Re: [PATCH bpf v4 1/2] selftests/bpf: Add F_SETFL for fcntl in test_sockmap

[Geliang Tang <geliang@kernel.org>]

On Wed, 2024-04-17 at 17:28 -0700, Martin KaFai Lau wrote:
> On 4/17/24 1:14 AM, Geliang Tang wrote:
> > Hi Martin,
> > 
> > On Thu, Apr 11, 2024 at 11:10:49AM -0700, Martin KaFai Lau wrote:
> > > On 4/8/24 10:18 PM, Geliang Tang wrote:
> > > > From: Geliang Tang <tanggeliang@kylinos.cn>
> > > > 
> > > > Incorrect arguments are passed to fcntl() in test_sockmap.c
> > > > when invoking
> > > > it to set file status flags. If O_NONBLOCK is used as 2nd
> > > > argument and
> > > > passed into fcntl, -EINVAL will be returned (See do_fcntl() in
> > > > fs/fcntl.c).
> > > > The correct approach is to use F_SETFL as 2nd argument, and
> > > > O_NONBLOCK as
> > > > 3rd one.
> > > > 
> > > > In nonblock mode, if EWOULDBLOCK is received, continue
> > > > receiving, otherwise
> > > > some subtests of test_sockmap fail.
> > > > 
> > > > Fixes: 16962b2404ac ("bpf: sockmap, add selftests")
> > > > Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
> > > > Acked-by: Yonghong Song <yonghong.song@linux.dev>
> > > > ---
> > > >    tools/testing/selftests/bpf/test_sockmap.c | 5 ++++-
> > > >    1 file changed, 4 insertions(+), 1 deletion(-)
> > > > 
> > > > diff --git a/tools/testing/selftests/bpf/test_sockmap.c
> > > > b/tools/testing/selftests/bpf/test_sockmap.c
> > > > index 024a0faafb3b..4feed253fca2 100644
> > > > --- a/tools/testing/selftests/bpf/test_sockmap.c
> > > > +++ b/tools/testing/selftests/bpf/test_sockmap.c
> > > > @@ -603,7 +603,9 @@ static int msg_loop(int fd, int iov_count,
> > > > int iov_length, int cnt,
> > > >    		struct timeval timeout;
> > > >    		fd_set w;
> > > > -		fcntl(fd, fd_flags);
> > > > +		if (fcntl(fd, F_SETFL, fd_flags))
> > > > +			goto out_errno;
> > > > +
> > > >    		/* Account for pop bytes noting each iteration
> > > > of apply will
> > > >    		 * call msg_pop_data helper so we need to
> > > > account for this
> > > >    		 * by calculating the number of apply
> > > > iterations. Note user
> > > > @@ -678,6 +680,7 @@ static int msg_loop(int fd, int iov_count,
> > > > int iov_length, int cnt,
> > > >    					perror("recv
> > > > failed()");
> > > >    					goto out_errno;
> > > >    				}
> > > > +				continue;
> > > 
> > >  From looking at it again, there is a select() earlier, so it
> > > should not hit
> > > EWOULDBLOCK.
> > 
> > Can the patch in the attachment be accepted? It can work, but I'm
> > not sure
> > if it has changed the behavior of this test. Anyway, I would like
> > to hear
> > your opinion.
> 
> I don't know what is the correct expectation also. John and JakubS,
> can you take 
> a look?

Hello,
New version v5 has been sent. Please review it for me.

Thanks,
-Geliang