[LLVMdev] How to make Polly ignore some non-affine memory accesses

[Tobias Grosser]

On 11/21/2011 03:44 AM, Marcello Maggioni wrote:
> 2011/11/21 Tobias Grosser<tobias at grosser.es>:
>> On 11/20/2011 04:36 PM, Marcello Maggioni wrote:
>>>
>>> Sorry for the noobish question, but what kind of subscripts generate a
>>> SCEVCouldNotCompute  from the SCEV engine?
>>> I tried for a while but I wasn't able to trigger that
>>
>> Hi Marcello,
>>
>> the SCEV returns SCEVCouldNotCompute in case it cannot analyze an expression
>> or if the analysis would be to complicated. I am currently
>> not sure if this may actually happen when calling getSCEV(), because
>> getSCEV() could just return a SCEVUnknown referencing the Value itself.
>> Maybe SCEVCouldNotCompute is just generated by functions like
>> SE->getBackedgeTakenCount()?
>>
>> In case you cannot generate a test case that yields to this, I don't think
>> there is a need to try more. Even without such a test case the patch should
>> be OK.
>>
>> Cheers
>> Tobi
>>
>
> Ok, thanks, I'll try other test cases to check if everything is ok,
> better being sure.
>
> About SCOPs with a "bitcast instruction" inside them I see that
> ScopDetection discards those.
> What problems can cause bitcasts to the Scop? Is it related to aliasing?

I don't know. At the moment I wrote the scop detection I was not sure 
that they don't cause any problems, so I conservatively discarded them.
Meanwhile I actually believe the don't introduce any problems and could 
be allowed. However, before changing this, I would like to set up a 
nightly tester that ensures the llvm test suite is compiled correctly.
This would allow us to understand if the support for this construct 
introduces any regressions.

Cheers
Tobi