[Mesa-dev] automake-gallium testing

Matt Turner mattst88 at gmail.com
Mon Oct 8 11:13:15 PDT 2012


On Mon, Oct 1, 2012 at 8:08 AM, Andreas Boll <andreas.boll.dev at gmail.com> wrote:
> 2012/9/21 Matt Turner <mattst88 at gmail.com>:
>> On Mon, Sep 17, 2012 at 8:22 AM, Andreas Boll
>> <andreas.boll.dev at gmail.com> wrote:
>>> Yep, vdpau is also affected.
>>>
>>> $ LD_LIBRARY_PATH="/home/andreas/mesa-testing/lib/vdpau"
>>> VDPAU_DRIVER=r600' vdpauinfo
>>> display: :0   screen: 0
>>> Failed to open VDPAU backend
>>> /home/andreas/mesa-testing/lib/vdpau/libvdpau_r600.so: undefined
>>> symbol: xcb_dri2_id
>>> Error creating VDPAU device: 1
>>
>> Thanks. I think this should be fixed now. I feel like I'm starting to
>> go blind checking nm output against scanelf -n.
>
> Hi Matt,
>
> I've tested your automake-gallium2 branch and found another regression
> in the vdpau state tracker.
>
> I've bisected to this commit:
>
> 5a20b8689989faa027f7ad29bb23b33a2d81d4b9 is the first bad commit
> commit 5a20b8689989faa027f7ad29bb23b33a2d81d4b9
> Author: Matt Turner <mattst88 at gmail.com>
> Date:   Tue Sep 4 13:13:51 2012 -0700
>
>     Remove vdpau hack
>
> Steps to reproduce:
>
> $ ./autogen.sh --prefix=$HOME/mesa-testing \
>         --with-gallium-drivers=r600,swrast \
>         --with-dri-drivers= \
>         --enable-texture-float \
>         --enable-glx-tls \
>         --enable-gallium-g3dvl \
>         --enable-vdpau \
>         --enable-xvmc
>
> $ make
> $ make install
>
> $ LD_LIBRARY_PATH="/home/andreas/mesa-testing/lib/vdpau"
> VDPAU_DRIVER=r600 vdpauinfo
> display: :0   screen: 0
> /home/andreas/mesa-testing/lib/vdpau/libvdpau_r600.so.1: undefined
> symbol: vdp_imp_device_create_x11
> Error creating VDPAU device: 1
>
> Andreas.

I think there's intermediate breakage caused by this commit (which I
can probably hide by folding into another commit), but can you confirm
that you don't have this issue at the end of the series?


More information about the mesa-dev mailing list