[arch-general] Suspend seems not to work with nVidia Nouveau driver

[Javier Vasquez]

On 1/3/12, Javier Vasquez <j.e.vasquez.v at gmail.com> wrote:
> On 1/3/12, C Anthony Risinger <anthony at xtfx.me> wrote:
>> On Tue, Jan 3, 2012 at 2:46 PM, Jonathan Vasquez <jvasquez1011 at gmail.com>
>> wrote:
>>> On Tue, Jan 3, 2012 at 3:43 PM, Alex Ferrando <alferpal at gmail.com>
>>> wrote:
>>>> On 03/01/12 21:25, Javier Vasquez wrote:
>>>>>
>>>>> % 'grep' resume /etc/mkinitcpio.conf HOOKS="base udev autodetect pata
>>>>> scsi
>>>>> sata usb filesystems keymap usbinput resume"
>>>>
>>>> Last time I checked (2 secs ago) resume hook should go before
>>>> filesystems.
>>>
>>> I have this but my comp as I said before doesn't even get to suspend
>>> in the first place since it gets stuck at the black screen just before
>>> suspending:
>>>
>>> HOOKS="base udev scsi sata lvm2 resume filesystems usbinput"
>>
>> is there a reason to not use the `autodetect` hook?
>>
>> it's not clear to me that everyone in this thread is talking about
>> suspend-to-disk, ie. `hibernation` ... the OP i believe was referring
>> to suspend-to-RAM, though perhaps i am mistaken ... AFAIK the term
>> "suspend" is generally used for the RAM variant.
>>
>> `resume` hook is for hibernation only.  it should run immediately
>> after the swap partition holding the frozen image becomes available.
>> whatever drivers/hooks needed to access the swap device should of
>> course run first, in general this just means `udev` + `autodetect` --
>> if the swap partition is on an LVM2 partition, *then* `lvm2` hook is
>> also ran -- again, run the minimum needed to access to the swap
>> partition, then follow it by the `resume` hook.  i don't remember for
>> sure, but IIRC so long as the `udev` hook is ran before `resume`
>> (always), you *should* be able to use the device UUID or label.
>>
>> the `filesystem` hook is install-only (no actual hook) ... it just
>> adds all the FS modules (usually filtered by `autodetect` hook) -- the
>> order won't make a difference -- in general it should be last or near
>> the end.  IIRC `resume` should normally follow `udev` unless the setup
>> is a bit more complex (eg, LVM2).
>>
>> i don't really use hibernation, but there is likely a way to increase
>> verbosity by modifying the initramfs hook and rebuilding the image.
>> FTW, i use nouveau with suspend-to-RAM often, without issue.
>>
>> --
>>
>> C Anthony
>
> As I have things set, I have no problems with "suspend to RAM"
> (actually what gets written into /sys/power/state is "mem", and I use
> "acpitool -s" for that in a similar script I use for suspend to
> disk)...
>
> I use suspend to disk to resume work after a night, or days, in which
> case suspend to RAM is not good given its power consumption, not to
> mention consuming power and the environment, :-)
>
> So, as I haven't experienced problems with suspend to RAM, my
> supposition, and perhaps other's, was that we were talking about
> suspend to disk...
>
> BTW, you were right about the resume hook, only thing is that it's
> documented in the pm-utils wiki which I do not use, :-)  Not sure if
> that'll help me specifying the resume swap partition with UUID type of
> specification instead of plain partition, I'll have to try out, :-)

Still don't know if referred to suspend to ram or disk, but  I moved
"resume" before "filesystems" in the hooks array, and I experience
just the same thing, as mentioned by someone else...

-- 
Javier.