x86*cpuid update.

diff --git a/crypto/x86_64cpuid.pl b/crypto/x86_64cpuid.pl
index f90bbc9ab0089777cb95bdafc5073b7a215c6ead..e29792a1573d7d131aca055d1ff04d855c07c63a 100644 (file)
--- a/crypto/x86_64cpuid.pl
+++ b/crypto/x86_64cpuid.pl
@@ -134,8 +134,6 @@ OPENSSL_ia32_cpuid:
 
        mov     \$1,%eax
        cpuid
-       bt      \$28,%edx               # test hyper-threading bit
-       jnc     .Ldone
        cmp     \$0,%r9d
        jne     .Lnotintel
        or      \$1<<20,%edx            # use reserved bit to engage RC4_CHAR
@@ -144,6 +142,8 @@ OPENSSL_ia32_cpuid:
        je      .Lnotintel
        or      \$1<<30,%edx            # use reserved bit to skip unrolled loop
 .Lnotintel:
+       bt      \$28,%edx               # test hyper-threading bit
+       jnc     .Ldone
        shr     \$16,%ebx
        cmp     \$1,%bl                 # see if cache is shared
        ja      .Ldone

diff --git a/crypto/x86cpuid.pl b/crypto/x86cpuid.pl
index ce9b8ea4dd3568952cc4b7bb0e29779fb91546d7..0fb0657fa7ec0020d395e4451833d521d452f59d 100644 (file)
--- a/crypto/x86cpuid.pl
+++ b/crypto/x86cpuid.pl
@@ -19,7 +19,7 @@ for (@ARGV) { $sse2=1 if (/-DOPENSSL_IA32_SSE2/); }
        &pop    ("eax");
        &xor    ("ecx","eax");
        &bt     ("ecx",21);
-       &jnc    (&label("nocpuid"));
+       &jnc    (&label("done"));
        &xor    ("eax","eax");
        &cpuid  ();
        &xor    ("eax","eax");
@@ -34,17 +34,20 @@ for (@ARGV) { $sse2=1 if (/-DOPENSSL_IA32_SSE2/); }
        &or     ("ebp","eax");
        &mov    ("eax",1);
        &cpuid  ();
-       &bt     ("edx",28);             # test hyper-threading bit
-       &jnc    (&label("nocpuid"));
        &cmp    ("ebp",0);
-       &jne    (&label("notintel"));
+       &jne    (&label("notP4"));
+       &and    (&HB("eax"),15);        # familiy ID
+       &cmp    (&HB("eax"),15);        # P4?
+       &jne    (&label("notP4"));
        &or     ("edx",1<<20);          # use reserved bit to engage RC4_CHAR
-&set_label("notintel");
+&set_label("notP4");
+       &bt     ("edx",28);             # test hyper-threading bit
+       &jnc    (&label("done"));
        &shr    ("ebx",16);
        &cmp    (&LB("ebx"),1);         # see if cache is shared(*)
-       &ja     (&label("nocpuid"));
+       &ja     (&label("done"));
        &and    ("edx",0xefffffff);     # clear hyper-threading bit if not
-&set_label("nocpuid");
+&set_label("done");
        &mov    ("eax","edx");
        &mov    ("edx","ecx");
 &function_end("OPENSSL_ia32_cpuid");