VPP代码阅读中文注解---cpu.h

#include <vppinfra/format.h>

/*
 * multiarchitecture support. Adding new entry will produce
 * new graph node function variant optimized for specific cpu
 * microarchitecture.
 * Order is important for runtime selection, as 1st match wins...
 */

本文，主要用于多cpu架构的支持。顺序是比较重要的，第一个匹配的CPU架构将被选中。

 

#if __x86_64__ && CLIB_DEBUG == 0
#define foreach_march_variant(macro, x) \
  macro(avx2,  x, "arch=core-avx2")
#else
#define foreach_march_variant(macro, x)
#endif


#if __GNUC__ > 4  && !__clang__
#define CLIB_CPU_OPTIMIZED __attribute__ ((optimize ("O3")))
#else
#define CLIB_CPU_OPTIMIZED
#endif


#define CLIB_MULTIARCH_ARCH_CHECK(arch, fn, tgt)			\
  if (clib_cpu_supports_ ## arch())					\
    return & fn ## _ ##arch;

#define CLIB_MULTIARCH_SELECT_FN(fn,...)                               \
  __VA_ARGS__ void * fn ## _multiarch_select(void)                     \
{                                                                      \
  foreach_march_variant(CLIB_MULTIARCH_ARCH_CHECK, fn)                 \
  return & fn;                                                         \
}

根据CPU是否支持avx2返回普通版本的函数，或者是返回带avx2后缀的函数。

 

#ifdef CLIB_MARCH_VARIANT
#define __CLIB_MULTIARCH_FN(a,b) a##_##b
#define _CLIB_MULTIARCH_FN(a,b) __CLIB_MULTIARCH_FN(a,b)
#define CLIB_MULTIARCH_FN(fn) _CLIB_MULTIARCH_FN(fn,CLIB_MARCH_VARIANT)
#else
#define CLIB_MULTIARCH_FN(fn) fn
#endif

#define CLIB_MARCH_SFX CLIB_MULTIARCH_FN

支持多CPU架构和不支持多cpu架构的函数

#define foreach_x86_64_flags \
_ (sse3,     1, ecx, 0)   \
_ (ssse3,    1, ecx, 9)   \
_ (sse41,    1, ecx, 19)  \
_ (sse42,    1, ecx, 20)  \
_ (avx,      1, ecx, 28)  \
_ (avx2,     7, ebx, 5)   \
_ (avx512f,  7, ebx, 16)  \
_ (x86_aes,  1, ecx, 25)  \
_ (sha,      7, ebx, 29)  \
_ (invariant_tsc, 0x80000007, edx, 8)

#include "cpuid.h"

static inline int
clib_get_cpuid (const u32 lev, u32 * eax, u32 * ebx, u32 * ecx, u32 * edx)
{
  if ((u32) __get_cpuid_max (0x80000000 & lev, 0) < lev)
    return 0;
  if (lev == 7)
    __cpuid_count (lev, 0, *eax, *ebx, *ecx, *edx);
  else
    __cpuid (lev, *eax, *ebx, *ecx, *edx);
  return 1;
}


#define _(flag, func, reg, bit) \
static inline int							\
clib_cpu_supports_ ## flag()						\
{									\
  u32 __attribute__((unused)) eax, ebx = 0, ecx = 0, edx  = 0;		\
  clib_get_cpuid (func, &eax, &ebx, &ecx, &edx);			\
									\
  return ((reg & (1 << bit)) != 0);					\
}
foreach_x86_64_flags
#undef _

X86_64 CPU下面的子架构支持情况。

 

#define foreach_aarch64_flags \
_ (fp,          0) \
_ (asimd,       1) \
_ (evtstrm,     2) \
_ (aarch64_aes, 3) \
_ (pmull,       4) \
_ (sha1,        5) \
_ (sha2,        6) \
_ (crc32,       7) \
_ (atomics,     8) \
_ (fphp,        9) \
_ (asimdhp,    10) \
_ (cpuid,      11) \
_ (asimdrdm,   12) \
_ (jscvt,      13) \
_ (fcma,       14) \
_ (lrcpc,      15) \
_ (dcpop,      16) \
_ (sha3,       17) \
_ (sm3,        18) \
_ (sm4,        19) \
_ (asimddp,    20) \
_ (sha512,     21) \
_ (sve,        22)

#include <sys/auxv.h>
#define _(flag, bit) \
static inline int							\
clib_cpu_supports_ ## flag()						\
{									\
  unsigned long hwcap = getauxval(AT_HWCAP);				\
  return (hwcap & (1 << bit));						\
}
  foreach_aarch64_flags
#undef _

aarch64 cpu子架构支持情况

/*
 * aes is the only feature with the same name in both flag lists
 * handle this by prefixing it with the arch name, and handling it
 * with the custom function below
 */
  static inline int
clib_cpu_supports_aes ()
{
#if defined (__aarch64__)
  return clib_cpu_supports_x86_aes ();
#elif defined (__aarch64__)
  return clib_cpu_supports_aarch64_aes ();
#else
  return 0;
#endif
}

