This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH][RFC] Make expansion of balanced binary trees of switches on tree level.
On 08/29/2017 01:42 PM, Richard Biener wrote:
> On Thu, Aug 24, 2017 at 5:35 PM, Martin Liška <mliska@suse.cz> wrote:
>> On 08/18/2017 12:25 PM, Martin Liška wrote:
>>> On 08/18/2017 11:30 AM, Richard Biener wrote:
>>>> On Tue, Aug 15, 2017 at 2:37 PM, Martin Liška <mliska@suse.cz> wrote:
>>>>> On 08/14/2017 10:32 AM, Richard Biener wrote:
>>>>>> Hmm, but the existing "lowering" part is called from the
>>>>>> switch-conversion pass. So
>>>>>> I'm not sure a new file is good.
>>>>>
>>>>> Good, I'm not against having that in a single file. So new version of the patch
>>>>> does that.
>>>>>
>>>>> Patch can bootstrap on ppc64le-redhat-linux and survives regression tests.
>>>>>
>>>>> Ready to be installed?
>>>>
>>>> Hmm, I see you duplicate add_case_node for example. Is that just temporary?
>>>> If not can you please factor out the data structure and common code?
>>>> (case.[Ch]?)
>>>
>>> You are right. As we'll generate just jump table in stmt.c the proper fix is to remove
>>> all usages of 'case_node' in the file because simple iteration of labels will work fine.
>>> Let me do it incrementally to minimize fall out :)
>>
>> Hello.
>>
>> So lesson learned. I should follow your recommendation and make the clean-up in stmt.c. I didn't
>> so adding new variant of case_node with a different size caused bootstrap failure on aarch64 and
>> it was quite hard to debug. So sending updated version of the patch which has cleaned up stmt.c.
>>
>> Patch can bootstrap on ppc64le-redhat-linux and survives regression tests. Same of aarch64-linux-gnu.
>>
>> Ready to be installed?
>
> No ChangeLog entry for tree-switch-conversion.c? At least you added
> make_pass_lower_switch
> and friends.
Hi.
Yes, sorry for the missing entries. Fixed and installed as r251412.
Hopefully fall out will be small.
Martin
>
> Ok with a little more verbose changelog.
>
> Thanks and sorry for the delay,
> Richard.
>
>> Martin
>>>
>>> Martin
>>>
>>>>
>>>> Thanks,
>>>> Richard.
>>>>
>>>>> Martin
>>>
>>