When we try to start a CPU with a WAIT PSW, we have to take care that
TCG will actually try to continue executing instructions.
We must therefore really only unhalt the CPU if we don't have a WAIT
PSW. Also document the special order for restart interrupts, which
load a new PSW and change the state to operating.
To keep KVM working, simply don't have a look at the WAIT bit when
loading the PSW. Otherwise the behavior of a restart interrupt when
a CPU stopped would be changed.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <
20170928203708.9376-31-david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
break;
case CPU_STATE_OPERATING:
case CPU_STATE_LOAD:
- /* unhalt the cpu for common infrastructure */
- s390_cpu_unhalt(cpu);
+ /*
+ * Starting a CPU with a PSW WAIT bit set:
+ * KVM: handles this internally and triggers another WAIT exit.
+ * TCG: will actually try to continue to run. Don't unhalt, will
+ * be done when the CPU actually has work (an interrupt).
+ */
+ if (!tcg_enabled() || !(cpu->env.psw.mask & PSW_MASK_WAIT)) {
+ s390_cpu_unhalt(cpu);
+ }
break;
default:
error_report("Requested CPU state is not a valid S390 CPU state: %u",
case CPU_STATE_STOPPED:
/* the restart irq has to be delivered prior to any other pending irq */
cpu_synchronize_state(cs);
- do_restart_interrupt(&cpu->env);
+ /*
+ * Set OPERATING (and unhalting) before loading the restart PSW.
+ * load_psw() will then properly halt the CPU again if necessary (TCG).
+ */
s390_cpu_set_state(CPU_STATE_OPERATING, cpu);
+ do_restart_interrupt(&cpu->env);
break;
case CPU_STATE_OPERATING:
cpu_inject_restart(cpu);