On Fri, Oct 9, 2015 at 3:42 PM, Steve Ellcey <sell...@imgtec.com> wrote:
> On Fri, 2015-10-09 at 15:36 -0700, H.J. Lu wrote:
>
>>
>> I am not sure what you were asking. I tried:
>>
>> make check-g++ RUNTESTFLAGS="--target_board='unix{-m32\
>> -mstackrealign}' old-deja.exp=spec*.C"
>> ...
>>
>> === g++ Summary ===
>>
>> # of expected passes 495
>> # of expected failures 3
>>
>> make check-g++ RUNTESTFLAGS="--target_board='unix{-m32\
>> -mstackrealign}' old-deja.exp=eh*.C"
>> ...
>> === g++ Summary ===
>>
>> # of expected passes 372
>
> OK, that was what I was wondering about. I wasn't sure if you had run
> the entire test suite with -mstackrealign or only the tests in
> g++.dg/torture/stackalign.
>
When we were developing dynamic stack realignment, we defaulted
GCC itself to always realign stack. Tests under g++.dg/torture/stackalign
were extra failures when stack was realigned. We added those tests
to make sure that stack realignment won't regress.
You can checkout svn branches/stack branch to see what we
did.
--
H.J.
