)
---
v2:
- cover f2fs_zero_post_eof_page w/ invalidate_lock
fs/f2fs/file.c | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+)
fs/f2fs/file.c +38 -0
@@ -35,6 +35,17 @@#include <trace/events/f2fs.h> #include <uapi/linux/f2fs.h> +static void f2fs_zero_post_eof_page (struct inode *inode,loff_t new_size) +{ +loff_t old_size =i_size_read (inode); + +if (old_size > new_size) +return ; + +/* zero or drop pages only in range of [old_size, new_size] */ +truncate_pagecache (inode, old_size); +} +static vm_fault_t f2fs_filemap_fault (struct vm_fault *vmf) {struct inode *inode =file_inode (vmf->vma->vm_file);
@@ -103,8 +114,13 @@ static vm_fault_t f2fs_vm_page_mkwrite(struct vm_fault *vmf)f2fs_bug_on (sbi,f2fs_has_inline_data (inode)); +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, (folio->index +1 ) <<PAGE_SHIFT ); +filemap_invalidate_unlock (inode->i_mapping); +file_update_time (vmf->vma->vm_file);filemap_invalidate_lock_shared (inode->i_mapping); +folio_lock (folio); if (unlikely(folio->mapping != inode->i_mapping || folio_pos(folio) > i_size_read(inode) ||
@@ -1109,6 +1125,8 @@ int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,f2fs_down_write (&fi->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (inode->i_mapping); +if (attr->ia_size > old_size) +f2fs_zero_post_eof_page (inode, old_size);truncate_setsize (inode, attr->ia_size);if (attr->ia_size <= old_size)
@@ -1227,6 +1245,10 @@ static int f2fs_punch_hole(struct inode *inode, loff_t offset, loff_t len)if (ret)return ret; +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (inode->i_mapping); + pg_start = ((unsigned long long ) offset) >>PAGE_SHIFT ; pg_end = ((unsigned long long ) offset + len) >>PAGE_SHIFT ;
@@ -1510,6 +1532,8 @@ static int f2fs_do_collapse(struct inode *inode, loff_t offset, loff_t len)f2fs_down_write (&F2FS_I (inode)->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +f2fs_lock_op (sbi);f2fs_drop_extent_tree (inode);truncate_pagecache (inode, offset);
@@ -1631,6 +1655,10 @@ static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len,if (ret)return ret; +filemap_invalidate_lock (mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (mapping); + pg_start = ((unsigned long long ) offset) >>PAGE_SHIFT ; pg_end = ((unsigned long long ) offset + len) >>PAGE_SHIFT ;
@@ -1762,6 +1790,8 @@ static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len)/* avoid gc operation during block exchange */ f2fs_down_write (&F2FS_I (inode)->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (mapping); + +f2fs_zero_post_eof_page (inode, offset + len);truncate_pagecache (inode, offset);while (!ret && idx > pg_start) {
@@ -1819,6 +1849,10 @@ static int f2fs_expand_inode_data(struct inode *inode, loff_t offset,if (err)return err; +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (inode->i_mapping); +f2fs_balance_fs (sbi, true); pg_start = ((unsigned long long )offset) >>PAGE_SHIFT ;
@@ -4860,6 +4894,10 @@ static ssize_t f2fs_write_checks(struct kiocb *iocb, struct iov_iter *from) err =file_modified (file);if (err)return err; + +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, iocb->ki_pos +iov_iter_count (from)); +filemap_invalidate_unlock (inode->i_mapping);return count; }
Signature · 1 line
