Provide crypto_cpu_caps_init() for amd64.

This is a CPU capability detection implementation in C, with minimal
inline assembly (for cpuid and xgetbv). This replaces the assembly
mess generated by x86_64cpuid.pl. Rather than populating OPENSSL_ia32cap_P
directly with CPUID output, just set the bits that the remaining
perlasm checks (namely AESNI, AVX, FXSR, INTEL, HT, MMX, PCLMUL, SSE, SSE2
and SSSE3).

ok joshua@ tb@

1 files changed, 6 insertions, 1 deletions

diff --git a/src/lib/libcrypto/x86_arch.h b/src/lib/libcrypto/x86_arch.h
index 5b2cf97546..e9e9d48960 100644
--- a/src/lib/libcrypto/x86_arch.h
+++ b/src/lib/libcrypto/x86_arch.h
@@ -1,4 +1,4 @@
-/*      $OpenBSD: x86_arch.h,v 1.1 2016/11/04 17:30:30 miod Exp $       */
+/*      $OpenBSD: x86_arch.h,v 1.2 2024/10/18 13:36:24 jsing Exp $      */
 /*
  * Copyright (c) 2016 Miodrag Vallat.
  *
@@ -76,15 +76,20 @@
 #define IA32CAP_MASK1_SSSE3     (1 << IA32CAP_BIT1_SSSE3)
 #define IA32CAP_MASK1_FMA3      (1 << IA32CAP_BIT1_FMA3)
 #define IA32CAP_MASK1_AESNI     (1 << IA32CAP_BIT1_AESNI)
+#define IA32CAP_MASK1_OSXSAVE   (1 << IA32CAP_BIT1_OSXSAVE)
 #define IA32CAP_MASK1_AVX       (1 << IA32CAP_BIT1_AVX)
 
 #define IA32CAP_MASK1_AMD_XOP   (1 << IA32CAP_BIT1_AMD_XOP)
 
 /* bit masks for OPENSSL_cpu_caps() */
+#define CPUCAP_MASK_HT          IA32CAP_MASK0_HT
 #define CPUCAP_MASK_MMX         IA32CAP_MASK0_MMX
 #define CPUCAP_MASK_FXSR        IA32CAP_MASK0_FXSR
 #define CPUCAP_MASK_SSE         IA32CAP_MASK0_SSE
+#define CPUCAP_MASK_SSE2        IA32CAP_MASK0_SSE2
+#define CPUCAP_MASK_INTEL       IA32CAP_MASK0_INTEL
 #define CPUCAP_MASK_INTELP4     IA32CAP_MASK0_INTELP4
 #define CPUCAP_MASK_PCLMUL      (1ULL << (32 + IA32CAP_BIT1_PCLMUL))
 #define CPUCAP_MASK_SSSE3       (1ULL << (32 + IA32CAP_BIT1_SSSE3))
 #define CPUCAP_MASK_AESNI       (1ULL << (32 + IA32CAP_BIT1_AESNI))
+#define CPUCAP_MASK_AVX         (1ULL << (32 + IA32CAP_BIT1_AVX))