Re: [directfb-dev] Re: Bug#341597: [ppc, d-i-gtk] installation report on b&w G3
On Sat, Dec 03, 2005 at 03:00:43PM +0100, Attilio Fiandrotti wrote:
> Sven Luther wrote:
> >On Sat, Dec 03, 2005 at 12:50:19AM +0200, Guillem Jover wrote:
> >
> >>Hi,
> >>
> >>On Fri, Dec 02, 2005 at 10:18:40AM +0100, Attilio Fiandrotti wrote:
> >>
> >>>If the bug is fixed when hw-acceleration is disabled i think a good idea
> >>>would be making the libdirectfb or the gtk-rootskel udeb containing anto
> >>>/etc/directfbrc configuration file that contains
> >>>
> >>>#prevents DFB from using hardware acceleration
> >>>no-hardware
> >>>#lets user dumping screen by pressing "Stamp" key
> >>>screenshot-dir=/
> >
> >
> >Mmm, Attilio, i don't think this file should be hardcoded like that, but
> >better provided by rootskel-gtk, or generated from rootskel-gtk at early
> >runtime, as we may wish to make it configurable from /proc/cmdline, and do
> >some more advanced matching depending on graphic card used or
> >architecture, as
> >i proposed yesterday.
>
> I think disabling hardware acceleration could be useful in many cases,
> so we should evaluate what's the best choiche among:
>
> *Disablig hw accel by default inserting a static entry in
> /etc/directfbrc file provided by gtk-rootskel.
> -pro: simplest option, no per-videocard hw acceleration disabilitation
> mechanism required; for our purposes we do not need hw acceleration;
> reported bug like this could be easily imputable to DFB's unaccelerated
> video driver so that we culd easily know where to dig in the case of DFB
> crashes.
> -con: this would prevent further potential bugs in accelerated DFB
> drivers from being discovered.
This would at least need user-forced-acceleration-reenabling at runtime, in
order to ease testing.
> *DFB HW accel enabled by default, disabled by default for known broken
> accelerated modes and user-disabilitable at boot-time using ad-hoc
> parameters.
> -pro: hw acceleration enabled by default would make the g-i a good test
> bench for DFB
> -con: per-card hw accel disabilitation mechanism has still to be
> developed; users may be disoriented by crashing installations if they do
> not know that DFB's hw acceleration can be disabled (remember that many
> oh whom betatested the g-i tought it was based on XFree! )
It is trivial though, maybe 5 lines of shell or so. As for the rest, you just
document it.
> If we decide to keep the hw-acceleration enabled by default we'll need
> co-operation from directfb-dev, were bugreports related to DFB crashes
> that we may receive should be forwarded.
Seems logic.
> As soon as the g-i is released it, and henche DFB, will receive a lot of
> testing from many users all around th world and this could be a good
> testbench for DFB too.
:) Like enabling yaird by default in the debian kernels did bring them a bug
report storm :)
Friendly,
Sven Luther
Reply to: