Hello,
While processing __adeos_handle_event enables interrupts for the ectual
event handler periods. Interrupts might happen then and only logged for
domains lower than the current domain in the __adeos_handle_event loop.
The logged irqs would normally get replayed at the next real interrupt
if the domain that caused the event was lower than the ones the
interrupts were logged in or if it wasn't lower, at the next
suspend_domain. On powersave-enabled ppc machines this might cause a
chain where the cpu goes napping and wakes at the next timer tick which,
having possibly been missed at *handle_event, happens after maximum
decrementer period (~minutes).
The following patch seems to help in the described case, but I've also
observed another case where interrupt seems to get dropped altogether.
Performance didn't seem to suffer much from the patch. If anything, the
latency/cruncher results got better. This is for the non-threaded
domains case.
-- Heikki Lindholm
--- kernel/adeos.c.orig 2005-08-31 12:01:30.002707784 +0300
+++ kernel/adeos.c 2005-08-31 13:10:25.148070136 +0300
@@ -196,6 +196,7 @@
adeos_declare_cpuid;
adevinfo_t evinfo;
int propagate = 1;
+ int notfirst = 0;
adeos_lock_cpu(flags);
@@ -204,7 +205,20 @@
list_for_each_safe(pos,npos,&__adeos_pipeline) {
next_domain = list_entry(pos,adomain_t,p_link);
-
+ /* did the previous unlock_cpu gap cause interrupts here? */
+ if (notfirst && next_domain != adp_root && !test_bit(IPIPE_STALL_FLAG,
+ &next_domain->cpudata[cpuid].status) &&
+ next_domain->cpudata[cpuid].irq_pending_hi != 0)
+ {
+ adp_cpu_current[cpuid] = next_domain;
+ __adeos_sync_stage(IPIPE_IRQMASK_ANY);
+
+ if (adp_cpu_current[cpuid] != next_domain)
+ /* Something has changed the current domain under our
+ * feet recycling the register set; take note. */
+ this_domain = adp_cpu_current[cpuid];
+ }
+
if (next_domain->events[event].handler != NULL)
{
adp_cpu_current[cpuid] = next_domain;
