Re: [HEADSUP] amd64 suspend/resume code to be comitted

Paul B. Mahol wrote:
> On 3/26/09, Daniel Thiele <dthiele_at_gmx.net> wrote:
>> Jung-uk Kim wrote:
>>> On Wednesday 25 March 2009 05:28 pm, Gustau Perez wrote:
>>>>> Please try the attached patch, which just disables restoring VGA
>>>>> state while resuming.
>>>>   Tried your patch again, and this time using hw.acpi.reset_video,
>>>> hw.acpi.suspend_bounce and friends. The same result, the screen
>>>> remains completely black.
>>>>
>>>>    Before appliying the patch, when resuming the screen showed
>>>> messages about the resume of some hardware (usb, firewire, etc ...)
>>>> and then it went black.
>>>>
>>>>   Is there anything I can test ?
>>> Then, it is something else, e.g., acpi_video(4).  You can try setting
>>> debug.acpi.disabled="video" in /boot/loader.conf for example.
>>>
>> I have the same problem here with a ThinkPad T400. This is a model with
>> the Switchable Graphics feature, but I set the corresponding BIOS option
>> to use the integrated Intel GMA X4500 only. Suspend and resume seem to
>> work except for video output. I am using a vanilla CURRENT from March 25
>> (so I have not tried the vga_isa.diff patch yet).
>>
>> As in Gustau's case, after resume the screen stays just black. The
>> system, though, seems to be up and running, since I get console's
>> audible bell when I expect it and I can edit files, blindly of course.
>> Setting debug.acpi.disabled="video" in /boot/loader.conf produces the
>> same result: A black screen right after resume.
>>
>> I also tried setting hw.acpi.reset_video to 1 but the the machine does
>> not resume at all. When I, in this case, additionally turn on
>> debug.acpi.resume_beep all I get is a long beep that only stops, if I
>> power-cycle the machine.
>>
>> Also, at first I tried to use an USB flash drive to test suspend/resume
>> on amd64, which results in a panic on suspend. Unfortunately, I don't
>> have any dumps, but the "tracing pid" output from "where" at the
>> debugger prompt reported PID 12, wich seems to be [intr]. Or at leas it
>> seem to be intr whenever I look up PID 12. Right now I am (ab)using my
>> swap partition, which works fine.
>>
>> So, if there are any other knobs to fiddle with or patches to try out
>> to investigate this further (at least the black screen issue) just let
>> me know.
> 
> The only way I managed (without suspending while inside X11) my HP
> nx7300 on UP i386 to
> resume with video working is kldloading vesa before suspend.
> 

Cool, thanks a lot for that hint! That helped. I did not find the vesa
module in the amd64's /boot/kernel directory nor the man-page, but
loading the i915 module seems to do the trick.


Daniel