Re: io_uring_enter() with opcode IORING_OP_RECV ignores MSG_WAITALL in msg_flags

[Jens Axboe <axboe@kernel.dk>]

(please don't top post, replies go below the text you're replying to)

On 3/23/22 6:32 AM, Constantine Gavrilov wrote:
> Yes, I have a real test case. I cannot share it vebratim, but with a
> little effort I believe I can come with a simple code of
> client/server.
> 
> It seems the issue shall be directly seen from the implementation, but
> if it is not so, I will provide a sample code.

If you can, that would be useful. I took a quick look, and recv and
recvmsg handle this similarly. Are you seeing a short return, or is the
data wrong?

A bug report with a test case is always infinitely more vauable than one
that does not have one. It serves two purposes:

1) It more accurately tells us what the submitter thinks is wrong (eg
   the "this is what I expected, but this is what happened").

2) It means that we don't have to write one, which saves a lot of time.
   Ideally we end up putting it into the regression tests, which helps
   to guarantee it won't regress there again.

While you may think that we can just look at the code and fix it, a fix
needs a regression test too. And that now means that we have to write
that too...

> Forgot to mention that the issue is seen of Fedora kernel version
> 5.16.12-200.fc35.x86_64.

Thanks, forgot to ask, that's useful to know.

-- 
Jens Axboe