[Libdlo] GDM black screen / user and count meaning

[Alexander Todorov]

На 23.12.2010 09:02, Bernie Thompson написа:
> Hi Alexander,
>
> I took a first run at more complete DPMS/FB_BLANK support.  Can you
> grab the latest udlfb from git.plugable.com and verify if blanking
> works as expected, and if it helps resolve the original problem you
> were seeing?
>

Hello Bernie,
sorry for the late reply but I've been quite busy.

I've updated to the latest git code and removed my work arounds for the problem. 
Then I've executed a series of 10 reboots and shutdown/start on my test system 
which has 2 displays. In all cases the GDM screen showed up successfully. When 
it starts I notice that the screens flicker 2 times.

1st ot goes from green to black while showing the mouse pointer as busy. I 
assume this is when Xorg starts.
2nd is shortly after the first. It goes from black to color showing the GDM 
login screen.

/var/log/messages has been pretty consistent in what it reports:
Jan 14 23:09:34 mini kernel: udlfb: open /dev/fb1 user=1 fb_info=de801400 count=1
Jan 14 23:09:34 mini kernel: udlfb: released /dev/fb1 user=1 count=0
Jan 14 23:09:34 mini kernel: udlfb: open /dev/fb1 user=1 fb_info=de801400 count=1
Jan 14 23:09:34 mini kernel: udlfb: open /dev/fb1 user=1 fb_info=de801400 count=2
Jan 14 23:09:34 mini kernel: udlfb: mmap() framebuffer addr:3777011712 size:1572864
Jan 14 23:09:34 mini kernel: udlfb: /dev/fb1 FB_BLANK mode 0 --> 1
Jan 14 23:09:34 mini kernel: udlfb: open /dev/fb2 user=1 fb_info=de802400 count=1
Jan 14 23:09:34 mini kernel: udlfb: released /dev/fb2 user=1 count=0
Jan 14 23:09:34 mini kernel: udlfb: open /dev/fb2 user=1 fb_info=de802400 count=1
Jan 14 23:09:34 mini kernel: udlfb: open /dev/fb2 user=1 fb_info=de802400 count=2
Jan 14 23:09:34 mini kernel: udlfb: mmap() framebuffer addr:3780173824 size:1572864
Jan 14 23:09:34 mini kernel: udlfb: /dev/fb2 FB_BLANK mode 0 --> 1
Jan 14 23:09:35 mini kernel: udlfb: /dev/fb1 FB_BLANK mode 1 --> 0
Jan 14 23:09:35 mini kernel: udlfb: /dev/fb2 FB_BLANK mode 1 --> 0

i.e. first the display goes from mode 0 to mode 1 then from mode 1 to mode 0.

I will have to test it on the system which has 15 displays to be 100% confident 
but I think this fixes my original issue. Thanks for the patch.

Regards,
Alexander.