Messages in this thread | | | Date | Mon, 18 Jun 2012 17:17:24 +0300 | From | Avi Kivity <> | Subject | Re: [PATCHv7 3/8] kvm_para: guest side for eoi avoidance |
| |
On 06/14/2012 04:53 PM, Michael S. Tsirkin wrote: > The idea is simple: there's a bit, per APIC, in guest memory, > that tells the guest that it does not need EOI. > Guest tests it using a single est and clear operation - this is > necessary so that host can detect interrupt nesting - and if set, it can > skip the EOI MSR. > > I run a simple microbenchmark to show exit reduction > (note: for testing, need to apply follow-up patch > 'kvm: host side for eoi optimization' + a qemu patch > I posted separately, on host): > > > diff --git a/arch/x86/include/asm/bitops.h b/arch/x86/include/asm/bitops.h > index a6983b2..47f9eff 100644 > --- a/arch/x86/include/asm/bitops.h > +++ b/arch/x86/include/asm/bitops.h > @@ -28,11 +28,13 @@ > #if __GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 1) > /* Technically wrong, but this avoids compilation errors on some gcc > versions. */ > -#define BITOP_ADDR(x) "=m" (*(volatile long *) (x)) > +#define BITOP_ADDR_CONSTRAINT "=m" > #else > -#define BITOP_ADDR(x) "+m" (*(volatile long *) (x)) > +#define BITOP_ADDR_CONSTRAINT "+m" > #endif > > +#define BITOP_ADDR(x) BITOP_ADDR_CONSTRAINT (*(volatile long *) (x)) > + > #define ADDR BITOP_ADDR(addr)
What's this doing here?
> > +/* size alignment is implied but just to make it explicit. */ > +static DEFINE_PER_CPU(unsigned long, kvm_apic_eoi) __aligned(2) = > + KVM_PV_EOI_DISABLED;
You're actually breaking the alignment. ulong has 8 byte alignment sometimes and you can make it cross cache boundary this way.
> + > void __cpuinit kvm_guest_cpu_init(void) > { > if (!kvm_para_available()) > @@ -300,11 +320,17 @@ void __cpuinit kvm_guest_cpu_init(void) > smp_processor_id()); > } > > + if (kvm_para_has_feature(KVM_FEATURE_PV_EOI)) { > + __get_cpu_var(kvm_apic_eoi) = 0; > + wrmsrl(MSR_KVM_PV_EOI_EN, __pa(&__get_cpu_var(kvm_apic_eoi)) | > + KVM_MSR_ENABLED);
Bad formatting.
> + } > + > if (has_steal_clock) > kvm_register_steal_time(); > } >
Please check that the kexec path also disables pveoi.
-- error compiling committee.c: too many arguments to function
| |