[LLVMdev] BasicAA unable to analyze recursive PHI nodes

Daniel Berlin dberlin at dberlin.org
Wed Jun 10 17:11:02 PDT 2015


On Wed, Jun 10, 2015 at 4:09 PM, Tobias Edler von Koch
<tobias at codeaurora.org> wrote:
> It probably will, but it isn't on by default (right?)

True, but so what?
> and has quite a
> lot more overhead.

I didn't see numbers one way or the other. It probably has some
overhead, if it's the right solution, i'm sure it can be ade faster.

>  I think this is a common enough case for it to be
> handled by BasicAA.
>

Can you provide some evidence?

IE hard numbers of how often it occurs on some set of benchmarks or
something, and what the compile time cost is?

It's really hard to say whether this is the right tradeoff without any data.

> On Wed, 10 Jun 2015 15:52:07 -0700 Sanjoy Das
> <sanjoy at playingwithpointers.com> wrote:
>
>> Sorry for missing this the first time around, but I think -scev-aa
>> does what you want.  I've commented on the phabricator review.
>>
>> On Wed, Jun 10, 2015 at 3:32 PM, Tobias Edler von Koch
>> <tobias at codeaurora.org> wrote:
>> > On Tue, 2 Jun 2015 11:32:13 -0500 Tobias Edler von Koch
>> > <tobias at codeaurora.org> wrote:
>> >
>> >> Hi all,
>> >>
>> >> I came across the following limitation in our BasicAliasAnalysis. This
>> >> happens with the following IR pattern:
>> >>
>> >>   %x = phi [ %incptr, ... ] [ %var, ... ]
>> >>   %incptr = getelementptr %x, 1
>> >>
>> >> We will basically always return MayAlias for %x and any other value
>> >> because aliasPHI recurses on the first value and gives up.
>> >
>> > Patch now posted as http://reviews.llvm.org/D10368
>> >
>> > Tobias
>> >
>> > --
>> > Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
>> > Linux Foundation Collaborative Project.
>> > _______________________________________________
>> > LLVM Developers mailing list
>> > LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
>
> --
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
> Linux Foundation Collaborative Project.
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev



More information about the llvm-dev mailing list