[llvm] r265265 - [CodeGenPrepare] Fix r265264.

[whitequark via llvm-commits]

On 2016-04-03 21:58, Renato Golin wrote:
> Hi Peter,
> 
> It did not fix the ARM bots... :)

Should be fixed in r265271--I restored the old behavior for non-fcmp
instructions and X86+ARM tests pass locally, too.

> 
> http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15/builds/11146
> 
> http://lab.llvm.org:8011/builders/clang-cmake-thumbv7-a15/builds/11223
> 
> http://lab.llvm.org:8011/builders/clang-cmake-aarch64-quick/builds/5918
> 
> http://lab.llvm.org:8011/builders/clang-cmake-aarch64-42vma/builds/7192
> 
> cheers,
> --renato
> 
> On 3 April 2016 at 18:11, Peter Zotov via llvm-commits
> <llvm-commits at lists.llvm.org> wrote:
>> Author: whitequark
>> Date: Sun Apr  3 12:11:53 2016
>> New Revision: 265265
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=265265&view=rev
>> Log:
>> [CodeGenPrepare] Fix r265264.
>> 
>> The case where there was no TargetLowering was not handled,
>> leading to null pointer dereferences.
>> 
>> Modified:
>>     llvm/trunk/lib/CodeGen/CodeGenPrepare.cpp
>> 
>> Modified: llvm/trunk/lib/CodeGen/CodeGenPrepare.cpp
>> URL: 
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/CodeGenPrepare.cpp?rev=265265&r1=265264&r2=265265&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/CodeGen/CodeGenPrepare.cpp (original)
>> +++ llvm/trunk/lib/CodeGen/CodeGenPrepare.cpp Sun Apr  3 12:11:53 2016
>> @@ -910,8 +910,8 @@ static bool SinkCmpExpression(CmpInst *C
>>    return MadeChange;
>>  }
>> 
>> -static bool OptimizeCmpExpression(CmpInst *CI, const TargetLowering 
>> &TLI) {
>> -  if (SinkCmpExpression(CI, TLI))
>> +static bool OptimizeCmpExpression(CmpInst *CI, const TargetLowering 
>> *TLI) {
>> +  if (TLI && SinkCmpExpression(CI, *TLI))
>>      return true;
>> 
>>    if (CombineUAddWithOverflow(CI))
>> @@ -5177,7 +5177,7 @@ bool CodeGenPrepare::optimizeInst(Instru
>> 
>>    if (CmpInst *CI = dyn_cast<CmpInst>(I))
>>      if (!TLI || !TLI->hasMultipleConditionRegisters())
>> -      return OptimizeCmpExpression(CI, *TLI);
>> +      return OptimizeCmpExpression(CI, TLI);
>> 
>>    if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
>>      stripInvariantGroupMetadata(*LI);
>> 
>> 
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits

-- 
whitequark