Skip to contentSkip to conversation

fix generic non-SMP builds

PATCH In review dann frazier 3 participants 3 messages Mar 14, 2005 – Jun 8, 2005

lore
Share & tools
lore

dann frazier

author
#0 lore raw

PATCH 1/1 fix generic non-SMP builds

Here is a forward port of Jesse Barnes' fixes for generic, non-SMP builds to today's linux-ia64-test-2.6.12 tree. I built & booted both SMP & non-SMP versions on a zx1 box.

Signed-off-by dann frazier <[email protected]>
arch/ia64/kernel/Makefile +1 -0
diff -urN linux-ia64-test-2.6.12-20050313.orig/arch/ia64/kernel/Makefile linux-ia64-test-2.6.12-20050313/arch/ia64/kernel/Makefile
@@ -17,6 +17,7 @@
 obj-$(CONFIG_IOSAPIC)		+= iosapic.o
 obj-$(CONFIG_MODULES)		+= module.o
 obj-$(CONFIG_SMP)		+= smp.o smpboot.o domain.o
+obj-$(CONFIG_NUMA)		+= numa.o
 obj-$(CONFIG_PERFMON)		+= perfmon_default_smpl.o
 obj-$(CONFIG_IA64_CYCLONE)	+= cyclone.o
 obj-$(CONFIG_IA64_MCA_RECOVERY)	+= mca_recovery.o
arch/ia64/kernel/acpi.c +3 -1
diff -urN linux-ia64-test-2.6.12-20050313.orig/arch/ia64/kernel/acpi.c linux-ia64-test-2.6.12-20050313/arch/ia64/kernel/acpi.c
@@ -642,9 +642,11 @@
 			if (smp_boot_data.cpu_phys_id[cpu] != hard_smp_processor_id())
 				node_cpuid[i++].phys_id = smp_boot_data.cpu_phys_id[cpu];
 	}
-	build_cpu_to_node_map();
 # endif
 #endif
+#ifdef CONFIG_ACPI_NUMA
+	build_cpu_to_node_map();
+#endif
 	/* Make boot-up look pretty */
 	printk(KERN_INFO "%d CPUs available, %d CPUs total\n", available_cpus, total_cpus);
 	return 0;
arch/ia64/kernel/numa.c +57 -0

57 lines not on this page · raw message

arch/ia64/kernel/smpboot.c +0 -41

47 lines not on this page · raw message

arch/ia64/mm/discontig.c +48 -22

121 lines not on this page · raw message

include/asm-ia64/sn/sn_cpuid.h +0 -5
diff -urN linux-ia64-test-2.6.12-20050313.orig/include/asm-ia64/sn/sn_cpuid.h linux-ia64-test-2.6.12-20050313/include/asm-ia64/sn/sn_cpuid.h
@@ -81,11 +81,6 @@
  *
  */
 
-#ifndef CONFIG_SMP
-#define cpu_physical_id(cpuid)			((ia64_getreg(_IA64_REG_CR_LID) >> 16) & 0xffff)
-#endif
-
-
 #define get_node_number(addr)			NASID_GET(addr)
 
 /*
Signature · 1 line
dann frazier <[email protected]>
#1 lore raw

arch/ia64/kernel/numa.c line 31

Quoted 4 lines · dann frazier
arch/ia64/kernel/numa.c +4 -0
+u8 cpu_to_node_map[NR_CPUS] __cacheline_aligned;
+EXPORT_SYMBOL(cpu_to_node_map);
+
+cpumask_t node_to_cpu_mask[MAX_NUMNODES] __cacheline_aligned;

I'm trying to figure out why these are caheline aligned... they're ready-only for after boot (hotplug aside), right? Do we see that a lot of cpu/node local references to them occur? OTOH, the wastage isn't that big either way...

arch/ia64/kernel/numa.c line 50

Quoted 5 lines
arch/ia64/kernel/numa.c +5 -0
+	for(cpu = 0; cpu < NR_CPUS; ++cpu) {
+		node = -1;
+		for (i = 0; i < NR_CPUS; ++i)
+			if (cpu_physical_id(cpu) = node_cpuid[i].phys_id) {
+				node = node_cpuid[i].nid;

Is there a for_each_node(n) now?

arch/ia64/mm/discontig.c line 561

Quoted 2 lines
arch/ia64/mm/discontig.c +2 -0
+	for (cpu = 0; cpu < NR_CPUS; cpu++)
+		per_cpu(local_per_cpu_offset, cpu) = __per_cpu_offset[cpu];

for_each_cpu here I think.

Not sure when Tony had this queued up, so we may have to rediff in a week or two, otherwise it looks good.

Jess

#2 lore raw
In reply to dann frazier

On Sun, Mar 13, 2005 at 06:57:47PM -0700, dann frazier wrote:

| Here is a forward port of Jesse Barnes' fixes for generic, non-SMP
| builds to today's linux-ia64-test-2.6.12 tree.  I built & booted both
| SMP & non-SMP versions on a zx1 box.
| 
| Signed-off-by: dann frazier <[email protected]>

Dann, care to resend this? I see nonsmp, generic builds are still busted.

Greg