PATCH [2/2] - KFB power management

[Manish Regmi]

Hi,

On 3/2/07, Jordan Crouse <jordan.crouse at amd.com> wrote:
> This is the very early beginnings of KFB power management.  By using the
> previous patch to keep track of KD_TEXT and KD_GRAPHICS mode, we can
> make assumptions about what we need to do to save the state of the
> system.  To get the ball rolling, we do what it takes to restore
> the system back to the console prompt (ie, when the current VT is in
> KD_TEXT mode).  If we are in KD_TEXT mode, we are lucky, in that we
> can just blast the current mode back into the registers when we come back
> up - so no state needs to be saved.  As you can tell by the printks() and
> FIXMEs - that won't work when X is involved - when we aren't in control of
> the mode, we'll need to save off an incredible number of registers, including
> ones we don't even access in the KFB, like the GP and VID registers.  I don't
> want that extensive work to gate the simple suspend/resume path, so this
> patch will do in the mean time.

just curious... Where are these patches commited? I could not find it
in olpc-2.6 :(
I am unable to see the changes in the git tree i cloned form
olpc-2.6. It is upto date. just pulled it now.
I also tried the geode branch. I dont have the clone of geode tree but
gitweb does not show the additions.

do i need to patch my tree?
Thanks....

> +#ifdef CONFIG_PM
> +static int gxfb_suspend(struct pci_dev *pdev,  pm_message_t state)
> +{
> +       struct fb_info *info = pci_get_drvdata(pdev);
> +       struct geodefb_par *par = info->par;
> +
> +       if (pdev->dev.power.power_state.event == state.event)
> +               return 0;
> +
> +       if (state.event == PM_SUSPEND_MEM) {
> +
> +               /* If the FB is active, then somebody other then the console
> +                * owns it - we need to save the entire state of the
> +                * graphics engine to put back upon resume.  Otherwise,
> +                * we know what mode we are in right now, so we do nothing
> +                * here and resume that mode upon resume
> +                */
> +
> +               if (par->fbactive) {
> +                       printk(KERN_INFO "gxfb:  FIXME - I need to save the state!\n");
> +               }
> +
> +
> +               acquire_console_sem();
> +               /* FIXME: Blank the screen here? */
> +               fb_set_suspend(info, 1);
> +               release_console_sem();
> +       }
> +
> +       pdev->dev.power.power_state = state;
> +       return 0;

i did not understand why the fbactive trick is needed. will it be
needed after we save all  "Graphics Processor Registers" and
family....

Thanks
-- 
--------------------------------------------------------
Manish Regmi
Volunteer
OLPC Nepal

http://manish-cs.blogspot.com/