The logic in tcp_timer() was inverted. fwd_out should expose the host
ports in the ns. Therfore it must read the ports on the host and then
bind them in the netns. The same for fwd_in which checks ports in the
ns and then exposes them on the host.
Note that this only fixes tcp ports, udp does not seems to work at all
right now with the auto mode.
Signed-off-by: Paul Holzinger <pholzing(a)redhat.com>
---
tcp.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/tcp.c b/tcp.c
index 0214087..0d0ad13 100644
--- a/tcp.c
+++ b/tcp.c
@@ -89,7 +89,7 @@
* No port translation is needed for connections initiated remotely or by the
* local host: source port from socket is reused while establishing connections
* to the guest.
- *
+ *
* For connections initiated by the guest, it's not possible to force the same
* source port as connections are established by the host kernel: that's the
* only port translation needed.
@@ -173,7 +173,7 @@
* new socket is created and mapped in connection tracking table, setting
* MSS and window clamping from header and option of the observed SYN segment
*
- *
+ *
* Aging and timeout
* -----------------
*
@@ -560,7 +560,7 @@ static struct tcp6_l2_flags_buf_t {
#endif
struct tap_hdr taph; /* 14 2 */
struct ipv6hdr ip6h; /* 32 20 */
- struct tcphdr th /* 72 */ __attribute__ ((aligned(4))); /* 60 */
+ struct tcphdr th /* 72 */ __attribute__ ((aligned(4))); /* 60 */
char opts[OPT_MSS_LEN + OPT_WS_LEN + 1];
#ifdef __AVX2__
} __attribute__ ((packed, aligned(32)))
@@ -3308,14 +3308,14 @@ void tcp_timer(struct ctx *c, const struct timespec *ts)
struct tcp_port_detect_arg detect_arg = { c, 0 };
struct tcp_port_rebind_arg rebind_arg = { c, 0 };
- if (c->tcp.fwd_in.mode == FWD_AUTO) {
+ if (c->tcp.fwd_out.mode == FWD_AUTO) {
detect_arg.detect_in_ns = 0;
tcp_port_detect(&detect_arg);
rebind_arg.bind_in_ns = 1;
NS_CALL(tcp_port_rebind, &rebind_arg);
}
- if (c->tcp.fwd_out.mode == FWD_AUTO) {
+ if (c->tcp.fwd_in.mode == FWD_AUTO) {
detect_arg.detect_in_ns = 1;
NS_CALL(tcp_port_detect, &detect_arg);
rebind_arg.bind_in_ns = 0;
--
2.39.2
Thanks for the patch! On Mon, 20 Mar 2023 19:10:34 +0100 Paul Holzinger <pholzing(a)redhat.com> wrote:The logic in tcp_timer() was inverted. fwd_out should expose the host ports in the ns. Therfore it must read the ports on the host and then bind them in the netns. The same for fwd_in which checks ports in the ns and then exposes them on the host. Note that this only fixes tcp ports, udp does not seems to work at all right now with the auto mode.Note that for UDP there's no periodic scan, "auto" just checks bound ports when pasta starts: -u, --udp-ports spec Configure UDP port forwarding to namespace. spec is as described for TCP above, and the list of ports is derived from listening sockets reported by /proc/net/udp and /proc/net/udp6, see proc(5), when pasta starts (not periodically).Signed-off-by: Paul Holzinger <pholzing(a)redhat.com>Fixes: 1128fa03fe73 ("Improve types and names for port forwarding configuration") (I'll add that tag). The patch itself looks good to me. I'm now looking at other parts (tcp_sock_init()) where we seem to have the same kind of swap. Unfortunately this is only covered by the Podman demo as a test, which has been disabled for a while now: https://bugs.passt.top/show_bug.cgi?id=29 and as David is meanwhile working to improve the test framework, we can probably wait a bit to introduce a new test. Meanwhile I'll check this part manually on related changes. -- Stefano
