llvm dev - Nov 2012 - [LLVMdev] Forward references of globals in .ll files

I'm experiencing a weird issue during .ll file parsing, and I'm not sure
if
it's a bug in the .ll parser, or if I'm just not understanding the IR.

Take the following IR:

@a = addrspace(1) global i8 0
@a2 = global i8 addrspace(1)* @a

This parses fine.  But if I rearrange the statements to:

@a2 = global i8 addrspace(1)* @a
@a = addrspace(1) global i8 0

then I get an ugly assertion failure:

Assertion failed: (InitVal->getType() == getType()->getElementType()
&&
"Initializer type must match GlobalVariable type"), function
setInitializer, file
/Users/jholewinski/projects/llvm/src/llvm/lib/VMCore/Globals.cpp, line 168.

Looking at the debugger, I see that InitVal (the initializer for @a2) has
type i8*, which looks incorrect as it drops the addrspace from the
definition of @a.

Before I go too far into this, should this IR work?  It looks like the
parser is failing to properly process the forward declaration of @a, but I
just want to be sure.

-- 

Thanks,

Justin Holewinski
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20121109/eda6c257/attachment.html>

On Fri, Nov 9, 2012 at 5:01 PM, Justin Holewinski
<justin.holewinski at gmail.com> wrote:
> I'm experiencing a weird issue during .ll file parsing, and I'm not
sure if
> it's a bug in the .ll parser, or if I'm just not understanding the
IR.
>
> Take the following IR:
>
> @a = addrspace(1) global i8 0
> @a2 = global i8 addrspace(1)* @a
>
> This parses fine.  But if I rearrange the statements to:
>
> @a2 = global i8 addrspace(1)* @a
> @a = addrspace(1) global i8 0
>
> then I get an ugly assertion failure:
>
> Assertion failed: (InitVal->getType() == getType()->getElementType()
&&
> "Initializer type must match GlobalVariable type"), function
setInitializer,
> file /Users/jholewinski/projects/llvm/src/llvm/lib/VMCore/Globals.cpp, line
> 168.
>
> Looking at the debugger, I see that InitVal (the initializer for @a2) has
> type i8*, which looks incorrect as it drops the addrspace from the
> definition of @a.
>
> Before I go too far into this, should this IR work?  It looks like the
> parser is failing to properly process the forward declaration of @a, but I
> just want to be sure.
Looks like it should work.

-Eli

Does not work and same error even on old r157485 build.

- D.

2012/11/10 Eli Friedman <eli.friedman at
gmail.com>:
> On Fri, Nov 9, 2012 at 5:01 PM, Justin Holewinski
> <justin.holewinski at gmail.com> wrote:
>> I'm experiencing a weird issue during .ll file parsing, and I'm
not sure if
>> it's a bug in the .ll parser, or if I'm just not understanding
the IR.
>>
>> Take the following IR:
>>
>> @a = addrspace(1) global i8 0
>> @a2 = global i8 addrspace(1)* @a
>>
>> This parses fine.  But if I rearrange the statements to:
>>
>> @a2 = global i8 addrspace(1)* @a
>> @a = addrspace(1) global i8 0
>>
>> then I get an ugly assertion failure:
>>
>> Assertion failed: (InitVal->getType() ==
getType()->getElementType() &&
>> "Initializer type must match GlobalVariable type"), function
setInitializer,
>> file /Users/jholewinski/projects/llvm/src/llvm/lib/VMCore/Globals.cpp,
line
>> 168.
>>
>> Looking at the debugger, I see that InitVal (the initializer for @a2)
has
>> type i8*, which looks incorrect as it drops the addrspace from the
>> definition of @a.
>>
>> Before I go too far into this, should this IR work?  It looks like the
>> parser is failing to properly process the forward declaration of @a,
but I
>> just want to be sure.
>
> Looks like it should work.
>
> -Eli
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev