about summary refs log tree commit diff
diff options
context:
space:
mode:
authorH.J. Lu <hjl.tools@gmail.com>2016-05-13 07:18:25 -0700
committerH.J. Lu <hjl.tools@gmail.com>2016-05-13 07:18:35 -0700
commit9e4ec3e81614164314bffe648e9446f17adc5fb7 (patch)
treee6cf84d13c33651d11aac638e7b46afe5b66bec0
parent4cf6c72fd2a482e7499c29162349810029632c3f (diff)
downloadglibc-9e4ec3e81614164314bffe648e9446f17adc5fb7.tar.gz
glibc-9e4ec3e81614164314bffe648e9446f17adc5fb7.tar.xz
glibc-9e4ec3e81614164314bffe648e9446f17adc5fb7.zip
Support non-inclusive caches on Intel processors
	* sysdeps/x86/cacheinfo.c (init_cacheinfo): Check and support
	non-inclusive caches on Intel processors.
-rw-r--r--ChangeLog5
-rw-r--r--sysdeps/x86/cacheinfo.c12
2 files changed, 16 insertions, 1 deletions
diff --git a/ChangeLog b/ChangeLog
index 1f8db94e4a..05d0fe7b5e 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,8 @@
+2016-05-13  H.J. Lu  <hongjiu.lu@intel.com>
+
+	* sysdeps/x86/cacheinfo.c (init_cacheinfo): Check and support
+	non-inclusive caches on Intel processors.
+
 2016-05-13  Florian Weimer  <fweimer@redhat.com>
 
 	* malloc/malloc.c (dumped_main_arena_start)
diff --git a/sysdeps/x86/cacheinfo.c b/sysdeps/x86/cacheinfo.c
index 143b3333a8..8408624ea4 100644
--- a/sysdeps/x86/cacheinfo.c
+++ b/sysdeps/x86/cacheinfo.c
@@ -492,6 +492,9 @@ init_cacheinfo (void)
     {
       data = handle_intel (_SC_LEVEL1_DCACHE_SIZE, max_cpuid);
 
+      long int core = handle_intel (_SC_LEVEL2_CACHE_SIZE, max_cpuid);
+      bool inclusive_cache = true;
+
       /* Try L3 first.  */
       level  = 3;
       shared = handle_intel (_SC_LEVEL3_CACHE_SIZE, max_cpuid);
@@ -500,7 +503,7 @@ init_cacheinfo (void)
 	{
 	  /* Try L2 otherwise.  */
 	  level  = 2;
-	  shared = handle_intel (_SC_LEVEL2_CACHE_SIZE, max_cpuid);
+	  shared = core;
 	}
 
       /* Figure out the number of logical threads that share the
@@ -526,6 +529,9 @@ init_cacheinfo (void)
 	    }
 	  while (((eax >> 5) & 0x7) != level);
 
+	  /* Check if cache is inclusive of lower cache levels.  */
+	  inclusive_cache = (edx & 0x2) != 0;
+
 	  threads = (eax >> 14) & 0x3ff;
 
 	  /* If max_cpuid >= 11, THREADS is the maximum number of
@@ -592,6 +598,10 @@ init_cacheinfo (void)
 	 threads.  */
       if (shared > 0 && threads > 0)
 	shared /= threads;
+
+      /* Account for non-inclusive L2 and L3 caches.  */
+      if (level == 3 && !inclusive_cache)
+	shared += core;
     }
   /* This spells out "AuthenticAMD".  */
   else if (is_amd)