Re: 2.6.25-rc2 System no longer powers off after suspend-to-disk. Screen becomes green.

[Jesse Barnes <jesse.barnes@intel.com>]

On Wednesday, February 20, 2008 9:17 am Jeff Chua wrote:
> On Feb 20, 2008 2:19 PM, Jeff Chua
>
> > I'll try the "idle=poll" to see if that works and will try some printk
>
> I don't know what exactly the i915_suspend() and i915_resume() are
> supposed to do because it works better without them.
>
> After inserting "return 0;" right at the top of those two functions,
> suspend (and power-off properly), and resume (without green screen) works
> just fine.
>
> I would like to know what they're for.

They're for saving and restoring GPU state across suspend/resume.  They're 
particularly useful if your machine doesn't re-POST at resume time.  In that 
case your GPU may be totally uninitialized, so either the kernel or X has to 
set it up for you (X only does that partially).

> Tested suspend-to-ram, and suspend-to-disk, both console and X on notebook
> internal LCD display, all works without these two functions.
>
> But, anyway, got down to just one line in i915_drv.c causing the hang
> during suspend. "pci_set_power_state(dev->pdev, PCI_D3hot);".

Interesting, which chipset do you have?  AFAIK that shouldn't cause a hang.

> And green screen problem during resume is caused by i915_restore_vga(dev);

I know I fixed that problem in at least one configuration...  Can you try:
  # echo test > /sys/power/disk
  # echo disk > /sys/power/state
and see if that also turns your screen green?

Also, getting a GPU register dump would be helpful.  The intel_reg_dumper tool 
is built as part of the xf86-video-driver build 
(git://anongit.freedesktop.org/git/xorg/driver/xf86-video-intel), can you 
pull that down and try it out?

Thanks,
Jesse

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/