llvm dev - May 2016 - Back end with special loop instructions

Hello.
     I'm writing a back end for my research SIMD processor that has an
assembly language
that is blocked structured, with one-level loops. An example program with my
assembly
language:
       REPEAT_X_TIMES(Param2)
         R0 = LS[offset_A];
       END_REPEAT;

     The LLVM code somewhat equivalent to the above ASM program is:
       vector.body:
         %index = phi i64 [ %index.unr, %vector.body.preheader.split.split ], [ 
%index.next.3, %vector.body ]
         %20 = getelementptr inbounds i32, i32* %A, i64 %index
         %21 = bitcast i32* %20 to <16 x i32>*
         %wide.load = load <16 x i32>, <16 x i32>* %21, align 4
         br i1 %48, label %middle.block.unr-lcssa, label %vector.body,
!llvm.loop !3

     How do you suggest to attack this problem?
     I guess I need to provide custom matching code in the Select() function of
the back
end prior to the SelectCode() invocation in order to translate the label
vector.body in
LLVM/Machine Instr(DAG) before selection to a REPEAT_X_TIMES(...) instruction -
are you
aware if anybody else has done such a thing? Also, the br LLVM instruction will
be
translated to an END_REPEAT.

   Thank you,
     Alex

Hi Alex,

You might find it useful to look at how lib/Target/PowerPC/PPCCTRLoops.cpp
works.

 -Hal

----- Original Message -----
> From: "Alex Susu via llvm-dev" <llvm-dev at lists.llvm.org>
> To: "llvm-dev" <llvm-dev at lists.llvm.org>
> Sent: Monday, May 30, 2016 5:09:37 PM
> Subject: [llvm-dev] Back end with special loop instructions
> 
> Hello.
>      I'm writing a back end for my research SIMD processor that has
>      an assembly language
> that is blocked structured, with one-level loops. An example program
> with my assembly
> language:
>        REPEAT_X_TIMES(Param2)
>          R0 = LS[offset_A];
>        END_REPEAT;
> 
>      The LLVM code somewhat equivalent to the above ASM program is:
>        vector.body:
>          %index = phi i64 [ %index.unr,
>          %vector.body.preheader.split.split ], [
> %index.next.3, %vector.body ]
>          %20 = getelementptr inbounds i32, i32* %A, i64 %index
>          %21 = bitcast i32* %20 to <16 x i32>*
>          %wide.load = load <16 x i32>, <16 x i32>* %21, align 4
>          br i1 %48, label %middle.block.unr-lcssa, label
>          %vector.body, !llvm.loop !3
> 
>      How do you suggest to attack this problem?
>      I guess I need to provide custom matching code in the Select()
>      function of the back
> end prior to the SelectCode() invocation in order to translate the
> label vector.body in
> LLVM/Machine Instr(DAG) before selection to a REPEAT_X_TIMES(...)
> instruction - are you
> aware if anybody else has done such a thing? Also, the br LLVM
> instruction will be
> translated to an END_REPEAT.
> 
>    Thank you,
>      Alex
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
> 
-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory

Hello.
     Hal, the source file you mention (lib/Target/PowerPC/PPCCTRLoops.cpp) makes
use of
LLVM IR intrinsics, in this case defined at 
[LLVM_repo]/llvm/include/llvm/IR/IntrinsicsPowerPC.td, such as:
           // Intrinsics used to generate ctr-based loops. These should only be
	  // generated by the PowerPC backend!
	  def int_ppc_mtctr : Intrinsic<[], [llvm_anyint_ty], []>;
	  def int_ppc_is_decremented_ctr_nonzero : Intrinsic<[llvm_i1_ty], [],
[]>;

     And these intrinsics are instantiated in the LLVM program, for example with
the
following code in PPCCTRLoops.cpp:
       IRBuilder<> CountBuilder(Preheader->getTerminator());
       Module *M = Preheader->getParent()->getParent();
       Value *MTCTRFunc = Intrinsic::getDeclaration(M,
Intrinsic::ppc_mtctr,CountType);
       CountBuilder.CreateCall(MTCTRFunc, ECValue);

     I have defined also some intrinsics for my loop instructions in my file 
Intrinsics_Connex.td: 1 intrinsic for REPEAT_X_TIMES and 1 for END_REPEAT.
       /* following Intrinsics.td:
       class Intrinsic<list<LLVMType> ret_types,
                 list<LLVMType> param_types = [],
                 list<IntrinsicProperty> properties = [],
                 string name = "">
       */
       def int_connex_repeat_x_times : Intrinsic<[], [], []>;
       def int_connex_end_repeat : Intrinsic<[llvm_i1_ty], [], []>;
    and added C++ code doing CreateCall() like the one above.