2.49.0
Chao Yu via Linux-f2fs-devel <[email protected]>
Quoted 51 lines · 于2025年6月4日周三 17:01写道:
fstest reports a f2fs bug:
generic/363 42s ... [failed, exit status 1]- output mismatch (see /share/git/fstests/results//generic/363.out.bad)
--- tests/generic/363.out 2025-01-12 21:57:40.271440542 +0800 +++ /share/git/fstests/results//generic/363.out.bad 2025-05-19 19:55:58.000000000 +0800 @@ -1,2 +1,78 @@ QA output created by 363 fsx -q -S 0 -e 1 -N 100000 +READ BAD DATA: offset = 0xd6fb, size = 0xf044, fname = /mnt/f2fs/junk +OFFSET GOOD BAD RANGE +0x1540d 0x0000 0x2a25 0x0 +operation# (mod 256) for the bad data may be 37 +0x1540e 0x0000 0x2527 0x1 ... (Run 'diff -u /share/git/fstests/tests/generic/363.out /share/git/fstests/results//generic/363.out.bad' to see the entire diff)Ran: generic/363
Failures: generic/363
Failed 1 of 1 testsThe root cause is user can update post-eof page via mmap, however, f2fs missed to zero post-eof page in below operations, so, once it expands
i_size, then it will include dummy data locates previous post-eof page, so during below operations, we need to zero post-eof page.Operations which can include dummy data after previous
i_sizeafter expandingi_size:
- write
- mapwrite
- truncate
- fallocate
* preallocate
*zero_range
*insert_range
*collapse_range
-clone_range(doesn’t support in f2fs)
-copy_range(doesn’t support in f2fs)Signed-off-by Chao Yu <[email protected]>---
v2:
- coverf2fs_zero_post_eof_pagew/invalidate_lockfs/f2fs/file.c | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+)fs/f2fs/file.c +0
-0
fs/f2fs/file.c line 42
Quoted 8 lines
fs/f2fs/file.c +5
-0@@ -35,6 +35,17 @@#include <trace/events/f2fs.h> #include <uapi/linux/f2fs.h> +static void f2fs_zero_post_eof_page (struct inode *inode,loff_t new_size) +{ +loff_t old_size =i_size_read (inode); + +if (old_size > new_size)
Hi Chao,
should it be old_size >= new_size?
Quoted 24 lines
fs/f2fs/file.c +11
-0+return ; + +/* zero or drop pages only in range of [old_size, new_size] */ +truncate_pagecache (inode, old_size); +} +static vm_fault_t f2fs_filemap_fault (struct vm_fault *vmf) {struct inode *inode =file_inode (vmf->vma->vm_file);@@ -103,8 +114,13 @@ static vm_fault_t f2fs_vm_page_mkwrite(struct vm_fault *vmf)f2fs_bug_on (sbi,f2fs_has_inline_data (inode)); +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, (folio->index +1 ) <<PAGE_SHIFT ); +filemap_invalidate_unlock (inode->i_mapping); +file_update_time (vmf->vma->vm_file);filemap_invalidate_lock_shared (inode->i_mapping); +folio_lock (folio); if (unlikely(folio->mapping != inode->i_mapping || folio_pos(folio) > i_size_read(inode) ||
fs/f2fs/file.c int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, · line 1129
Quoted 5 lines
fs/f2fs/file.c +2
-0@@ -1109,6 +1125,8 @@ int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,f2fs_down_write (&fi->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (inode->i_mapping); +if (attr->ia_size > old_size) +f2fs_zero_post_eof_page (inode, old_size);
I'm a little suspicious about the logic here, here old_size is from i_size_read(inode),
and it will compare with the "old_size" in f2fs_zero_post_eof_page, it is also from i_size_read(inode),
so is this actually meaningless?
Thanks!
