[LLVMdev] (GEP) Index validity

[Matthijs Kooijman]

Hi Evan,


> I don't think this is right. According to llvm documentation:
>
> The index types specified for the 'getelementptr' instruction depend on the 
> pointer type that is being indexed into. Pointer and array types can use a 
> 32-bit or 64-bit integer type but the value will always be sign extended to 
> 64-bits. Structure and packed structure types require i32 constants.
That's correct. I'm not saying to allow other integer types for GEP
instructions. The first two methods below are methods in StructType and
SequentialType. Regardless of any limitations posed by a GEP instruction, I
argue that "i8 2" is a perfectly valid index for a struct. The fact that it is
not valid for a GEP instruction, should be checked somewhere else (GEPInst
constructor or GEPInst::getIndexedValue probably).

The last method in the patch is directly related to GEP indices, so asserting
only the constness might not be enough, though it would not hurt to make this
assert slightly more general (ie, don't check the bitwidth as well).

Gr.

Matthijs
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20080710/7ee7033c/attachment.sig>