[PATCH net-next v7 0/2] Add F_SETFL for fcntl in test_sockmap

[PATCH net-next v7 0/2] Add F_SETFL for fcntl in test_sockmap

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

v7:
 - a better fix for tls_sw_recvmsg.

v6:
 - add a fix for tls_sw_recvmsg().

v5:
 - add a new patch "Check recv lengths in test_sockmap" instead of using
   "continue" in msg_loop.

v4:
 - address Martin's comments for v3. (thanks.)
 - add Yonghong's "Acked-by" tags. (thanks.)
 - update subject-prefix from "bpf-next" to "bpf".

Patch 1, v3 of "selftests/bpf: Add F_SETFL for fcntl":
- detect nonblock flag automatically, then test_sockmap can run in both
block and nonblock modes.
- use continue instead of again in v2.

Patch 2, fix for umount cgroup2 error.

Geliang Tang (2):
  tls: receive msg again for sk_redirect
  selftests/bpf: Add F_SETFL for fcntl in test_sockmap

 net/tls/tls_sw.c                           | 3 +++
 tools/testing/selftests/bpf/test_sockmap.c | 4 +++-
 2 files changed, 6 insertions(+), 1 deletion(-)

-- 
2.43.0

[PATCH net-next v7 1/2] tls: receive msg again for sk_redirect

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

tls_sw doesn't work for sk_redirect in nonblock mode, sk_msg_recvmsg()
returns 0 in that case in tls_sw_recvmsg(). This patch fixes this by using
"continue" to receive msg again instead of ending it if strp isn't ready
and rx_list is empty.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/tls/tls_sw.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/tls/tls_sw.c b/net/tls/tls_sw.c
index 305a412785f5..ae8bbe7dc8ec 100644
--- a/net/tls/tls_sw.c
+++ b/net/tls/tls_sw.c
@@ -2017,6 +2017,9 @@ int tls_sw_recvmsg(struct sock *sk,
 					len -= chunk;
 					continue;
 				}
+				if (!chunk && !tls_strp_msg_ready(ctx) &&
+				    skb_queue_empty_lockless(&ctx->rx_list))
+					continue;
 			}
 			goto recv_end;
 		}
-- 
2.43.0

[PATCH net-next v7 2/2] selftests/bpf: Add F_SETFL for fcntl in test_sockmap

[Geliang Tang]

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.

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 | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/bpf/test_sockmap.c b/tools/testing/selftests/bpf/test_sockmap.c
index 9cba4ec844a5..99d3ca8e44bb 100644
--- a/tools/testing/selftests/bpf/test_sockmap.c
+++ b/tools/testing/selftests/bpf/test_sockmap.c
@@ -604,7 +604,9 @@ 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
-- 
2.43.0

Re: [PATCH net-next v7 1/2] tls: receive msg again for sk_redirect

[Jakub Kicinski]

On Wed, 12 Jun 2024 17:19:03 +0800 Geliang Tang wrote:
> tls_sw doesn't work for sk_redirect in nonblock mode, sk_msg_recvmsg()
> returns 0 in that case in tls_sw_recvmsg(). This patch fixes this by using
> "continue" to receive msg again instead of ending it if strp isn't ready
> and rx_list is empty.

Can you explain what user-visible behavior problem you're trying to fix?
sk_msg_recvmsg() returns 0, but tls_sw_recvmsg() will return -EAGAIN as 
a whole, so everything seems in order.