[llvm] r179890 - Try explicitly setting the target triple to see if this gets it to pass on ARM.

[Benjamin Kramer]

On 20.04.2013, at 00:52, Hal Finkel <hfinkel at anl.gov> wrote:

> ----- Original Message -----
>> From: "Benjamin Kramer" <benny.kra at gmail.com>
>> To: "Hal Finkel" <hfinkel at anl.gov>
>> Cc: "Bill Wendling" <isanbard at gmail.com>, llvm-commits at cs.uiuc.edu
>> Sent: Friday, April 19, 2013 5:41:33 PM
>> Subject: Re: [llvm] r179890 - Try explicitly setting the target triple to see if	this gets it to pass on ARM.
>> 
>> 
>> On 20.04.2013, at 00:27, Hal Finkel <hfinkel at anl.gov> wrote:
>> 
>>> Bill,
>>> 
>>> I'm seeing this fail on ppc64/Linux (Release+Asserts):
>>> 
>>> llvm-trunk/test/Transforms/MergeFunc/inttoptr.ll:49:10: error:
>>> expected string not found in input
>>> ; CHECK: %[[V2:[0-9]+]] = bitcast %.qux.2585* %{{.*}} to
>>> %.qux.2496*
>>>        ^
>>> <stdin>:1:1: note: scanning from here
>>> ; ModuleID = '<stdin>'
>>> ^
>>> <stdin>:42:13: note: possible intended match here
>>> %tmp = getelementptr inbounds %.qux.2585* %this, i32 0
>>>           ^
>>> 
>>> Maybe we should make V2 match any generic variable name?
>> 
>> Hmm, this looks like the function isn't turned into a thunk at all,
>> something strange seems to be going on.
>> 
>> The target shouldn't have an effect on value names here, neither does
>> changing the triple on the command line.
> 
> Yep, on my machine this outputs:
> 
> define internal hidden i8* @func35(%.qux.2585* nocapture %this) align 2 {
> bb:
>  %tmp = getelementptr inbounds %.qux.2585* %this, i32 0, i32 2
>  %tmp1 = load i8** %tmp, align 4
>  ret i8* %tmp1
> }
> 
> which seems essentially identical to the input. The debug output is:
> Subtarget features: SSELevel 0, 3DNowLevel 0, 64bit 0
> size of module: 6
> size of worklist: 6
> Inserting as unique: func1
> Inserting as unique: func10
> Inserting as unique: func29
> Inserting as unique: func33
> Inserting as unique: func34
> Inserting as unique: func35
> size of FnSet: 6

Try r179905. This was a scary bug :[

- Ben
> 
> -Hal
> 
>> 
>> - Ben
>> 
>>> 
>>> -Hal
>>> 
>>> ----- Original Message -----
>>>> From: "Bill Wendling" <isanbard at gmail.com>
>>>> To: llvm-commits at cs.uiuc.edu
>>>> Sent: Friday, April 19, 2013 4:24:51 PM
>>>> Subject: [llvm] r179890 - Try explicitly setting the target triple
>>>> to see if	this gets it to pass on ARM.
>>>> 
>>>> Author: void
>>>> Date: Fri Apr 19 16:24:51 2013
>>>> New Revision: 179890
>>>> 
>>>> URL: http://llvm.org/viewvc/llvm-project?rev=179890&view=rev
>>>> Log:
>>>> Try explicitly setting the target triple to see if this gets it to
>>>> pass on ARM.
>>>> 
>>>> Modified:
>>>>   llvm/trunk/test/Transforms/MergeFunc/inttoptr.ll
>>>> 
>>>> Modified: llvm/trunk/test/Transforms/MergeFunc/inttoptr.ll
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/MergeFunc/inttoptr.ll?rev=179890&r1=179889&r2=179890&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/test/Transforms/MergeFunc/inttoptr.ll (original)
>>>> +++ llvm/trunk/test/Transforms/MergeFunc/inttoptr.ll Fri Apr 19
>>>> 16:24:51 2013
>>>> @@ -1,4 +1,4 @@
>>>> -; RUN: opt -mergefunc -S < %s | FileCheck %s
>>>> +; RUN: opt -mtriple i386-pc-linux-gnu -mergefunc -S < %s |
>>>> FileCheck
>>>> %s
>>>> ; PR15185
>>>> target datalayout =
>>>> "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32-S128"
>>>> target triple = "i386-pc-linux-gnu"
>>>> 
>>>> 
>>>> _______________________________________________
>>>> llvm-commits mailing list
>>>> llvm-commits at cs.uiuc.edu
>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>> 
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> 
>>