From: Mike Larkin Subject: Re: psp(4): longer timeouts To: tech@openbsd.org Date: Wed, 2 Oct 2024 09:42:01 -0700 On Tue, Oct 01, 2024 at 02:37:07PM +0200, Hans-Jörg Höxer wrote: > Hi, > > psp(4) on my EPYC 9124 seems to need more time to respond. With both > polling during autoconf and interrupt during normal operation. > > Take care, > HJ. ok mlarkin > ------------------------------------------------------------------------ > commit 91ace94c0af584a1284f25ed72b3129a1cfde483 > Author: Hans-Joerg Hoexer > Date: Wed Aug 7 17:54:43 2024 +0200 > > psp(4): longer timeouts > > On my EPYC 9124 psp(4) seems to need more to time to respond. > Otherwise it will not attach. > > diff --git a/sys/dev/ic/psp.c b/sys/dev/ic/psp.c > index 5bbbf2de5ee..84c8379d741 100644 > --- a/sys/dev/ic/psp.c > +++ b/sys/dev/ic/psp.c > @@ -199,7 +199,7 @@ ccp_wait(struct psp_softc *sc, uint32_t *status, int poll) > > if (poll) { > count = 0; > - while (count++ < 10) { > + while (count++ < 100) { > cmdword = bus_space_read_4(sc->sc_iot, sc->sc_ioh, > PSP_REG_CMDRESP); > if (cmdword & PSP_CMDRESP_RESPONSE) > @@ -211,7 +211,7 @@ ccp_wait(struct psp_softc *sc, uint32_t *status, int poll) > return (1); > } > > - if (tsleep_nsec(sc, PWAIT, "psp", SEC_TO_NSEC(1)) == EWOULDBLOCK) > + if (tsleep_nsec(sc, PWAIT, "psp", SEC_TO_NSEC(2)) == EWOULDBLOCK) > return (1); > > done: