linux-hardening - Re: [PATCH] btrfs: Annotate structs with __counted

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <7747116F-BA20-455C-84AF-80D0EAD8B152@toblux.com>
Date: Wed, 14 Aug 2024 21:01:42 +0200
From: Thorsten Blum <thorsten.blum@...lux.com>
To: Nathan Chancellor <nathan@...nel.org>
Cc: Johannes Thumshirn <Johannes.Thumshirn@....com>,
 "clm@...com" <clm@...com>,
 "josef@...icpanda.com" <josef@...icpanda.com>,
 "dsterba@...e.com" <dsterba@...e.com>,
 "kees@...nel.org" <kees@...nel.org>,
 "gustavoars@...nel.org" <gustavoars@...nel.org>,
 "linux-btrfs@...r.kernel.org" <linux-btrfs@...r.kernel.org>,
 "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
 "linux-hardening@...r.kernel.org" <linux-hardening@...r.kernel.org>
Subject: Re: [PATCH] btrfs: Annotate structs with __counted_by()

On 14. Aug 2024, at 20:02, Nathan Chancellor <nathan@...nel.org> wrote:
> On Mon, Aug 12, 2024 at 02:03:44PM +0200, Thorsten Blum wrote:
>> On 12. Aug 2024, at 12:54, Johannes Thumshirn <Johannes.Thumshirn@....com> wrote:
>>> On 12.08.24 12:37, Thorsten Blum wrote:
>>>> Add the __counted_by compiler attribute to the flexible array member
>>>> stripes to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and
>>>> CONFIG_FORTIFY_SOURCE.
>>>> 
>>>> Signed-off-by: Thorsten Blum <thorsten.blum@...lux.com>
>>>> ---
>>>> fs/btrfs/volumes.h | 4 ++--
>>>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>>> 
>>>> diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
>>>> index 37a09ebb34dd..f28fa318036b 100644
>>>> --- a/fs/btrfs/volumes.h
>>>> +++ b/fs/btrfs/volumes.h
>>>> @@ -551,7 +551,7 @@ struct btrfs_io_context {
>>>>  * stripes[data_stripes + 1]: The Q stripe (only for RAID6).
>>>>  */
>>>>  u64 full_stripe_logical;
>>>> - struct btrfs_io_stripe stripes[];
>>>> + struct btrfs_io_stripe stripes[] __counted_by(num_stripes);
>>>> };
>>>> 
>>>> struct btrfs_device_info {
>>>> @@ -591,7 +591,7 @@ struct btrfs_chunk_map {
>>>>  int io_width;
>>>>  int num_stripes;
>>>>  int sub_stripes;
>>>> - struct btrfs_io_stripe stripes[];
>>>> + struct btrfs_io_stripe stripes[] __counted_by(num_stripes);
>>>> };
>>>> 
>>>> #define btrfs_chunk_map_size(n) (sizeof(struct btrfs_chunk_map) + \
>>> 
>>> Looks good to me,
>>> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@....com>
>>> 
>>> Out of curiosity, have you encountered any issues with this patch applied?
>> 
>> I only compile-tested it.
> 
> This change is now in next-20240814 and I see a UBSAN warning at runtime
> as a result because the assignment of ->num_stripes happens after
> accessing ->stripes[] (which breaks one of the requirements for using
> __counted_by [1]), meaning that UBSAN thinks this is a zero sized array
> due to bioc being allocated with kzalloc().
> 
>  [   24.992264] ------------[ cut here ]------------
>  [   25.009196] UBSAN: array-index-out-of-bounds in fs/btrfs/volumes.c:6602:11
>  [   25.021963] index 1 is out of range for type 'struct btrfs_io_stripe[] __counted_by(num_stripes)' (aka 'struct btrfs_io_stripe[]')
>  [   25.036463] CPU: 28 UID: 0 PID: 1171 Comm: systemd-random- Not tainted 6.11.0-rc3-next-20240814 #1
>  [   25.048172] Hardware name: ADLINK Ampere Altra Developer Platform/Ampere Altra Developer Platform, BIOS TianoCore 2.04.100.11 (SYS: 2.06.20220308) 11/06/2
>  [   25.064754] Call trace:
>  [   25.069965]  dump_backtrace+0x114/0x19c
>  [   25.076564]  show_stack+0x28/0x3c
>  [   25.082642]  dump_stack_lvl+0x48/0x94
>  [   25.089068]  __ubsan_handle_out_of_bounds+0x10c/0x184
>  [   25.096883]  btrfs_map_block+0x540/0xb3c
>  [   25.103570]  btrfs_submit_bio+0xf8/0x654
>  [   25.110256]  write_one_eb+0x290/0x444
>  [   25.116682]  btree_write_cache_pages+0x44c/0x5a8
>  [   25.124063]  btree_writepages+0x2c/0x8c
>  [   25.130662]  do_writepages+0x10c/0x34c
>  [   25.137175]  filemap_fdatawrite_wbc+0x84/0xb0
>  [   25.144295]  filemap_fdatawrite_range+0x74/0xac
>  [   25.151589]  btrfs_write_marked_extents+0xa0/0x140
>  [   25.159143]  btrfs_sync_log+0x298/0xa98
>  [   25.165743]  btrfs_sync_file+0x440/0x608
>  [   25.172429]  __arm64_sys_fsync+0x90/0xd4
>  [   25.179115]  invoke_syscall+0x8c/0x11c
>  [   25.185628]  el0_svc_common
>  [   25.191185]  do_el0_svc+0x2c/0x3c
>  [   25.197264]  el0_svc+0x48/0xf0
>  [   25.203083]  el0t_64_sync_handler+0x98/0x108
>  [   25.210118]  el0t_64_sync+0x19c/0x1a0
>  [   25.216552] ---[ end trace ]---
> 
> The fix might be as simple as something like
> 
> diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
> index 4a259bdaa21c..0cabc2ebde71 100644
> --- a/fs/btrfs/volumes.c
> +++ b/fs/btrfs/volumes.c
> @@ -6561,6 +6561,7 @@ int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
> }
> bioc->map_type = map->type;
> 
> + bioc->num_stripes = io_geom.num_stripes;
> /*
> * For RAID56 full map, we need to make sure the stripes[] follows the
> * rule that data stripes are all ordered, then followed with P and Q
> @@ -6621,7 +6622,6 @@ int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
> }
> 
> *bioc_ret = bioc;
> - bioc->num_stripes = io_geom.num_stripes;
> bioc->max_errors = io_geom.max_errors;
> bioc->mirror_num = io_geom.mirror_num;
> 
> 
> but I am not sure of the implications of this change on quick glance
> with regards to error handling and such.
> 
> [1]: https://people.kernel.org/gustavoars/how-to-use-the-new-counted_by-attribute-in-c-and-linux
> 
> Cheers,
> Nathan

My patch should probably be reverted as I somehow missed quite a few
things and need more time to rework this properly.

Sorry about that and thanks for reporting this!