[LLVMdev] Why always abort in verifyFunction?

Michael.Kang blackfin.kang at gmail.com
Mon Jun 11 19:11:01 PDT 2012


On Mon, Jun 11, 2012 at 5:44 PM, Joey Gouly <joel.gouly at gmail.com> wrote:

> Hi Xiu,
>
> The Verifier pass runs a PreVerifier pass, which does not honour the
> action argument,
> and will always abort on a broken module, (Line 106,
> lib/VMCore/Verifier.cpp)
>

So the argument can not be used as described as the official document? I
just want to make sure  of that and probably we
can find another way to deal with broken module.

Thanks
MK


>
> Perhaps you should file a bug against this, to allow you to not abort if
> you so wish.
>
> Joey
>
> On 11 June 2012 09:41, Guowei Xu <myesis at gmail.com> wrote:
>
>> Hello everyone:
>>
>>         I have a little question about the second argument *action* of
>> verifyFunction.
>>
>>         In docs:
>>
>> *Enumerator: * *AbortProcessAction*
>>
>> verifyModule will print to stderr and abort()
>>  *PrintMessageAction*
>>
>> verifyModule will print to stderr and return true
>>  *ReturnStatusAction*
>>
>> verifyModule will just return true
>>
>>        But it still abort when I pass
>> PrintMessageAction/ReturnStatusAction to function.
>>
>> Log:
>>        Basic Block in function 'jitmain' does not have terminator!
>>        label %load_store
>>        LLVM ERROR: Broken module, no Basic Block terminator!
>>
>>
>>        Is there something wrong?
>>
>>
>> Best regards,
>> -myesis
>>
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>
>>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>


-- 
www.skyeye.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20120612/0ab2f5ba/attachment.html>


More information about the llvm-dev mailing list