Re: ixv(4): porting Virtual Function driver for Intel 82599 series.

2022-12-20 Thread Yuichiro NAITO
I received an issue that ixv(4) doesn't detect linkdowns in personal email. When linkdown happens, the PF (Primary Function) driver interrupts all VFs (Virtual Function) via `mailbox`. But ixv(4) doesn't receive the interrupt. According to NetBSD, this problem has been fixed by following

Re: ixv(4): porting Virtual Function driver for Intel 82599 series.

2022-12-04 Thread Yuichiro NAITO
I updated my patch to remove AIM code. I can't see any performance improvement with AIM enabled. I ran packet forwarding test with Cisco TRex traffic Generator (*1). AIM doesn't improves performance in this test. *1: https://trex-tgn.cisco.com/ With my latest patch, I've got following maximum

Re: ixv(4): porting Virtual Function driver for Intel 82599 series.

2022-11-20 Thread Yuichiro NAITO
Thank you so much for reviewing my patch. On 11/21/22 08:02, Christian Weisgerber wrote: Yuichiro NAITO: +static void +ixv_set_multi(struct ix_softc *sc) +{ [...] + if ((ifp->if_flags & IFF_PROMISC) == 0 && ac->ac_multirangecnt <= 0 && + ac->ac_multicnt <=

Re: ixv(4): porting Virtual Function driver for Intel 82599 series.

2022-11-20 Thread Christian Weisgerber
Yuichiro NAITO: > +static void > +ixv_set_multi(struct ix_softc *sc) > +{ [...] > + if ((ifp->if_flags & IFF_PROMISC) == 0 && ac->ac_multirangecnt <= 0 && > + ac->ac_multicnt <= MAX_NUM_MULTICAST_ADDRESSES) { > + ETHER_FIRST_MULTI(step, >arpcom, enm); > +