Skip to contentSkip to conversation

small mca cleanup

PATCH In review Christoph Egger 2 participants 5 messages Aug 16, 2007

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 small mca cleanup

Hi!

The MCG_CAP MSR never returns a negative count of available error-reporting banks. Thus make nr_mce_banks unsigned.

While here, do some other minor cleanups.

Signature · 12 lines
AMD Saxony, Dresden, Germany
Operating System Research Center

Legal Information:
AMD Saxony Limited Liability Company & Co. KG
Sitz (Geschäftsanschrift):
   Wilschdorfer Landstr. 101, 01109 Dresden, Deutschland
Registergericht Dresden: HRA 4896
vertretungsberechtigter Komplementär:
   AMD Saxony LLC (Sitz Wilmington, Delaware, USA)
Geschäftsführer der AMD Saxony LLC:
   Dr. Hans-R. Deppe, Thomas McCoy
xen/arch/x86/cpu/mcheck/mce.c +34 -20
diff -r 256160ff19b7 xen/arch/x86/cpu/mcheck/mce.c
@@ -15,7 +15,7 @@
 #include "mce.h"
 
 int mce_disabled = 0;
-int nr_mce_banks;
+unsigned int nr_mce_banks;
 
 /* Handle unconfigured int18 (should never happen) */
 static fastcall void unexpected_machine_check(struct cpu_user_regs * regs, long error_code)
@@ -29,35 +29,49 @@ void fastcall (*machine_check_vector)(st
 /* This has to be run for each processor */
 void mcheck_init(struct cpuinfo_x86 *c)
 {
-	if (mce_disabled==1)
+	uint32_t mca, mce;
+
+	if (mce_disabled == 1) {
+		printk("MCE support disabled by bootparam\n");
 		return;
+	}
+
+	/* Xen requires an i686 class CPU. So this
+	 * check should be almost always true here. */
+	mca = cpu_has(c, X86_FEATURE_MCA);
+	mce = cpu_has(c, X86_FEATURE_MCE);
+
+	if (!mca && !mce) {
+		printk(KERN_INFO "CPU%i: No machine check support available\n",
+			smp_processor_id());
+		return;
+	}
 
 	switch (c->x86_vendor) {
-		case X86_VENDOR_AMD:
-			if (c->x86==6 || c->x86==15)
-				amd_mcheck_init(c);
-			break;
+	case X86_VENDOR_AMD:
+		amd_mcheck_init(c);
+		break;
 
-		case X86_VENDOR_INTEL:
+	case X86_VENDOR_INTEL:
 #ifndef CONFIG_X86_64
-			if (c->x86==5)
-				intel_p5_mcheck_init(c);
-			if (c->x86==6)
-				intel_p6_mcheck_init(c);
+		if (c->x86==5)
+			intel_p5_mcheck_init(c);
+		if (c->x86==6)
+			intel_p6_mcheck_init(c);
 #endif
-			if (c->x86==15)
-				intel_p4_mcheck_init(c);
-			break;
+		if (c->x86==15)
+			intel_p4_mcheck_init(c);
+		break;
 
 #ifndef CONFIG_X86_64
-		case X86_VENDOR_CENTAUR:
-			if (c->x86==5)
-				winchip_mcheck_init(c);
-			break;
+	case X86_VENDOR_CENTAUR:
+		if (c->x86==5)
+			winchip_mcheck_init(c);
+		break;
 #endif
 
-		default:
-			break;
+	default:
+		break;
 	}
 }
 
xen/arch/x86/cpu/mcheck/mce.h +1 -2
diff -r 256160ff19b7 xen/arch/x86/cpu/mcheck/mce.h
@@ -10,5 +10,4 @@ extern fastcall void (*machine_check_vec
 extern fastcall void (*machine_check_vector)(struct cpu_user_regs *, long error_code);
 
 extern int mce_disabled __initdata;
-extern int nr_mce_banks;
-
+extern unsigned int nr_mce_banks;

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#1 lore raw

The file you modify is from upstream Linux, and that's where this patch belongs.

-- Keir

Quoted 11 lines · Christoph Egger

Hi!

The MCG_CAP MSR never returns a negative count of available error-reporting banks. Thus make nr_mce_banks unsigned.

