K5 writing to the framebuffer topmost (or mugging Xorg)

[twobob]

Hi.

When running a fbcon SDL application. In normal Kindle 5 mode with cvm and Xorg active, I am unsure how to bring it to the "front"

simply passing a STOP signal to cvm Xorg doesn't do the trick.

Is there a way to influence this seeming interference with the FB or am I just being dim.

SDL_VIDEODRIVER=fbcon SDL_NOMOUSE=1 prboom -nofullscreen -noaccel -width 512 -height 384 -nosound

even with a while true; do eips '';usleep 80000;done
will only bring up a black screen until Xorg or awesome are actually killed (as it stands) I would prefer a more elegant solution

Answers on a postcard.

Thanks all.

[geekmaster]

Stopping Xorg doesn't work because it keeps ownership of the framebuffer, and SDL waits for it to be freed. The eink algorithmic art scripts can do it because they do not take ownership, or even wait for framebuffer ready.

That is why I ran my SDL code from diags SSH when I played with it back in December, when I first got my K5. No Xorg to choke on. The difficulty of using SDL from K5 main is why I temporarily "abandoned" it in favor of direct framebuffer access. Not using SDL has benefits that can be seen in my gmplay program. Faster, and runs "everywhere".

And that is why I recommended replacing video SDL calls in prboom with my GMLIB code from my newtrix demo. But we still need SDL working (someday) if we want DosBox and Qemu working on our eink kindles.

[twobob]

Have the SDL "working" as far as any support for apps is concerned.
Dosbox was on my list, but until the sticky situation with the FB is addressed all that is fluff and pipe smoking.

Grabbed the newtrix again and tend to agree about the FB hotwiring.
not really my field but likely bits are identified.

Expecting everyone to jump to Diags to run this stuff is - well - untenable.

Xorg is reinstantiated as is awesome after a full kill. perhaps a more abstract look at the issue would be beneficial. Can something be convinced not reboot when awesome dies? a service change for example. Thoughts of that nature.

[knc1]

If X11 insists on running . . .
Can it be re-started to use a dummy frame buffer?
Then let SDL use the real one.

[twobob]

Thats a great idea if the whole "rebooting framework" thing can be sorted. in fact hotswapping the underlying device is sort of what I was mooting at before when Niluje helped out with some details.

Don't see why it can't be fooled into thinking something else is going. Yes, it doesn't work right now that way AFAIK but no harm in fiddling. thanks KNC1