Disable CONFIG_SMP for IA64_HP_SIM

Disable CONFIG_SMP for IA64_HP_SIM

[Ian Wienand]

[-- Attachment #1.1: Type: text/plain, Size: 429 bytes --]

CONFIG_SMP depends on CONFIG_IOSAPIC to build (maybe this is the real
problem, I don't know).  

CONFIG_IOSAPIC excludes IA64_HP_SIM.  Some sample configs [1] have SMP
enabled for the simulator, and it broke my autobuild, so maybe making
it explicit would be good?

-i
ianw@gelato.unsw.edu.au
http://www.gelato.unsw.edu.au

[1] http://www.hpl.hp.com/research/linux/ski/get.php?file=2.6.5-config&name=config.2.5.5



[-- Attachment #1.2: Kconfig-nosimsmp.patch --]
[-- Type: text/plain, Size: 417 bytes --]

===== arch/ia64/Kconfig 1.73 vs edited =====
--- 1.73/arch/ia64/Kconfig	Sat May 15 12:00:12 2004
+++ edited/arch/ia64/Kconfig	Wed May 19 10:52:57 2004
@@ -225,6 +225,7 @@
 
 config SMP
 	bool "Symmetric multi-processing support"
+	depends on !IA64_HP_SIM
 	help
 	  This enables support for systems with more than one CPU. If you have
 	  a system with only one CPU, say N.  If you have a system with more

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Disable CONFIG_SMP for IA64_HP_SIM

[David Mosberger]

>>>>> On Wed, 19 May 2004 11:23:26 +1000, Ian Wienand <ianw@gelato.unsw.edu.au> said:

  Ian> CONFIG_SMP depends on CONFIG_IOSAPIC to build (maybe this is
  Ian> the real problem, I don't know).

  Ian> CONFIG_IOSAPIC excludes IA64_HP_SIM.  Some sample configs [1]
  Ian> have SMP enabled for the simulator, and it broke my autobuild,
  Ian> so maybe making it explicit would be good?

SMP must work with the simulator.  It's often useful for
testing-purposes, even though Ski simulates only one CPU.

CONFIG_SMP probably shouldn't depend on CONFIG_IOSAPIC.

	--david

Re: Disable CONFIG_SMP for IA64_HP_SIM

[Matthew Wilcox]

On Tue, May 18, 2004 at 06:35:12PM -0700, David Mosberger wrote:
> CONFIG_SMP probably shouldn't depend on CONFIG_IOSAPIC.

I agree -- I think the SGI machines don't have an IOSAPIC.

-- 
"Next the statesmen will invent cheap lies, putting the blame upon 
the nation that is attacked, and every man will be glad of those
conscience-soothing falsities, and will diligently study them, and refuse
to examine any refutations of them; and thus he will by and by convince 
himself that the war is just, and will thank God for the better sleep 
he enjoys after this process of grotesque self-deception." -- Mark Twain

Re: Disable CONFIG_SMP for IA64_HP_SIM

[Ian Wienand]

[-- Attachment #1.1: Type: text/plain, Size: 448 bytes --]

On Wed, May 19, 2004 at 02:59:13AM +0100, Matthew Wilcox wrote:
> On Tue, May 18, 2004 at 06:35:12PM -0700, David Mosberger wrote:
> > CONFIG_SMP probably shouldn't depend on CONFIG_IOSAPIC.
> 
> I agree -- I think the SGI machines don't have an IOSAPIC.

I don't think it would effect the SGI machines as they have their own
arch/sn/kernel/irq.c?

My suggestion is attached.

-i
ianw@gelato.unsw.edu.au
http://www.gelato.unsw.edu.au

[-- Attachment #1.2: irqcpumask.patch --]
[-- Type: text/plain, Size: 1230 bytes --]

===== arch/ia64/kernel/iosapic.c 1.42 vs edited =====
--- 1.42/arch/ia64/kernel/iosapic.c	Sat May 15 12:00:12 2004
+++ edited/arch/ia64/kernel/iosapic.c	Wed May 19 12:58:40 2004
@@ -100,7 +100,7 @@
 #endif
 
 static spinlock_t iosapic_lock = SPIN_LOCK_UNLOCKED;
-cpumask_t	__cacheline_aligned pending_irq_cpumask[NR_IRQS];
+extern cpumask_t __cacheline_aligned pending_irq_cpumask[NR_IRQS];
 
 /* These tables map IA-64 vectors to the IOSAPIC pin that generates this vector. */
 
===== arch/ia64/kernel/irq.c 1.39 vs edited =====
--- 1.39/arch/ia64/kernel/irq.c	Sat May 15 12:00:12 2004
+++ edited/arch/ia64/kernel/irq.c	Wed May 19 13:00:18 2004
@@ -56,8 +56,6 @@
 #include <asm/delay.h>
 #include <asm/irq.h>
 
-extern cpumask_t    __cacheline_aligned pending_irq_cpumask[NR_IRQS];
-
 /*
  * Linux has a controller-independent x86 interrupt architecture.
  * every controller has a 'controller-template', that is used
@@ -84,6 +82,11 @@
 		.lock = SPIN_LOCK_UNLOCKED
 	}
 };
+
+/* 
+ * This is updated when the user sets irq affinity via /proc
+ */ 
+cpumask_t    __cacheline_aligned pending_irq_cpumask[NR_IRQS];
 
 #ifdef CONFIG_IA64_GENERIC
 irq_desc_t * __ia64_irq_desc (unsigned int irq)

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Disable CONFIG_SMP for IA64_HP_SIM

[Jesse Barnes]

On Tuesday, May 18, 2004 11:10 pm, Ian Wienand wrote:
> On Wed, May 19, 2004 at 02:59:13AM +0100, Matthew Wilcox wrote:
> > On Tue, May 18, 2004 at 06:35:12PM -0700, David Mosberger wrote:
> > > CONFIG_SMP probably shouldn't depend on CONFIG_IOSAPIC.
> >
> > I agree -- I think the SGI machines don't have an IOSAPIC.
>
> I don't think it would effect the SGI machines as they have their own
> arch/sn/kernel/irq.c?

I'll try it out now.  At one point, we needed iosapic.c since some files 
referenced symbols there in unused codepaths.  I think that's since been 
fixed though.

Jesse

Re: Disable CONFIG_SMP for IA64_HP_SIM

[Jesse Barnes]

On Tuesday, May 18, 2004 11:10 pm, Ian Wienand wrote:
> On Wed, May 19, 2004 at 02:59:13AM +0100, Matthew Wilcox wrote:
> > On Tue, May 18, 2004 at 06:35:12PM -0700, David Mosberger wrote:
> > > CONFIG_SMP probably shouldn't depend on CONFIG_IOSAPIC.
> >
> > I agree -- I think the SGI machines don't have an IOSAPIC.
>
> I don't think it would effect the SGI machines as they have their own
> arch/sn/kernel/irq.c?

The patch looks fine.  sn2 still needs iosapic though (ACPI support relies on 
it), but that's a seperate issue.

Jesse