[PATCH] SubRegister liveness tracking

[Sean Silva]

On Mon, Oct 7, 2013 at 10:14 PM, Matthias Braun <mbraun at apple.com> wrote:

> fixed.
>


That was not the only instance.

+  // register unit lane mask lists.

+      // we have subranges => reg must have subregs => regunits must have
masks

(I'm sure there are others as well)

Please audit the patch set for other cases where this happens.

-- Sean Silva


>
>
> On Oct 7, 2013, at 6:30 PM, Sean Silva <silvas at purdue.edu> wrote:
>
> +        // when tracking subregister liveness the main range must start
> new
> +        // values on partial register writes, even if there is no read.
>
> <http://llvm.org/docs/CodingStandards.html#commenting> "When writing
> comments, write them as English prose, which means they should use proper
> capitalization, punctuation, etc."
>
> -- Sean Silva
>
>
> On Mon, Oct 7, 2013 at 5:47 PM, Matthias Braun <mbraun at apple.com> wrote:
>
>> Let’s try this again. Hopefully without empty patches this time…
>>
>>
>>
>> On Oct 7, 2013, at 2:03 PM, Matthias Braun <mbraun at apple.com> wrote:
>>
>> > Here is the full patchset to enable subregister liveness tracking in
>> llvm. I did not split all these patches into separate mails because they
>> have to be applied in order and they only really make sense when applied
>> together. For a high-level discussion of the changes, see the mail I sent
>> to the llvmdev mailing list.
>> > The patches have to be applied after applying the set of cleanups here:
>> http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130902/187212.html
>> > and the last two small patches for ‘covers’ and ‘advanceTo’.
>> >
>> > Greetings
>> >       Matthias
>> >
>> >
>> <0001-Add-support-to-track-liveness-of-subregisters-in-a-L.patch><0002-Compute-subregister-ranges-in-LiveIntervalAnalysis.patch><0003-Update-SubRanges-in-LiveIntervalAnalysis-shrinkToUse.patch><0004-Adapt-LiveIntervalAnalysis-handleMove-for-subregiste.patch><0005-Adapt-LiveRangeEdit-eliminateDeadDef-for-subregister.patch><0006-Adapt-LiveIntervalAnalysis-repairIntervalsInRange-to.patch><0007-Add-a-flag-to-enable-disable-subregister-liveness.patch><0008-Introduce-LiveQuery-accessor-for-dead-or-live-out-va.patch><0009-Add-subregister-aware-variants-of-LiveIntervalAnalys.patch><0010-Make-subrange-shrinkToUse-variant-public.patch><0011-Preserve-subregister-liveranges-in-register-coalesce.patch><0012-Tablegen-erate-lanemasks-for-register-units.patch><0013-LiveIntervalUnion-allow-specification-of-liverange-w.patch><0014-Respect-subregister-liveness-when-allocating-registe.patch><0015-Do-not-add-implicit-defs-uses-when-tracking-subregis.patch><0016-MachineVerifier-Allow-LiveInterval-segments-to-end-a.patch><0017-MachineVerifier-allow-physreg-use-if-just-a-subreg-i.patch><0018-Enable-subreg-liveness-on-ARM.patch>_______________________________________________
>> > llvm-commits mailing list
>> > llvm-commits at cs.uiuc.edu
>> > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20131009/6568a330/attachment.html>