Quoted 74 lines
fs/f2fs/file.c +20
-1truncate_setsize (inode, attr->ia_size);if (attr->ia_size <= old_size)@@ -1227,6 +1245,10 @@ static int f2fs_punch_hole(struct inode *inode, loff_t offset, loff_t len)if (ret)return ret; +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (inode->i_mapping); + pg_start = ((unsigned long long ) offset) >>PAGE_SHIFT ; pg_end = ((unsigned long long ) offset + len) >>PAGE_SHIFT ;@@ -1510,6 +1532,8 @@ static int f2fs_do_collapse(struct inode *inode, loff_t offset, loff_t len)f2fs_down_write (&F2FS_I (inode)->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +f2fs_lock_op (sbi);f2fs_drop_extent_tree (inode);truncate_pagecache (inode, offset);@@ -1631,6 +1655,10 @@ static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len,if (ret)return ret; +filemap_invalidate_lock (mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (mapping); + pg_start = ((unsigned long long ) offset) >>PAGE_SHIFT ; pg_end = ((unsigned long long ) offset + len) >>PAGE_SHIFT ;@@ -1762,6 +1790,8 @@ static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len)/* avoid gc operation during block exchange */ f2fs_down_write (&F2FS_I (inode)->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (mapping); + +f2fs_zero_post_eof_page (inode, offset + len);truncate_pagecache (inode, offset);while (!ret && idx > pg_start) {@@ -1819,6 +1849,10 @@ static int f2fs_expand_inode_data(struct inode *inode, loff_t offset,if (err)return err; +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (inode->i_mapping); +f2fs_balance_fs (sbi, true); pg_start = ((unsigned long long )offset) >>PAGE_SHIFT ;@@ -4860,6 +4894,10 @@ static ssize_t f2fs_write_checks(struct kiocb *iocb, struct iov_iter *from) err =file_modified (file);if (err)return err; + +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, iocb->ki_pos +iov_iter_count (from)); +filemap_invalidate_unlock (inode->i_mapping);return count; } --2.49.0
_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
Chao Yu
maintainerQuoted 56 lines · Zhiguo Niu
Chao Yu via Linux-f2fs-devel <[email protected]>
Quoted 54 lines · 于2025年6月4日周三 17:01写道:
fstest reports a f2fs bug:
generic/363 42s ... [failed, exit status 1]- output mismatch (see /share/git/fstests/results//generic/363.out.bad)
--- tests/generic/363.out 2025-01-12 21:57:40.271440542 +0800 +++ /share/git/fstests/results//generic/363.out.bad 2025-05-19 19:55:58.000000000 +0800 @@ -1,2 +1,78 @@ QA output created by 363 fsx -q -S 0 -e 1 -N 100000 +READ BAD DATA: offset = 0xd6fb, size = 0xf044, fname = /mnt/f2fs/junk +OFFSET GOOD BAD RANGE +0x1540d 0x0000 0x2a25 0x0 +operation# (mod 256) for the bad data may be 37 +0x1540e 0x0000 0x2527 0x1 ... (Run 'diff -u /share/git/fstests/tests/generic/363.out /share/git/fstests/results//generic/363.out.bad' to see the entire diff)Ran: generic/363
Failures: generic/363
Failed 1 of 1 testsThe root cause is user can update post-eof page via mmap, however, f2fs missed to zero post-eof page in below operations, so, once it expands
i_size, then it will include dummy data locates previous post-eof page, so during below operations, we need to zero post-eof page.Operations which can include dummy data after previous
i_sizeafter expandingi_size:
- write
- mapwrite
- truncate
- fallocate
* preallocate
*zero_range
*insert_range
*collapse_range
-clone_range(doesn’t support in f2fs)
-copy_range(doesn’t support in f2fs)Signed-off-by Chao Yu <[email protected]>---
v2:
- coverf2fs_zero_post_eof_pagew/invalidate_lockfs/f2fs/file.c | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+)fs/f2fs/file.c +0
-0@@ -35,6 +35,17 @@#include <trace/events/f2fs.h> #include <uapi/linux/f2fs.h>
Quoted 7 lines
fs/f2fs/file.c line 42
Quoted 5 lines
fs/f2fs/file.c +5
-0+static void f2fs_zero_post_eof_page (struct inode *inode,loff_t new_size) +{ +loff_t old_size =i_size_read (inode); + +if (old_size > new_size)Hi Chao,
should it beold_size>=new_size?
Zhiguo,
Yes, you're right.
Quoted 28 lines
Quoted 28 lines
fs/f2fs/file.c +11
-0+return ; + +/* zero or drop pages only in range of [old_size, new_size] */ +truncate_pagecache (inode, old_size); +} +static vm_fault_t f2fs_filemap_fault (struct vm_fault *vmf) {struct inode *inode =file_inode (vmf->vma->vm_file);@@ -103,8 +114,13 @@ static vm_fault_t f2fs_vm_page_mkwrite(struct vm_fault *vmf)f2fs_bug_on (sbi,f2fs_has_inline_data (inode)); +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, (folio->index +1 ) <<PAGE_SHIFT ); +filemap_invalidate_unlock (inode->i_mapping); +file_update_time (vmf->vma->vm_file);filemap_invalidate_lock_shared (inode->i_mapping); +folio_lock (folio); if (unlikely(folio->mapping != inode->i_mapping || folio_pos(folio) > i_size_read(inode) ||@@ -1109,6 +1125,8 @@ int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,f2fs_down_write (&fi->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (inode->i_mapping);
Quoted 7 lines
fs/f2fs/file.c int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, · line 1129
Quoted 2 lines
fs/f2fs/file.c +2
-0+if (attr->ia_size > old_size) +f2fs_zero_post_eof_page (inode, old_size);I'm a little suspicious about the logic here, here
old_sizeis fromi_size_read(inode),
and it will compare with the "old_size" inf2fs_zero_post_eof_page, it is also fromi_size_read(inode),
so is this actually meaningless?
Oh, I need to pass attr->ia_size instead of old_size.
Let me revise in v3, anyway, thanks for the review.
Thanks,
Quoted 75 lines
Thanks!
Quoted 74 lines
fs/f2fs/file.c +20
-1truncate_setsize (inode, attr->ia_size);if (attr->ia_size <= old_size)@@ -1227,6 +1245,10 @@ static int f2fs_punch_hole(struct inode *inode, loff_t offset, loff_t len)if (ret)return ret; +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (inode->i_mapping); + pg_start = ((unsigned long long ) offset) >>PAGE_SHIFT ; pg_end = ((unsigned long long ) offset + len) >>PAGE_SHIFT ;@@ -1510,6 +1532,8 @@ static int f2fs_do_collapse(struct inode *inode, loff_t offset, loff_t len)f2fs_down_write (&F2FS_I (inode)->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +f2fs_lock_op (sbi);f2fs_drop_extent_tree (inode);truncate_pagecache (inode, offset);@@ -1631,6 +1655,10 @@ static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len,if (ret)return ret; +filemap_invalidate_lock (mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (mapping); + pg_start = ((unsigned long long ) offset) >>PAGE_SHIFT ; pg_end = ((unsigned long long ) offset + len) >>PAGE_SHIFT ;@@ -1762,6 +1790,8 @@ static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len)/* avoid gc operation during block exchange */ f2fs_down_write (&F2FS_I (inode)->i_gc_rwsem[WRITE ]);filemap_invalidate_lock (mapping); + +f2fs_zero_post_eof_page (inode, offset + len);truncate_pagecache (inode, offset);while (!ret && idx > pg_start) {@@ -1819,6 +1849,10 @@ static int f2fs_expand_inode_data(struct inode *inode, loff_t offset,if (err)return err; +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, offset + len); +filemap_invalidate_unlock (inode->i_mapping); +f2fs_balance_fs (sbi, true); pg_start = ((unsigned long long )offset) >>PAGE_SHIFT ;@@ -4860,6 +4894,10 @@ static ssize_t f2fs_write_checks(struct kiocb *iocb, struct iov_iter *from) err =file_modified (file);if (err)return err; + +filemap_invalidate_lock (inode->i_mapping); +f2fs_zero_post_eof_page (inode, iocb->ki_pos +iov_iter_count (from)); +filemap_invalidate_unlock (inode->i_mapping);return count; } --2.49.0
_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel