[PATCH][2.4] generic support for systems with more than 8 CPUs (2/2)

[PATCH][2.4] generic support for systems with more than 8 CPUs (2/2)

[Pallipadi, Venkatesh]

[-- Attachment #1: Type: text/plain, Size: 3202 bytes --]

2/2 : switching to physical mode APIC setup in case of more than 8 CPU system

Thanks,
-Venkatesh

diff -urN linux-2.4.21-pre2.org/Documentation/Configure.help linux-2.4.21-pre2/Documentation/Configure.help
--- linux-2.4.21-pre2.org/Documentation/Configure.help	2002-12-17 20:05:15.000000000 -0800
+++ linux-2.4.21-pre2/Documentation/Configure.help	2002-12-17 20:11:23.000000000 -0800
@@ -262,6 +262,13 @@
 
   If you don't have this computer, you may safely say N.
 
+Other more than 8 logical CPU system support
+CONFIG_X86_MANY_CPU
+  This option is required to support systems with more than 8 logical CPUs.
+
+  If you don't have such a computer, you may safely say N.
+  You dont have to choose this for IBM Summit/x440 systems.
+
 IO-APIC support on uniprocessors
 CONFIG_X86_UP_IOAPIC
   An IO-APIC (I/O Advanced Programmable Interrupt Controller) is an
diff -urN linux-2.4.21-pre2.org/arch/i386/config.in linux-2.4.21-pre2/arch/i386/config.in
--- linux-2.4.21-pre2.org/arch/i386/config.in	2002-12-17 20:05:15.000000000 -0800
+++ linux-2.4.21-pre2/arch/i386/config.in	2002-12-17 20:11:23.000000000 -0800
@@ -229,6 +229,10 @@
          define_bool CONFIG_X86_CLUSTERED_APIC y
       fi
    fi
+   bool 'Other more than 8 logical CPU system support' CONFIG_X86_MANY_CPU
+   if [ "$CONFIG_X86_MANY_CPU" = "y" ]; then
+      define_bool CONFIG_X86_CLUSTERED_APIC y
+   fi
 fi
 
 if [ "$CONFIG_X86_NUMA" != "y" ]; then
diff -urN linux-2.4.21-pre2.org/arch/i386/kernel/mpparse.c linux-2.4.21-pre2/arch/i386/kernel/mpparse.c
--- linux-2.4.21-pre2.org/arch/i386/kernel/mpparse.c	2002-12-17 20:12:21.000000000 -0800
+++ linux-2.4.21-pre2/arch/i386/kernel/mpparse.c	2002-12-17 20:11:23.000000000 -0800
@@ -590,15 +590,6 @@
 	}
 
 
-	printk("Enabling APIC mode: ");
-	if(clustered_apic_mode == CLUSTERED_APIC_NUMAQ)
-		printk("Clustered Logical.	");
-	else if(clustered_apic_mode == CLUSTERED_APIC_XAPIC)
-		printk("Physical.	");
-	else
-		printk("Flat.	");
-	printk("Using %d I/O APICs\n",nr_ioapics);
-
 	if (!num_processors)
 		printk(KERN_ERR "SMP mptable: no processors registered!\n");
 	return num_processors;
@@ -834,6 +825,33 @@
 
 	printk("Processors: %d\n", num_processors);
 	printk("xAPIC support %s present\n", (xapic_support?"is":"is not"));
+
+#ifdef CONFIG_X86_CLUSTERED_APIC
+	/*
+	 * Switch to Physical destination mode in case of generic
+	 * more than 8 CPU system, which has xAPIC support
+	 */
+#define FLAT_APIC_CPU_MAX	8
+	if ((clustered_apic_mode == CLUSTERED_APIC_NONE) &&
+	    (xapic_support) &&
+	    (num_processors > FLAT_APIC_CPU_MAX)) {
+		clustered_apic_mode = CLUSTERED_APIC_XAPIC;
+		apic_broadcast_id = APIC_BROADCAST_ID_XAPIC;
+		int_dest_addr_mode = APIC_DEST_PHYSICAL;
+		int_delivery_mode = dest_Fixed;
+		esr_disable = 1;
+	}
+#endif
+
+	printk("Enabling APIC mode: ");
+	if (clustered_apic_mode == CLUSTERED_APIC_NUMAQ)
+		printk("Clustered Logical.	");
+	else if (clustered_apic_mode == CLUSTERED_APIC_XAPIC)
+		printk("Physical.	");
+	else
+		printk("Flat.	");
+	printk("Using %d I/O APICs\n",nr_ioapics);
+
 	/*
 	 * Only use the first configuration found.
 	 */




[-- Attachment #2: physapic_2.4.21-pre2.patch --]
[-- Type: application/octet-stream, Size: 3008 bytes --]

diff -urN linux-2.4.21-pre2.org/Documentation/Configure.help linux-2.4.21-pre2/Documentation/Configure.help
--- linux-2.4.21-pre2.org/Documentation/Configure.help	2002-12-17 20:05:15.000000000 -0800
+++ linux-2.4.21-pre2/Documentation/Configure.help	2002-12-17 20:11:23.000000000 -0800
@@ -262,6 +262,13 @@
 
   If you don't have this computer, you may safely say N.
 
