On Fri, Nov 27, 2020 at 12:29:35PM +0100, Jan Beulich wrote:
> On 27.11.2020 11:59, Roger Pau Monné wrote:
> > --- a/xen/arch/x86/hvm/irq.c
> > +++ b/xen/arch/x86/hvm/irq.c
> > @@ -187,6 +187,10 @@ void hvm_gsi_assert(struct domain *d, unsigned int gsi)
> > * to know if the GSI is pending or not.
> > */
> > spin_lock(&d->arch.hvm.irq_lock);
> > + if ( gsi == TRACK_IRQ )
> > + debugtrace_printk("hvm_gsi_assert irq %u trig %u assert count
> > %u\n",
> > + gsi, trig, hvm_irq->gsi_assert_count[gsi]);
>
> This produces
>
> 81961 hvm_gsi_assert irq 34 trig 1 assert count 1
>
> Since the logging occurs ahead of the call to assert_gsi(), it
> means we don't signal anything to Dom0, because according to our
> records there's still an IRQ in flight. Unfortunately we only
> see the tail of the trace, so it's not possible to tell how / when
> we got into this state.
>
> Manuel - is this the only patch you have in place? Or did you keep
> any prior ones? Iirc there once was one where Roger also suppressed
> some de-assert call.
Yes, I have some of the previous patches (otherwise Xen panics).
Attached is the diffs I currently have
--
Manuel Bouyer <[email protected]>
NetBSD: 26 ans d'experience feront toujours la difference
--
diff --git a/xen/arch/x86/hvm/irq.c b/xen/arch/x86/hvm/irq.c
index 38ac5fb6c7..9db3dcc957 100644
--- a/xen/arch/x86/hvm/irq.c
+++ b/xen/arch/x86/hvm/irq.c
@@ -187,6 +187,10 @@ void hvm_gsi_assert(struct domain *d, unsigned int gsi)
* to know if the GSI is pending or not.
*/
spin_lock(&d->arch.hvm.irq_lock);
+ if ( gsi == TRACK_IRQ )
+ debugtrace_printk("hvm_gsi_assert irq %u trig %u assert count %u\n",
+ gsi, trig, hvm_irq->gsi_assert_count[gsi]);
+
if ( trig == VIOAPIC_EDGE_TRIG || !hvm_irq->gsi_assert_count[gsi] )
{
if ( trig == VIOAPIC_LEVEL_TRIG )
diff --git a/xen/arch/x86/hvm/vioapic.c b/xen/arch/x86/hvm/vioapic.c
index 67d4a6237f..e6748e0649 100644
--- a/xen/arch/x86/hvm/vioapic.c
+++ b/xen/arch/x86/hvm/vioapic.c
@@ -257,7 +257,11 @@ static void vioapic_write_redirent(
vlapic_adjust_i8259_target(d);
}
else if ( ent.fields.trig_mode == VIOAPIC_EDGE_TRIG )
+ {
+ if ( gsi == TRACK_IRQ )
+ debugtrace_printk("vIO-APIC set edge trigger irq %u\n", gsi);
pent->fields.remote_irr = 0;
+ }
else if ( !ent.fields.mask &&
!ent.fields.remote_irr &&
hvm_irq->gsi_assert_count[idx] )
@@ -278,6 +282,10 @@ static void vioapic_write_redirent(
*/
int ret = vioapic_hwdom_map_gsi(gsi, ent.fields.trig_mode,
ent.fields.polarity);
+
+ if ( gsi == TRACK_IRQ )
+ debugtrace_printk("vIO-APIC UNMASK irq %u\n", gsi);
+
if ( ret )
{
gprintk(XENLOG_ERR,
@@ -285,6 +293,9 @@ static void vioapic_write_redirent(
unmasked = 0;
}
}
+ else if ( is_hardware_domain(d) && gsi == TRACK_IRQ )
+ debugtrace_printk("vIO-APIC MASK irq %u\n", gsi);
+
if ( gsi == 0 || unmasked )
pt_may_unmask_irq(d, NULL);
@@ -405,6 +416,10 @@ static void vioapic_deliver(struct hvm_vioapic *vioapic,
unsigned int pin)
ASSERT(spin_is_locked(&d->arch.hvm.irq_lock));
+ if ( irq == TRACK_IRQ )
+ debugtrace_printk("vIO-APIC deliver irq %u vector %u\n",
+ irq, vector);
+
HVM_DBG_LOG(DBG_LEVEL_IOAPIC,
"dest=%x dest_mode=%x delivery_mode=%x "
"vector=%x trig_mode=%x",
diff --git a/xen/arch/x86/io_apic.c b/xen/arch/x86/io_apic.c
index e66fa99ec7..c28025657d 100644
--- a/xen/arch/x86/io_apic.c
+++ b/xen/arch/x86/io_apic.c
@@ -1641,6 +1641,9 @@ static void mask_and_ack_level_ioapic_irq(struct irq_desc
*desc)
unsigned long v;
int i;
+ if ( desc->irq == TRACK_IRQ )
+ debugtrace_printk("ACK irq %u\n", desc->irq);
