ext4: Fix crash due to getting bogus eh_depth value on big-endian systems
authorPaul Mackerras <paulus@samba.org>
Mon, 12 Dec 2011 16:00:56 +0000 (11:00 -0500)
committerTheodore Ts'o <tytso@mit.edu>
Mon, 12 Dec 2011 16:00:56 +0000 (11:00 -0500)
commitb4611abfa98af8351bf32e8b9ecc9d3384931c37
treea1ff17e10c37207e83bc602cf1fc8045d778c8a5
parentb5a7e97039a80fae673ccc115ce595d5b88fb4ee
ext4: Fix crash due to getting bogus eh_depth value on big-endian systems

Commit 1939dd84b3 ("ext4: cleanup ext4_ext_grow_indepth code") added a
reference to ext4_extent_header.eh_depth, but forget to pass the value
read through le16_to_cpu.  The result is a crash on big-endian
machines, such as this crash on a POWER7 server:

attempt to access beyond end of device
sda8: rw=0, want=776392648163376, limit=168558560
Unable to handle kernel paging request for data at address 0x6b6b6b6b6b6b6bcb
Faulting instruction address: 0xc0000000001f5f38
cpu 0x14: Vector: 300 (Data Access) at [c000001bd1aaecf0]
    pc: c0000000001f5f38: .__brelse+0x18/0x60
    lr: c0000000002e07a4: .ext4_ext_drop_refs+0x44/0x80
    sp: c000001bd1aaef70
   msr: 9000000000009032
   dar: 6b6b6b6b6b6b6bcb
 dsisr: 40000000
  current = 0xc000001bd15b8010
  paca    = 0xc00000000ffe4600
    pid   = 19911, comm = flush-8:0
enter ? for help
[c000001bd1aaeff0c0000000002e07a4 .ext4_ext_drop_refs+0x44/0x80
[c000001bd1aaf090c0000000002e0c58 .ext4_ext_find_extent+0x408/0x4c0
[c000001bd1aaf180c0000000002e145c .ext4_ext_insert_extent+0x2bc/0x14c0
[c000001bd1aaf2c0c0000000002e3fb8 .ext4_ext_map_blocks+0x628/0x1710
[c000001bd1aaf420c0000000002b2974 .ext4_map_blocks+0x224/0x310
[c000001bd1aaf4d0c0000000002b7f2c .mpage_da_map_and_submit+0xbc/0x490
[c000001bd1aaf5a0c0000000002b8688 .write_cache_pages_da+0x2c8/0x430
[c000001bd1aaf720c0000000002b8b28 .ext4_da_writepages+0x338/0x670
[c000001bd1aaf8d0c000000000157280 .do_writepages+0x40/0x90
[c000001bd1aaf940c0000000001ea830 .writeback_single_inode+0xe0/0x530
[c000001bd1aafa00c0000000001eb680 .writeback_sb_inodes+0x210/0x300
[c000001bd1aafb20c0000000001ebc84 .__writeback_inodes_wb+0xd4/0x140
[c000001bd1aafbe0c0000000001ebfec .wb_writeback+0x2fc/0x3e0
[c000001bd1aafce0c0000000001ed770 .wb_do_writeback+0x2f0/0x300
[c000001bd1aafdf0c0000000001ed848 .bdi_writeback_thread+0xc8/0x340
[c000001bd1aafed0c0000000000c5494 .kthread+0xb4/0xc0
[c000001bd1aaff90c000000000021f48 .kernel_thread+0x54/0x70

This is due to getting ext_depth(inode) == 0x101 and therefore running
off the end of the path array in ext4_ext_drop_refs into following
unallocated structures.

This fixes it by adding the necessary le16_to_cpu.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
fs/ext4/extents.c