Expand OPENSS_ia32cap to 64 bits.

6 files changed, 44 insertions, 27 deletions

diff --git a/crypto/cryptlib.c b/crypto/cryptlib.c
index 44eb2bbc97..520e42d624 100644
--- a/crypto/cryptlib.c
+++ b/crypto/cryptlib.c
@@ -659,30 +659,40 @@ const char *CRYPTO_get_lock_name(int type)
 
 #if	defined(__i386)   || defined(__i386__)   || defined(_M_IX86) || \
 	defined(__INTEL__) || \
-	defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64)
+	defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || defined(_M_X64)
 
-unsigned long  OPENSSL_ia32cap_P=0;
-unsigned long *OPENSSL_ia32cap_loc(void) { return &OPENSSL_ia32cap_P; }
+unsigned int  OPENSSL_ia32cap_P[2];
+unsigned int *OPENSSL_ia32cap_loc(void) { return OPENSSL_ia32cap_P; }
 
 #if defined(OPENSSL_CPUID_OBJ) && !defined(OPENSSL_NO_ASM) && !defined(I386_ONLY)
 #define OPENSSL_CPUID_SETUP
+#if defined(_WIN32)
+typedef unsigned __int64 IA32CAP;
+#define strtoull _strtoui64
+#else
+typedef unsigned long long IA32CAP;
+#endif
 void OPENSSL_cpuid_setup(void)
 { static int trigger=0;
-  unsigned long OPENSSL_ia32_cpuid(void);
+  IA32CAP OPENSSL_ia32_cpuid(void);
+  IA32CAP vec;
   char *env;
 
     if (trigger)	return;
 
     trigger=1;
     if ((env=getenv("OPENSSL_ia32cap")))
-	OPENSSL_ia32cap_P = strtoul(env,NULL,0)|(1<<10);
+	vec = strtoull(env,NULL,0);
     else
-	OPENSSL_ia32cap_P = OPENSSL_ia32_cpuid()|(1<<10);
+	vec = OPENSSL_ia32_cpuid();
+
     /*
      * |(1<<10) sets a reserved bit to signal that variable
      * was initialized already... This is to avoid interference
      * with cpuid snippets in ELF .init segment.
      */
+    OPENSSL_ia32cap_P[0] = (unsigned int)vec|(1<<10);
+    OPENSSL_ia32cap_P[1] = (unsigned int)(vec>>32);
 }
 #endif
 
diff --git a/crypto/cryptlib.h b/crypto/cryptlib.h
index 5ceaa964b5..4761a894a8 100644
--- a/crypto/cryptlib.h
+++ b/crypto/cryptlib.h
@@ -99,7 +99,7 @@ extern "C" {
 #define HEX_SIZE(type)		(sizeof(type)*2)
 
 void OPENSSL_cpuid_setup(void);
-extern unsigned long OPENSSL_ia32cap_P;
+extern unsigned int OPENSSL_ia32cap_P[];
 void OPENSSL_showfatal(const char *,...);
 void *OPENSSL_stderr(void);
 extern int OPENSSL_NONPIC_relocated;
diff --git a/crypto/perlasm/x86gas.pl b/crypto/perlasm/x86gas.pl
index 6eab727fd4..b470507730 100644
--- a/crypto/perlasm/x86gas.pl
+++ b/crypto/perlasm/x86gas.pl
@@ -150,7 +150,7 @@ sub ::public_label
 
