On 30.11.2022 18:57, Shigeru Yoshida wrote:
ntfs_trim_fs() should loop with wnd->nwnd, not wnd->nbits. KASAN
detects this as an out-of-bounds access like below:
==================================================================
BUG: KASAN: slab-out-of-bounds in ntfs_trim_fs (fs/ntfs3/bitmap.c:1434)
Read of size 2 at addr ffff8881745b4f02 by task repro/19678
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-2.fc37 04/01/2014
Call Trace:
<TASK>
dump_stack_lvl (lib/dump_stack.c:107 (discriminator 4))
print_report (mm/kasan/report.c:285 mm/kasan/report.c:395)
? __virt_addr_valid (arch/x86/mm/physaddr.c:66)
? __phys_addr (arch/x86/mm/physaddr.c:32 (discriminator 4))
? ntfs_trim_fs (fs/ntfs3/bitmap.c:1434)
? ntfs_trim_fs (fs/ntfs3/bitmap.c:1434)
kasan_report (mm/kasan/report.c:162 mm/kasan/report.c:497)
? ntfs_trim_fs (fs/ntfs3/bitmap.c:1434)
ntfs_trim_fs (fs/ntfs3/bitmap.c:1434)
ntfs_ioctl (fs/ntfs3/file.c:41 fs/ntfs3/file.c:57)
? ntfs_fiemap (fs/ntfs3/file.c:51)
? bpf_lsm_file_ioctl (./include/linux/lsm_hook_defs.h:165)
? ntfs_fiemap (fs/ntfs3/file.c:51)
__x64_sys_ioctl (fs/ioctl.c:52 fs/ioctl.c:870 fs/ioctl.c:856 fs/ioctl.c:856)
do_syscall_64 (arch/x86/entry/common.c:50 arch/x86/entry/common.c:80)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:120)
Signed-off-by: Shigeru Yoshida <syoshida@xxxxxxxxxx>
---
fs/ntfs3/bitmap.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/ntfs3/bitmap.c b/fs/ntfs3/bitmap.c
index e92bbd754365..1930640be31a 100644
--- a/fs/ntfs3/bitmap.c
+++ b/fs/ntfs3/bitmap.c
@@ -1424,7 +1424,7 @@ int ntfs_trim_fs(struct ntfs_sb_info *sbi, struct fstrim_range *range)
down_read_nested(&wnd->rw_lock, BITMAP_MUTEX_CLUSTERS);
- for (; iw < wnd->nbits; iw++, wbit = 0) {
+ for (; iw < wnd->nwnd; iw++, wbit = 0) {
CLST lcn_wnd = iw * wbits;
struct buffer_head *bh;
Thanks for work, this bug has already been fixed:
https://lore.kernel.org/ntfs3/20221001070024.1366018-1-abdun.nihaal@xxxxxxxxx/
