llvm dev - Nov 2008 - [LLVMdev] Invalid comparison instruction generation

I have a simple program that generates correct intermediate
representation. However, when working on my backend, and my lowering
function gets called. The comparison operation is flipped via an invalid
transformation. i.e. gt ==> le, lt ==> ge etc..

define void @test_fc_if_gt(double %x, double %y, double addrspace(11)*
%result) {

entry:

        %x.addr = alloca double         ; <double*> [#uses=3]

        %y.addr = alloca double         ; <double*> [#uses=2]

        %result.addr = alloca double addrspace(11)*             ;
<double addrspace(11)**> [#uses=2]

        store double %x, double* %x.addr

        store double %y, double* %y.addr

        store double addrspace(11)* %result, double addrspace(11)**
%result.addr

        %tmp = load double* %x.addr             ; <double> [#uses=1]

        %tmp1 = load double* %y.addr            ; <double> [#uses=1]

        %cmp = fcmp ogt double %tmp, %tmp1              ; <i1> [#uses=1]

        br i1 %cmp, label %ifthen, label %ifend

 

ifthen:         ; preds = %entry

        %tmp2 = load double addrspace(11)** %result.addr
; <double addrspace(11)*> [#uses=1]

        %tmp3 = load double* %x.addr            ; <double> [#uses=1]

        store double %tmp3, double addrspace(11)* %tmp2

        br label %ifend

 

ifend:          ; preds = %ifthen, %entry

        ret void

}

 

 

With the above kernel run through llc with -march=x86
-view-dag-combine1-dags I still see the ogt as the comparison operation,
but when I run it with llc -march=x86 -view-legalize-dags the ogt node
has been transformed into a ule.

 

So, my question is, how do I get llvm to stop doing invalid translation
of comparison instructions? This problem affects my custom backend and I
have reproduced it with the x86 backend.

 

 

Micah Villmow

Systems Engineer

Advanced Technology & Performance

Advanced Micro Devices Inc.

4555 Great America Pkwy,

Santa Clara, CA. 95054

P: 408-572-6219

F: 408-572-6596

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20081110/0bf10a19/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: good-graph.dot
Type: application/octet-stream
Size: 8955 bytes
Desc: good-graph.dot
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20081110/0bf10a19/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: bad-graph.dot
Type: application/octet-stream
Size: 7503 bytes
Desc: bad-graph.dot
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20081110/0bf10a19/attachment-0001.obj>

On Mon, Nov 10, 2008 at 3:06 PM, Villmow, Micah <Micah.Villmow at amd.com>
wrote:
> With the above kernel run through llc with -march=x86
> -view-dag-combine1-dags I still see the ogt as the comparison operation,
but
> when I run it with llc -march=x86 -view-legalize-dags the ogt node has been
> transformed into a ule.
Okay... I can see that in the attached graph.
> So, my question is, how do I get llvm to stop doing invalid translation of
> comparison instructions? This problem affects my custom backend and I have
> reproduced it with the x86 backend.
It seems valid to me... what makes you think it's invalid?

-Eli

Eli, 
Using the variables from the original IR,
assuming tmp == tmp1 and assume the value is not nan
ogt(tmp, tmp1) is !isnan(tmp) && !isnan(tmp1) && tmp > tmp1,
or false
ule(tmp, tmp1) is isnan(tmp) || isnan(tmp1) || tmp <= tmp1, or true

So, this is invalid, or am I misunderstanding what ogt and ule stand
for?

Assuming this is valid, why convert comparison instructions instead of
just passing them through as they originally exist? The backend I am
targeting does not support all comparison instructions and trying to
guess which instruction LLVM converted the current comparison
instruction from and then converting to a supported instruction is not
as simple as it can be.
For example, I need to convert all ogt instructions to an olt
instruction with LHS and RHS swapped, but since ogt is converted to ule,
do I convert all ule into olt and swap? 

Thanks,

Micah

-----Original Message-----
From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu]
On Behalf Of Eli Friedman
Sent: Monday, November 10, 2008 4:38 PM
To: LLVM Developers Mailing List
Subject: Re: [LLVMdev] Invalid comparison instruction generation

On Mon, Nov 10, 2008 at 3:06 PM, Villmow, Micah <Micah.Villmow at amd.com>
wrote:
> With the above kernel run through llc with -march=x86
> -view-dag-combine1-dags I still see the ogt as the comparison
operation, but
> when I run it with llc -march=x86 -view-legalize-dags the ogt node has
been
> transformed into a ule.
Okay... I can see that in the attached graph.
> So, my question is, how do I get llvm to stop doing invalid
translation of
> comparison instructions? This problem affects my custom backend and I
have
> reproduced it with the x86 backend.
It seems valid to me... what makes you think it's invalid?

-Eli
_______________________________________________
LLVM Developers mailing list
LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev