[PATCH] Finish assembly support for Intel SHA Extensions

[Craig Topper]

Oops you're right I missed that. Sorry!

On Friday, September 13, 2013, Langmuir, Ben wrote:

>  I thought that having the defm for the new instructions inside the same
> let block as sha1rnds4 would suffice.  Is that not the case? That seems to
> be how the AES instructions are set up just below this.****
>
> ** **
>
> Thanks for the tip about blendvpd – I’ll look at adding InstAliases.****
>
> ** **
>
> Ben****
>
> ** **
>
> *From:* Craig Topper [mailto:craig.topper at gmail.com <javascript:_e({},
> 'cvml', 'craig.topper at gmail.com');>]
> *Sent:* Thursday, September 12, 2013 11:34 PM
> *To:* reviews+D1663+public+238ef3f1c2db68b0 at llvm-reviews.chandlerc.com<javascript:_e({}, 'cvml',
> 'reviews%2BD1663%2Bpublic%2B238ef3f1c2db68b0 at llvm-reviews.chandlerc.com');>
> *Cc:* Langmuir, Ben; llvm-commits at cs.uiuc.edu <javascript:_e({}, 'cvml',
> 'llvm-commits at cs.uiuc.edu');>; Nadav Rotem
> *Subject:* Re: [PATCH] Finish assembly support for Intel SHA Extensions***
> *
>
> ** **
>
> I think you're missing  let Constraints = "$src1 = $dst", hasSideEffects
> = 0, Predicates = [HasSHA] on the new instructions aren't you?****
>
> ** **
>
> Also please add InstAliases for the SHA256RNDS2 instruction so that users
> can specify the implicit xmm0 operand. See the aliases for blendvpd for
> an example.****
>
> ** **
>
> ** **
>
> 2013/9/12 Ben Langmuir <ben.langmuir at intel.com>****
>
> Hi craig.topper,
>
> Add the remaining Intel SHA instructions and tests.  All of these
> instructions
> seem to have the same form, except that sha256rnds2 has an implicit read
> dependency on xmm0.
>
> http://llvm-reviews.chandlerc.com/D1663
>
> Files:
>   lib/Target/X86/X86InstrSSE.td
>   test/MC/Disassembler/X86/x86-64.txt
>   test/MC/X86/x86_64-encoding.s
>
> Index: lib/Target/X86/X86InstrSSE.td
> ===================================================================
> --- lib/Target/X86/X86InstrSSE.td
> +++ lib/Target/X86/X86InstrSSE.td
> @@ -7324,6 +7324,17 @@
>  // SHA-NI Instructions
>
>  //===----------------------------------------------------------------------===//
>
> +multiclass SHAI_binop<bits<8> Opc, string OpcodeStr> {
> +  def rr : I<Opc, MRMSrcReg, (outs VR128:$dst),
> +             (ins VR128:$src1, VR128:$src2),
> +             !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"), []>,
> T8;
> +
> +  let mayLoad = 1 in
> +  def rm : I<Opc, MRMSrcMem, (outs VR128:$dst),
> +             (ins VR128:$src1, i128mem:$src2),
> +             !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"), []>,
> T8;
> +}
> +
>  let Constraints = "$src1 = $dst", hasSideEffects = 0, Predicates =
> [HasSHA] in {
>    def SHA1RNDS4rri : Ii8<0xCC, MRMSrcReg, (outs VR128:$dst),
>                           (ins VR128:$src1, VR128:$src2, i8imm:$src3),
> @@ -7334,6 +7345,16 @@
>                           (ins VR128:$src1, i128mem:$src2, i8imm:$src3),
>                           "sha1rnds4\t{$src3, $src2, $dst|$dst, $src2,
> $src3}",
>                           []>, TA;
> +
> +  defm SHA1NEXTE : SHAI_binop<0xC8, "sha1nexte">;
> +  defm SHA1MSG1  : SHAI_binop<0xC9, "sha1msg1">;
> +  defm SHA1MSG2  : SHAI_binop<0xCA, "sha1msg2">;
> +
> +  let Uses=[XMM0] in
> +  defm SHA256RNDS2 : SHAI_binop<0xCB, "sha256rnds2">;
> +
> +  defm SHA256MSG1 : SHAI_binop<0xCC, "sha256msg1">;
> +  defm SHA256MSG2 : SHAI_binop<0xCD, "sha256msg2">;
>  }
>
>
>  //===----------------------------------------------------------------------===//
> Index: test/MC/Disassembler/X86/x86-64.txt
> ===================================================================
> --- test/MC/Disassembler/X86/x86-64.txt
> +++ test/MC/Disassembler/X86/x86-64.txt
> @@ -162,4 +162,40 @@
>  0x0f 0x3a 0xcc 0xd1 0x01
>
>  # CHECK: sha1rnds4 $1, (%rax), %xmm2
> -0x0f 0x3a 0xcc 0x10 0x01
> \ No newline at end of file
> +0x0f 0x3a 0xcc 0x10 0x01
> +
> +# CHECK: sha1nexte %xmm1, %xmm2
> +0x0f 0x38 0xc8 0xd1
> +
> +# CHECK: sha1nexte (%rax), %xmm2
> +0x0f 0x38 0xc8 0x10
> +
> +# CHECK: sha1msg1 %xmm1, %xmm2
> +0x0f 0x38 0xc9 0xd1
> +
> +# CHECK: sha1msg1 (%rax), %xmm2
> +0x0f 0x38 0xc9 0x10
> +
> +# CHECK: sha1msg2 %xmm1, %xmm2
> +0x0f 0x38 0xca 0xd1
> +
> +# CHECK: sha1msg2 (%rax), %xmm2
> +0x0f 0x38 0xca 0x10
> +
> +# CHECK: sha256rnds2 (%rax), %xmm2
> +0x0f 0x38 0xcb 0x10
> +
> +# CHECK: sha256rnds2 %xmm1, %xmm2
> +0x0f 0x38 0xcb 0xd1
> +
> +# CHECK: sha256msg1 %xmm1, %xmm2
> +0x0f 0x38 0xcc 0xd1
> +
> +# CHECK: sha256msg1 (%rax), %xmm2
> +0x0f 0x38 0xcc 0x10
> +
> +# CHECK: sha256msg2 %xmm1, %xmm2
> +0x0f 0x38 0xcd 0xd1
> +
> +# CHECK: sha256msg2 (%rax), %xmm2
> +0x0f 0x38 0xcd 0x10
> \ No newline at end of file
> Index: test/MC/X86/x86_64-encoding.s
> ===================================================================
> --- test/MC/X86/x86_64-encoding.s
> +++ test/MC/X86/x86_64-encoding.s
> @@ -128,6 +128,50 @@
>  // CHECK:   encoding: [0x0f,0x3a,0xcc,0x10,0x01]
>  sha1rnds4 $1, (%rax), %xmm2
>


-- 
~Craig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130913/6a039a17/attachment.html>