     I'm looking now at http://llvm.org/docs/ExtendingLLVM.html on how to
specify the
instruction selection of this intrinsic. They write there:
       "Once the intrinsic has been added to the system, you must add code
generator
support for it. Generally you must do the following steps:
       Add support to the .td file for the target(s) of your choice in
lib/Target/*/*.td.
       This is usually a matter of adding a pattern to the .td file that matches
the
intrinsic, though it may obviously require adding the instructions you want to
generate as
well. There are lots of examples in the PowerPC and X86 backend to follow."

     Then in my pass I create a call to the respective intrinsic:
      /* See
http://llvm.org/docs/doxygen/html/classllvm_1_1TargetIntrinsicInfo.html :
"Create or insert an LLVM Function declaration for an intrinsic, and return
it."
       *   This creates a line in the LLVM program like: declare void 
@llvm.connex.repeat.x.times() #2 .
       *     This line is required, otherwise llc will complain:
       *        <<error: use of undefined value
'@llvm.connex.repeat.x.times'
             call void @llvm.connex.repeat.x.times()>>
       */
       Value *repeatFunc = Intrinsic::getDeclaration(M,
                                                
Intrinsic::connex_repeat_x_times);

       // See http://llvm.org/docs/doxygen/html/classllvm_1_1IRBuilder.html
       aB.CreateCall(repeatFunc); //, ECValue);


     Then, in the back end, in InstrInfo.td I write:
       let hasSideEffects = 1, isCodeGenOnly = 1 in {
       //let Pattern = [(int_connex_repeat_x_times)] in
       def REPEAT_X_TIMES : ImmediateInstruction< 0b111111,
                                             (outs),
                                             (ins),
                                             "REPEAT_X_TIMES(1001)",
                                             [(int_connex_repeat_x_times)] >;
       }


     Then, when I run opt and llc I obtain the expected behavior.

   Thank you,
     Alex



On 5/31/2016 2:30 AM, Hal Finkel wrote:
> Hi Alex,
>
> You might find it useful to look at how lib/Target/PowerPC/PPCCTRLoops.cpp
works.
>
>   -Hal
>
> ----- Original Message -----
>> From: "Alex Susu via llvm-dev" <llvm-dev at
lists.llvm.org>
>> To: "llvm-dev" <llvm-dev at lists.llvm.org>
>> Sent: Monday, May 30, 2016 5:09:37 PM
>> Subject: [llvm-dev] Back end with special loop instructions
>>
>> Hello.
>>       I'm writing a back end for my research SIMD processor that
has
>>       an assembly language
>> that is blocked structured, with one-level loops. An example program
>> with my assembly
>> language:
>>         REPEAT_X_TIMES(Param2)
>>           R0 = LS[offset_A];
>>         END_REPEAT;
>>
>>       The LLVM code somewhat equivalent to the above ASM program is:
>>         vector.body:
>>           %index = phi i64 [ %index.unr,
>>           %vector.body.preheader.split.split ], [
>> %index.next.3, %vector.body ]
>>           %20 = getelementptr inbounds i32, i32* %A, i64 %index
>>           %21 = bitcast i32* %20 to <16 x i32>*
>>           %wide.load = load <16 x i32>, <16 x i32>* %21,
align 4
>>           br i1 %48, label %middle.block.unr-lcssa, label
>>           %vector.body, !llvm.loop !3
>>
>>       How do you suggest to attack this problem?
>>       I guess I need to provide custom matching code in the Select()
>>       function of the back
>> end prior to the SelectCode() invocation in order to translate the
>> label vector.body in
>> LLVM/Machine Instr(DAG) before selection to a REPEAT_X_TIMES(...)
>> instruction - are you
>> aware if anybody else has done such a thing? Also, the br LLVM
>> instruction will be
>> translated to an END_REPEAT.
>>
>>     Thank you,
>>       Alex
>> _______________________________________________
>> LLVM Developers mailing list
>> llvm-dev at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>
>