llvm dev - Oct 2019 - register spilling and printing live variables

Hello,

I have studied register allocation in theoretical aspects and exploring the
same in the implementation level.

I need a minimal testcase for register spilling to analyze spilling
procedure in llvm. I tried with a testcase taking 20 variables but all the
20 variables are getting stored in the stack using %rbp. Maybe my live
variable analysis is wrong. Please help me with a minimal testcase to
explore register spilling in llvm,

Is there any way to print the maximum live variables in source code. So
that we can find out the minimum registers required for execution. Also, is
it possible to print the register map of each variable in source code.

Any help would be appreciated.
Thank you in advance.

Regards,
Priyanka
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20191025/e8c8abe6/attachment.html>

I don't have an answer, but if you can share your test case, it may help.


On Thu, Oct 24, 2019 at 11:10 PM Priyanka Panigrahi via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> Hello,
>
> I have studied register allocation in theoretical aspects and exploring
> the same in the implementation level.
>
> I need a minimal testcase for register spilling to analyze spilling
> procedure in llvm. I tried with a testcase taking 20 variables but all the
> 20 variables are getting stored in the stack using %rbp. Maybe my live
> variable analysis is wrong. Please help me with a minimal testcase to
> explore register spilling in llvm,
>
> Is there any way to print the maximum live variables in source code. So
> that we can find out the minimum registers required for execution. Also, is
> it possible to print the register map of each variable in source code.
>
> Any help would be appreciated.
> Thank you in advance.
>
> Regards,
> Priyanka
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20191025/77652834/attachment.html>

Hi

On Fri, Oct 25, 2019 at 11:40 AM Priyanka Panigrahi via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> Hello,
>
> I have studied register allocation in theoretical aspects and exploring
> the same in the implementation level.
>
> I need a minimal testcase for register spilling to analyze spilling
> procedure in llvm. I tried with a testcase taking 20 variables but all the
> 20 variables are getting stored in the stack using %rbp. Maybe my live
> variable analysis is wrong. Please help me with a minimal testcase to
> explore register spilling in llvm,
>
You can share your testcase, people can have a look at it
>
> Is there any way to print the maximum live variables in source code. So
> that we can find out the minimum registers required for execution. Also, is
> it possible to print the register map of each variable in source code.
>
*-debug-only=regalloc *will dump a lot of information including live
intervals and register maps.
Something like this.

********** REWRITE VIRTUAL REGISTERS **********
********** Function: main
********** REGISTER MAP **********
[%0 -> $r12] GR64_with_sub_8bit
[%1 -> $r15] GR64_with_sub_8bit
[%9 -> $rsi] GR64_with_sub_8bit
[%10 -> $rdx] GR64_with_sub_8bit
[%14 -> $xmm1] FR32
[%17 -> $xmm1] FR32
[%20 -> $xmm1] FR32
[%27 -> $r15] GR64
[%35 -> $rdi] GR64
[%38 -> $xmm0] FR64
[%39 -> $xmm0] FR32
[%44 -> $xmm0] FR32
[%45 -> $xmm0] FR64
[%56 -> $rcx] GR64_with_sub_8bit
[%60 -> $rdi] GR64_with_sub_8bit
[%66 -> $rax] GR64_with_sub_8bit
[%73 -> $xmm0] FR32
[%74 -> $xmm0] FR64
[%82 -> $rbx] GR64_NOSP
[%83 -> $rbx] GR64_NOSP
[%87 -> $rdi] GR64_NOSP
[%88 -> $xmm0] FR32
[%90 -> $rbx] GR64_NOSP
[%91 -> $rbx] GR64_NOSP
[%93 -> $r14] GR64_with_sub_8bit


>
> Any help would be appreciated.
> Thank you in advance.
>
> Regards,
> Priyanka
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>

-- 
Regards,
DTharun
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20191028/42c20f0e/attachment.html>

Thank you for ur reply.
Here is a simple testcase:

    int a1,a2,a3,a4,a5,a6,a7,a8,a9,a10;
    int a11,a12,a13,a14,a15,a16,a17,a18,a19,a20;

    a1=a2=a3=a4=a5=a6=a7=a8=a9=a10=10;
    a11=a12=a13=a14=a15=a16=a17=a18=a19=a20=20;


if(a1==a2==a3==a4==a5==a6==a7==a8==a9==a10==a11==a12==a13==a14==a15==a16==a17==a18==a19==a20)
    {
        printf("equal \n");
    }
    else
    {
        printf("not equal \n");
    }

All the variables are stored in the stack.

I have already tried the option *-debug-only=regalloc, *but I am not able
to understand the mapping and live variable analysis properly.

I want to analyze, how spilling takes place.

Thank you for your time.

Regards,
Priyanka


On Mon, Oct 28, 2019 at 12:48 PM Dangeti Tharun kumar <
cs15mtech11002 at iith.ac.in> wrote:
> Hi
>
> On Fri, Oct 25, 2019 at 11:40 AM Priyanka Panigrahi via llvm-dev <
> llvm-dev at lists.llvm.org> wrote:
>
>> Hello,
>>
>> I have studied register allocation in theoretical aspects and exploring
>> the same in the implementation level.
>>
>> I need a minimal testcase for register spilling to analyze spilling
>> procedure in llvm. I tried with a testcase taking 20 variables but all
the
>> 20 variables are getting stored in the stack using %rbp. Maybe my live
>> variable analysis is wrong. Please help me with a minimal testcase to
>> explore register spilling in llvm,
>>
>
> You can share your testcase, people can have a look at it
>
>>
>> Is there any way to print the maximum live variables in source code. So
>> that we can find out the minimum registers required for execution.
Also, is
>> it possible to print the register map of each variable in source code.
>>
>
> *-debug-only=regalloc *will dump a lot of information including live
> intervals and register maps.
> Something like this.
>
> ********** REWRITE VIRTUAL REGISTERS **********
> ********** Function: main
> ********** REGISTER MAP **********
> [%0 -> $r12] GR64_with_sub_8bit
> [%1 -> $r15] GR64_with_sub_8bit
> [%9 -> $rsi] GR64_with_sub_8bit
> [%10 -> $rdx] GR64_with_sub_8bit
> [%14 -> $xmm1] FR32
> [%17 -> $xmm1] FR32
> [%20 -> $xmm1] FR32
> [%27 -> $r15] GR64
> [%35 -> $rdi] GR64
> [%38 -> $xmm0] FR64
> [%39 -> $xmm0] FR32
> [%44 -> $xmm0] FR32
> [%45 -> $xmm0] FR64
> [%56 -> $rcx] GR64_with_sub_8bit
> [%60 -> $rdi] GR64_with_sub_8bit
> [%66 -> $rax] GR64_with_sub_8bit
> [%73 -> $xmm0] FR32
> [%74 -> $xmm0] FR64
> [%82 -> $rbx] GR64_NOSP
> [%83 -> $rbx] GR64_NOSP
> [%87 -> $rdi] GR64_NOSP
> [%88 -> $xmm0] FR32
> [%90 -> $rbx] GR64_NOSP
> [%91 -> $rbx] GR64_NOSP
> [%93 -> $r14] GR64_with_sub_8bit
>
>
>
>>
>> Any help would be appreciated.
>> Thank you in advance.
>>
>> Regards,
>> Priyanka
>> _______________________________________________
>> LLVM Developers mailing list
>> llvm-dev at lists.llvm.org
>> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>
>
>
> --
> Regards,
> DTharun
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20191028/eab61fda/attachment.html>