Antonino Daplas wrote:
On 6/3/07, Petr Vandrovec [off-list ref] wrote:
quoted
Antonino A. Daplas wrote:
quoted
- the pseudo_palette is only 16 elements long.
- do not write to the pseudo_palette if regno (array index) is more
than 15
quoted
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
NACK. If you want to remove entry #16, you need to also fix all uses of
it - matroxfb_crtc2 uses it, matroxfb_accel as well. Before hardware
cursor was removed from fbcon it was used by hardware cursor - now it is
only set, but never used. Removing entry without removing code to set
entry #16 memory corruption will occur.
Petr
Okay, I see it now. Andrew, drop the patch for matroxfb.
Petr,
Do you have plans for the 17th entry?
No. It was used in the past when hardware XOR cursor was used (so
palette 4bpp/8bpp needed XOR mask 0x0F, while truecolor modes needed
0x7FFF for 15bpp/0xFFFF for 16bpp/0xFFFFFF for 32bpp). Now matroxfb
uses software cursor (which just repaints complete character) and I have
no plans to resurrect it.
So feel free to remove code which sets that entry. I believe that
passing matroxfb sources through "grep -v 'cmap\[16\]\|pal\[.*F'"
"fixes" it...
petr@gwy:/usr/src/linus/linux-2.6.22-rc3/drivers/video/matrox$ grep
'cmap\[16\]\|pal\[.*F' *
matroxfb_accel.c: pal[i] = 0xFFFFFFFF;
matroxfb_accel.c: pal[i] = 0x0F0F0F0F;
matroxfb_accel.c:
ACCESS_FBINFO(cmap[16]) = 0x7FFF7FFF;
matroxfb_accel.c:
ACCESS_FBINFO(cmap[16]) = 0xFFFFFFFF;
matroxfb_accel.c: ACCESS_FBINFO(cmap[16])
= 0xFFFFFFFF;
matroxfb_accel.c: ACCESS_FBINFO(cmap[16])
= 0xFFFFFFFF;
matroxfb_crtc2.c: m2info->cmap[16] = 0xFFFFFFFF;
Petr
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/