Closed GoogleCodeExporter closed 9 years ago
attached vbt.c, small c program that scans the vesa bios area
0c0000-100000 in a memory image file (/dev/realmodemem)
for 16 bit words that look like vga modes with different
displacement between the width and height value to look
for mode tables.
can be also used to change the modes.
example:
# dump tables columns
vbt /dev/realmodemem
# change entry at ca800 to 1600x1200 (artificial example)
vbt /dev/realmodemem ca800 1600 ca800+8 1200
Original comment by cinap_le...@felloff.net
on 4 Nov 2011 at 10:51
Attachments:
I've used vbt to determine some combinations of addresses that, when written
to, manage to alter the output of aux/vga -p. So far, while aux/vga -p now
includes vesa modes that match my desired resolution, attempting to set that
resolution with aux/vga -m vesa -l $resolution has resulted in interested
failures, including: blank screen, wrong resolution, inverted colors, etc. So
far, this brute force approach has not yet yielded a working VESA mode
1920x1080.
Original comment by stanley....@gmail.com
on 6 Nov 2011 at 6:29
These are some of the addresses that, when written to, yielded changes:
http://plan9.stanleylieber.com/hardware/thinkcentre/m55/8810-d3u/hits3
The first line is an address from the +2 table. The second line is an address
from the +8 table. The following lines are the new modes that appeared after
writing the appropriate strings to the addresses from the two tables. Each
section is separated by a blank line.
Original comment by stanley....@gmail.com
on 6 Nov 2011 at 6:31
915resolution tool for Intel chipsets:
http://915resolution.mango-lang.org
Original comment by stanley....@gmail.com
on 9 Nov 2011 at 6:20
[deleted comment]
I made some progress with an NVidia GeForce 8400 GS:
http://www.flickr.com/photos/stanleylieber/sets/72157627976638191
Original comment by stanley....@gmail.com
on 11 Nov 2011 at 6:29
just for the record, the kernel panic seen if the image
is due nvidia cards not starting the framebuffer on the
beginning of ther pci membar, but somewhere below. plan9
tries to map the whole framebuffer memory on first init
because we can't undo/resize this kernel mapping.
it failed to detect the whole size and just calculated
the size from the current graphics mode. this in generaly
would'v affected mode switches to a bigger resolution/depth.
this was all fixed in changeset: R18198808ffaa
so with current 9front, that mode from the photo will probably
just work now.
Original comment by cinap_le...@felloff.net
on 27 Jan 2012 at 5:56
Original comment by cinap_le...@felloff.net
on 22 May 2012 at 5:19
Original issue reported on code.google.com by
stanley....@gmail.com
on 3 Nov 2011 at 8:13