Patch to BitcodeReader

[Joe Abbey]

Committed in r190634

Joe

On Sep 12, 2013, at 5:49 PM, Joe Abbey <joe.abbey at gmail.com> wrote:

> Sure Tom I can commit this for you.
> 
> Joe
> 
> On Sep 12, 2013, at 5:26 PM, Tom Roeder <tmroeder at google.com> wrote:
> 
>> Thanks!
>> 
>> I don't have commit access; could one of you please commit for me?
>> 
>> Tom
>> 
>> 
>> On Thu, Sep 12, 2013 at 2:14 PM, Joe Abbey <joe.abbey at gmail.com> wrote:
>> Tom,
>> 
>> This all looks good please review my updates to your patch for LLVM coding style.
>> 
>> Joe
>> 
>> 
>> 
>> 
>> On Sep 12, 2013, at 5:10 PM, Tobias Grosser <tobias at grosser.es> wrote:
>> 
>> > On 09/12/2013 11:03 PM, Tom Roeder wrote:
>> >> Here's a version with a comment and the fixed CHECK statement.
>> >
>> > Nice.
>> >
>> > Just a tiny comment I overlooked before:
>> >
>> >> Index: lib/Bitcode/Reader/BitcodeReader.cpp
>> >> ===================================================================
>> >> --- lib/Bitcode/Reader/BitcodeReader.cpp     (revision 190562)
>> >> +++ lib/Bitcode/Reader/BitcodeReader.cpp     (working copy)
>> >> @@ -1389,11 +1389,23 @@
>> >>      }
>> >>      case bitc::CST_CODE_CE_SELECT:  // CE_SELECT: [opval#, opval#, opval#]
>> >>        if (Record.size() < 3) return Error("Invalid CE_SELECT record");
>> >> -      V = ConstantExpr::getSelect(
>> >> -                          ValueList.getConstantFwdRef(Record[0],
>> >> -                                                      Type::getInt1Ty(Context)),
>> >> -                          ValueList.getConstantFwdRef(Record[1],CurTy),
>> >> -                          ValueList.getConstantFwdRef(Record[2],CurTy));
>> >> +
>> >> +      // If CurTy is a vector of length n, then Record[0] must be a <n x i1>
>> >> +      // vector. Otherwise, it must be a single bit.
>> >> +      if (VectorType *VTy = dyn_cast<VectorType>(CurTy)) {
>> >> +        unsigned int EltCount = VTy->getNumElements();
>> >> +        Type *SelectorTy = VectorType::get(Type::getInt1Ty(Context), EltCount);
>> >> +
>> >> +        V = ConstantExpr::getSelect(ValueList.getConstantFwdRef(Record[0],
>> >> +                                                                SelectorTy),
>> >> +                                    ValueList.getConstantFwdRef(Record[1],CurTy),
>> >> +                                    ValueList.getConstantFwdRef(Record[2],CurTy));
>> >> +      } else {
>> >> +        V = ConstantExpr::getSelect(ValueList.getConstantFwdRef(Record[0],
>> >> +                                                                Type::getInt1Ty(Context)),
>> >> +                                  ValueList.getConstantFwdRef(Record[1],CurTy),
>> >> +                                  ValueList.getConstantFwdRef(Record[2],CurTy));
>> >
>> >                                     ^^ Those lines are not properly
>> >                                        aligned.
>> >
>> > Otherwise, the patch looks good. Can you commit this patch yourself or do you need me to commit it?
>> >
>> > Cheers,
>> > Tobias
>> > _______________________________________________
>> > llvm-commits mailing list
>> > llvm-commits at cs.uiuc.edu
>> > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> 
>> 
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130912/72d8239d/attachment.html>