On Sat, 2017-09-09 at 10:08 +0200, Cédric Le Goater wrote: > On 09/09/2017 09:39 AM, Benjamin Herrenschmidt wrote: > > On Wed, 2017-07-05 at 19:13 +0200, Cédric Le Goater wrote: > > > Signed-off-by: Cédric Le Goater <c...@kaod.org> > > > --- > > > hw/intc/xive.c | 21 +++++++++++++++++++++ > > > 1 file changed, 21 insertions(+) > > > > > > diff --git a/hw/intc/xive.c b/hw/intc/xive.c > > > index c3c1e9c9db2d..cda1fa18e44d 100644 > > > --- a/hw/intc/xive.c > > > +++ b/hw/intc/xive.c > > > @@ -53,6 +53,21 @@ static uint64_t xive_icp_accept(XiveICPState *xicp) > > > return (nsr << 8) | xicp->tima_os[TM_CPPR]; > > > } > > > > > > +static uint8_t ipb_to_pipr(uint8_t ibp) > > > +{ > > > + return ibp ? clz32((uint32_t)ibp << 24) : 0xff; > > > +} > > > > The PIPR needs to be updated also on accept etc... anything that change > > IPBs or CPPR really. > > I did forget the update in accept ... thanks. > > > Also I just learned something from the designers: The IPIR is clamped > > to CPPR. > > > > So basically the value in the PIPR is: > > > > v = leftmost_bit_of(ipb) (or 0xff); > > pipr = v < cppr ? v : cppr; > > > > which means it's never actually 0xff ... surprise ! > > ah. that is not what the specs say but it shouldn't be a problem. > I will change the code to reflect that.
Right, I've asked the HW guys to update the spec. Cheers, Ben. > C. > > > That also means I need to fix my implementation of H_IPOLL in KVM. > > > > > +static void xive_icp_notify(XiveICPState *xicp) > > > +{ > > > + xicp->tima_os[TM_PIPR] = ipb_to_pipr(xicp->tima_os[TM_IPB]); > > > + > > > + if (xicp->tima_os[TM_PIPR] < xicp->tima_os[TM_CPPR]) { > > > + xicp->tima_os[TM_NSR] |= TM_QW1_NSR_EO; > > > + qemu_irq_raise(ICP(xicp)->output); > > > + } > > > +} > > > + > > > static void xive_icp_set_cppr(XiveICPState *xicp, uint8_t cppr) > > > { > > > if (cppr > XIVE_PRIORITY_MAX) { > > > @@ -60,6 +75,10 @@ static void xive_icp_set_cppr(XiveICPState *xicp, > > > uint8_t cppr) > > > } > > > > > > xicp->tima_os[TM_CPPR] = cppr; > > > + > > > + /* CPPR has changed, inform the ICP which might raise an > > > + * exception */ > > > + xive_icp_notify(xicp); > > > } > > > > > > /* > > > @@ -339,6 +358,8 @@ static void xive_icp_irq(XiveICSState *xs, int lisn) > > > } else { > > > qemu_log_mask(LOG_UNIMP, "XIVE: w7 format1 not implemented\n"); > > > } > > > + > > > + xive_icp_notify(xicp); > > > } > > > > > > /*