[LLVMdev] question about enabling cfl-aa and collecting a57 numbers (original) (raw)

Hal Finkel hfinkel at anl.gov
Fri Jan 30 07:56:58 PST 2015

----- Original Message -----

From: "George Burgess IV" <george.burgess.iv at gmail.com> To: "Daniel Berlin" <dberlin at dberlin.org> Cc: "Chandler Carruth" <chandlerc at google.com>, "Hal Finkel" <hfinkel at anl.gov>, "Jiangning Liu" <Jiangning.Liu at arm.com>, "LLVM Developers Mailing List" <llvmdev at cs.uiuc.edu> Sent: Friday, January 30, 2015 8:15:55 AM Subject: Re: [LLVMdev] question about enabling cfl-aa and collecting a57 numbers

I'm not exactly thrilled about the size of this diff -- I'll happily break it up into more manageable bits later today, because some of it is test fixes, another bit is a minor bug fix, etc.

Yes, please break it into independent parts.

Important bit (WRT ConstantExpr): moved the loop body from buildGraphFrom into a new function. The body has a few tweaks to call constexprToEdges on all ConstantExprs that we encounter. constexprToEdges, naturally, interprets a ConstantExpr (and all nested ConstantExprs) and places the results into a SmallVector. I'm assuming this method of handling ConstantExprs isn't 100% correct because I was told that handling them correctly would be more difficult than I think it is. I can't quite figure out why, so examples of cases that break my code would be greatly appreciated. :)

I had thought that the case that Danny had looked at had a constant GEP, and so this constant might alias with other global pointers. How is that handled now?

Thanks again, Hal

George On Mon, Jan 26, 2015 at 2:43 PM, George Burgess IV <_ _george.burgess.iv at gmail.com > wrote:

Inline George On Jan 26, 2015, at 1:05 PM, Daniel Berlin < dberlin at dberlin.org > wrote: George, given that, can you just build constexpr handling (it's not as easy as you think) as a separate funciton and have it use it in the right places? Will do. :)

FWIW, my current list of CFLAA issues is: 1. Unknown values (results from ptrtoint, incoming pointers, etc) are not treated as unknown. These should be done through graph edge (so that they can be one way, otherwise, you will unify everything :P) 2. Constexpr handling ^^^ These are correctness issues. I'm pretty sure there are a few more but i haven't finished auditing 3. In a number of places we treat non-pointers as memory-locations and unify them with pointers. This introduces a lot of spurious aliasing. 4. More generally, we induce a lot of spurious aliasing through things at different dereference levels. In these cases, one may to the other, but, for example, if we have a foo***, and a foo* (and neither pointers to unknown things or escapes), the only way for foo *** to alias foo* is if there is a graph path with two dereferences between them. We seem to get this wrong sometimes. Agreed on all four. Though naturally it should be fixed, I’d like to see how much of an issue #4 ends up being when we properly deal with #3.

On Sun Jan 25 2015 at 6:44:07 PM Chandler Carruth <_ _chandlerc at google.com > wrote: On Sun, Jan 25, 2015 at 6:37 PM, George Burgess IV <_ _george.burgess.iv at gmail.com > wrote: > Fixing that still gives a wrong result, i haven't started to track > down what else is going on here. Running with the attached diff + a modified buildGraphFrom to handle the constexpr GEPs, we seem to flag everything in test2.ll (conservatively) correctly. Is store the only place we can expect to see these constexpr analogs, or is just about anywhere fair game? Any Value can be a ConstantExpr, so all operands to instructions are fair game.

-- Hal Finkel Assistant Computational Scientist Leadership Computing Facility Argonne National Laboratory

More information about the llvm-dev mailing list