[LLVMdev] question about enabling cfl-aa and collecting a57 numbers

Daniel Berlin dberlin at dberlin.org
Wed Jan 21 13:48:25 PST 2015


On Wed Jan 21 2015 at 12:30:50 PM Hal Finkel <hfinkel at anl.gov> wrote:

> ----- Original Message -----
> > From: "Daniel Berlin" <dberlin at dberlin.org>
> > To: "Hal Finkel" <hfinkel at anl.gov>
> > Cc: "Jiangning Liu" <Jiangning.Liu at arm.com>, "George Burgess IV" <
> george.burgess.iv at gmail.com>, "LLVM Developers
> > Mailing List" <llvmdev at cs.uiuc.edu>, "Nick Lewycky" <nlewycky at google.com
> >
> > Sent: Wednesday, January 21, 2015 1:10:07 PM
> > Subject: Re: [LLVMdev] question about enabling cfl-aa and collecting a57
> numbers
> >
> > Updated testcases to have MayAlias/note issues as FIXME.
> >
>
> Okay, thanks! This LGTM, but we should probably split the delegation fixes
> from the others and commit as two separate patches (especially because Ana
> noted some potential miscompiles caused by the other improvements).
>

I think she mentioned the miscompiles due to us returning partialalias.
But in any case, i'm happy to, but just to note they are all required to
get the LICM issue fixed :)


>
> Regarding this:
>
> @@ -768,7 +774,10 @@ static Optional<StratifiedAttr>
> valueToAttrIndex(Value *Val) {
>      return AttrGlobalIndex;
>
>    if (auto *Arg = dyn_cast<Argument>(Val))
> -    if (!Arg->hasNoAliasAttr())
> +    // Only pointer arguments should have the argument attribute,
> +    // because things can't escape through scalars without us seeing a
> +    // cast, and thus, interaction with them doesn't matter.
> +    if (!Arg->hasNoAliasAttr() && Arg->getType()->isPointerTy())
>        return argNumberToAttrIndex(Arg->getArgNo());
>    return NoneType();
>  }
>
> when we do see the inttoptr case, we add an edge from the source to the
> destination.


Correct.


> If we've not noted potential aliasing of the non-pointer-typed argument,
> then does this end up looking like a unique global?
>

No. It will end up looking like something that points to nothing.
Even without this change, it will end up looking like something that points
to nothing, it will just have an attribute that says "argument". :)

You can come up with cases where even with this attribute set, it will get
the wrong answer. It just happens to have code that, through luck, gets the
right answer in a lot of cases:
  (That is this code:

 if (AttrsA.any() && AttrsB.any())
     return AliasAnalysis::MayAlias;
)

So there is a bug here, but it's not caused by this code.

The bug here is that we can't ever know what happens as the result of
inttoptr.  We never do math, and the tracking we do is never going to be
sufficient to determine the range of possible pointers for an inttoptr in
all cases (in theory, it could point to anything anywhere in the program.
If we knew the sizes of *all* objects, and any binary operator performed on
it was evaluable, we could do a little better. If we knew the value came
from a ptrtoint, we could do better, etc).
Same with ptrtoint.

The result of both of these instructions should start to be "we have no
idea what the pointer that comes from inttoptr or goes to ptrtoint points
to", and we should return mayalias for anything that interacts with them.
We don't do that right now.
We are just hiding it mildly well.


> Speaking of which, the code has checks for global variables in several
> places. Do these need to be for globals that are not aliases and don't have
> weak linkage?
>

It's more a question of whether they are in SSA form than if they are
globals.

It's effectively using Globals/Arguments as a way to say "don't know" in
some cases, where it should really just say "don't know".

There is a bunch of code i now have marked for cleanup and bugfixes around
these issues (constant vs global handling, handling of non-pointer values,
etc).

As mentioned, the above is necessary to fix the LICM issue (and is correct,
even if somewhere else is wrong. For reference, GCC does the identical
thing to what i'm saying :P), but i'm happy to move it to a separate fix
(that includes fixes for the other argument/unknown related issues) if you
like.




