Re: [orca-list] the attached file crash at-spi2-registr

[Mike Gorse <mgorse alum wpi edu>]

Hi all,

There were two places that had similar code that was trying to free the 
wrong data, and I had only fixed one of them. I've committed a fix for the 
other instance along with a fix for another problem with deregistering 
applications that I found while running it under valgrind.

-Mike

On Thu, 6 Dec 2012, Josà Vilmar EstÃcio de Souza wrote:

> Hi Mike.
>
> Perhaps I am doing something wrong, but the problem persists in my machine.
>
> I built from master and installed, reboot my machine, repeat the same test 
> and receive the same error.
> I tried three times.
>
> Program received signal SIGABRT, Aborted.
> 0x00007fa236e6efd5 in raise () from /usr/lib/libc.so.6
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x00007f229197a9b2 in malloc_consolidate () from /usr/lib/libc.so.6
>
> Program received signal SIGABRT, Aborted.
> 0x00007ff2ba5a7fd5 in raise () from /usr/lib/libc.so.6
>
> One way to reproduce the error is scroll in the file very quickly, pressing 
> in random way up and down.
> Thanks.
>
>
>
> On 12/06/2012 01:57 PM, Mike Gorse wrote:
> > Hi JosÃ,
> >
> > I've just pushed a commit that I think will fix the crash--I cannot
> > reproduce it anymore, anyway, so build the latest at-spi2-core from
> > git.gnome.org if you'd like to test.
> >
> > Thanks,
> > -Mike
> >
> > On Wed, 5 Dec 2012, Josà Vilmar EstÃcio de Souza wrote:
> >
> > > Hi Mike.
> > > Do you want that I open a bug regarding this problem?
> > > Thanks.
> > > On 12/05/2012 08:37 PM, Mike Gorse wrote:
> > > > Hi JosÃ,
> > > >
> > > > > I don't know where to post this problem, sorry.
> > > > > I hope that someone can point me to the correct place.
> > > >
> > > > http://bugzilla.gnome.org is the place to file bugs against AT-SPI.
> > > > You'de select AT-SPI as the product, and, in this case, you should
> > > > select at-spi2-core as the component, since the bug is in the registry
> > > > daemon.
> > > >
> > > > > To reproduce my problem try the following steps:
> > > >
> > > > Yeah, I can reproduce it. It looks as though it relates to the code that
> > > > tries to detect when processes are hung, and in some case something is
> > > > being freed when it shouldn't or something similar. I need to
> > > > investigate more.
> > > >
> > > > > I can not use ctrl to stop speech and if I switch to terminal and
> > > > > begin to type, orca reads the entire line for each key pressed.
> > > >
> > > > In 2.7.x, applications will automatically re-register, and ATs will
> > > > automatically re-register keystroke listeners, if the registry daemon
> > > > dies and then comes back, so hopefully this is less of an issue now, but
> > > > the crash should be fixed in any case.
> > > >
> > > > Also, something somewhere doesn't handle extremely long lines very well
> > > > (I'm not sure off-hand whether to blame Orca, AT-SPI, gtk+, or some
> > > > combination), but that is a separate issue.
> > > >
> > > > Thanks,
> > > > -Mike
> > >
> > > --
> > > {}S Josà Vilmar EstÃcio de Souza
>
> --
> {}S Josà Vilmar EstÃcio de Souza