Re: [PATCH net v2] net/9p: fix uninit-value in p9_client_rpc()

[Nikita Zhandarovich <n.zhandarovich@fintech.ru>]

On 4/8/24 07:10, Nikita Zhandarovich wrote:
> Syzbot with the help of KMSAN reported the following error:
> 
> BUG: KMSAN: uninit-value in trace_9p_client_res include/trace/events/9p.h:146 [inline]
> BUG: KMSAN: uninit-value in p9_client_rpc+0x1314/0x1340 net/9p/client.c:754
>  trace_9p_client_res include/trace/events/9p.h:146 [inline]
>  p9_client_rpc+0x1314/0x1340 net/9p/client.c:754
>  p9_client_create+0x1551/0x1ff0 net/9p/client.c:1031
>  v9fs_session_init+0x1b9/0x28e0 fs/9p/v9fs.c:410
>  v9fs_mount+0xe2/0x12b0 fs/9p/vfs_super.c:122
>  legacy_get_tree+0x114/0x290 fs/fs_context.c:662
>  vfs_get_tree+0xa7/0x570 fs/super.c:1797
>  do_new_mount+0x71f/0x15e0 fs/namespace.c:3352
>  path_mount+0x742/0x1f20 fs/namespace.c:3679
>  do_mount fs/namespace.c:3692 [inline]
>  __do_sys_mount fs/namespace.c:3898 [inline]
>  __se_sys_mount+0x725/0x810 fs/namespace.c:3875
>  __x64_sys_mount+0xe4/0x150 fs/namespace.c:3875
>  do_syscall_64+0xd5/0x1f0
>  entry_SYSCALL_64_after_hwframe+0x6d/0x75
> 
> Uninit was created at:
>  __alloc_pages+0x9d6/0xe70 mm/page_alloc.c:4598
>  __alloc_pages_node include/linux/gfp.h:238 [inline]
>  alloc_pages_node include/linux/gfp.h:261 [inline]
>  alloc_slab_page mm/slub.c:2175 [inline]
>  allocate_slab mm/slub.c:2338 [inline]
>  new_slab+0x2de/0x1400 mm/slub.c:2391
>  ___slab_alloc+0x1184/0x33d0 mm/slub.c:3525
>  __slab_alloc mm/slub.c:3610 [inline]
>  __slab_alloc_node mm/slub.c:3663 [inline]
>  slab_alloc_node mm/slub.c:3835 [inline]
>  kmem_cache_alloc+0x6d3/0xbe0 mm/slub.c:3852
>  p9_tag_alloc net/9p/client.c:278 [inline]
>  p9_client_prepare_req+0x20a/0x1770 net/9p/client.c:641
>  p9_client_rpc+0x27e/0x1340 net/9p/client.c:688
>  p9_client_create+0x1551/0x1ff0 net/9p/client.c:1031
>  v9fs_session_init+0x1b9/0x28e0 fs/9p/v9fs.c:410
>  v9fs_mount+0xe2/0x12b0 fs/9p/vfs_super.c:122
>  legacy_get_tree+0x114/0x290 fs/fs_context.c:662
>  vfs_get_tree+0xa7/0x570 fs/super.c:1797
>  do_new_mount+0x71f/0x15e0 fs/namespace.c:3352
>  path_mount+0x742/0x1f20 fs/namespace.c:3679
>  do_mount fs/namespace.c:3692 [inline]
>  __do_sys_mount fs/namespace.c:3898 [inline]
>  __se_sys_mount+0x725/0x810 fs/namespace.c:3875
>  __x64_sys_mount+0xe4/0x150 fs/namespace.c:3875
>  do_syscall_64+0xd5/0x1f0
>  entry_SYSCALL_64_after_hwframe+0x6d/0x75
> 
> If p9_check_errors() fails early in p9_client_rpc(), req->rc.tag
> will not be properly initialized. However, trace_9p_client_res()
> ends up trying to print it out anyway before p9_client_rpc()
> finishes.
> 
> Fix this issue by assigning default values to p9_fcall fields
> such as 'tag' and (just in case KMSAN unearths something new) 'id'
> during the tag allocation stage.
> 
> Reported-and-tested-by: syzbot+ff14db38f56329ef68df@syzkaller.appspotmail.com
> Fixes: 348b59012e5c ("net/9p: Convert net/9p protocol dumps to tracepoints")
> Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru>
> ---
> v2: change fc->tag init value from 0 to P9_NOTAG per Dominique
> Martinet's <asmadeus@codewreck.org> helpful suggestion.
> Link: https://lore.kernel.org/all/ZhNVMivKCCq6wir0@codewreck.org/
> 
>  net/9p/client.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/net/9p/client.c b/net/9p/client.c
> index f7e90b4769bb..b05f73c291b4 100644
> --- a/net/9p/client.c
> +++ b/net/9p/client.c
> @@ -235,6 +235,8 @@ static int p9_fcall_init(struct p9_client *c, struct p9_fcall *fc,
>  	if (!fc->sdata)
>  		return -ENOMEM;
>  	fc->capacity = alloc_msize;
> +	fc->id = 0;
> +	fc->tag = P9_NOTAG;
>  	return 0;
>  }
>  

Hi Dominique,

Gentle ping on this issue as it is still open. Thanks in advance :)

Regards,
Nikita