Hi,
Attached is a patch against 2.5.49 + James Simmons' latest fbdev.diff.
This fixes fbcon oopsing at load time which is due to the fb_cursor's
palette index entries being improperly updated.
A new kind of 'emacs glitch' appeared, though. This can be described as
a block of text that is incompletely copied, ie when "tabbing" a line
of text. This is present with all hardware I tested, soft accel and
hardware accel, which indicates a problem in the higher layers (probably
fbcon). vgacon works okay.
Other changes:
1. optimization of fbcon_accel_putcs()
2. enabling logo displays at all packed pixel formats (as long as
fb_imageblit is supported)
3. Various fbcon_accel_cursor() fixes which would have resulted in
wrong cursor colors or an invisible cursor.
Tony
> This fixes fbcon oopsing at load time which is due to the fb_cursor's
> palette index entries being improperly updated.
Fixed.
> A new kind of 'emacs glitch' appeared, though. This can be described as
> a block of text that is incompletely copied, ie when "tabbing" a line
> of text. This is present with all hardware I tested, soft accel and
> hardware accel, which indicates a problem in the higher layers (probably
> fbcon). vgacon works okay.
Its is fbcon. If you play with the low level driver to draw things you
noticed everything is okay. I have to track it down.
> Other changes:
>
> 1. optimization of fbcon_accel_putcs()
Applied. If you look at my patch I attempted to make fbcon_pixmap dynamic
instead of a static memeory area. It oops so I have that code commented
out for now. Actually I applied it when the fix came out rigth away.
> 2. enabling logo displays at all packed pixel formats (as long as
> fb_imageblit is supported)
This didn't apply well. Only thing missing.
> 3. Various fbcon_accel_cursor() fixes which would have resulted in
> wrong cursor colors or an invisible cursor.
Applied. Worked now :-)