On Tue, Mar 19, 2024 at 04:46:13PM +0530, Disha Goel wrote:
> On power systems with 64k block size (where default page size is 64k) we
> encountered a kernel oops due to an integer overflow issue when writing
> near the last logical block of a file. The allocator could allocate a
> range where the end exceeds the maximum supported logical block
> (UINT32_MAX), leading to a subsequent BUG_ON. This issue has been
> addressed in the upstream kernel with commit 2dcf5fde6dff
> ("ext4: prevent the normalized size from exceeding EXT_MAX_BLOCKS").
>
> ==================================================
> kernel BUG at fs/ext4/mballoc.c:4448!
> Oops: Exception in kernel mode, sig: 5 [#1]
> CPU: 8 PID: 2880554 Comm: xfs_io
> NIP ext4_mb_use_inode_pa+0x110/0x160 [ext4]
> LR ext4_mb_use_inode_pa+0xac/0x160 [ext4]
> Call Trace:
> ext4_mb_new_inode_pa+0x134/0x3a0 [ext4]
> ext4_mb_try_best_found+0x158/0x280 [ext4]
> ext4_mb_regular_allocator+0x16c/0x940 [ext4]
> ext4_mb_new_blocks+0x610/0x960 [ext4]
> ext4_ext_map_blocks+0x858/0xa90 [ext4]
> ext4_map_blocks+0x218/0x800 [ext4]
> ext4_iomap_alloc+0x10c/0x260 [ext4]
> ext4_iomap_begin+0xfc/0x1f0 [ext4]
> iomap_iter+0xf0/0x190
> __iomap_dio_rw+0x208/0x690
> iomap_dio_rw+0x20/0x80
> ext4_dio_write_iter+0x210/0x4d0 [ext4]
> vfs_write+0x364/0x4e0
> sys_pwrite64+0xd4/0x120
> system_call_exception+0x164/0x310
> system_call_vectored_common+0xe8/0x278
> ==================================================
>
> This test has been extended to provide a hint about the relevant fix
> in case of failure.
>
> Signed-off-by: Disha Goel <disgoel@xxxxxxxxxxxxx>
> ---
Looks good to me, will merge it if no more review points from ext4 list.
Reviewed-by: Zorro Lang <zlang@xxxxxxxxxx>
> tests/generic/735 | 13 ++++++++-----
> 1 file changed, 8 insertions(+), 5 deletions(-)
>
> diff --git a/tests/generic/735 b/tests/generic/735
> index 75b23d5e..0ba111a6 100755
> --- a/tests/generic/735
> +++ b/tests/generic/735
> @@ -6,16 +6,19 @@
> #
> # Append writes to a file with logical block numbers close to 0xffffffff
> # and observe if a kernel crash is caused by ext4_lblk_t overflow triggering
> -# BUG_ON at ext4_mb_new_inode_pa(). This is a regression test for commit
> -# bc056e7163ac ("ext4: fix BUG in ext4_mb_new_inode_pa() due to overflow")
> -#
> +# BUG_ON at ext4_mb_new_inode_pa(). This is a regression test for
> +# commit bc056e7163ac ("ext4: fix BUG in ext4_mb_new_inode_pa() due to overflow")
> +# commit 2dcf5fde6dff ("ext4: prevent the normalized size from exceeding EXT_MAX_BLOCKS")
> +
> . ./common/preamble
> . ./common/populate
> _begin_fstest auto quick insert prealloc
>
> # real QA test starts here
> -[[ "$FSTYP" =~ ext[0-9]+ ]] && _fixed_by_kernel_commit bc056e7163ac \
> - "ext4: fix BUG in ext4_mb_new_inode_pa() due to overflow"
> +if [[ "$FSTYP" =~ ext[0-9]+ ]]; then
> + _fixed_by_kernel_commit bc056e7163ac "ext4: fix BUG in ext4_mb_new_inode_pa() due to overflow"
> + _fixed_by_kernel_commit 2dcf5fde6dff "ext4: prevent the normalized size from exceeding EXT_MAX_BLOCKS"
> +fi
>
> _require_odirect
> _require_xfs_io_command "falloc"
> --
> 2.39.1
>
>
