[PATCH] x86_64: x86_64/i386 fix Intel cache detection code assumption about threads...

Fix the Intel cache detection code assumption that number of threads
sharing the cache will either be equal to number of HT or core siblings.

This also cleans up the code in general a bit.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

diff --git a/arch/i386/kernel/cpu/intel_cacheinfo.c b/arch/i386/kernel/cpu/intel_cacheinfo.c
index e66d14099564daa4aadf15cbc2a205e628fc4c7c..fbfd374aa336aff467612054031a19356b41b1a7 100644 (file)
--- a/arch/i386/kernel/cpu/intel_cacheinfo.c
+++ b/arch/i386/kernel/cpu/intel_cacheinfo.c
@@ -293,29 +293,45 @@ static struct _cpuid4_info *cpuid4_info[NR_CPUS];
 #ifdef CONFIG_SMP
 static void __cpuinit cache_shared_cpu_map_setup(unsigned int cpu, int index)
 {
-       struct _cpuid4_info     *this_leaf;
+       struct _cpuid4_info     *this_leaf, *sibling_leaf;
        unsigned long num_threads_sharing;
-#ifdef CONFIG_X86_HT
-       struct cpuinfo_x86 *c = cpu_data + cpu;
-#endif
+       int index_msb, i;
+       struct cpuinfo_x86 *c = cpu_data;
 
        this_leaf = CPUID4_INFO_IDX(cpu, index);
        num_threads_sharing = 1 + this_leaf->eax.split.num_threads_sharing;
 
        if (num_threads_sharing == 1)
                cpu_set(cpu, this_leaf->shared_cpu_map);
-#ifdef CONFIG_X86_HT
-       else if (num_threads_sharing == smp_num_siblings)
-               this_leaf->shared_cpu_map = cpu_sibling_map[cpu];
-       else if (num_threads_sharing == (c->x86_max_cores * smp_num_siblings))
-               this_leaf->shared_cpu_map = cpu_core_map[cpu];
-       else
-               printk(KERN_DEBUG "Number of CPUs sharing cache didn't match "
-                               "any known set of CPUs\n");
-#endif
+       else {
+               index_msb = get_count_order(num_threads_sharing);
+
+               for_each_online_cpu(i) {
+                       if (c[i].apicid >> index_msb ==
+                           c[cpu].apicid >> index_msb) {
+                               cpu_set(i, this_leaf->shared_cpu_map);
+                               if (i != cpu && cpuid4_info[i])  {
+                                       sibling_leaf = CPUID4_INFO_IDX(i, index);
+                                       cpu_set(cpu, sibling_leaf->shared_cpu_map);
+                               }
+                       }
+               }
+       }
+}
+static void __devinit cache_remove_shared_cpu_map(unsigned int cpu, int index)
+{
+       struct _cpuid4_info     *this_leaf, *sibling_leaf;
+       int sibling;
+
+       this_leaf = CPUID4_INFO_IDX(cpu, index);
+       for_each_cpu_mask(sibling, this_leaf->shared_cpu_map) {
+               sibling_leaf = CPUID4_INFO_IDX(sibling, index); 
+               cpu_clear(cpu, sibling_leaf->shared_cpu_map);
+       }
 }
 #else
 static void __init cache_shared_cpu_map_setup(unsigned int cpu, int index) {}
+static void __init cache_remove_shared_cpu_map(unsigned int cpu, int index) {}
 #endif
 
 static void free_cache_attributes(unsigned int cpu)
@@ -574,8 +590,10 @@ static void __cpuexit cache_remove_dev(struct sys_device * sys_dev)
        unsigned int cpu = sys_dev->id;
        unsigned long i;
 
-       for (i = 0; i < num_cache_leaves; i++)
+       for (i = 0; i < num_cache_leaves; i++) {
+               cache_remove_shared_cpu_map(cpu, i);
                kobject_unregister(&(INDEX_KOBJECT_PTR(cpu,i)->kobj));
+       }
        kobject_unregister(cache_kobject[cpu]);
        cpuid4_cache_sysfs_exit(cpu);
        return;