Sorry for ignoring __GFP_NOFAIL flag.
Please ignore this patch.
On 2020/11/5 9:33, Zhiqiang Liu wrote:
> ping ...
>
> On 2020/10/22 21:13, Zhiqiang Liu wrote:
>>
>> In fuse_simple_request func, we will call fuse_request_alloc func to alloc
>> one request from fuse_req_cachep when args->force is true. However, the
>> return value of fuse_request_alloc func is not checked whether it is NULL.
>> If allocating request fails, access-NULL-pointer problem will occur.
>>
>> Here, we check the return value of fuse_request_alloc func.
>>
>> Fixes: 7213394c4e18 ("fuse: simplify request allocation")
>> Signed-off-by: Zhiqiang Liu <liuzhiqiang26@xxxxxxxxxx>
>> Signed-off-by: Haotian Li <lihaotian9@xxxxxxxxxx>
>> ---
>> fs/fuse/dev.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
>> index 02b3c36b3676..f7dd33ae8e31 100644
>> --- a/fs/fuse/dev.c
>> +++ b/fs/fuse/dev.c
>> @@ -481,6 +481,8 @@ ssize_t fuse_simple_request(struct fuse_conn *fc, struct fuse_args *args)
>> if (args->force) {
>> atomic_inc(&fc->num_waiting);
>> req = fuse_request_alloc(GFP_KERNEL | __GFP_NOFAIL);
>> + if (!req)
>> + return -ENOMEM;
>>
>> if (!args->nocreds)
>> fuse_force_creds(fc, req);
>>
>
>
> .
>