当前CPU是否支持AES

static inline int
clib_cpu_march_priority_avx512 ()
{
  if (clib_cpu_supports_avx512f ())
    return 20;
  return -1;
}

static inline int
clib_cpu_march_priority_avx2 ()
{
  if (clib_cpu_supports_avx2 ())
    return 10;
  return -1;
}

是否支持AVX2， 或者是否支持AVX512。

static inline u32
clib_cpu_implementer ()
{
  char buf[128];
  static u32 implementer = -1;

  if (-1 != implementer)
    return implementer;

  FILE *fp = fopen ("/proc/cpuinfo", "r");
  if (!fp)
    return implementer;

  while (!feof (fp))
    {
      if (!fgets (buf, sizeof (buf), fp))
	break;
      buf[127] = '\0';
      if (strstr (buf, "CPU implementer"))
	implementer = (u32) strtol (memchr (buf, ':', 128) + 2, NULL, 0);
      if (-1 != implementer)
	break;
    }
  fclose (fp);

  return implementer;
}

从cpuinfo中读取cpu 实现者编号，首次读取后将结果缓存起来，后续读取时，直接取缓存结果。

 

static inline u32
clib_cpu_part ()
{
  char buf[128];
  static u32 part = -1;

  if (-1 != part)
    return part;

  FILE *fp = fopen ("/proc/cpuinfo", "r");
  if (!fp)
    return part;

  while (!feof (fp))
    {
      if (!fgets (buf, sizeof (buf), fp))
	break;
      buf[127] = '\0';
      if (strstr (buf, "CPU part"))
	part = (u32) strtol (memchr (buf, ':', 128) + 2, NULL, 0);
      if (-1 != part)
	break;
    }
  fclose (fp);

  return part;
}

从CPU INFO中读取cpu part编号。首次读取后缓存结果，后续读取直接使用缓存的值。

 

#define AARCH64_CPU_IMPLEMENTER_THUNERDERX2 0x43
#define AARCH64_CPU_PART_THUNERDERX2        0x0af
#define AARCH64_CPU_IMPLEMENTER_QDF24XX     0x51
#define AARCH64_CPU_PART_QDF24XX            0xc00
#define AARCH64_CPU_IMPLEMENTER_CORTEXA72   0x41
#define AARCH64_CPU_PART_CORTEXA72          0xd08

static inline int
clib_cpu_march_priority_thunderx2t99 ()
{
  if ((AARCH64_CPU_IMPLEMENTER_THUNERDERX2 == clib_cpu_implementer ()) &&
      (AARCH64_CPU_PART_THUNERDERX2 == clib_cpu_part ()))
    return 20;
  return -1;
}

static inline int
clib_cpu_march_priority_qdf24xx ()
{
  if ((AARCH64_CPU_IMPLEMENTER_QDF24XX == clib_cpu_implementer ()) &&
      (AARCH64_CPU_PART_QDF24XX == clib_cpu_part ()))
    return 20;
  return -1;
}

static inline int
clib_cpu_march_priority_cortexa72 ()
{
  if ((AARCH64_CPU_IMPLEMENTER_CORTEXA72 == clib_cpu_implementer ()) &&
      (AARCH64_CPU_PART_CORTEXA72 == clib_cpu_part ()))
    return 10;
  return -1;
}

3种实现者和对应的PART。给他们制定一个优先级。

#ifdef CLIB_MARCH_VARIANT
#define CLIB_MARCH_FN_PRIORITY() CLIB_MARCH_SFX(clib_cpu_march_priority)()
#else
#define CLIB_MARCH_FN_PRIORITY() 0
#endif

优先级

 

#define CLIB_MARCH_FN_CONSTRUCTOR(fn)					\
static void __clib_constructor 						\
CLIB_MARCH_SFX(fn ## _march_constructor) (void)				\
{									\
  if (CLIB_MARCH_FN_PRIORITY() > fn ## _selected_priority)		\
    {									\
      fn ## _selected = & CLIB_MARCH_SFX (fn ## _ma);			\
      fn ## _selected_priority = CLIB_MARCH_FN_PRIORITY();		\
    }									\
}									\

#ifndef CLIB_MARCH_VARIANT
#define CLIB_MARCH_FN(fn, rtype, _args...)				\
  static rtype CLIB_CPU_OPTIMIZED CLIB_MARCH_SFX (fn ## _ma)(_args);	\
  rtype (*fn ## _selected) (_args) = & CLIB_MARCH_SFX (fn ## _ma);	\
  int fn ## _selected_priority = 0;					\
  static inline rtype CLIB_CPU_OPTIMIZED 				\
  CLIB_MARCH_SFX (fn ## _ma)(_args)
#else
#define CLIB_MARCH_FN(fn, rtype, _args...)				\
  static rtype CLIB_CPU_OPTIMIZED CLIB_MARCH_SFX (fn ## _ma)(_args);	\
  extern int (*fn ## _selected) (_args);				\
  extern int fn ## _selected_priority;					\
  CLIB_MARCH_FN_CONSTRUCTOR (fn)					\
  static rtype CLIB_CPU_OPTIMIZED CLIB_MARCH_SFX (fn ## _ma)(_args)
#endif

如果当前CPU架构的优先级高于已经选择的CPU架构，则重新选择当前的CPU架构。