Re: [PATCH v4 1/2] io_uring: avoid whole io_wq_work copy for requests completed inline

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 31/05/2020 16:57, Xiaoguang Wang wrote:
>> There is another thing:
>>
>> io_submit_sqes()
>>      -> io_close() (let ->flush == NULL)
>>          -> __io_close_finish()
>>              -> filp_close(req->close.put_file, *req->work.files*);
>>
>> where req->work.files is garbage.
> I think this bug is independent of my patch. Without my patches, if close request

It looks like it's ok to pass NULL, at least Jens did it here and I see an
occurrence of ->flush(NULL). And it's usually not referenced or completely
ignored. I'll check later.

> will be submitted and completed inline, req->work.files will be NULL, it's still
> problematic, should we use current->files here?

-- 
Pavel Begunkov



[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux