Le Wed, 11 May 2022 12:36:38 +0300, Vladimir Oltean <olteanv@xxxxxxxxx> a écrit : > On Tue, May 10, 2022 at 10:34:58AM +0200, Clément Léger wrote: > > > By the way, does this switch pass > > > tools/testing/selftests/drivers/net/dsa/no_forwarding.sh? > > > > Unfortunately, the board I have only has 2 ports availables and thus, I > > can only test one bridge or two separated ports at a time... I *should* > > receive a 4 ports one in a near future but that not yet sure. > > 2 switch ports or 2 ports in total? h1 and h2 can be non-switch ports > (should work with USB-Ethernet adapters etc). Ok, I finally got the tests running. They seems to work for the standalone variant. TEST: Standalone switch ports: Unicast non-IP untagged [ OK ] TEST: Standalone switch ports: Multicast non-IP untagged [ OK ] TEST: Standalone switch ports: Broadcast non-IP untagged [ OK ] TEST: Standalone switch ports: Unicast IPv4 untagged [ OK ] TEST: Standalone switch ports: Multicast IPv4 untagged [ OK ] TEST: Standalone switch ports: Unicast IPv6 untagged [ OK ] TEST: Standalone switch ports: Multicast IPv6 untagged [ OK ] I disabled tests two_bridges and one_bridge_two_pvids since the switch driver does only support 1 bridge and do not have vlan support yet. I also ran ./test_bridge_fdb_stress.sh which did reveal some sleep in spin_lock() issue due to poll_io_timeout() being used. I switch the FDB lock to use a mutex instead of a spinlock. -- Clément Léger, Embedded Linux and Kernel engineer at Bootlin https://bootlin.com
