Re: [Qemu-devel] [PATCH v4 5/5] i8259: fix dynamically masking slave IRQs with IMR register

[Paolo Bonzini <pbonzini@redhat.com>]

Il 04/09/2012 16:29, Maciej W. Rozycki ha scritto:
>  So first of all, the *output* of the 8259A is always edge triggered, 
> regardless of whether it's the master or one of the slaves (only one slave 
> is used in the PC/AT architecture, but up to eight are supported; the 
> PC/XT had none).  

I swear I read all your message :) but this seems to be the crux.  It means
that something like this ought to fix the bug too.  Matthew, can you post
your code or test it?

diff --git a/hw/i8259.c b/hw/i8259.c
index 53daf78..3dc1dff 100644
--- a/hw/i8259.c
+++ b/hw/i8259.c
@@ -104,12 +104,11 @@ static void pic_update_irq(PICCommonState *s)
     int irq;
 
     irq = pic_get_irq(s);
+    qemu_irq_lower(s->int_out[0]);
     if (irq >= 0) {
         DPRINTF("pic%d: imr=%x irr=%x padd=%d\n",
                 s->master ? 0 : 1, s->imr, s->irr, s->priority_add);
         qemu_irq_raise(s->int_out[0]);
-    } else {
-        qemu_irq_lower(s->int_out[0]);
     }
 }
 
The logic of the in-kernel 8259 is a bit different, but something
like this should do it, too:

diff --git a/arch/x86/kvm/i8259.c b/arch/x86/kvm/i8259.c
index 81cf4fa..feb6d5b 100644
--- a/arch/x86/kvm/i8259.c
+++ b/arch/x86/kvm/i8259.c
@@ -174,9 +174,11 @@ static void pic_update_irq(struct kvm_pic *s)
 		/*
 		 * if irq request by slave pic, signal master PIC
 		 */
+		pic_set_irq1(&s->pics[0], 2, 0);
 		pic_set_irq1(&s->pics[0], 2, 1);
+	} else if (s->pics[0].irr & (1 << 2))
 		pic_set_irq1(&s->pics[0], 2, 0);
-	}
+
 	irq = pic_get_irq(&s->pics[0]);
 	pic_irq_request(s->kvm, irq >= 0);
 }

Both patches untested.

Paolo