> Thanks again,
> Hal
>
> >
> >
> >
> >
> > On Tue Jan 20 2015 at 3:54:10 PM Hal Finkel < hfinkel at anl.gov >
> > wrote:
> >
> >
> > ----- Original Message -----
> > > From: "Daniel Berlin" < dberlin at dberlin.org >
> > > To: "Hal Finkel" < hfinkel at anl.gov >
> > > Cc: "Jiangning Liu" < Jiangning.Liu at arm.com >, "George Burgess IV"
> > > < george.burgess.iv at gmail.com >, "LLVM Developers
> > > Mailing List" < llvmdev at cs.uiuc.edu >, "Nick Lewycky" <
> > > nlewycky at google.com >
> > > Sent: Tuesday, January 20, 2015 1:48:44 PM
> > > Subject: Re: [LLVMdev] question about enabling cfl-aa and
> > > collecting a57 numbers
> > >
> > > So, I can make all these testcases work, but it's a little tricky
> > > (it
> > > involves tracking some things, like GEP byte range, and then
> > > checking bases and using getObjectSize, much like BasicAA does).
> > >
> > >
> > > Because i really don't want to put that much "not well tested" code
> > > in a bugfix, and honestly, i'm not sure we will catch any cases
> > > here
> > > that BasicAA does not, i've attached a change to XFAIL these
> > > testcases, and updated the code to return MayAlias.
> >
> > Okay. I think you might as well just update the test cases to want
> > MayAlias, and put a FIXME comment explaining that they could be
> > PartialAlias. As far as I know, there is no code in LLVM that really
> > handles a PartialAlias differently than a MayAlias or MustAlias, and
> > so while there may be some benefit here, I'm not sure it will be
> > worth the effort.
> >
> > >
> > > I will build and test a patch to get these back to PartialAlias,
> > > but
> > > this patch will at least get us to not be "giving wrong answers". I
> > > will also see if we catch anything with it that BasicAA does not,
> > > because if we don't, it doesn't seem worth it to me.
> >
> > My guess is that BasicAA will get almost all of the interesting
> > PartialAlias cases, and as I said, we essentially ignore them
> > anyway.
> >
> > As a side note, there is this one place in lib/Analysis/
> > MemoryDependenceAnalysis.cpp that could use some attention:
> >
> > #if 0 // FIXME: Temporarily disabled. GVN is cleverly rewriting loads
> > // in terms of clobbering loads, but since it does this by looking
> > // at the clobbering load directly, it doesn't know about any
> > // phi translation that may have happened along the way.
> >
> > // If we have a partial alias, then return this as a clobber for the
> > // client to handle.
> > if (R == AliasAnalysis::PartialAlias)
> > return MemDepResult::getClobber(Inst) ;
> > #endif
> >
> > >
> > >
> > > Conservative new patch attached.
> > >
> > >
> > >
> > > (Note that i still updated the testcases, because we will *never*
> > > be
> > > able to legally return PartialAlias as they were written)
> > >
> >
> > Yes, sounds good.
> >
> > -Hal
> >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > On Sun Jan 18 2015 at 2:12:47 PM Daniel Berlin <
> > > dberlin at dberlin.org
> > > > wrote:
> > >
> > >
> > >
> > > On Sat Jan 17 2015 at 3:15:27 PM Hal Finkel < hfinkel at anl.gov >
> > > wrote:
> > >
> > >
> > > ----- Original Message -----
> > > > From: "Daniel Berlin" < dberlin at dberlin.org >
> > > > To: "Hal Finkel" < hfinkel at anl.gov >
> > > > Cc: "Jiangning Liu" < Jiangning.Liu at arm.com >, "George Burgess
> > > > IV"
> > > > < george.burgess.iv at gmail.com >, "LLVM Developers
> > > > Mailing List" < llvmdev at cs.uiuc.edu >, "Nick Lewycky" <
> > > > nlewycky at google.com >
> > > > Sent: Saturday, January 17, 2015 1:08:10 PM
> > > > Subject: Re: [LLVMdev] question about enabling cfl-aa and
> > > > collecting a57 numbers
> > > >
> > > >
> > > >
> > > >
> > > > On Sat Jan 17 2015 at 12:03:33 AM Hal Finkel < hfinkel at anl.gov >
> > > > wrote:
> > > >
> > > >
> > > > Hi Danny,
> > > >
> > > > // Add TypeBasedAliasAnalysis before BasicAliasAnalysis so that
> > > > // BasicAliasAnalysis wins if they disagree. This is intended to
> > > > help
> > > > // support "obvious" type-punning idioms.
> > > > - if (UseCFLAA)
> > > > - addPass( createCFLAliasAnalysisPass());
> > > > addPass( createTypeBasedAliasAnalysisPa ss());
> > > > addPass( createScopedNoAliasAAPass());
> > > > + if (UseCFLAA)
> > > > + addPass( createCFLAliasAnalysisPass());
> > > > addPass( createBasicAliasAnalysisPass() );
> > > >
> > > > Do we really want to change the order here? I had originally
> > > > placed
> > > > it after the metadata-based passes thinking that the compile-time
> > > > would be better (guessing that the metadata queries would be
> > > > faster
> > > > than the CFL queries, so if the metadata could quickly return a
> > > > NoAlias, then we'd cut out unecessary CFL queries). Perhaps this
> > > > is
> > > > an irrelevant effect, but we should have some documented
> > > > rationale.
> > > >
> > > >
> > > >
> > > > Yeah, this was a mistake (Chandler had suggested it was right
