On 9/20/22 18:59, Tetsuo Handa wrote:
syzbot is reporting shift-out-of-bounds in true_sectors_per_clst() [1], for
commit a3b774342fa752a5 ("fs/ntfs3: validate BOOT sectors_per_clusters")
did not address that (0 - boot->sectors_per_clusters) < 0 because "u8" was
chosen for type of boot->sectors_per_clusters because 0x80 needs to be
positive in order to support 64K clusters. Use "s8" cast in order to make
sure that (0 - (s8) boot->sectors_per_clusters) > 0.
Link: https://syzkaller.appspot.com/bug?extid=1631f09646bc214d2e76 [1]
Reported-by: syzbot <syzbot+1631f09646bc214d2e76@xxxxxxxxxxxxxxxxxxxxxxxxx>
Signed-off-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
Tested-by: syzbot <syzbot+1631f09646bc214d2e76@xxxxxxxxxxxxxxxxxxxxxxxxx>
Fixes: a3b774342fa752a5 ("fs/ntfs3: validate BOOT sectors_per_clusters")
diff --git a/fs/ntfs3/super.c b/fs/ntfs3/super.c
index 47012c9bf505..c7ffd21fb255 100644
--- a/fs/ntfs3/super.c
+++ b/fs/ntfs3/super.c
@@ -672,7 +672,7 @@ static u32 true_sectors_per_clst(const struct NTFS_BOOT *boot)
if (boot->sectors_per_clusters <= 0x80)
return boot->sectors_per_clusters;
if (boot->sectors_per_clusters >= 0xf4) /* limit shift to 2MB max */
- return 1U << (0 - boot->sectors_per_clusters);
+ return 1U << (0 - (s8) boot->sectors_per_clusters);
return -EINVAL;
}
Hello
Thanks for patch, but there was already a similar patch by Shigeru Yoshida,
so I chose it.
Sorry about that, thanks again for your work.
