Re: [syzbot] [nilfs?] KMSAN: uninit-value in nilfs_add_checksums_on_logs (2)

[Ryusuke Konishi]

On Sat, Apr 6, 2024 at 8:00 PM syzbot
<syzbot+47a017c46edb25eff048@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
>
> syzbot has found a reproducer for the following issue on:
>
> HEAD commit:    e8b0ccb2a787 Merge tag '9p-for-6.9-rc3' of https://github..
> git tree:       upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=115eb623180000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=5112b3f484393436
> dashboard link: https://syzkaller.appspot.com/bug?extid=47a017c46edb25eff048
> compiler:       Debian clang version 15.0.6, GNU ld (GNU Binutils for Debian) 2.40
> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=156679a1180000
> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=10f27ef6180000
>
> Downloadable assets:
> disk image: https://storage.googleapis.com/syzbot-assets/cf4b0d1e3b2d/disk-e8b0ccb2.raw.xz
> vmlinux: https://storage.googleapis.com/syzbot-assets/422cac6cc940/vmlinux-e8b0ccb2.xz
> kernel image: https://storage.googleapis.com/syzbot-assets/9a4df48e199b/bzImage-e8b0ccb2.xz
> mounted in repro: https://storage.googleapis.com/syzbot-assets/69e1e69e7522/mount_0.gz
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+47a017c46edb25eff048@xxxxxxxxxxxxxxxxxxxxxxxxx
>
> =====================================================
> BUG: KMSAN: uninit-value in crc32_body lib/crc32.c:110 [inline]
> BUG: KMSAN: uninit-value in crc32_le_generic lib/crc32.c:179 [inline]
> BUG: KMSAN: uninit-value in crc32_le_base+0x43c/0xd80 lib/crc32.c:197
>  crc32_body lib/crc32.c:110 [inline]
>  crc32_le_generic lib/crc32.c:179 [inline]
>  crc32_le_base+0x43c/0xd80 lib/crc32.c:197
>  nilfs_segbuf_fill_in_data_crc fs/nilfs2/segbuf.c:224 [inline]
>  nilfs_add_checksums_on_logs+0xb80/0xe40 fs/nilfs2/segbuf.c:327
>  nilfs_segctor_do_construct+0x9876/0xdeb0 fs/nilfs2/segment.c:2078
>  nilfs_segctor_construct+0x1eb/0xe30 fs/nilfs2/segment.c:2381
>  nilfs_segctor_thread_construct fs/nilfs2/segment.c:2489 [inline]
>  nilfs_segctor_thread+0xc50/0x11e0 fs/nilfs2/segment.c:2573
>  kthread+0x3e2/0x540 kernel/kthread.c:388
>  ret_from_fork+0x6d/0x90 arch/x86/kernel/process.c:147
>  ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:243
>
> Uninit was stored to memory at:
>  memcpy_from_iter lib/iov_iter.c:73 [inline]
>  iterate_bvec include/linux/iov_iter.h:122 [inline]
>  iterate_and_advance2 include/linux/iov_iter.h:249 [inline]
>  iterate_and_advance include/linux/iov_iter.h:271 [inline]
>  __copy_from_iter lib/iov_iter.c:249 [inline]
>  copy_page_from_iter_atomic+0x12b7/0x2b60 lib/iov_iter.c:481
>  generic_perform_write+0x4c1/0xc60 mm/filemap.c:3982
>  __generic_file_write_iter+0x20a/0x460 mm/filemap.c:4069
>  generic_file_write_iter+0x103/0x5b0 mm/filemap.c:4095
>  __kernel_write_iter+0x68b/0xc40 fs/read_write.c:523
>  dump_emit_page fs/coredump.c:890 [inline]
>  dump_user_range+0x8dc/0xee0 fs/coredump.c:951
>  elf_core_dump+0x520f/0x59c0 fs/binfmt_elf.c:2077
>  do_coredump+0x32d5/0x4920 fs/coredump.c:764
>  get_signal+0x267e/0x2d00 kernel/signal.c:2896
>  arch_do_signal_or_restart+0x53/0xcb0 arch/x86/kernel/signal.c:310
>  exit_to_user_mode_loop kernel/entry/common.c:111 [inline]
>  exit_to_user_mode_prepare include/linux/entry-common.h:328 [inline]
>  __syscall_exit_to_user_mode_work kernel/entry/common.c:207 [inline]
>  syscall_exit_to_user_mode+0x5d/0x160 kernel/entry/common.c:218
>  do_syscall_64+0xe4/0x1f0 arch/x86/entry/common.c:89
>  entry_SYSCALL_64_after_hwframe+0x72/0x7a
>
> Uninit was created at:
>  __alloc_pages+0x9d6/0xe70 mm/page_alloc.c:4598
>  alloc_pages_mpol+0x299/0x990 mm/mempolicy.c:2264
>  alloc_pages+0x1bf/0x1e0 mm/mempolicy.c:2335
>  dump_user_range+0x4a/0xee0 fs/coredump.c:935
>  elf_core_dump+0x520f/0x59c0 fs/binfmt_elf.c:2077
>  do_coredump+0x32d5/0x4920 fs/coredump.c:764
>  get_signal+0x267e/0x2d00 kernel/signal.c:2896
>  arch_do_signal_or_restart+0x53/0xcb0 arch/x86/kernel/signal.c:310
>  exit_to_user_mode_loop kernel/entry/common.c:111 [inline]
>  exit_to_user_mode_prepare include/linux/entry-common.h:328 [inline]
>  __syscall_exit_to_user_mode_work kernel/entry/common.c:207 [inline]
>  syscall_exit_to_user_mode+0x5d/0x160 kernel/entry/common.c:218
>  do_syscall_64+0xe4/0x1f0 arch/x86/entry/common.c:89
>  entry_SYSCALL_64_after_hwframe+0x72/0x7a
>
> CPU: 0 PID: 5014 Comm: segctord Not tainted 6.9.0-rc2-syzkaller-00207-ge8b0ccb2a787 #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/27/2024
> =====================================================
>
>
> ---
> If you want syzbot to run the reproducer, reply with:
> #syz test: git://repo/address.git branch-or-commit-hash
> If you attach or paste a git patch, syzbot will apply it before testing.

#syz fix: x86: call instrumentation hooks from copy_mc.c

This is one of the false positive warnings that the memory dumped by
elf_core_dump() was mixed into the file system side via
copy_mc_to_kernel() of x86, which was called with the following call
path and did not support KMSAN until recently:

 elf_core_dump
   dump_user_range
      dump_page_copy
          copy_mc_to_kernel
      dump_emit_page
          ...

Given the syzbot CPU information, we can confirm that the x86 ERMS
feature flag is set, a condition that is affected by the issue.

The above commit, which was merged during the merge window for v6.10,
made copy_mc_to_kernel() on x86 KMSAN-compatible and should have fixed
this issue.

Ryusuke Konishi