+Other more than 8 logical CPU system support
+CONFIG_X86_MANY_CPU
+  This option is required to support systems with more than 8 logical CPUs.
+
+  If you don't have such a computer, you may safely say N.
+  You dont have to choose this for IBM Summit/x440 systems.
+
 IO-APIC support on uniprocessors
 CONFIG_X86_UP_IOAPIC
   An IO-APIC (I/O Advanced Programmable Interrupt Controller) is an
diff -urN linux-2.4.21-pre2.org/arch/i386/config.in linux-2.4.21-pre2/arch/i386/config.in
--- linux-2.4.21-pre2.org/arch/i386/config.in	2002-12-17 20:05:15.000000000 -0800
+++ linux-2.4.21-pre2/arch/i386/config.in	2002-12-17 20:11:23.000000000 -0800
@@ -229,6 +229,10 @@
          define_bool CONFIG_X86_CLUSTERED_APIC y
       fi
    fi
+   bool 'Other more than 8 logical CPU system support' CONFIG_X86_MANY_CPU
+   if [ "$CONFIG_X86_MANY_CPU" = "y" ]; then
+      define_bool CONFIG_X86_CLUSTERED_APIC y
+   fi
 fi
 
 if [ "$CONFIG_X86_NUMA" != "y" ]; then
diff -urN linux-2.4.21-pre2.org/arch/i386/kernel/mpparse.c linux-2.4.21-pre2/arch/i386/kernel/mpparse.c
--- linux-2.4.21-pre2.org/arch/i386/kernel/mpparse.c	2002-12-17 20:12:21.000000000 -0800
+++ linux-2.4.21-pre2/arch/i386/kernel/mpparse.c	2002-12-17 20:11:23.000000000 -0800
@@ -590,15 +590,6 @@
 	}
 
 
-	printk("Enabling APIC mode: ");
-	if(clustered_apic_mode == CLUSTERED_APIC_NUMAQ)
-		printk("Clustered Logical.	");
-	else if(clustered_apic_mode == CLUSTERED_APIC_XAPIC)
-		printk("Physical.	");
-	else
-		printk("Flat.	");
-	printk("Using %d I/O APICs\n",nr_ioapics);
-
 	if (!num_processors)
 		printk(KERN_ERR "SMP mptable: no processors registered!\n");
 	return num_processors;
@@ -834,6 +825,33 @@
 
 	printk("Processors: %d\n", num_processors);
 	printk("xAPIC support %s present\n", (xapic_support?"is":"is not"));
+
+#ifdef CONFIG_X86_CLUSTERED_APIC
+	/*
+	 * Switch to Physical destination mode in case of generic
+	 * more than 8 CPU system, which has xAPIC support
+	 */
+#define FLAT_APIC_CPU_MAX	8
+	if ((clustered_apic_mode == CLUSTERED_APIC_NONE) &&
+	    (xapic_support) &&
+	    (num_processors > FLAT_APIC_CPU_MAX)) {
+		clustered_apic_mode = CLUSTERED_APIC_XAPIC;
+		apic_broadcast_id = APIC_BROADCAST_ID_XAPIC;
+		int_dest_addr_mode = APIC_DEST_PHYSICAL;
+		int_delivery_mode = dest_Fixed;
+		esr_disable = 1;
+	}
+#endif
+
+	printk("Enabling APIC mode: ");
+	if (clustered_apic_mode == CLUSTERED_APIC_NUMAQ)
+		printk("Clustered Logical.	");
+	else if (clustered_apic_mode == CLUSTERED_APIC_XAPIC)
+		printk("Physical.	");
+	else
+		printk("Flat.	");
+	printk("Using %d I/O APICs\n",nr_ioapics);
+
 	/*
 	 * Only use the first configuration found.
 	 */

Re: [PATCH][2.4] generic support for systems with more than 8 CPUs (2/2)

[William Lee Irwin III]

On Sat, Dec 21, 2002 at 10:59:10PM -0800, Pallipadi, Venkatesh wrote:
> 2/2 : switching to physical mode APIC setup in case of more than 8 CPU system
[...]
> -	printk("Enabling APIC mode: ");
> -	if(clustered_apic_mode == CLUSTERED_APIC_NUMAQ)
> -		printk("Clustered Logical.	");
> -	else if(clustered_apic_mode == CLUSTERED_APIC_XAPIC)
> -		printk("Physical.	");
> -	else
> -		printk("Flat.	");
> -	printk("Using %d I/O APICs\n",nr_ioapics);

IIRC NUMA-Q can be dynamically detected at boot by means of an MP OEM
table's presence, in particular if there's a matching string in the 8B
OEM record in the OEM table, with a value of "IBM NUMA" IIRC. This is
probably a line or two's worth of change to mpparse.c and declaring a
variable for clustered_apic_mode. If it were difficult to detect, I
wouldn't have suggested implementing it (though do so at your leisure). =)

I still think this is 2.5 material + backport once it gets testing there.


