Re: [Question] about async buffered reads feature

[Jens Axboe <axboe@kernel.dk>]

On 9/25/20 3:23 PM, Jens Axboe wrote:
> On 9/25/20 3:18 AM, Hao_Xu wrote:
>> Hi Jens,
>> I'm doing tests about this feature: [PATCHSET RFC 0/11] Add support for 
>> async buffered reads
>> But currently with fio testing, I found the code doesn't go to the 
>> essential places in the function generic_file_buffered_read:
>>
>>            if (iocb->ki_flags & IOCB_WAITQ) {
>>                    if (written) {
>>                            put_page(page);
>>                            goto out;
>>                    }
>>                    error = wait_on_page_locked_async(page,
>>                                                    iocb->ki_waitq);
>>            } else {
>>
>> and
>>
>>    page_not_up_to_date:
>>           /* Get exclusive access to the page ... */
>>           if (iocb->ki_flags & IOCB_WAITQ)
>>                   error = lock_page_async(page, iocb->ki_waitq);
>>           else
> 
> Can you try with this added? Looks like a regression got introduced...

Oops, needs 'ret' cleared too. This should be better!

diff --git a/fs/io_uring.c b/fs/io_uring.c
index ad828fa19af4..11b8e428300d 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -3172,10 +3172,8 @@ static int io_read(struct io_kiocb *req, bool force_nonblock,
 			goto done;
 		/* some cases will consume bytes even on error returns */
 		iov_iter_revert(iter, iov_count - iov_iter_count(iter));
-		ret = io_setup_async_rw(req, iovec, inline_vecs, iter, false);
-		if (ret)
-			goto out_free;
-		return -EAGAIN;
+		ret = 0;
+		goto copy_iov;
 	} else if (ret < 0) {
 		/* make sure -ERESTARTSYS -> -EINTR is done */
 		goto done;

-- 
Jens Axboe