This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH][RFC] Sanity checking for -freorder-blocks-and-partition failures


Hello Teresa,

It seems to me that it's better if you commit it along with your set
of fixes. My patch doesn't fix any bugs, it just exposes them :-)

Ciao!
Steven



On Fri, Nov 9, 2012 at 9:09 PM, Teresa Johnson <tejohnson@google.com> wrote:
>
> Hi Steven,
>
> I've spent this week trying to clean up all the issues exposed by this new verification patch. Some of the issues I described in the email thread on my related patch (http://gcc.gnu.org/ml/gcc-patches/2012-11/msg00287.html) and earlier in this thread. It also exposed more issues of the type described in the last message regarding my patch (the link I included here), where transformations were being applied but the partitions not being correctly fixed up. Things look clean now across SPEC2006 int C benchmarks at peak, gcc regression tests and our internal benchmarks. I need to update from head, retest and clean things up though before sending the new patch. But do you want to go ahead and commit this patch? I guess it should be fine to commit asynchronously with mine since -freorder-blocks-and-partition is off by default and not working anyway. I assume it can still go in since it was proposed already and is related to some outstanding bugs?
>
> Thanks,
> Teresa
>
>
> On Wed, Oct 31, 2012 at 1:06 PM, Teresa Johnson <tejohnson@google.com> wrote:
>>
>> On Wed, Oct 31, 2012 at 12:58 PM, Christophe Lyon
>> <christophe.lyon@st.com> wrote:
>> > On 30.10.2012 17:59, Teresa Johnson wrote:
>> >>
>> >> On Tue, Oct 30, 2012 at 9:26 AM, Steven Bosscher <stevenb.gcc@gmail.com>
>> >> wrote:
>> >>>
>> >>> Hello,
>> >>>
>> >>> Hot/cold partitioning is apparently a hot topic all of a sudden, which
>> >>> is a good thing of course, because it's in need of some TLC.
>> >>>
>> >>> The attached patch adds another check the RTL cfg checking
>> >>> (verify_flow_info) for the partitioning: A hot block can never be
>> >>> dominated by a cold block (because the dominated block must also be
>> >>> cold). This trips in PR55121.
>> >>>
>> >>> I haven't tested this with any profiling tests, but it's bound to
>> >>> break things. From my POV, whatever gets broken by this patch was
>> >>> already broken to begin with :-)   If you're in CC, it's because I
>> >>> hope you can help test this patch.
>> >>
>> >> I will try testing your patch on top of mine with our fdo benchmarks.
>> >> For the others on the cc list, you may need to include my patch as
>> >> well for testing. Without it, -freorder-blocks-and-partition was DOA
>> >> for me. For my patch, see
>> >> http://gcc.gnu.org/ml/gcc-patches/2012-10/msg02692.html
>> >>
>> >> Teresa
>> >>
>> > I have tried Steven's patch an indeed it reported numerous errors while
>> > compiling spec2k.
>> >
>> > I tried Teresa's patch too, but it changed nothing in my tests. The patches
>> > already posted by Matt are still necessary and Teresa's patch does not
>> > improve my tests.
>>
>> With checking enabled I am seeing additional failures that my fixes
>> are not addressing. Looking into those now.
>> Can someone point me to Matt's patches? Is that this one:
>> http://gcc.gnu.org/ml/gcc-patches/2012-09/msg00274.html
>> or are there others?
>>
>> Thanks,
>> Teresa
>>
>> >
>> > I am out of office at the moment, so it's a little bit inconvenient to
>> > investigate deeper the reasons for all the errors reported by Steven's
>> > patch. Anyway it looks like it's really needed :)
>> > I also noticed that some compilations failed with an ICE in calc_dfs_tree at
>> > dominance.c:395.
>> >
>> >
>> > Christophe.
>> >
>>
>>
>>
>> --
>> Teresa Johnson | Software Engineer | tejohnson@google.com | 408-460-2413
>
>
>
>
> --
> Teresa Johnson | Software Engineer |  tejohnson@google.com |  408-460-2413
>


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]