> -----Original Message----- > From: kvm-ppc-owner@xxxxxxxxxxxxxxx > [mailto:kvm-ppc-owner@xxxxxxxxxxxxxxx] On Behalf Of Liu Yu-B13201 > Sent: Thursday, April 09, 2009 11:45 AM > To: Hollis Blanchard > Cc: kvm-ppc@xxxxxxxxxxxxxxx > Subject: RE: [PATCH] Map guest (ts,tid) to individual shadow tid > > > -----Original Message----- > > From: Hollis Blanchard [mailto:hollisb@xxxxxxxxxx] > > Sent: Thursday, April 09, 2009 12:43 AM > > To: Liu Yu-B13201 > > Cc: kvm-ppc@xxxxxxxxxxxxxxx > > Subject: Re: [PATCH] Map guest (ts,tid) to individual shadow tid > > > > On Tuesday 07 April 2009 21:11:11 Liu Yu-B13201 wrote: > > > > > > > -----Original Message----- > > > > From: Hollis Blanchard [mailto:hollisb@xxxxxxxxxx] > > > > Sent: Tuesday, April 07, 2009 11:41 PM > > > > To: Liu Yu-B13201 > > > > Cc: kvm-ppc@xxxxxxxxxxxxxxx > > > > Subject: Re: [PATCH] Map guest (ts,tid) to individual shadow tid > > > > > > > > On Tue, 2009-04-07 at 17:51 +0800, Liu Yu wrote: > > > > > Hi guys, > > > > > > > > > > Thanks to Hollis's idea. > > > > > This patch help to make KVMPPC be more friendly to OSes > > > > other than Linux. > > > > > > > > Wow, that was quick. :) > > > > > > > > Have you tested this code? Doesn't this break the assumption in > > > > kvmppc_e500_tlb1_invalidate() (and probably elsewhere) > > that stid == > > > > gtid? > > > > > > Yes, have taken a simple test. > > > > Once we can reduce the number of TLB flushes (see below) it will be > > interesting to see if there's a performance impact. > > > > >Good catch, it needs to handle it here in > > kvmppc_e500_tlb1_invalidate(). > > >Thanks. > > >But it's ok for now, because TLB1 only contains tid=0 > > entries, and tid=0 > > >always be mapped to stid=0. > > >That's why the test is fine... > > > > OK. Still, it makes me nervous to break such a simple > > assumption. We should > > introduce nice accessors to make it difficult to code it > > wrong in the future. > > I'm honestly surprised that kvmppc_e500_tlb1_invalidate() is > > the only affected > > site. > > It is the only site. We don't pay much attention on shadow tid. > For 500, shadow tid is inherit from guest tid. > As guest tlb1 mapping is broken into 4K shadow mappings, > the kvmppc_e500_tlb1_invalidate() need to check tid to find > all 4k shadow mappings related to an guest mapping. > > Actually, I have been thinking about remove all shadow tlb > like 440 did. > This maybe helpful to succedent work such as huge tlb mapping. > After doing that, kvmppc_e500_tlb1_invalidate() won't has > this assumption. > > Anyway, the patch is an RFC, not aimed at getting applied. > Just make something to discuss. :) > Well, I missed another site... Host pid should be updated to stid in two case: 1. guest accesses to SPRN_PID 2. guest switch btween kernel and userspace. (if we map guest kernel [tid=0] to non zero stid) -- To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
