From: Andrew M. <ak...@os...> - 2005-05-25 23:50:42
|
"Antonino A. Daplas" <ad...@ho...> wrote: > > On Tuesday 22 March 2005 04:44, Andrew Morton wrote: > > Guys, we seem to have a tridentfb regression in 2.6.12-rc1. > > > > viking <vi...@fl...> wrote: > > > Also (and this is something I've noticed since I tried tridentfb) the > > > video buffer never seems to get cleared, except when I change virtual > > > terminals, then change back to the terminal concerned. THEN it's cleared. > > > So for the moment, I've gone back to using VESAfb. > > > > Smells like a fillrect problem. > > 1. First, can you do an fbset and make sure that yres_virtual and yres are > the same? If you run fbset -i, in the line that starts with 'geometry', the > 2nd and 4th number should be the same (or have a difference not more than > 7). This disables ypan. > > mode "800x600-85" > # D: 56.252 MHz, H: 53.676 kHz, V: 85.065 Hz > geometry 800 600 832 32767 8 > ^^^ ^^^ <- same number > > If not the same, do: > > fbset -vyres n > > where n: is equal to the vertical resolution > > 2. If you still get a non-clearing display, then check if > CONFIG_FB_TRIDENT_ACCEL is set to y. If it is, set it to n. Recompile and > reboot. > > 3. Please send your .config, dmesg, lspci -vvv, and fbset -i. > > 4. What is the last kernel version where tridentfb worked for you? > > BTW, there are no major changes in tridentfb for months, so I'm surprised > about this regression. > Guys, is this problem still present in 2.6.12-rc5? Thanks. |