> > > > earlier, but we were both wrong :P)
> > > >
> > > >
> > > >
> > > > ; CHECK: PartialAlias: i16* %bigbase0, i8* %phi
> > > > -define i8 @test0(i8* %base, i1 %x) {
> > > > +define i8 @test0(i1 %x) {
> > > > entry:
> > > > + %base = alloca i8, align 4
> > > > %baseplusone = getelementptr i8* %base, i64 1
> > > > br i1 %x, label %red, label %green
> > > > red:
> > > > @@ -25,8 +26,9 @@ green:
> > > > }
> > > >
> > > > why should this return PartialAlias? %ohi does partially overlap,
> > > > so
> > > > this correct, but what happens when the overlap is partial or
> > > > control dependent?
> > > > So, after talking with some people offline, they convinced me in
> > > > SSA
> > > > form, the name would change in these situations, and the only
> > > > situations you can get into trouble is with things "based on
> > > > pointer
> > > > arguments" (because you have no idea what their initial state
> > > > is),
> > > > or "globals" (because they are not in SSA form)
> > > > I could not come up with a case otherwise
> > >
> > > Okay; that part of the code could really use some more commentary.
> > > I'd really appreciate it if you should put these thoughts in
> > > written
> > > form that could be added as comments.
> > >
> > >
> > >
> > >
> > >
> > > Will do
> > >
> > >
> > >
> > > > But i'm welcome to hear if you think this is wrong.
> > > >
> > > > FWIW: I bootstrapped/tested the compiler with an assert that
> > > > triggered if CFL-AA was going to return PartialAlias and BasicAA
> > > > would have returned NoAlias, and it did not trigger with this
> > > > change.
> > > >
> > > >
> > > > (It would have triggered before this set of changes)
> > > >
> > > > Not proof of course, but it at least tells me it's not "obviously
> > > > wrong" :)
> > > >
> > > >
> > >
> > > That's good :) -- but, not exactly what I was concerned about. Our
> > > general convention has been that PartialAlias is a "strong" result,
> > > like MustAlias, but implies that AA has proved that only a partial
> > > overlap will occur.
> > >
> > > So, in this test case we get the right result:
> > >
> > > ; CHECK: PartialAlias: i16* %bigbase0, i8* %phi
> > > define i8 @test0(i1 %x) {
> > > entry:
> > > %base = alloca i8, align 4
> > > %baseplusone = getelementptr i8* %base, i64 1
> > > br i1 %x, label %red, label %green
> > > red:
> > > br label %green
> > > green:
> > > %phi = phi i8* [ %baseplusone, %red ], [ %base, %entry ]
> > > store i8 0, i8* %phi
> > >
> > > %bigbase0 = bitcast i8* %base to i16*
> > > store i16 -1, i16* %bigbase0
> > >
> > > %loaded = load i8* %phi
> > > ret i8 %loaded
> > > }
> > >
> > > because %phi will have a partial overlap with %bigbase0, the only
> > > uncertainty is whether the overlap is with the low byte or the high
> > > byte. But if I modify the test to be this:
> > >
> > > define i8 @test0x(i1 %x) {
> > > entry:
> > > %base = alloca i8, align 4
> > > %baseplustwo = getelementptr i8* %base, i64 2
> > > br i1 %x, label %red, label %green
> > > red:
> > > br label %green
> > > green:
> > > %phi = phi i8* [ %baseplustwo, %red ], [ %base, %entry ]
> > > store i8 0, i8* %phi
> > >
> > > %bigbase0 = bitcast i8* %base to i16*
> > > store i16 -1, i16* %bigbase0
> > >
> > > %loaded = load i8* %phi
> > > ret i8 %loaded
> > > }
> > >
> > > I still get this result:
> > > PartialAlias: i16* %bigbase0, i8* %phi
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > but now %phi might not overlap %bigbase0 at all (although, when it
> > > does, there is a partial overlap), so we should just return
> > > MayAlias
> > > (perhaps without delegation because this is a definitive result?).
> > >
> > >
> > >
> > >
> > > Yeah, i have to do some size checking, let me see if we have the
> > > info
> > > and i'll update the patch.
> > >
> > >
> > >
> > >
> > > Otherwise, my view is that we should always delegate MayAlias,
> > > because we have no idea what order the passes are in or what pass
> > > someone has inserted where :)
> > >
> > >
> > > (WIW: I believe the same about everything except MustAlias and
> > > NoAlias, but currently we don't delegate PartialAlias.
> > > We claim PartialAlias is a definitive result, but it really isn't.
> > > Right now we have TBAA that will give NoAlias results on things
> > > other
> > > passes claim are PartialAlias, and that result is correct. That's
> > > just in our default, we have no idea what other people do. Even if
> > > you ignore TBAA, plenty of other compilers have noalias/mustalias
> > > metadata that would have the same effect.
> >
> > --
> > Hal Finkel
> > Assistant Computational Scientist
> > Leadership Computing Facility
> > Argonne National Laboratory
> >
>
> --
> Hal Finkel
> Assistant Computational Scientist
> Leadership Computing Facility
> Argonne National Laboratory
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20150121/e6d972e4/attachment.html>


More information about the llvm-dev mailing list