On 26.04.2012, at 23:45, Benjamin Herrenschmidt wrote:
>
>> +static void kvmppc_fill_pt_regs(struct pt_regs *regs)
>> +{
>> + ulong r1, ip, msr, lr;
>> +
>> + asm("mr %0, 1" : "=r"(r1));
>> + asm("mflr %0" : "=r"(lr));
>> + asm("mfmsr %0" : "=r"(msr));
>> + asm("bl 1f; 1: mflr %0" : "=r"(ip));
>> +
>> + memset(regs, 0, sizeof(*regs));
>> + regs->gpr[1] = r1;
>> + regs->nip = ip;
>> + regs->msr = msr;
>> + regs->link = lr;
>> +}
>
> That is -very- gross ... I suppose it works but yuck :-)
>
>> +static void kvmppc_restart_interrupt(struct kvm_vcpu *vcpu,
>> + unsigned int exit_nr)
>> +{
>> + struct pt_regs regs;
>> +
>> + switch (exit_nr) {
>> + case BOOK3S_INTERRUPT_EXTERNAL:
>> + case BOOK3S_INTERRUPT_EXTERNAL_LEVEL:
>> + case BOOK3S_INTERRUPT_EXTERNAL_HV:
>> + kvmppc_fill_pt_regs(®s);
>> + soft_irq_disable();
>> + do_IRQ(®s);
>> + soft_irq_enable();
>
> What are those soft_irq_disable/enable ? They look like sometimes
> local_irq_disable/enable and sometimes something else ?
Yeah, on ppc64 these are local_irq_en/disable. On ppc32, they are nops, because there we don't distinguish between soft and hard irq enabled.
>
> If you are hard disabled already (which you should be) and want to
> "mark" things as soft disabled, I suppose that will work except that
> you'll be missing PACA_IRQ_HARD_DIS in irq_happened, so
> local_irq_enable() will not hard-enable.
I jump through this hoop because in kernel/posix-cpu-timers.c:1288 the BUG_ON triggers, since irqs_disabled() is false, because they are hard, but not soft disabled (which is a combination that iiuc doesn't happen in normal Linux).
>
>> + break;
>> + case BOOK3S_INTERRUPT_DECREMENTER:
>> + case BOOK3S_INTERRUPT_HV_DECREMENTER:
>> + kvmppc_fill_pt_regs(®s);
>> + soft_irq_disable();
>> + timer_interrupt(®s);
>> + soft_irq_enable();
>> + break;
>
> Same.
>
>> + case BOOK3S_INTERRUPT_MACHINE_CHECK:
>> + /* FIXME */
>> + break;
>> + case BOOK3S_INTERRUPT_PERFMON:
>> + kvmppc_fill_pt_regs(®s);
>> + soft_irq_disable();
>> + performance_monitor_exception(®s);
>> + soft_irq_enable();
>> + break;
>
> Same.
>
>> + }
>> +}
>> +
>> int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
>> unsigned int exit_nr)
>> {
>> @@ -548,6 +602,10 @@ int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
>> run->exit_reason = KVM_EXIT_UNKNOWN;
>> run->ready_for_interrupt_injection = 1;
>>
>> + /* restart interrupts if they were meant for the host */
>> + kvmppc_restart_interrupt(vcpu, exit_nr);
>> + __hard_irq_enable();
>> +
>
> I suppose that's to work around the above comment about missing
> PACA_IRQ_HARD_DIS ?
This is to actually enable interrupts for real, regardless of ppc64 and ppc32. In fact, the previous code was pretty buggy - it was running the handlers with interrupts disabled ;).
Alex
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
