From: Tetsu Yamamoto <tyamamot@redhat.com>
Date: Thu, 3 Jan 2008 10:36:20 -0500
Subject: [Xen] ia64: stop all CPUs on HV panic
Message-id: 477D00F4.7070107@redhat.com
O-Subject: [RHEL5.2 PATCH] [Xen] Dom0/DomU may continue to work after hypervisor panic
Bugzilla: 426129

This patch fixes BZ#426129.
https://bugzilla.redhat.com/show_bug.cgi?id=426129

This is backported from the upstream to stop all CPUs on hypervisor panic.
http://xenbits.xensource.com/ext/xen-ia64-unstable.hg?rev/124f7a7f5f84

I've tested these patches with kernel-2.6.18-58.el5, and confirmed that
Dom0 is stopped on hypervisor panic.

Please review and ACK.

Regards,

Tetsu Yamamoto

# HG changeset patch
# User Alex Williamson <alex.williamson@hp.com>
# Date 1197393757 25200
# Node ID 124f7a7f5f84f69c1d6430ace1affd8f6cc00626
# Parent  35b2c54f59d5ab9c8fa414be86f2668da4149b6a
[IA64] Stop all cpus at panic

Current panic() of hypervisor doesn't stop all cpus.
So domains can work after hypervisor panic (when using
noreboot option).  If dom0 continues to work after
xen panic, the system may get into serious problems.

This patch fixes the issue.

I chose cpu_halt() to stop other cpus by smp_send_stop().
I think it should work fine for kexec.  machine_halt()
is only spinloop for leaving the panic calltrace.

Signed-off-by: Akio Takebe <takebe_akio@jp.fujitsu.com>

Acked-by: "Stephen C. Tweedie" <sct@redhat.com>
Acked-by: Jarod Wilson <jwilson@redhat.com>
Acked-by: Bill Burns <bburns@redhat.com>

diff --git a/arch/ia64/linux-xen/smp.c b/arch/ia64/linux-xen/smp.c
index f9ae4f2..5ebe45c 100644
--- a/arch/ia64/linux-xen/smp.c
+++ b/arch/ia64/linux-xen/smp.c
@@ -122,9 +122,7 @@ stop_this_cpu (void)
 	cpu_clear(smp_processor_id(), cpu_online_map);
 	max_xtp();
 	local_irq_disable();
-#ifndef XEN
 	cpu_halt();
-#endif
 }
 
 void
@@ -132,9 +130,7 @@ cpu_die(void)
 {
 	max_xtp();
 	local_irq_disable();
-#ifndef XEN
 	cpu_halt();
-#endif
 	/* Should never be here */
 	BUG();
 	for (;;);