[llvm-commits] Please review: /llvm/trunk: include/llvm/CodeGen/DFAPacketizer.h lib/CodeGen/DFAPacketizer.cpp
Sirish Pande
spande at codeaurora.org
Mon Apr 30 07:48:10 PDT 2012
Andy,
Could you please review the patch and let me know if it's good to be
committed? This patch is a pre-req for the rest of the Hexagon patches.
ping :-)
Thanks.
Sirish
On 4/25/2012 1:36 PM, Sirish Pande wrote:
> Hi Andy,
>
> I tried using MISUnits map but it requires more work than I estimated
> - given that we use the map created in DFAPacketizer.cpp is used
> extensively in our HexagonVLIWPacketizer.cpp.
> So, I was wondering if I could just commit the patch with the map
> created in DFAPacketizer.cpp. It still incorporates all the other
> suggestions that you have made.
>
> I will have another patch down the line that will use your mapping
> functionality. I think another patch is good place for it because that
> patch will have some rework of DFAPacketizer as ScheduleDAGInstrs.h is
> in the include directory now. In the rework of DFAPacketizer, we
> don't need to create an instance of DefaultVLIWScheduler in the ctor
> of VLIWPacketizerList. That will make things much simpler.
>
> Let me know what you think. Also I have attached for your review.
>
> Sirish
>
> On 4/24/2012 3:49 PM, Andrew Trick wrote:
>> On Apr 24, 2012, at 11:43 AM, Sirish Pande<spande at codeaurora.org>
>> wrote:
>>
>>> I have attached Target Independent fix required for Hexagon VLIW
>>> packetizer.
>>>
>>> Like Andrew suggested, it does not have any warnings, make check-all
>>> passes all test cases for *all* targets.
>>>
>>> I have also incorporated Andrew's suggestions like moving exitRegion
>>> after packetization, using finishBlock and using map mi to su
>>> functionality.
>>>
>>> Please review. And, if there are any questions or comments, please
>>> let me know.
>> Thanks for copying me.
>>
>> I don't understand why you need to call initSUnits. That's supposed
>> to be a helper for buildSchedGraph. Any reason you can't use the
>> MISUnits map that it creates?
>>
>> If you can just remove that call and everything works, then go ahead
>> and commit.
>>
>> -Andy
>>
>>>
>>> --
>>> Qualcomm Innovation Center, Inc is a member of Code Aurora Forum
>>>
>>> <HexagonPacketizerTargetIndependentFix.patch>
>
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
--
Qualcomm Innovation Center, Inc is a member of Code Aurora Forum
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120430/e1c7eff7/attachment.html>
More information about the llvm-commits
mailing list