sha/asm/keccak1600-x86_64.pl: remove redundant moves.

[openssl.git] / crypto / x86_64cpuid.pl

diff --git a/crypto/x86_64cpuid.pl b/crypto/x86_64cpuid.pl
index 285ad1ba7dba7bfe3bbf620ff984e19615636166..2467af7e9e70b695b74611d78fd79df36aeae590 100644 (file)
--- a/crypto/x86_64cpuid.pl
+++ b/crypto/x86_64cpuid.pl
@@ -63,10 +63,12 @@ OPENSSL_rdtsc:
 .type  OPENSSL_ia32_cpuid,\@function,1
 .align 16
 OPENSSL_ia32_cpuid:
+.cfi_startproc
        mov     %rbx,%r8                # save %rbx
+.cfi_register  %rbx,%r8
 
        xor     %eax,%eax
-       mov     %eax,8(%rdi)            # clear 3rd word
+       mov     %eax,8(%rdi)            # clear extended feature flags
        cpuid
        mov     %eax,%r11d              # max value for standard query level
 
@@ -134,14 +136,6 @@ OPENSSL_ia32_cpuid:
        shr     \$14,%r10d
        and     \$0xfff,%r10d           # number of cores -1 per L1D
 
-       cmp     \$7,%r11d
-       jb      .Lnocacheinfo
-
-       mov     \$7,%eax
-       xor     %ecx,%ecx
-       cpuid
-       mov     %ebx,8(%rdi)
-
 .Lnocacheinfo:
        mov     \$1,%eax
        cpuid
@@ -171,23 +165,42 @@ OPENSSL_ia32_cpuid:
        or      %ecx,%r9d               # merge AMD XOP flag
 
        mov     %edx,%r10d              # %r9d:%r10d is copy of %ecx:%edx
+
+       cmp     \$7,%r11d
+       jb      .Lno_extended_info
+       mov     \$7,%eax
+       xor     %ecx,%ecx
+       cpuid
+       mov     %ebx,8(%rdi)            # save extended feature flags
+.Lno_extended_info:
+
        bt      \$27,%r9d               # check OSXSAVE bit
        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
        mov     %r8,%rbx                # restore %rbx
+.cfi_restore   %rbx
        or      %r9,%rax
        ret
+.cfi_endproc
 .size  OPENSSL_ia32_cpuid,.-OPENSSL_ia32_cpuid
 
 .globl  OPENSSL_cleanse
@@ -393,36 +406,67 @@ 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