Using VESA to restore display settings on resume

[Kevin Oberman]

> From: Vladimir Grebenschikov <vova at fbsd.ru>
> Date: Thu, 03 Mar 2005 00:36:51 +0300
> Sender: Vladimir Grebenschikov <vova at vbook.fbsd.ru>
> 
> ÷ ÓÒ, 02/03/2005 × 11:18 -0800, Kevin Oberman ÐÉÛÅÔ:
> >> From: Vladimir Grebenschikov <vova at fbsd.ru>
> >> Date: Tue, 01 Mar 2005 23:31:15 +0300
> >> Sender: owner-freebsd-mobile at freebsd.org
> >> 
> >> ÷ ÐÎ, 28/02/2005 × 20:26 +0000, Ian Dowse ÐÉÛÅÔ:
> >> >In message <200502281016.aa49779 at salmon.maths.tcd.ie>, Ian Dowse writes> :
> >> >>In message <1109545918.1537.8.camel at localhost>, Vladimir Grebenschikov>  writes:
> >> >>>Also if I try this patch with vesa module loaded it just carsh in vm8> 6
> >> >>>mode on zzz (curproc acpiconf).
> >> >>
> >> >>It is probably necessary to get this working with VESA to have much
> >> >>of a chance of it helping, since otherwise only the basic VGA
> >> >>registers will be saved and restored.
> >> >>
> >> >>One possible reason for the vm86 crash is that it appears our VESA
> >> >>code cannot handle saving more than 4k of state data. Does it make
> >> >>any difference if you revert the part of the patch that changed
> >> >>STATE_MOST to STATE_ALL? i.e. change the two uses of STATE_ALL in
> >> >>vesa.c back to STATE_MOST. I found that with SAVE_MOST, I needed
> >> >>to switch to X and back before the console would display, but the
> >> >>backlight did come on immediately.
> >> >
> >> >I've updated the patch at
> >> >
> >> >	http://people.freebsd.org/~iedowse/vesa_restore.diff
> >> >
> >> >so that the VESA code allows up to 8k of state storage and will
> >> >fail if the BIOS claims to need more space. Could you try this with
> >> >the vesa module loaded and with both 0 and 1 for hw.acpi.reset_video?
> >> 
> >> I've rechecked patch with fresh kernel build and firewire console.
> >> 
> >> It does not panics and does not awake :( 
> >> As in previous variant it sleeps ok, but on awake it only spin up HDD.
> >> No any other activity seen, it does not react on keys, firewire console
> >> is dead.
> >> It does not answer on network activity.
> >> 
> >> Probably It hangs while try to restore console, and I should try to
> >> disable console at all ? 
> >
> >Is the disk active LED on? Many systems have had trouble with this,
> >although I thought Soren had fixed this.
> 
> Both led and sound, it spin-up, shows some activity by both led and
> sound, and then stop any activity.
> 
> Probably I should try kernel without ata ? (only md for root) 

Only if the LED comes on and stays on is it an ATA issue. If it seems to
come back but never finishes, it is something else.

Can you connect to the system over the net after resume? If so, it's
probably a display issue, as you seem to have suspected. I was just
bringing up the ATA issue in case you really had two problems. Seems
like you do not.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman at es.net			Phone: +1 510 486-8634