From: Mark Nelson <mark.nelson@inktank.com>
Date: Mon, 24 Jun 2013 03:18:03 +0000 (-0500)
Subject: Initial Intel SSE4 crc32c implementation.
X-Git-Tag: v0.67-rc1~158^2~1
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=7c59288d9168ddef3b3dc570464ae9a1f180d18c;p=ceph.git

Initial Intel SSE4 crc32c implementation.

Signed-off-by: Mark Nelson <mark.nelson@inktank.com>
---

diff --git a/src/common/crc32c-intel.c b/src/common/crc32c-intel.c
new file mode 100644
index 000000000000..c1fded99fddb
--- /dev/null
+++ b/src/common/crc32c-intel.c
@@ -0,0 +1,111 @@
+#include <inttypes.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <signal.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+/*
+ *  * Based on a posting to lkml by Austin Zhang <austin.zhang@intel.com>
+ *   * Further based on the fio crc32c-intel.c implementation by Jens Axboe.
+ *    *
+ *     * Using hardware provided CRC32 instruction to accelerate the CRC32 disposal.
+ *      * CRC32C polynomial:0x1EDC6F41(BE)/0x82F63B78(LE)
+ *       * CRC32 is a new instruction in Intel SSE4.2, the reference can be found at:
+ *        * http://www.intel.com/products/processor/manuals/
+ *         * Intel(R) 64 and IA-32 Architectures Software Developer's Manual
+ *          * Volume 2A: Instruction Set Reference, A-M
+ *           */
+
+int crc32c_intel_available = 0;
+
+/* TODO: Need some kind of ifdef here for arch... */
+
+#if BITS_PER_LONG == 64
+#define REX_PRE "0x48, "
+#define SCALE_F 8
+#else
+#define REX_PRE
+#define SCALE_F 4
+#endif
+
+static int crc32c_probed;
+
+static uint32_t crc32c_intel_le_hw_byte(uint32_t crc, unsigned char const *data,
+                                        unsigned length)
+{
+        while (length--) {
+                __asm__ __volatile__(
+                        ".byte 0xf2, 0xf, 0x38, 0xf0, 0xf1"
+                        :"=S"(crc)
+                        :"0"(crc), "c"(*data)
+                );
+                data++;
+        }
+
+        return crc;
+}
+
+/*
+ *  * Steps through buffer one byte at at time, calculates reflected
+ *   * crc using table.
+ *    */
+uint32_t crc32c_intel(uint32_t crc, unsigned char const *data, unsigned length)
+{
+        unsigned int iquotient = length / SCALE_F;
+        unsigned int iremainder = length % SCALE_F;
+#if BITS_PER_LONG == 64
+        uint64_t *ptmp = (uint64_t *) data;
+#else
+        uint32_t *ptmp = (uint32_t *) data;
+#endif
+
+        while (iquotient--) {
+                __asm__ __volatile__(
+                        ".byte 0xf2, " REX_PRE "0xf, 0x38, 0xf1, 0xf1;"
+                        :"=S"(crc)
+                        :"0"(crc), "c"(*ptmp)
+                );
+                ptmp++;
+        }
+
+        if (iremainder)
+                crc = crc32c_intel_le_hw_byte(crc, (unsigned char *)ptmp,
+                                 iremainder);
+
+        return crc;
+}
+
+static void do_cpuid(unsigned int *eax, unsigned int *ebx, unsigned int *ecx,
+                     unsigned int *edx)
+{
+        int id = *eax;
+
+        asm("movl %4, %%eax;"
+            "cpuid;"
+            "movl %%eax, %0;"
+            "movl %%ebx, %1;"
+            "movl %%ecx, %2;"
+            "movl %%edx, %3;"
+                : "=r" (*eax), "=r" (*ebx), "=r" (*ecx), "=r" (*edx)
+                : "r" (id)
+                : "eax", "ebx", "ecx", "edx");
+}
+
+void crc32c_intel_probe(void)
+{
+        if (!crc32c_probed) {
+                unsigned int eax, ebx, ecx, edx;
+
+                eax = 1;
+
+                do_cpuid(&eax, &ebx, &ecx, &edx);
+                crc32c_intel_available = (ecx & (1 << 20)) != 0;
+                crc32c_probed = 1;
+        }
+}
+
+uint32_t ceph_crc32c_le(uint32_t crc, unsigned char const *data, unsigned length)
+{
+        return crc32c_intel(crc, data, length);
+}