S3-level ACPI suspend

View: New views
11 Messages — Rating Filter:   Alert me  

S3-level ACPI suspend

by b4283 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Hi,

I've had some nice experiences with Windows(XP)'s suspend functionality,
and it seems to work on all machines I've worked with.

and the other day I just had a successful resume on my machine with
"s2ram -f -m -p", but it takes like a minute for it to fully recover to
a terminal (tty/1).

(ps: my computer is a desktop with AMD RS780 chipset, which isn't
supported by KMS yet)

Unlike Windows ones, which only take like 3-5 seconds to wake up, linux
takes much longer, I don't know if this is common...

My question is: What's the average resume time of yours, and is it
possible to be any faster ?

Re: S3-level ACPI suspend

by Sven-Hendrik Haase :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On 26.10.2009 16:58, b4283 wrote:

> Hi,
>
> I've had some nice experiences with Windows(XP)'s suspend functionality,
> and it seems to work on all machines I've worked with.
>
> and the other day I just had a successful resume on my machine with
> "s2ram -f -m -p", but it takes like a minute for it to fully recover to
> a terminal (tty/1).
>
> (ps: my computer is a desktop with AMD RS780 chipset, which isn't
> supported by KMS yet)
>
> Unlike Windows ones, which only take like 3-5 seconds to wake up, linux
> takes much longer, I don't know if this is common...
>
> My question is: What's the average resume time of yours, and is it
> possible to be any faster ?
>
>  
My somewhat slow laptop (Pentium M 1.60Ghz, 512MB RAM) resumes fully
from in about 5 seconds (yes, on Arch Linux). Your figure seems very
high at one full minute. Did you check the logs? Maybe it is also
loading some stuff from the swap file which it will do in case it
couldn't fit it all in RAM.

Re: S3-level ACPI suspend

by Karol Babioch :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Hi,

On Mo, 2009-10-26 at 23:58 +0800, b4283 wrote:
> My question is: What's the average resume time of yours, and is it
> possible to be any faster ?

It takes something like 10 seconds or so for me. I noticed that you are
using s2ram, have you tried to use pm-utils
(http://wiki.archlinux.org/index.php/Pm-utils)?

--
Best regards,
Karol Babioch <karol@...>


signature.asc (853 bytes) Download Attachment

Re: S3-level ACPI suspend

by Alexander Lam :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

My netbook (Intel Atom 1.6Ghz, 1GB ram) comes up from suspend-to-ram in
about 3 seconds.

I use pmutils to suspend.


--
Alexander Lam

Re: S3-level ACPI suspend

by b4283 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Alexander Lam 提到:
> My netbook (Intel Atom 1.6Ghz, 1GB ram) comes up from suspend-to-ram in
> about 3 seconds.
>
> I use pmutils to suspend.

So it's just me then. I've read the wiki about Pm-utils, but it isn't
very thorough though.

Trying it anyway.

Parent Message unknown Re: S3-level ACPI suspend

by b4283 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Still no luck with Pm-utils. It takes about a minute before the keyboard
 lights and mouse leds to go on, and Caps-Lock key works, so it isn't
hanging.

I could use the SysRq method to reboot my system, so I guess maybe
everything is actually working -- except video. Using Ctrl-Alt-[1-7] to
switch between ttys doesn't work, because my monitor doesn't even
receive any signal.

Re: S3-level ACPI suspend

by b4283 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

I think I nailed it.

With luck, I was able to resume the computer with "s2ram -f -a 1", even
it's within Xorg. But it still takes a long time to resume, then I found
out about my drives hooked on an old IDE extension card stopped working.
So I removed the module "pata_hpt37x", and...

Wala, 11 seconds !

Now it's only issue is to automated this umount+rmmod and modprobe+mount
thing...

Re: S3-level ACPI suspend

by Karol Babioch :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Di, 2009-10-27 at 16:33 +0800, b4283 wrote:
> Now it's only issue is to automated this umount+rmmod and modprobe
> +mount
> thing.

Using pm-utils it is quite easy to do such things:
http://wiki.archlinux.org/index.php/Pm-utils

There must be something similar for s2ram. According to the wiki
(http://wiki.archlinux.org/index.php/Suspend_to_RAM) the file
"hibernate.conf" should make it possible to say which modules need to be
unloaded.

--
Best regards,
Karol Babioch <karol@...>


signature.asc (853 bytes) Download Attachment

Re: S3-level ACPI suspend

by Adrian C. (anrxc) :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Tue, 27 Oct 2009, Karol Babioch wrote:

> There must be something similar for s2ram. According to the wiki
> (http://wiki.archlinux.org/index.php/Suspend_to_RAM) the file
> "hibernate.conf" should make it possible to say which modules need to
> be unloaded.

There is no need. Pm-utils is a frame-work which supports uswsusp
(s2ram/s2disk), tuxonice and kernel (which is default). So user can just
setup /etc/pm/config.d/config like so:

   SLEEP_MODULE="uswsusp"
   SUSPEND_MODULES="pata_hpt37x"

Although uswsusp supports quirks and maintains a whitelist, it's better
to include pm-utils in the story since it is a standard. It has quirks,
reads HAL whitelist, and exposes a nice frame-work for writing your own
hooks.

--
Adrian C. (anrxc) | anrxc..sysphere.org | PGP ID: D20A0618
PGP FP: 02A5 628A D8EE 2A93 996E  929F D5CB 31B7 D20A 0618


Re: S3-level ACPI suspend

by Damjan Georgievski :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

> There is no need. Pm-utils is a frame-work which supports uswsusp
> (s2ram/s2disk), tuxonice and kernel (which is default). So user can just
> setup /etc/pm/config.d/config like so:
>
>  SLEEP_MODULE="uswsusp"
>  SUSPEND_MODULES="pata_hpt37x"
>
> Although uswsusp supports quirks and maintains a whitelist, it's better to
> include pm-utils in the story since it is a standard. It has quirks, reads
> HAL whitelist, and exposes a nice frame-work for writing your own hooks.

Does it support the hibernate-script feature "UnloadAllModules" ?
I find it necessary for suspend to ram



--
damjan

Re: S3-level ACPI suspend

by Adrian C. (anrxc) :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Wed, 4 Nov 2009, Damjan Georgievski wrote:

> Does it support the hibernate-script feature "UnloadAllModules" ? I
> find it necessary for suspend to ram

Not likely, at least not by SUSPEND_MODULES, modules hook calls a
function modunload which runs rmmod and knows how to handle dependant
modules.

Modules hook is called 75modules... if a hook with the same name is
found in /etc/pm it takes precedance. Maybe someone implemented that in
a personal hook, but probably not when hibernate-script does the work.


--
Adrian C. (anrxc) | anrxc..sysphere.org | PGP ID: D20A0618
PGP FP: 02A5 628A D8EE 2A93 996E  929F D5CB 31B7 D20A 0618