llvm dev - Mar 2017 - Well-formed @llvm.lifetime.start and @llvm.lifetime.end intrinsics

2017-03-31 0:19 GMT+02:00 Daniel Berlin <dberlin at
dberlin.org>:
>> I modified Polly's code generator to produce
>>
>>     if (c) {
>>       llvm.lifetime.start(&var)
>>     } else {
>>       llvm.lifetime.start(&var)
>>     }
>>     [...]
>>     llvm.lifetime.end(&var)
>>
>> and it does not miscompile anymore. However, Polly is currently is not
>> able to compute the lifetime in the general case.
>
>
> Of course, nothing is.
> But they are also metadata, so you could just drop them if you wanted.
This is actually exactly what Polly did, but only in the optimized
side of the loop versioning. The original code version is not touched,
which keeps the llvm.lifetime.start in it.

So of the bug is on Polly's side, I would "fix" it by deleting all
lifetime markers also in the original code.

Michael

On Thu, Mar 30, 2017 at 3:45 PM, Michael Kruse <llvmdev at meinersbur.de>
wrote:
> 2017-03-31 0:19 GMT+02:00 Daniel Berlin <dberlin at dberlin.org>:
> >> I modified Polly's code generator to produce
> >>
> >>     if (c) {
> >>       llvm.lifetime.start(&var)
> >>     } else {
> >>       llvm.lifetime.start(&var)
> >>     }
> >>     [...]
> >>     llvm.lifetime.end(&var)
> >>
> >> and it does not miscompile anymore. However, Polly is currently is
not
> >> able to compute the lifetime in the general case.
> >
> >
> > Of course, nothing is.
> > But they are also metadata, so you could just drop them if you wanted.
>
> This is actually exactly what Polly did, but only in the optimized
> side of the loop versioning. The original code version is not touched,
> which keeps the llvm.lifetime.start in it.
>
> So of the bug is on Polly's side, I would "fix" it by
deleting all
> lifetime markers also in the original code.
>
I'm curious, btw, what made it think the above is legal.
The intrinsics are marked as touching memory, having side effects, etc.

Do you guys specially ignore them or something?
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20170330/2e2f21e9/attachment.html>

(in particular, those things keep LICM from moving them)

On Thu, Mar 30, 2017 at 3:54 PM, Daniel Berlin <dberlin at dberlin.org>
wrote:
>
>
> On Thu, Mar 30, 2017 at 3:45 PM, Michael Kruse <llvmdev at
meinersbur.de>
> wrote:
>
>> 2017-03-31 0:19 GMT+02:00 Daniel Berlin <dberlin at dberlin.org>:
>> >> I modified Polly's code generator to produce
>> >>
>> >>     if (c) {
>> >>       llvm.lifetime.start(&var)
>> >>     } else {
>> >>       llvm.lifetime.start(&var)
>> >>     }
>> >>     [...]
>> >>     llvm.lifetime.end(&var)
>> >>
>> >> and it does not miscompile anymore. However, Polly is
currently is not
>> >> able to compute the lifetime in the general case.
>> >
>> >
>> > Of course, nothing is.
>> > But they are also metadata, so you could just drop them if you
wanted.
>>
>> This is actually exactly what Polly did, but only in the optimized
>> side of the loop versioning. The original code version is not touched,
>> which keeps the llvm.lifetime.start in it.
>>
>> So of the bug is on Polly's side, I would "fix" it by
deleting all
>> lifetime markers also in the original code.
>>
>
> I'm curious, btw, what made it think the above is legal.
> The intrinsics are marked as touching memory, having side effects, etc.
>
> Do you guys specially ignore them or something?
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20170330/d27549bc/attachment.html>

2017-03-31 0:54 GMT+02:00 Daniel Berlin <dberlin at
dberlin.org>:
> I'm curious, btw, what made it think the above is legal.
> The intrinsics are marked as touching memory, having side effects, etc.
The precedes my involvement with Polly.

Polly has a list of intrinsics that can be safely ignored, e.g.
llvm.dbg.value/llvm.gbg.declare. The lifetime markers are also in this
list. Ignored intrinsic are ... well ... ignored when generating the
optimized code.

Michael