ext4: fix kernel BUG in 'ext4_write_inline_data_end()'
authorYe Bin <yebin10@huawei.com>
Tue, 6 Dec 2022 14:41:34 +0000 (22:41 +0800)
committerTheodore Ts'o <tytso@mit.edu>
Fri, 9 Dec 2022 02:49:25 +0000 (21:49 -0500)
Syzbot report follow issue:
------------[ cut here ]------------
kernel BUG at fs/ext4/inline.c:227!
invalid opcode: 0000 [#1] PREEMPT SMP KASAN
CPU: 1 PID: 3629 Comm: syz-executor212 Not tainted 6.1.0-rc5-syzkaller-00018-g59d0d52c30d4 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022
RIP: 0010:ext4_write_inline_data+0x344/0x3e0 fs/ext4/inline.c:227
RSP: 0018:ffffc90003b3f368 EFLAGS: 00010293
RAX: 0000000000000000 RBX: ffff8880704e16c0 RCX: 0000000000000000
RDX: ffff888021763a80 RSI: ffffffff821e31a4 RDI: 0000000000000006
RBP: 000000000006818e R08: 0000000000000006 R09: 0000000000068199
R10: 0000000000000079 R11: 0000000000000000 R12: 000000000000000b
R13: 0000000000068199 R14: ffffc90003b3f408 R15: ffff8880704e1c82
FS:  000055555723e3c0(0000) GS:ffff8880b9b00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fffe8ac9080 CR3: 0000000079f81000 CR4: 0000000000350ee0
Call Trace:
 <TASK>
 ext4_write_inline_data_end+0x2a3/0x12f0 fs/ext4/inline.c:768
 ext4_write_end+0x242/0xdd0 fs/ext4/inode.c:1313
 ext4_da_write_end+0x3ed/0xa30 fs/ext4/inode.c:3063
 generic_perform_write+0x316/0x570 mm/filemap.c:3764
 ext4_buffered_write_iter+0x15b/0x460 fs/ext4/file.c:285
 ext4_file_write_iter+0x8bc/0x16e0 fs/ext4/file.c:700
 call_write_iter include/linux/fs.h:2191 [inline]
 do_iter_readv_writev+0x20b/0x3b0 fs/read_write.c:735
 do_iter_write+0x182/0x700 fs/read_write.c:861
 vfs_iter_write+0x74/0xa0 fs/read_write.c:902
 iter_file_splice_write+0x745/0xc90 fs/splice.c:686
 do_splice_from fs/splice.c:764 [inline]
 direct_splice_actor+0x114/0x180 fs/splice.c:931
 splice_direct_to_actor+0x335/0x8a0 fs/splice.c:886
 do_splice_direct+0x1ab/0x280 fs/splice.c:974
 do_sendfile+0xb19/0x1270 fs/read_write.c:1255
 __do_sys_sendfile64 fs/read_write.c:1323 [inline]
 __se_sys_sendfile64 fs/read_write.c:1309 [inline]
 __x64_sys_sendfile64+0x1d0/0x210 fs/read_write.c:1309
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x39/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x63/0xcd
---[ end trace 0000000000000000 ]---

Above issue may happens as follows:
ext4_da_write_begin
  ext4_da_write_inline_data_begin
    ext4_da_convert_inline_data_to_extent
      ext4_clear_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA);
ext4_da_write_end

ext4_run_li_request
  ext4_mb_prefetch
    ext4_read_block_bitmap_nowait
      ext4_validate_block_bitmap
        ext4_mark_group_bitmap_corrupted(sb, block_group, EXT4_GROUP_INFO_BBITMAP_CORRUPT)
 percpu_counter_sub(&sbi->s_freeclusters_counter,grp->bb_free);
  -> sbi->s_freeclusters_counter become zero
ext4_da_write_begin
  if (ext4_nonda_switch(inode->i_sb)) -> As freeclusters_counter is zero will return true
    *fsdata = (void *)FALL_BACK_TO_NONDELALLOC;
    ext4_write_begin
ext4_da_write_end
  if (write_mode == FALL_BACK_TO_NONDELALLOC)
    ext4_write_end
      if (inline_data)
        ext4_write_inline_data_end
  ext4_write_inline_data
    BUG_ON(pos + len > EXT4_I(inode)->i_inline_size);
           -> As inode is already convert to extent, so 'pos + len' > inline_size
   -> then trigger BUG.

To solve this issue, instead of checking ext4_has_inline_data() which
is only cleared after data has been written back, check the
EXT4_STATE_MAY_INLINE_DATA flag in ext4_write_end().

Fixes: f19d5870cbf7 ("ext4: add normal write support for inline data")
Reported-by: syzbot+4faa160fa96bfba639f8@syzkaller.appspotmail.com
Reported-by: Jun Nie <jun.nie@linaro.org>
Signed-off-by: Ye Bin <yebin10@huawei.com>
Link: https://lore.kernel.org/r/20221206144134.1919987-1-yebin@huaweicloud.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org
fs/ext4/inode.c

index 181bc161b1ac3dcf4cca539911345f07f7384f7b..a0f4d4197a0b710db296272fe9e4a0bd4e0abd6c 100644 (file)
@@ -1315,7 +1315,8 @@ static int ext4_write_end(struct file *file,
 
        trace_ext4_write_end(inode, pos, len, copied);
 
-       if (ext4_has_inline_data(inode))
+       if (ext4_has_inline_data(inode) &&
+           ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA))
                return ext4_write_inline_data_end(inode, pos, len, copied, page);
 
        copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);