[PATCH] don't assume iosapic interrupt controller in mca.c

[PATCH] don't assume iosapic interrupt controller in mca.c

[Jesse Barnes]

Any problems with this patch?  Sn2 doesn't have an iosapic interrupt
controller, so we'd rather not hardcode that type into mca.c for the CPE
interrupt vector.

Thanks,
Jesse

=== arch/ia64/kernel/mca.c 1.59 vs edited ==--- 1.59/arch/ia64/kernel/mca.c	Thu Feb 12 16:14:34 2004
+++ edited/arch/ia64/kernel/mca.c	Mon Mar  1 14:43:35 2004
@@ -103,8 +103,6 @@
 
 static ia64_mc_info_t		ia64_mc_info;
 
-extern struct hw_interrupt_type	irq_type_iosapic_level;
-
 struct ia64_mca_tlb_info ia64_mca_tlb_list[NR_CPUS];
 
 #define MAX_CPE_POLL_INTERVAL (15*60*HZ) /* 15 minutes */
@@ -1253,7 +1251,6 @@
 				if (irq_to_vector(irq) = cpev) {
 					desc = irq_descp(irq);
 					desc->status |= IRQ_PER_CPU;
-					desc->handler = &irq_type_iosapic_level;
 					setup_irq(irq, &mca_cpe_irqaction);
 				}
 			ia64_mca_register_cpev(cpev);

[PATCH] don't assume iosapic interrupt controllers

[Jesse Barnes]

Some ia64 machines don't use an iosapic interrupt controller (a flagrant
violation of the DIG64 spec, gasp!), so don't assume it's there in
mca.c.  This patch got no complaints last time I posted it, so I'm
proposing it for inclusion.

Thanks,
Jesse


=== arch/ia64/kernel/mca.c 1.59 vs edited ==--- 1.59/arch/ia64/kernel/mca.c	Thu Feb 12 16:14:34 2004
+++ edited/arch/ia64/kernel/mca.c	Mon Mar  1 14:43:35 2004
@@ -103,8 +103,6 @@
 
 static ia64_mc_info_t		ia64_mc_info;
 
-extern struct hw_interrupt_type	irq_type_iosapic_level;
-
 struct ia64_mca_tlb_info ia64_mca_tlb_list[NR_CPUS];
 
 #define MAX_CPE_POLL_INTERVAL (15*60*HZ) /* 15 minutes */
@@ -1253,7 +1251,6 @@
 				if (irq_to_vector(irq) = cpev) {
 					desc = irq_descp(irq);
 					desc->status |= IRQ_PER_CPU;
-					desc->handler = &irq_type_iosapic_level;
 					setup_irq(irq, &mca_cpe_irqaction);
 				}
 			ia64_mca_register_cpev(cpev);

Re: [PATCH] don't assume iosapic interrupt controllers

[David Mosberger]

>>>>> On Wed, 3 Mar 2004 09:28:35 -0800, jbarnes@sgi.com (Jesse Barnes) said:

  Jesse> Some ia64 machines don't use an iosapic interrupt controller
  Jesse> (a flagrant violation of the DIG64 spec, gasp!), so don't
  Jesse> assume it's there in mca.c.  This patch got no complaints
  Jesse> last time I posted it, so I'm proposing it for inclusion.

Your changelog explains why you want to remove the offending code.  It
doesn't explain why the change is safe for those machines that do have
I/O SAPIC.  Please provide a complete ChangeLog entry (along with the
patch).

Thanks,

	--david

[PATCH] don't assume iosapic interrupt controllers take 2

[Jesse Barnes]

Some ia64 machines don't use an iosapic interrupt controller (a flagrant
violation of the DIG64 spec, gasp!), so don't assume it's there in
mca.c.  Platforms that provide a ACPI_INTERRUPT_CPEI vector are
responsible for registering its interrupt controller type in platform
specific code, iosapic.c:iosapic_register_platform_intr() does this for
platforms with an iosapic where the CPEI vector is listed in the ACPI
namespace, and on sn2, all external interrupts are assigned the
irq_type_sn type, so this change of assumption should be safe.

Thanks,
Jesse


=== arch/ia64/kernel/mca.c 1.59 vs edited ==--- 1.59/arch/ia64/kernel/mca.c	Thu Feb 12 16:14:34 2004
+++ edited/arch/ia64/kernel/mca.c	Mon Mar  1 14:43:35 2004
@@ -103,8 +103,6 @@
 
 static ia64_mc_info_t		ia64_mc_info;
 
-extern struct hw_interrupt_type	irq_type_iosapic_level;
-
 struct ia64_mca_tlb_info ia64_mca_tlb_list[NR_CPUS];
 
 #define MAX_CPE_POLL_INTERVAL (15*60*HZ) /* 15 minutes */
@@ -1253,7 +1251,6 @@
 				if (irq_to_vector(irq) = cpev) {
 					desc = irq_descp(irq);
 					desc->status |= IRQ_PER_CPU;
-					desc->handler = &irq_type_iosapic_level;
 					setup_irq(irq, &mca_cpe_irqaction);
 				}
 			ia64_mca_register_cpev(cpev);

----- End forwarded message -----