+
irq_complete_move(desc);
if ( !directed_eoi_enabled )
@@ -1688,6 +1691,9 @@ static void mask_and_ack_level_ioapic_irq(struct irq_desc
*desc)
static void end_level_ioapic_irq_old(struct irq_desc *desc, u8 vector)
{
+ if ( desc->irq == TRACK_IRQ )
+ debugtrace_printk("END irq %u\n", desc->irq);
+
if ( directed_eoi_enabled )
{
if ( !(desc->status & (IRQ_DISABLED|IRQ_MOVE_PENDING)) )
diff --git a/xen/arch/x86/irq.c b/xen/arch/x86/irq.c
index 93c4fb9a79..c3a75d98a7 100644
--- a/xen/arch/x86/irq.c
+++ b/xen/arch/x86/irq.c
@@ -1109,6 +1109,10 @@ static void irq_guest_eoi_timer_fn(void *data)
unsigned int i, irq = desc - irq_desc;
irq_guest_action_t *action;
+ if ( desc->irq == TRACK_IRQ )
+ debugtrace_printk("irq_guest_eoi_timer_fn irq %u status %x\n",
+ desc->irq, desc->status);
+
spin_lock_irq(&desc->lock);
if ( !(desc->status & IRQ_GUEST) )
@@ -1118,6 +1122,10 @@ static void irq_guest_eoi_timer_fn(void *data)
ASSERT(action->ack_type != ACKTYPE_NONE);
+ if ( desc->irq == TRACK_IRQ )
+ debugtrace_printk("ack_type %u in_flight %u\n",
+ action->ack_type, action->in_flight);
+
/*
* Is no IRQ in flight at all, or another instance of this timer already
* running? Skip everything to avoid forcing an EOI early.
@@ -1837,6 +1845,12 @@ static void do_IRQ_guest(struct irq_desc *desc, unsigned
int vector)
unsigned int i;
struct pending_eoi *peoi = this_cpu(pending_eoi);
+ if ( desc->irq == TRACK_IRQ )
+ debugtrace_printk("do_IRQ_guest irq %u nr_guests %u ack_type %u
in_flight %u\n",
+ desc->irq, action->nr_guests, action->ack_type,
+ action->in_flight);
+
+
if ( unlikely(!action->nr_guests) )
{
/* An interrupt may slip through while freeing an ACKTYPE_EOI irq. */
diff --git a/xen/drivers/passthrough/io.c b/xen/drivers/passthrough/io.c
index 6b1305a3e5..86c2db9da0 100644
--- a/xen/drivers/passthrough/io.c
+++ b/xen/drivers/passthrough/io.c
@@ -174,7 +174,6 @@ static void pt_irq_time_out(void *data)
* In the identity mapped case the EOI can also be done now, this way
* the iteration over the list of domain pirqs is avoided.
*/
- hvm_gsi_deassert(irq_map->dom, dpci_pirq(irq_map)->pirq);
irq_map->flags |= HVM_IRQ_DPCI_EOI_LATCH;
pt_irq_guest_eoi(irq_map->dom, irq_map, NULL);
spin_unlock(&irq_map->dom->event_lock);
@@ -828,6 +827,9 @@ int hvm_do_IRQ_dpci(struct domain *d, struct pirq *pirq)
!pirq_dpci || !(pirq_dpci->flags & HVM_IRQ_DPCI_MAPPED) )
return 0;
+ if ( pirq->pirq == TRACK_IRQ )
+ debugtrace_printk("hvm_do_IRQ_dpci irq %u\n", pirq->pirq);
+
pirq_dpci->masked = 1;
raise_softirq_for(pirq_dpci);
return 1;
@@ -1010,6 +1012,9 @@ void hvm_dpci_eoi(struct domain *d, unsigned int
guest_gsi,
if ( !is_iommu_enabled(d) )
return;
+ if ( guest_gsi == TRACK_IRQ )
+ debugtrace_printk("hvm_dpci_eoi irq %u\n", guest_gsi);
+
if ( is_hardware_domain(d) )
{
spin_lock(&d->event_lock);
diff --git a/xen/drivers/vpci/msix.c b/xen/drivers/vpci/msix.c
index 64dd0a929c..3eb6102a61 100644
--- a/xen/drivers/vpci/msix.c
+++ b/xen/drivers/vpci/msix.c
@@ -370,7 +370,7 @@ static int msix_write(struct vcpu *v, unsigned long addr,
unsigned int len,
entry->updated = false;
}
- else
+ else if ( msix->enabled )
vpci_msix_arch_mask_entry(entry, pdev, entry->masked);
break;
diff --git a/xen/include/xen/irq.h b/xen/include/xen/irq.h
index 43d567fe44..871810134f 100644
--- a/xen/include/xen/irq.h
+++ b/xen/include/xen/irq.h
@@ -174,4 +174,6 @@ unsigned int arch_hwdom_irqs(domid_t);
void arch_evtchn_bind_pirq(struct domain *, int pirq);
#endif
+#define TRACK_IRQ 34
+
#endif /* __XEN_IRQ_H__ */
