Re: Bug #70805 (Segmentation faults whilst running Drupal 8 test suite)
| From: | Paul Dragoonis | Date: | Wed, 04 Nov 2015 10:27:47 +0000 |
| Subject: | Re: Bug #70805 (Segmentation faults whilst running Drupal 8 test suite) | ||
| References: | 1 2 3 4 | Groups: | php.internals |
| Request: | Send a blank email to internals+get-89067@lists.php.net to get a copy of this message | ||
Hey Bob,
Thanks, it was 2am and quite sleepy so wasn't considering the gen'd files.
Thanks for clarifying!
On Wed, Nov 4, 2015 at 2:26 AM, Bob Weinand <bobwei9@hotmail.com> wrote:
> Hey,
>
> zend_vm_execute.h is an auto-generated file, via zend_vm_gen.php. In
> reality the patch only fixes the code in exactly one location
> (zend_vm_def.h) and then regenerated zend_vm_execute.h.
>
> Bob
>
> > Am 04.11.2015 um 03:10 schrieb Paul Dragoonis <dragoonis@gmail.com>:
> >
> > Hey,
> >
> > Looking at the patch, the changes to zend_vm_def.h and zend_vm_execute.h
> > are duplicated in 10 locations. I'm wondering if we can consolidate this
> > into maintainable function/macro to handle this?
> >
> > On Wed, Nov 4, 2015 at 1:58 AM, Xinchen Hui <xinchen.h@zend.com> wrote:
> >
> >> Hey:
> >>
> >>
> >>
> >> On Wed, Nov 4, 2015 at 3:58 AM, Dmitry Stogov <dmitry@zend.com> wrote:
> >>
> >>> Hi,
> >>>
> >>> I think, I found the root problem of
> >> https://bugs.php.net/bug.php?id=70805
> >>>
> >>> unset($a) or unser($GLOBAL["a"]) triggered GC and destructors calls
> that
> >>> tried to release the same global variable $a once again. As result
> it's
> >>> reference counter was decremented twice and this caused use-after-free,
> >>> double-free, etc.
> >>>
> >>> The proposed cumulative fix for all related problems:
> >>>
> >>>
> >>> https://gist.github.com/dstogov/7aa9d24876e2b3fce8c5
> >>>
> >>> Xinchen, could you please review and verify this once again,
> >>> then add necessary tests and commit.
> >>>
> >> No problem, all issues we met are resovled , thanks :)
> >>
> >> tested and committed.
> >>
> >> and aslo thanks the fabian who provides us ssh access to a reproducible
> box
> >> (it's really hard to reproduce locally)
> >>
> >> thanks!
> >>
> >>>
> >>> Thanks. Dmitry.
> >>>
> >>
> >>
> >>
> >> --
> >> Xinchen Hui
> >> @Laruence
> >> http://www.laruence.com/
> >>
>
>