Re: [PATCH] mm/slub: fix to return errno if kmalloc() fails

From: Chao Yu <chao@kernel.org>
To: Christophe JAILLET <christophe.jaillet@wanadoo.fr>,
	"Vlastimil Babka (SUSE)" <vbabka@kernel.org>,
	linux-mm@kvack.org
Cc: akpm@linux-foundation.org, linux-kernel@vger.kernel.org,
	Chao Yu <chao.yu@oppo.com>,
	muchun.song@linux.dev
Subject: Re: [PATCH] mm/slub: fix to return errno if kmalloc() fails
Date: Tue, 13 Sep 2022 11:42:04 +0800	[thread overview]
Message-ID: <e14a916c-a597-5328-2d79-39274fbdad35@kernel.org> (raw)
In-Reply-To: <2025305d-16db-abdf-6cd3-1fb93371c2b4@wanadoo.fr>

On 2022/9/10 0:47, Christophe JAILLET wrote:
> Le 30/08/2022 à 16:10, Chao Yu a écrit :
>> From: Chao Yu <chao.yu@oppo.com>
>>
>> In create_unique_id(), kmalloc(, GFP_KERNEL) can fail due to
>> out-of-memory, if it fails, return errno correctly rather than
>> triggering panic via BUG_ON();
>>
>> kernel BUG at mm/slub.c:5893!
>> Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
>>
>> Call trace:
>>   sysfs_slab_add+0x258/0x260 mm/slub.c:5973
>>   __kmem_cache_create+0x60/0x118 mm/slub.c:4899
>>   create_cache mm/slab_common.c:229 [inline]
>>   kmem_cache_create_usercopy+0x19c/0x31c mm/slab_common.c:335
>>   kmem_cache_create+0x1c/0x28 mm/slab_common.c:390
>>   f2fs_kmem_cache_create fs/f2fs/f2fs.h:2766 [inline]
>>   f2fs_init_xattr_caches+0x78/0xb4 fs/f2fs/xattr.c:808
>>   f2fs_fill_super+0x1050/0x1e0c fs/f2fs/super.c:4149
>>   mount_bdev+0x1b8/0x210 fs/super.c:1400
>>   f2fs_mount+0x44/0x58 fs/f2fs/super.c:4512
>>   legacy_get_tree+0x30/0x74 fs/fs_context.c:610
>>   vfs_get_tree+0x40/0x140 fs/super.c:1530
>>   do_new_mount+0x1dc/0x4e4 fs/namespace.c:3040
>>   path_mount+0x358/0x914 fs/namespace.c:3370
>>   do_mount fs/namespace.c:3383 [inline]
>>   __do_sys_mount fs/namespace.c:3591 [inline]
>>   __se_sys_mount fs/namespace.c:3568 [inline]
>>   __arm64_sys_mount+0x2f8/0x408 fs/namespace.c:3568
>>
>> Cc: <stable@kernel.org>
>> Reported-by: syzbot+81684812ea68216e08c5@syzkaller.appspotmail.com
>> Signed-off-by: Chao Yu <chao.yu@oppo.com>
>> ---
>>   mm/slub.c | 5 ++++-
>>   1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/mm/slub.c b/mm/slub.c
>> index 862dbd9af4f5..e6f3727b9ad2 100644
>> --- a/mm/slub.c
>> +++ b/mm/slub.c
>> @@ -5890,7 +5890,8 @@ static char *create_unique_id(struct kmem_cache *s)
>>       char *name = kmalloc(ID_STR_LENGTH, GFP_KERNEL);
> 
> Hi,
> 
> looks that ID_STR_LENGTH could even be reduced to 32 or 16.
> 
> The 2nd BUG_ON at the end of the function could certainly be just removed as well or remplaced by a:
>         if (p > name + ID_STR_LENGTH - 1) {
>          kfree(name);
>          return -E<something>;
>      }

Hi Christophe, Vlastimil,

Should I include this in v3? or may be in another patch?

Thanks,

> 
> Just my 2c,
> 
> CJ
> 
>>       char *p = name;
>> -    BUG_ON(!name);
>> +    if (!name)
>> +        return ERR_PTR(-ENOMEM);
>>       *p++ = ':';
>>       /*
>> @@ -5948,6 +5949,8 @@ static int sysfs_slab_add(struct kmem_cache *s)
>>            * for the symlinks.
>>            */
>>           name = create_unique_id(s);
>> +        if (IS_ERR(name))
>> +            return PTR_ERR(name);
>>       }
>>       s->kobj.kset = kset;
> 