From: Jarod Wilson <jwilson@redhat.com>
Date: Mon, 25 Jun 2007 10:07:37 -0400
Subject: [PPC64] kdump: fix irq distribution on ppc970
Message-id: 467FCC29.4000804@redhat.com
O-Subject: [RHEL5.x PATCH] PPC64 kdump - fix irq distribution on ppc970 w/maxcpus=1
Bugzilla: 208659

This is a multi-part message in MIME format.
There's a problem with some PowerPC 970 systems where when one specifies
maxcpus=1, the kernel still tries to distribute IRQ's across all cpus in
the system. This is particularly problematic in the kexec/kdump context,
where we always try to bring the kdump environment up on a single cpu.

For RHEL5 GA, we worked around this by adding the extra kernel parameter
'noirqdistrib' to the ppc64 kdump config, but a proper fix recently made
its way into the upstream ppc kernel tree:

http://ozlabs.org/pipermail/linuxppc-dev/2007-June/037688.html

I've done the trivial backport to apply this atop our current RHEL5 tree
and successfully tested it on a previously impacted system and verified
that it does indeed fix the irq distribution problem (able to capture
multiple vmcores using kdump w/o the noirqdistrib parameter anymore).

This is the fix for both bug 208657 and bug 208659. Not a must-have for
5.1, can be delayed to 5.2, but here it is...

--
Jarod Wilson
jwilson@redhat.com

In some of the PPC970 based systems, interrupt would be distributed to
offline cpus also even when booted with "maxcpus=1". So check whether
cpu online map and cpu present map are equal or not. If they are equal
default_distrib_server is used as interrupt server otherwise boot cpu
(default_server) used as interrupt server.

In addition to this, if an interrupt is assigned to a specific cpu (ie
smp affinity) and if that cpu is not online, the earlier code used to
return the default_distrib_server as interrupt server. This patch
introduces an additional paramter to the get_irq function ie
strict_check, based on this parameter, if the cpu is not online either
default_distrib_server or -1 is returned.

Cc: Milton Miller <miltonm@bga.com>,
    Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Mohan Kumar M <mohan@in.ibm.com>

Trivial backport to RHEL-5 tree

Signed-off-by: Jarod Wilson <jwilson@redhat.com>

--

diff --git a/arch/powerpc/platforms/pseries/xics.c b/arch/powerpc/platforms/pseries/xics.c
index 29aec52..a1e133e 100644
--- a/arch/powerpc/platforms/pseries/xics.c
+++ b/arch/powerpc/platforms/pseries/xics.c
@@ -176,9 +176,9 @@ static inline void lpar_qirr_info(int n_cpu , u8 value)
 
 
 #ifdef CONFIG_SMP
-static int get_irq_server(unsigned int virq)
+static int get_irq_server(unsigned int virq, unsigned int strict_check)
 {
-	unsigned int server;
+	int server;
 	/* For the moment only implement delivery to all cpus or one cpu */
 	cpumask_t cpumask = irq_desc[virq].affinity;
 	cpumask_t tmp = CPU_MASK_NONE;
@@ -186,22 +186,25 @@ static int get_irq_server(unsigned int virq)
 	if (!distribute_irqs)
 		return default_server;
 
-	if (cpus_equal(cpumask, CPU_MASK_ALL)) {
-		server = default_distrib_server;
-	} else {
+	if (!cpus_equal(cpumask, CPU_MASK_ALL)) {
 		cpus_and(tmp, cpu_online_map, cpumask);
 
-		if (cpus_empty(tmp))
-			server = default_distrib_server;
-		else
-			server = get_hard_smp_processor_id(first_cpu(tmp));
+		server = first_cpu(tmp);
+
+		if (server < NR_CPUS)
+			return get_hard_smp_processor_id(server);
+
+		if (strict_check)
+			return -1;
 	}
 
-	return server;
+	if (cpus_equal(cpu_online_map, cpu_present_map))
+		return default_distrib_server;
 
+	return default_server;
 }
 #else
-static int get_irq_server(unsigned int virq)
+static int get_irq_server(unsigned int virq, unsigned int strict_check)
 {
 	return default_server;
 }
@@ -212,7 +215,7 @@ static void xics_unmask_irq(unsigned int virq)
 {
 	unsigned int irq;
 	int call_status;
-	unsigned int server;
+	int server;
 
 	pr_debug("xics: unmask virq %d\n", virq);
 
@@ -221,7 +224,7 @@ static void xics_unmask_irq(unsigned int virq)
 	if (irq == XICS_IPI || irq == XICS_IRQ_SPURIOUS)
 		return;
 
-	server = get_irq_server(virq);
+	server = get_irq_server(virq, 0);
 
 	call_status = rtas_call(ibm_set_xive, 3, 1, NULL, irq, server,
 				DEFAULT_PRIORITY);
@@ -256,7 +259,7 @@ static void xics_mask_real_irq(unsigned int irq)
 		return;
 	}
 
-	server = get_irq_server(irq);
+	server = get_irq_server(irq, 0);
 	/* Have to set XIVE to 0xff to be able to remove a slot */
 	call_status = rtas_call(ibm_set_xive, 3, 1, NULL, irq, server, 0xff);
 	if (call_status != 0) {
@@ -419,8 +422,7 @@ static void xics_set_affinity(unsigned int virq, cpumask_t cpumask)
 	unsigned int irq;
 	int status;
 	int xics_status[2];
-	unsigned long newmask;
-	cpumask_t tmp = CPU_MASK_NONE;
+	int irq_server;
 
 	irq = (unsigned int)irq_map[virq].hwirq;
 	if (irq == XICS_IPI || irq == XICS_IRQ_SPURIOUS)
@@ -434,18 +436,21 @@ static void xics_set_affinity(unsigned int virq, cpumask_t cpumask)
 		return;
 	}
 
-	/* For the moment only implement delivery to all cpus or one cpu */
-	if (cpus_equal(cpumask, CPU_MASK_ALL)) {
-		newmask = default_distrib_server;
-	} else {
-		cpus_and(tmp, cpu_online_map, cpumask);
-		if (cpus_empty(tmp))
-			return;
-		newmask = get_hard_smp_processor_id(first_cpu(tmp));
+	/*
+	 * For the moment only implement delivery to all cpus or one cpu.
+	 * Get current irq_server for the given irq
+	 */
+	irq_server = get_irq_server(irq, 1);
+	if (irq_server == -1) {
+		char cpulist[128];
+		cpumask_scnprintf(cpulist, sizeof(cpulist), cpumask);
+		printk(KERN_WARNING "xics_set_affinity: No online cpus in "
+				"the mask %s for irq %d\n", cpulist, virq);
+		return;
 	}
 
 	status = rtas_call(ibm_set_xive, 3, 1, NULL,
-				irq, newmask, xics_status[1]);
+				irq, irq_server, xics_status[1]);
 
 	if (status) {
 		printk(KERN_ERR "xics_set_affinity: irq=%u ibm,set-xive "