[v2,resend] apic: fixup fallthrough to PIC

Submitted by Mark Asselstine on April 4, 2013, 3:43 p.m.

Details

Message ID 1365090185-26920-1-git-send-email-mark.asselstine@windriver.com
State New
Headers show

Commit Message

Mark Asselstine April 4, 2013, 3:43 p.m.
Commit 0e21e12bb311c4c1095d0269dc2ef81196ccb60a [Don't route PIC
interrupts through the local APIC if the local APIC config says so.]
missed a check to ensure the local APIC is enabled. Since if the local
APIC is disabled it doesn't matter what the local APIC config says.

If this check isn't done and the guest has disabled the local APIC the
guest will receive a general protection fault, similar to what is seen
here:

https://lists.gnu.org/archive/html/qemu-devel/2012-12/msg02304.html

The GPF is caused by an attempt to service interrupt 0xffffffff. This
comes about since cpu_get_pic_interrupt() calls apic_accept_pic_intr()
(with the local APIC disabled apic_get_interrupt() returns -1).
apic_accept_pic_intr() returns 0 and thus the interrupt number which
is returned from cpu_get_pic_interrupt(), and which is attempted to be
serviced, is -1.

Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com>
---

*v2*  moved the checks to a single line

Resending since I would like to get this into the tree.

 hw/apic.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Patch hide | download patch | download mbox

diff --git a/hw/apic.c b/hw/apic.c
index d2395f0..e5e3435 100644
--- a/hw/apic.c
+++ b/hw/apic.c
@@ -590,7 +590,7 @@  int apic_accept_pic_intr(DeviceState *d)
     APICCommonState *s = DO_UPCAST(APICCommonState, busdev.qdev, d);
     uint32_t lvt0;
 
-    if (!s)
+    if (!s || !(s->spurious_vec & APIC_SV_ENABLE))
         return -1;
 
     lvt0 = s->lvt[APIC_LVT_LINT0];