While here, do some other minor cleanups.

#2 lore raw
Quoted 2 lines · On Thursday 16 August 2007 16:42:37 Keir Fraser wrote:

The file you modify is from upstream Linux, and that's where this patch belongs.

oh, I have many changes in my queue in files that are from Linux (and also new files, that are not in Linux).
They do one thing, that does not belong into Linux (or any other OS):
Notifying guests, preferably Dom0.

I'm about extracting the least intrusive changes first. We can still push changes into Linux that belong to there.

For more information where this is going to, please see the discussion "MCE/MCA concept" on this list. The thread started here:
http://lists.xensource.com/archives/html/xen-devel/2007-05/msg01015.html

Christoph

Quoted 12 lines

-- Keir

Quoted 8 lines · Christoph Egger

Hi!

The MCG_CAP MSR never returns a negative count of available error-reporting banks. Thus make nr_mce_banks unsigned.

While here, do some other minor cleanups.

Signature · 12 lines
AMD Saxony, Dresden, Germany
Operating System Research Center

Legal Information:
AMD Saxony Limited Liability Company & Co. KG
Sitz (Geschäftsanschrift):
   Wilschdorfer Landstr. 101, 01109 Dresden, Deutschland
Registergericht Dresden: HRA 4896
vertretungsberechtigter Komplementär:
   AMD Saxony LLC (Sitz Wilmington, Delaware, USA)
Geschäftsführer der AMD Saxony LLC:
   Dr. Hans-R. Deppe, Thomas McCoy
#3 lore raw

Oh yeah, sure: Xen-specific stuff and the general MCE/MCA reworking that you're talking about are one thing. But there's no point in minor fixes and stylistic cleanups going into Xen and bypassing upstream Linux.

-- Keir

Quoted 32 lines · Christoph Egger
Quoted 2 lines · On Thursday 16 August 2007 16:42:37 Keir Fraser wrote:

The file you modify is from upstream Linux, and that's where this patch belongs.

oh, I have many changes in my queue in files that are from Linux (and also new files, that are not in Linux).
They do one thing, that does not belong into Linux (or any other OS):
Notifying guests, preferably Dom0.

I'm about extracting the least intrusive changes first. We can still push changes into Linux that belong to there.

For more information where this is going to, please see the discussion "MCE/MCA concept" on this list. The thread started here:
http://lists.xensource.com/archives/html/xen-devel/2007-05/msg01015.html

Christoph

Quoted 12 lines

-- Keir

Quoted 8 lines · Christoph Egger

Hi!

The MCG_CAP MSR never returns a negative count of available error-reporting banks. Thus make nr_mce_banks unsigned.

While here, do some other minor cleanups.

#4 lore raw

To be fair -- if you posted this as part of a fuller patchset to move us away from Linux-synced files and towards a new fully thought-out method for handling MCA/MCE then I'd have no problems with it (but maybe I would with the patchset as a whole :-). It's just I see no sense in applying it in isolation.

-- Keir

Quoted 46 lines · Keir Fraser

Oh yeah, sure: Xen-specific stuff and the general MCE/MCA reworking that you're talking about are one thing. But there's no point in minor fixes and stylistic cleanups going into Xen and bypassing upstream Linux.

-- Keir

Quoted 9 lines · Christoph Egger
Quoted 2 lines · On Thursday 16 August 2007 16:42:37 Keir Fraser wrote:

The file you modify is from upstream Linux, and that's where this patch belongs.

oh, I have many changes in my queue in files that are from Linux (and also new files, that are not in Linux).
They do one thing, that does not belong into Linux (or any other OS):
Notifying guests, preferably Dom0.

Quoted 23 lines

I'm about extracting the least intrusive changes first. We can still push changes into Linux that belong to there.

For more information where this is going to, please see the discussion "MCE/MCA concept" on this list. The thread started here:
http://lists.xensource.com/archives/html/xen-devel/2007-05/msg01015.html

Christoph

Quoted 12 lines

-- Keir

Quoted 8 lines · Christoph Egger

Hi!

The MCG_CAP MSR never returns a negative count of available error-reporting banks. Thus make nr_mce_banks unsigned.

While here, do some other minor cleanups.

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel