|
|
|
|
|
|
|
|
|
|
xen-devel
Re: [Xen-devel] Testing Report for Xen 3.0.3 rc3
>We have tested Xen for 3.0.3 rc3. In the testing we mainly focus on
>retesting the issues of the rc2 & rc1.
>
>Following is the detail status:
>New issue exists in 3.0.3 rc3:
>1. FC6&FC5 IA32e SMP HVM Guest crash when booting SMP Guest
>Since FC5 IA32 SMP will still crash HVM Guest in 3.0.3 rc3, we checked
>FC6, found FC6 IA32e SMP also will crash HVM Guest. And we can boot up
>FC6 IA32 SMP Guest.
We're trying to reproduce this but cannot so far. I have a fix/workaround
in theory but need to confirm whether the guest is intentionally updating
the redirection table or if something bad is happening.
Can you please add the following debugging patch and post the output?
cheers,
S.
diff -r 3cc0b589c235 xen/arch/x86/hvm/vioapic.c
--- a/xen/arch/x86/hvm/vioapic.c Tue Oct 17 09:59:11 2006 +0100
+++ b/xen/arch/x86/hvm/vioapic.c Tue Oct 17 17:25:18 2006 +0100
@@ -179,11 +179,14 @@ static void hvm_vioapic_write_indirect(s
{
uint32_t redir_index = 0;
+ redir_index = (s->ioregsel - 0x10) >> 1;
+
HVM_DBG_LOG(DBG_LEVEL_IOAPIC, "hvm_vioapic_write_indirect "
"change redir index %x val %lx\n",
redir_index, val);
- redir_index = (s->ioregsel - 0x10) >> 1;
+ printk("hvm_vioapic_write_indirect: change redir index "
+ "%x val %lx\n", redir_index, val);
if (redir_index >= 0 && redir_index < IOAPIC_NUM_PINS) {
uint64_t redir_content;
@@ -455,6 +458,8 @@ static void ioapic_deliver(hvm_vioapic_t
if ( (irqno == 0) && (bit !=0) )
{
printk("PIT irq to bit %x\n", bit);
+ printk("deliver_bitmask:%x - deliver_mode:%x\n",
+ deliver_bitmask, delivery_mode);
domain_crash_synchronous();
}
#endif
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
|
|
|
|