Fix typo in x86_64-mont5.pl CFI directives

[openssl.git] / crypto / x86_64cpuid.pl

diff --git a/crypto/x86_64cpuid.pl b/crypto/x86_64cpuid.pl
index 4946688d922e9f2572d53f929840900897be457d..3082253aea947082d12b456e9399d20ea3c253aa 100644 (file)
--- a/crypto/x86_64cpuid.pl
+++ b/crypto/x86_64cpuid.pl
@@ -18,7 +18,7 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
 ( $xlate="${dir}perlasm/x86_64-xlate.pl" and -f $xlate) or
 die "can't locate x86_64-xlate.pl";
 
-open OUT,"| \"$^X\" $xlate $flavour $output";
+open OUT,"| \"$^X\" \"$xlate\" $flavour \"$output\"";
 *STDOUT=*OUT;
 
 ($arg1,$arg2,$arg3,$arg4)=$win64?("%rcx","%rdx","%r8", "%r9") :        # Win64 order
@@ -175,13 +175,21 @@ OPENSSL_ia32_cpuid:
        jnc     .Lclear_avx
        xor     %ecx,%ecx               # XCR0
        .byte   0x0f,0x01,0xd0          # xgetbv
+       and     \$0xe6,%eax             # isolate XMM, YMM and ZMM state support
+       cmp     \$0xe6,%eax
+       je      .Ldone
+       andl    \$0xfffeffff,8(%rdi)    # clear AVX512F, ~(1<<16)
+                                       # note that we don't touch other AVX512
+                                       # extensions, because they can be used
+                                       # with YMM (without opmasking though)
        and     \$6,%eax                # isolate XMM and YMM state support
        cmp     \$6,%eax
        je      .Ldone
 .Lclear_avx:
        mov     \$0xefffe7ff,%eax       # ~(1<<28|1<<12|1<<11)
        and     %eax,%r9d               # clear AVX, FMA and AMD XOP bits
-       andl    \$0xffffffdf,8(%rdi)    # cleax AVX2, ~(1<<5)
+       mov     \$0x3fdeffdf,%eax       # ~(1<<31|1<<30|1<<21|1<<16|1<<5)
+       and     %eax,8(%rdi)            # cleax AVX2 and AVX512* bits
 .Ldone:
        shl     \$32,%r9
        mov     %r10d,%eax
@@ -393,36 +401,67 @@ OPENSSL_instrument_bus2:
 ___
 }
 
+sub gen_random {
+my $rdop = shift;
 print<<___;
-.globl OPENSSL_ia32_rdrand
-.type  OPENSSL_ia32_rdrand,\@abi-omnipotent
+.globl OPENSSL_ia32_${rdop}
+.type  OPENSSL_ia32_${rdop},\@abi-omnipotent
 .align 16
-OPENSSL_ia32_rdrand:
+OPENSSL_ia32_${rdop}:
        mov     \$8,%ecx
-.Loop_rdrand:
-       rdrand  %rax
-       jc      .Lbreak_rdrand
-       loop    .Loop_rdrand
-.Lbreak_rdrand:
+.Loop_${rdop}:
+       ${rdop} %rax
+       jc      .Lbreak_${rdop}
+       loop    .Loop_${rdop}
+.Lbreak_${rdop}:
        cmp     \$0,%rax
        cmove   %rcx,%rax
        ret
-.size  OPENSSL_ia32_rdrand,.-OPENSSL_ia32_rdrand
+.size  OPENSSL_ia32_${rdop},.-OPENSSL_ia32_${rdop}
 
-.globl OPENSSL_ia32_rdseed
-.type  OPENSSL_ia32_rdseed,\@abi-omnipotent
+.globl OPENSSL_ia32_${rdop}_bytes
+.type  OPENSSL_ia32_${rdop}_bytes,\@abi-omnipotent
 .align 16
-OPENSSL_ia32_rdseed:
-       mov     \$8,%ecx
-.Loop_rdseed:
-       rdseed  %rax
-       jc      .Lbreak_rdseed
-       loop    .Loop_rdseed
-.Lbreak_rdseed:
-       cmp     \$0,%rax
-       cmove   %rcx,%rax
+OPENSSL_ia32_${rdop}_bytes:
+       xor     %rax, %rax      # return value
+       cmp     \$0,$arg2
+       je      .Ldone_${rdop}_bytes
+
+       mov     \$8,%r11
+.Loop_${rdop}_bytes:
+       ${rdop} %r10
+       jc      .Lbreak_${rdop}_bytes
+       dec     %r11
+       jnz     .Loop_${rdop}_bytes
+       jmp     .Ldone_${rdop}_bytes
+
+.align 16
+.Lbreak_${rdop}_bytes:
+       cmp     \$8,$arg2
+       jb      .Ltail_${rdop}_bytes
+       mov     %r10,($arg1)
+       lea     8($arg1),$arg1
+       add     \$8,%rax
+       sub     \$8,$arg2
+       jz      .Ldone_${rdop}_bytes
+       mov     \$8,%r11
+       jmp     .Loop_${rdop}_bytes
+
+.align 16
+.Ltail_${rdop}_bytes:
+       mov     %r10b,($arg1)
+       lea     1($arg1),$arg1
+       inc     %rax
+       shr     \$8,%r8
+       dec     $arg2
+       jnz     .Ltail_${rdop}_bytes
+
+.Ldone_${rdop}_bytes:
        ret
-.size  OPENSSL_ia32_rdseed,.-OPENSSL_ia32_rdseed
+.size  OPENSSL_ia32_${rdop}_bytes,.-OPENSSL_ia32_${rdop}_bytes
 ___
+}
+gen_random("rdrand");
+gen_random("rdseed");
 
 close STDOUT;  # flush