On 21/03/2023 09:18, Stefano Brivio wrote:Thanks for the patch! On Mon, 20 Mar 2023 19:10:34 +0100 Paul Holzinger <pholzing(a)redhat.com> wrote:Ok this makes sense then, is there a bug to track this? Because without it auto mode for UDP is useless for my Podman use case.The logic in tcp_timer() was inverted. fwd_out should expose the host ports in the ns. Therfore it must read the ports on the host and then bind them in the netns. The same for fwd_in which checks ports in the ns and then exposes them on the host. Note that this only fixes tcp ports, udp does not seems to work at all right now with the auto mode.Note that for UDP there's no periodic scan, "auto" just checks bound ports when pasta starts: -u, --udp-ports spec Configure UDP port forwarding to namespace. spec is as described for TCP above, and the list of ports is derived from listening sockets reported by /proc/net/udp and /proc/net/udp6, see proc(5), when pasta starts (not periodically).Signed-off-by: Paul Holzinger <pholzing(a)redhat.com>Fixes: 1128fa03fe73 ("Improve types and names for port forwarding configuration") (I'll add that tag). The patch itself looks good to me. I'm now looking at other parts (tcp_sock_init()) where we seem to have the same kind of swap. Unfortunately this is only covered by the Podman demo as a test, which has been disabled for a while now: https://bugs.passt.top/show_bug.cgi?id=29 and as David is meanwhile working to improve the test framework, we can probably wait a bit to introduce a new test. Meanwhile I'll check this part manually on related changes.
On Tue, 21 Mar 2023 14:55:17 +0100 Paul Holzinger <pholzing(a)redhat.com> wrote:On 21/03/2023 09:18, Stefano Brivio wrote:No, sorry, not yet, feel free to file one, or I can do that later today. Right now not even what the man page says works, I'll post a patch in a bit. -- StefanoThanks for the patch! On Mon, 20 Mar 2023 19:10:34 +0100 Paul Holzinger <pholzing(a)redhat.com> wrote:Ok this makes sense then, is there a bug to track this? Because without it auto mode for UDP is useless for my Podman use case.The logic in tcp_timer() was inverted. fwd_out should expose the host ports in the ns. Therfore it must read the ports on the host and then bind them in the netns. The same for fwd_in which checks ports in the ns and then exposes them on the host. Note that this only fixes tcp ports, udp does not seems to work at all right now with the auto mode.Note that for UDP there's no periodic scan, "auto" just checks bound ports when pasta starts: -u, --udp-ports spec Configure UDP port forwarding to namespace. spec is as described for TCP above, and the list of ports is derived from listening sockets reported by /proc/net/udp and /proc/net/udp6, see proc(5), when pasta starts (not periodically).
On Tue, 21 Mar 2023 15:02:30 +0100 Stefano Brivio <sbrivio(a)redhat.com> wrote:On Tue, 21 Mar 2023 14:55:17 +0100 Paul Holzinger <pholzing(a)redhat.com> wrote:There you go, https://bugs.passt.top/show_bug.cgi?id=45. -- StefanoOn 21/03/2023 09:18, Stefano Brivio wrote:No, sorry, not yet, feel free to file one, or I can do that later today.Thanks for the patch! On Mon, 20 Mar 2023 19:10:34 +0100 Paul Holzinger <pholzing(a)redhat.com> wrote:Ok this makes sense then, is there a bug to track this? Because without it auto mode for UDP is useless for my Podman use case.The logic in tcp_timer() was inverted. fwd_out should expose the host ports in the ns. Therfore it must read the ports on the host and then bind them in the netns. The same for fwd_in which checks ports in the ns and then exposes them on the host. Note that this only fixes tcp ports, udp does not seems to work at all right now with the auto mode.Note that for UDP there's no periodic scan, "auto" just checks bound ports when pasta starts: -u, --udp-ports spec Configure UDP port forwarding to namespace. spec is as described for TCP above, and the list of ports is derived from listening sockets reported by /proc/net/udp and /proc/net/udp6, see proc(5), when pasta starts (not periodically).
On Tue, 21 Mar 2023 09:18:48 +0100 Stefano Brivio <sbrivio(a)redhat.com> wrote:The patch itself looks good to me. I'm now looking at other parts (tcp_sock_init()) where we seem to have the same kind of swap.Not really, the rest was fine, I'll apply this one in a bit. -- Stefano