If we have a Guest/DomU with two or more of its VCPUs running
on same host CPU then it can quite likely happen that these
VCPUs fight for same spinlock and one of them will waste CPU
cycles in WFE instruction. This patch makes WFE instruction
trap for VCPU and forces VCPU to yield its timeslice.
The KVM ARM/ARM64 also does similar thing for handling WFE
instructions. (Please refer,
https://lists.cs.columbia.edu/pipermail/kvmarm/2013-November/006259.html)
In general, this patch is more of an optimization for an
oversubscribed system having number of VCPUs more than
underlying host CPUs.
Signed-off-by: Anup Patel <anup.patel@linaro.org>
Signed-off-by: Pranavkumar Sawargaonkar <pranavkumar@linaro.org>
Tested-by: Pranavkumar Sawargaonkar <pranavkumar@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- resolved conflict with "Adding helper function for WFI",
nuked stray hard tab ]
/* Setup hypervisor traps */
WRITE_SYSREG(HCR_PTW|HCR_BSU_INNER|HCR_AMO|HCR_IMO|HCR_FMO|HCR_VM|
- HCR_TWI|HCR_TSC|HCR_TAC|HCR_SWIO|HCR_TIDCP, HCR_EL2);
+ HCR_TWE|HCR_TWI|HCR_TSC|HCR_TAC|HCR_SWIO|HCR_TIDCP, HCR_EL2);
isb();
}
advance_pc(regs, hsr);
return;
}
- /* at the moment we only trap WFI */
- vcpu_block_unless_event_pending(current);
+ if ( hsr.wfi_wfe.ti ) {
+ /* Yield the VCPU for WFE */
+ vcpu_force_reschedule(current);
+ } else {
+ /* Block the VCPU for WFI */
+ vcpu_block_unless_event_pending(current);
+ }
advance_pc(regs, hsr);
break;
case HSR_EC_CP15_32:
unsigned long ec:6; /* Exception Class */
} cond;
+ struct hsr_wfi_wfe {
+ unsigned long ti:1; /* Trapped instruction */
+ unsigned long sbzp:19;
+ unsigned long cc:4; /* Condition Code */
+ unsigned long ccvalid:1;/* CC Valid */
+ unsigned long len:1; /* Instruction length */
+ unsigned long ec:6; /* Exception Class */
+ } wfi_wfe;
+
/* reg, reg0, reg1 are 4 bits on AArch32, the fifth bit is sbzp. */
struct hsr_cp32 {
unsigned long read:1; /* Direction */