[llvm-commits] [llvm-gcc-4.2] r104523 - /llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp

[Stuart Hastings]

On May 24, 2010, at 11:43 AM, Chris Lattner wrote:

> 
> On May 24, 2010, at 11:25 AM, Stuart Hastings wrote:
> 
>> Author: stuart
>> Date: Mon May 24 13:25:58 2010
>> New Revision: 104523
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=104523&view=rev
>> Log:
>> Fix and off-by-one bug in bitfield handling.  Radar 7980096.
> 
> Testcase please.

I've been thinking about that.  The signature difference made by the patch is that the bitfield load-byte should be one byte above the store (-word).  I'm looking at the existing bitfield test cases; is there an existing FrontendC test case that looks for something similar?

stuart

> 
> -Chris
> 
>> 
>> Modified:
>>   llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp
>> 
>> Modified: llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp?rev=104523&r1=104522&r2=104523&view=diff
>> ==============================================================================
>> --- llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp (original)
>> +++ llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp Mon May 24 13:25:58 2010
>> @@ -7304,7 +7304,7 @@
>>    // adjust FieldPtr so that it is close enough to the bitfield that
>>    // *FieldPtr contains the first needed bit.  Be careful to make sure that
>>    // the pointer remains appropriately aligned.
>> -    if (BitStart > LLVMValueBitSize) {
>> +    if (BitStart >= LLVMValueBitSize) {
>>      // In this case, we know that the alignment of the field is less than
>>      // the size of the field.  To get the pointer close enough, add some
>>      // number of alignment units to the pointer.
>> 
>> 
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>