Bill

Re: [PATCH][2.4] generic support for systems with more than 8 CPUs (2/2)

[Martin J. Bligh]

> IIRC NUMA-Q can be dynamically detected at boot by means of an MP OEM
> table's presence, in particular if there's a matching string in the 8B
> OEM record in the OEM table, with a value of "IBM NUMA" IIRC. This is
> probably a line or two's worth of change to mpparse.c and declaring a
> variable for clustered_apic_mode. If it were difficult to detect, I
> wouldn't have suggested implementing it (though do so at your leisure). =)
>
> I still think this is 2.5 material + backport once it gets testing there.

I think there are still some things around that are switched on #defines
for NUMA-Q. Also older machines will probably say Sequent instead of IBM
in the OEM table. Would need some testing ...

M.

Re: [PATCH][2.4] generic support for systems with more than 8 CPUs (2/2)

[William Lee Irwin III]

At some point in the past, I wrote:
>> IIRC NUMA-Q can be dynamically detected at boot by means of an MP OEM
>> table's presence, in particular if there's a matching string in the 8B
>> OEM record in the OEM table, with a value of "IBM NUMA" IIRC. This is
>> probably a line or two's worth of change to mpparse.c and declaring a
>> variable for clustered_apic_mode. If it were difficult to detect, I
>> wouldn't have suggested implementing it (though do so at your leisure). =)
>> I still think this is 2.5 material + backport once it gets testing there.

On Sun, Dec 22, 2002 at 09:47:28AM -0800, Martin J. Bligh wrote:
> I think there are still some things around that are switched on #defines
> for NUMA-Q. Also older machines will probably say Sequent instead of IBM
> in the OEM table. Would need some testing ...

Switching on #defines is easy to clean up; actually getting at the
older systems sounds like a PITA... OTOH OEM ID's are sort of too
simple to screw up aside from independent #defines that should be
visible issues on all systems. I'm not deeply concerned about this,
it merely appeared to dovetail into his goal of dynamic detection
and configuration of the APIC destination model. As far as we're
concerned, compile-time vs. runtime setup is probably not important.

Bill

Re: [PATCH][2.4] generic support for systems with more than 8 CPUs (2/2)

[Martin J. Bligh]

> +#ifdef CONFIG_X86_CLUSTERED_APIC
> +	/*
> +	 * Switch to Physical destination mode in case of generic
> +	 * more than 8 CPU system, which has xAPIC support
> +	 */
> +#define FLAT_APIC_CPU_MAX	8
> +	if ((clustered_apic_mode == CLUSTERED_APIC_NONE) &&
> +	    (xapic_support) &&
> +	    (num_processors > FLAT_APIC_CPU_MAX)) {
> +		clustered_apic_mode = CLUSTERED_APIC_XAPIC;
> +		apic_broadcast_id = APIC_BROADCAST_ID_XAPIC;
> +		int_dest_addr_mode = APIC_DEST_PHYSICAL;
> +		int_delivery_mode = dest_Fixed;
> +		esr_disable = 1;
> +	}
> +#endif

If you could stick the #define up in a header file somewhere, would
be more standard / neater. Otherwise that looks good to me ...

Thanks for cleaning all this stuff up - the new set is much less
invasive, and easier to read than the last set ;-)

M.

RE: [PATCH][2.4] generic support for systems with more than 8 CPUs (2/2)

[Pallipadi, Venkatesh]

> From: William Lee Irwin III [mailto:wli@holomorphy.com]
> 
> On Sat, Dec 21, 2002 at 10:59:10PM -0800, Pallipadi, Venkatesh wrote:
> > 2/2 : switching to physical mode APIC setup in case of more 
> than 8 CPU system
> [...]
> > -	printk("Enabling APIC mode: ");
> > -	if(clustered_apic_mode == CLUSTERED_APIC_NUMAQ)
> > -		printk("Clustered Logical.	");
> > -	else if(clustered_apic_mode == CLUSTERED_APIC_XAPIC)
> > -		printk("Physical.	");
> > -	else
> > -		printk("Flat.	");
> > -	printk("Using %d I/O APICs\n",nr_ioapics);
> 
> IIRC NUMA-Q can be dynamically detected at boot by means of an MP OEM
> table's presence, in particular if there's a matching string in the 8B
> OEM record in the OEM table, with a value of "IBM NUMA" IIRC. This is
> probably a line or two's worth of change to mpparse.c and declaring a
> variable for clustered_apic_mode. If it were difficult to detect, I
> wouldn't have suggested implementing it (though do so at your 
> leisure). =)

Yes. There is already some code in base that does this. For NUMAQ specifically
this check happens and clustered mode is selected for APIC. My patch has this 
additinal check (after the initial IBM OEM check), for non-NUMAQ systems with 
more than 8 CPUs and xAPIC support. These systems can not work with the 
default flat addressing mode. So, this patch sets up such systems in physical 
mode.

Thanks,
-Venkatesh