KMSAN: uninit-value in fat_subdirs

syzbot Applied syzbot 4 participants 5 messages Jan 7, 2022 – Jan 20, 2022 Block layer

lore

Share & tools

lore

syzbot

author syzbot

2022-01-07 11:40 UTC #0 lore raw

Hello,

syzbot found the following issue on:

HEAD commit:    81c325bbf94e kmsan: hooks: do not check memory in kmsan_in..
git tree:       https://github.com/google/kmsan.git master
console output: https://syzkaller.appspot.com/x/log.txt?x=11506cb3b00000
kernel config:  https://syzkaller.appspot.com/x/.config?x=2d8b9a11641dc9aa
dashboard link: https://syzkaller.appspot.com/bug?extid=ac94ae5f68b84197f41c
compiler:       clang version 14.0.0 (/usr/local/google/src/llvm-git-monorepo 2b554920f11c8b763cd9ed9003f4e19b919b8e1f), GNU ld (GNU Binutils for Debian) 2.35.2
userspace arch: i386

Unfortunately, I don't have any reproducer for this issue yet.

IMPORTANT: if you fix the issue, please add the following tag to the commit:

Reported-by [email protected]

loop3: detected capacity change from 0 to 7
=====================================================

BUG: KMSAN: uninit-value in fat_get_short_entry fs/fat/dir.c:875 [inline]
BUG: KMSAN: uninit-value in fat_subdirs+0x24b/0x460 fs/fat/dir.c:939
 fat_get_short_entry fs/fat/dir.c:875 [inline]
 fat_subdirs+0x24b/0x460 fs/fat/dir.c:939
 fat_read_root+0x92f/0xac0 fs/fat/inode.c:1413
 fat_fill_super+0x6e5b/0x7ff0 fs/fat/inode.c:1860
 vfat_fill_super+0xa6/0xc0 fs/fat/namei_vfat.c:1051
 mount_bdev+0x626/0x920 fs/super.c:1370
 vfat_mount+0xc9/0xe0 fs/fat/namei_vfat.c:1058
 legacy_get_tree+0x163/0x2e0 fs/fs_context.c:610
 vfs_get_tree+0xd8/0x5d0 fs/super.c:1500
 do_new_mount+0x7b5/0x16f0 fs/namespace.c:2988
 path_mount+0x1021/0x28b0 fs/namespace.c:3318
 do_mount fs/namespace.c:3331 [inline]
 __do_sys_mount fs/namespace.c:3539 [inline]
 __se_sys_mount+0x8a8/0x9d0 fs/namespace.c:3516
 __ia32_sys_mount+0x157/0x1b0 fs/namespace.c:3516
 do_syscall_32_irqs_on arch/x86/entry/common.c:114 [inline]
 __do_fast_syscall_32+0x96/0xf0 arch/x86/entry/common.c:180
 do_fast_syscall_32+0x34/0x70 arch/x86/entry/common.c:205
 do_SYSENTER_32+0x1b/0x20 arch/x86/entry/common.c:248
 entry_SYSENTER_compat_after_hwframe+0x4d/0x5c

Uninit was created at:

 __alloc_pages+0xbbf/0x1090 mm/page_alloc.c:5409
 alloc_pages+0x8a5/0xb80
 folio_alloc+0x7b/0x180 mm/mempolicy.c:2201
 filemap_alloc_folio mm/filemap.c:1036 [inline]
 __filemap_get_folio+0xee2/0x1b20 mm/filemap.c:1951
 pagecache_get_page+0xc6/0x3a0 mm/folio-compat.c:125
 find_or_create_page include/linux/pagemap.h:489 [inline]
 grow_dev_page+0x1b6/0xe00 fs/buffer.c:949
 grow_buffers fs/buffer.c:1014 [inline]
 __getblk_slow fs/buffer.c:1041 [inline]
 __getblk_gfp+0x410/0x620 fs/buffer.c:1334
 __bread_gfp+0xb3/0x810 fs/buffer.c:1379
 sb_bread include/linux/buffer_head.h:303 [inline]
 fat_fill_super+0x38d2/0x7ff0 fs/fat/inode.c:1685
 vfat_fill_super+0xa6/0xc0 fs/fat/namei_vfat.c:1051
 mount_bdev+0x626/0x920 fs/super.c:1370
 vfat_mount+0xc9/0xe0 fs/fat/namei_vfat.c:1058
 legacy_get_tree+0x163/0x2e0 fs/fs_context.c:610
 vfs_get_tree+0xd8/0x5d0 fs/super.c:1500
 do_new_mount+0x7b5/0x16f0 fs/namespace.c:2988
 path_mount+0x1021/0x28b0 fs/namespace.c:3318
 do_mount fs/namespace.c:3331 [inline]
 __do_sys_mount fs/namespace.c:3539 [inline]
 __se_sys_mount+0x8a8/0x9d0 fs/namespace.c:3516
 __ia32_sys_mount+0x157/0x1b0 fs/namespace.c:3516
 do_syscall_32_irqs_on arch/x86/entry/common.c:114 [inline]
 __do_fast_syscall_32+0x96/0xf0 arch/x86/entry/common.c:180
 do_fast_syscall_32+0x34/0x70 arch/x86/entry/common.c:205
 do_SYSENTER_32+0x1b/0x20 arch/x86/entry/common.c:248
 entry_SYSENTER_compat_after_hwframe+0x4d/0x5c

CPU: 1 PID: 5570 Comm: syz-executor.3 Tainted: G        W         5.16.0-rc5-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
=====================================================

---
This report is generated by a bot. It may contain errors. See https://goo.gl/tpsmEJ for more information about syzbot. syzbot engineers can be reached at [email protected].

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

OGAWA Hirofumi

2022-01-09 09:36 UTC #1 lore raw

PATCH 1/1 block: Fix wrong offset in bio_truncate()

bio_truncate() clears the buffer outside of last block of bdev, however current bio_truncate() is using the wrong offset of page. So it can return the uninitialized data.

This happened when both of truncated/corrupted FS and userspace (via bdev) are trying to read the last of bdev.

Reported-by [email protected]

Signed-off-by OGAWA Hirofumi <[email protected]>

---
 block/bio.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

block/bio.c +2 -1

index a6fb6a0..25f1ed2 100644
@@ -567,7 +567,8 @@ void bio_truncate(struct bio *bio, unsig
 				offset = new_size - done;
 			else
 				offset = 0;
-			zero_user(bv.bv_page, offset, bv.bv_len - offset);
+			zero_user(bv.bv_page, bv.bv_offset + offset,
+				  bv.bv_len - offset);
 			truncated = true;
 		}
 		done += bv.bv_len;

Signature · 1 line

OGAWA Hirofumi <[email protected]>

Ming Lei

Reviewed-by on v1 1/1

2022-01-10 08:02 UTC #2 lore raw

Re: [PATCH] block: Fix wrong offset in bio_truncate()

Quoted 18 lines · OGAWA Hirofumi

bio_truncate() clears the buffer outside of last block of bdev, however current bio_truncate() is using the wrong offset of page. So it can return the uninitialized data.
This happened when both of truncated/corrupted FS and userspace (via bdev) are trying to read the last of bdev.
Reported-by [email protected]
Signed-off-by OGAWA Hirofumi <[email protected]>
---
 block/bio.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
block/bio.c +0 -0
index a6fb6a0..25f1ed2 100644

block/bio.c void bio_truncate(struct bio *bio, unsig · line 571

Quoted 6 lines

block/bio.c +2 -1

@@ -567,7 +567,8 @@ void bio_truncate(struct bio *bio, unsig
 				offset = new_size - done;
 			else
 				offset = 0;
-			zero_user(bv.bv_page, offset, bv.bv_len - offset);
+			zero_user(bv.bv_page, bv.bv_offset + offset,
+				  bv.bv_len - offset);

Looks correct:

Reviewed-by Ming Lei <[email protected]>

Signature · 1 line

Ming

OGAWA Hirofumi

2022-01-20 04:43 UTC #3 lore raw

Re: [PATCH] block: Fix wrong offset in bio_truncate()

Quoted 28 lines · Ming Lei

Quoted 18 lines · OGAWA Hirofumi
bio_truncate() clears the buffer outside of last block of bdev, however current bio_truncate() is using the wrong offset of page. So it can return the uninitialized data.
This happened when both of truncated/corrupted FS and userspace (via bdev) are trying to read the last of bdev.
Reported-by [email protected]
Signed-off-by OGAWA Hirofumi <[email protected]>
---
 block/bio.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
block/bio.c +0 -0
index a6fb6a0..25f1ed2 100644
block/bio.c void bio_truncate(struct bio *bio, unsig · line 571
Quoted 6 lines
block/bio.c +2 -1
@@ -567,7 +567,8 @@ void bio_truncate(struct bio *bio, unsig
 				offset = new_size - done;
 			else
 				offset = 0;
-			zero_user(bv.bv_page, offset, bv.bv_len - offset);
+			zero_user(bv.bv_page, bv.bv_offset + offset,
+				  bv.bv_len - offset);
Looks correct:

Quoted 1 line

Reviewed-by Ming Lei <[email protected]>

ping?

Signature · 1 line

OGAWA Hirofumi <[email protected]>

Jens Axboe

maintainer Applied

2022-01-20 13:30 UTC #4 lore raw

Re: [PATCH] block: Fix wrong offset in bio_truncate()

In reply to OGAWA Hirofumi

Quoted 8 lines · OGAWA Hirofumi

bio_truncate() clears the buffer outside of last block of bdev, however current bio_truncate() is using the wrong offset of page. So it can return the uninitialized data.
This happened when both of truncated/corrupted FS and userspace (via bdev) are trying to read the last of bdev.
[...]

Applied, thanks!

[1/1] block: Fix wrong offset in bio_truncate()

      commit: 3ee859e384d453d6ac68bfd5971f630d9fa46ad3

Best regards,

Signature · 1 line

Jens Axboe