Firefox 49.0_5,1 crash on startup

[João Neves]

Hi,

Thanks all. In my case the actual problem was that my
virtualbox-ose-additions got updated to 5.1.6 but my VirtualBox on the host
was still on 5.1.4 and apparently that was enough to break 3D acceleration
in some non-obvious way.

Thanks for pointing me in the right direction!

/João

--
João Neves

On 21 September 2016 at 11:49, Fernando Herrero Carrón <elferdo at gmail.com>
wrote:

>
>
> 2016-09-21 11:41 GMT+02:00 Otacílio de Araújo Ramos Neto <
> otacilio.neto at bsd.com.br>:
>
>> Em qua, 21 de set de 2016 05:16, João Neves <sevenjp at gmail.com> escreveu:
>>
>> > Hi,
>> >
>> > Upgraded to Firefox 49.0_5,1 from ports on my 10.3 x86_64 installation
>> > running on VirtualBox, where a previous version was running OK recently
>> and
>> > now Firefox immediately fails to start up, no core dump or any other
>> > information.
>> >
>> > I enabled the DEBUG config option to try and get some sort of output,
>> but I
>> > only get:
>> >
>> > nsStringStats
>> >  => mAllocCount:              8
>> >  => mReallocCount:            1
>> >  => mFreeCount:               1  --  LEAKED 7 !!!
>> >  => mShareCount:              4
>> >  => mAdoptCount:              0
>> >  => mAdoptFreeCount:          0
>> >  => Process ID: 10931, Thread ID: 34403848192
>> >
>> > The only non-standard option I am currently using is DBUS=off.
>> >
>> > My make.conf is as follows:
>> >
>> > DEFAULT_VERSIONS+=ssl=libressl
>> > DEFAULT_VERSIONS+=linux=c6_64
>> >
>> > Cheers.
>> >
>> > --
>> > João Neves
>> >
>>
>>
>> Hello João
>>
>> This problem happens with me also. The fix was rebuild
>> virtualbox-ose-addtions with OpenGL support enabled. And enable this
>> support to OpenGL 3D acceleration in virtual machine configuration also.
>> To
>> confirm you can try run glxgears. If it crash the problem is exactly the
>> same that I have faced.
>>
>> []'s
>> -Otacilio
>>
>> >
>>
>
> Hi there,
>
> I experienced the same with firefox and darktable. The only obvious
> connection between the two is gtk3. Other applications using Tcl/Tk work
> fine. I was hoping to investigate it further, but good that someone
> mentioned it :-)
>