fix isal crc64 bug

deps/oblib/src/lib/checksum/ob_crc64.cpp

@@ -13,6 +13,8 @@
 #include <stdlib.h>
 #include "lib/checksum/ob_crc64.h"
 #include "lib/ob_define.h"
+#include "isa-l/crc64.h"
+#include "isa-l/crc.h"
 
 namespace oceanbase
 {
@@ -420,7 +422,7 @@ for RHEL4 support (GCC 3 doesn't support this instruction) */
 #define crc32_sse42_byte crc = __crc32cb(crc, (uint8_t)*buf); len--, buf++
 #endif /* defined(__GNUC__) && defined(__x86_64__) */
 
-inline static uint64_t crc64_sse42(uint64_t uCRC64,
+uint64_t crc64_sse42(uint64_t uCRC64,
                                    const char *buf, int64_t len)
 {
   uint64_t crc = uCRC64;
@@ -1102,6 +1104,14 @@ uint64_t fast_crc64_sse42_manually(uint64_t crc, const char *buf, int64_t len)
   return crc;
 }
 
+//If the CPU is intel, ISA-L library for CRC can be used
+uint64_t ob_crc64_isal(uint64_t uCRC64, const char* buf, int64_t cb)
+{
+  if (buf == NULL || cb <= 0){
+    return uCRC64;
+  }
+  return crc32_iscsi((unsigned char*)(buf), cb, uCRC64);
+}
 
 uint64_t crc64_sse42_dispatch(uint64_t crc, const char *buf, int64_t len)
 {
@@ -1110,13 +1120,27 @@ uint64_t crc64_sse42_dispatch(uint64_t crc, const char *buf, int64_t len)
   uint32_t b = 0;
   uint32_t c = 0;
   uint32_t d = 0;
+
+  uint32_t vendor_info[4];
+  __asm__("mov $0x0, %eax\n\t");
+  __asm__("cpuid\n\t");
+  __asm__("mov %%ebx, %0\n\t":"=r" (vendor_info[0]));
+  __asm__("mov %%edx, %0\n\t":"=r" (vendor_info[1]));
+  __asm__("mov %%ecx, %0\n\t":"=r" (vendor_info[2]));
+  vendor_info[3]='\0';
+
+  if (strcmp((char*)vendor_info, "GenuineIntel") == 0) {
+    ob_crc64_sse42_func = &ob_crc64_isal;
+    _OB_LOG(WARN, "Use ISAL for crc64 calculate");
+  } else{
     asm("cpuid" : "=a"(a), "=b"(b), "=c"(c), "=d"(d) : "0"(1));
     if ((c & (1 << 20)) != 0) {
       ob_crc64_sse42_func = &crc64_sse42;
-    _OB_LOG(INFO, "Use CPU crc32 instructs for crc64 calculate");
+      _OB_LOG(WARN, "Use CPU crc32 instructs for crc64 calculate");
     } else {
       ob_crc64_sse42_func = &fast_crc64_sse42_manually;
-    _OB_LOG(INFO, "Use manual crc32 table lookup for crc64 calculate");
+      _OB_LOG(WARN, "Use manual crc32 table lookup for crc64 calculate");
+    }
   }
   #elif defined(__aarch64__)
     #if 1

deps/oblib/src/lib/checksum/ob_crc64.h

@@ -68,6 +68,8 @@ inline uint64_t ob_crc64_sse42(const void *pv, int64_t cb)
   return (*ob_crc64_sse42_func)(0, static_cast<const char *>(pv), cb);
 }
 
+uint64_t ob_crc64_isal(uint64_t uCRC64, const char* buf, int64_t cb);
+uint64_t crc64_sse42(uint64_t uCRC64, const char* buf, int64_t len);
 uint64_t crc64_sse42_manually(uint64_t crc, const char *buf, int64_t len);
 uint64_t fast_crc64_sse42_manually(uint64_t crc, const char *buf, int64_t len);
 

deps/oblib/unittest/lib/checksum/test_crc64.cpp

@@ -41,13 +41,16 @@ TEST(TestCrc64, common)
   const int64_t STR_LEN = 10240;
   char *tmp_str = new char[STR_LEN];
   char *str = NULL;
+  uint64_t intermediate_hash = 0;
   for (int64_t i = 1; i < (STR_LEN - 1); ++i) {
     str = rand_str(tmp_str, i);
-    uint64_t manually_hash = crc64_sse42_manually(0, str, i);
+    uint64_t manually_hash = crc64_sse42_manually(intermediate_hash, str, i);
-    uint64_t fast_manually_hash = fast_crc64_sse42_manually(0, str, i);
+    uint64_t fast_manually_hash = fast_crc64_sse42_manually(intermediate_hash, str, i);
-    uint64_t sse42_hash = ob_crc64(0, str, i);
+    uint64_t sse42_hash = crc64_sse42(intermediate_hash, str, i);
+    uint64_t isal_hash = ob_crc64_isal(intermediate_hash, str, i);
     ASSERT_EQ(manually_hash, fast_manually_hash);
     ASSERT_EQ(manually_hash, sse42_hash);
+    ASSERT_EQ(manually_hash, isal_hash);
     //cout << "st = "<< tmp_str << endl;
     //cout << "crc64c = "<< manually_hash << endl;
   }
@@ -77,11 +80,19 @@ TEST(TestCrc64, test_speed)
 
     start = get_current_time_us();
     for (int64_t j = 0; j < COUNT; ++j) {
-      ob_crc64(0, tmp_str, i);
+      crc64_sse42(0, tmp_str, i);
     }
     end = get_current_time_us();
     cout << "          ob_crc64(sse42), execut_count = "<< COUNT << ", cost_us = " << end - start << " len = " << i << endl;
 
+    start = get_current_time_us();
+    for (int64_t j = 0; j < COUNT; ++j) {
+      ob_crc64_isal(0, tmp_str, i);
+    }
+    end = get_current_time_us();
+    cout << "          ob_crc64(ob_crc64_isal), execut_count = " << COUNT << ", cost_us = " << end - start << " len = " << i
+         << endl;    
+
     cout << endl;
     cout << endl;
   }