 sub ::file_end
 {   if (grep {/\b${nmdecor}OPENSSL_ia32cap_P\b/i} @out) {
-	my $tmp=".comm\t${nmdecor}OPENSSL_ia32cap_P,4";
+	my $tmp=".comm\t${nmdecor}OPENSSL_ia32cap_P,8";
 	if ($::elf)	{ push (@out,"$tmp,4\n"); }
 	else		{ push (@out,"$tmp\n"); }
     }
diff --git a/crypto/perlasm/x86masm.pl b/crypto/perlasm/x86masm.pl
index 4eca7bc367..3365114cd0 100644
--- a/crypto/perlasm/x86masm.pl
+++ b/crypto/perlasm/x86masm.pl
@@ -127,7 +127,7 @@ ___
     if (grep {/\b${nmdecor}OPENSSL_ia32cap_P\b/i} @out)
     {	my $comm=<<___;
 .bss	SEGMENT
-COMM	${nmdecor}OPENSSL_ia32cap_P:DWORD
+COMM	${nmdecor}OPENSSL_ia32cap_P:QWORD
 .bss	ENDS
 ___
 	# comment out OPENSSL_ia32cap_P declarations
diff --git a/crypto/perlasm/x86nasm.pl b/crypto/perlasm/x86nasm.pl
index ce2bed9bb2..1a384582bf 100644
--- a/crypto/perlasm/x86nasm.pl
+++ b/crypto/perlasm/x86nasm.pl
@@ -114,7 +114,7 @@ sub ::file_end
 {   if (grep {/\b${nmdecor}OPENSSL_ia32cap_P\b/i} @out)
     {	my $comm=<<___;
 ${drdecor}segment	.bss
-${drdecor}common	${nmdecor}OPENSSL_ia32cap_P 4
+${drdecor}common	${nmdecor}OPENSSL_ia32cap_P 8
 ___
 	# comment out OPENSSL_ia32cap_P declarations
 	grep {s/(^extern\s+${nmdecor}OPENSSL_ia32cap_P)/\;$1/} @out;
diff --git a/doc/crypto/OPENSSL_ia32cap.pod b/doc/crypto/OPENSSL_ia32cap.pod
index 2e659d34a5..b7d8a7618f 100644
--- a/doc/crypto/OPENSSL_ia32cap.pod
+++ b/doc/crypto/OPENSSL_ia32cap.pod
@@ -6,28 +6,29 @@ OPENSSL_ia32cap - finding the IA-32 processor capabilities
 
 =head1 SYNOPSIS
 
- unsigned long *OPENSSL_ia32cap_loc(void);
- #define OPENSSL_ia32cap (*(OPENSSL_ia32cap_loc()))
+ unsigned int *OPENSSL_ia32cap_loc(void);
+ #define OPENSSL_ia32cap ((OPENSSL_ia32cap_loc())[0])
 
 =head1 DESCRIPTION
 
 Value returned by OPENSSL_ia32cap_loc() is address of a variable
-containing IA-32 processor capabilities bit vector as it appears in EDX
-register after executing CPUID instruction with EAX=1 input value (see
-Intel Application Note #241618). Naturally it's meaningful on IA-32[E]
-platforms only. The variable is normally set up automatically upon
-toolkit initialization, but can be manipulated afterwards to modify
-crypto library behaviour. For the moment of this writing six bits are
-significant, namely:
-
-1. bit #28 denoting Hyperthreading, which is used to distiguish
+containing IA-32 processor capabilities bit vector as it appears in
+EDX:ECX register pair after executing CPUID instruction with EAX=1
+input value (see Intel Application Note #241618). Naturally it's
+meaningful on x86 and x86_64 platforms only. The variable is normally
+set up automatically upon toolkit initialization, but can be
+manipulated afterwards to modify crypto library behaviour. For the
+moment of this writing seven bits are significant, namely:
+
+1. bit #4 denoting presence of Time-Stamp Counter.
+2. bit #20, reserved by Intel, is used to choose between RC4 code
+   paths;
+3. bit #23 denoting MMX support;
+4. bit #25 denoting SSE support;
+5. bit #26 denoting SSE2 support;
+6. bit #28 denoting Hyperthreading, which is used to distiguish
    cores with shared cache;
-2. bit #26 denoting SSE2 support;
-3. bit #25 denoting SSE support;
-4. bit #23 denoting MMX support;
-5. bit #20, reserved by Intel, is used to choose between RC4 code
-   pathes;
-6. bit #4 denoting presence of Time-Stamp Counter.
+7. bit #57 denoting Intel AES instruction set extension;
 
 For example, clearing bit #26 at run-time disables high-performance
 SSE2 code present in the crypto library. You might have to do this if
@@ -40,4 +41,10 @@ OPENSSL_ia32cap=0x12900010 apps/openssl', to achieve same effect
 without modifying the application source code. Alternatively you can
 reconfigure the toolkit with no-sse2 option and recompile.
 
+Less intuituve is clearing bit #28. The truth is that it's not copied
+from CPUID output verbatim, but is adjusted to reflect whether or not
+the data cache is actually shared between logical cores. This in turn
+affects the decision on whether or not expensive countermeasures
+against cache-timing attacks are applied, most notably in AES assembler
+module.
 =cut