Re: kernel-2.6.17-1.2139_FC5 won't sleep

On Wed, Jun 28, 2006 at 07:24:42PM -0400, Matthew Saltzman wrote:
On Wed, 28 Jun 2006, David A. De Graaf wrote:

On Tue, Jun 27, 2006 at 02:33:51PM -0400, Matthew Saltzman wrote:
On Tue, 27 Jun 2006, David A. De Graaf wrote:

I haven't accurately measured the comparative rates of battery
discharge while asleep vs. awake because it takes a long time to do
the experiment. I will, though.
I do have the subjective recollection that when the earlier non-ACPI
system slept, the battery drain was very, very low.

OK try the following:
options radeonfb radeon_force_sleep=1

OK that fix is known to solve the Radeon power-consumption-in-suspend
problem. It looks like there is still something else going on and using
the radeonfb driver doesn't fix it.

I've measured the time to discharge two ways using acpitool -b
periodically:
1) The laptop running quiescently, with screensaver operative
2) The laptop in sleep mode, and manually woken periodically
Both runs used kernel 2.6.16-1.2133_FC5 (because the 2139 version
won't work) and the radeonfb radeon_force_sleep=1 option was used.

Running screensaver Sleep mode
Time Charge Time Charge
10:58:43 AM 100 03:11:34 PM 98.54
11:18:43 AM 83.53 03:12:00 PM 98.51
11:38:43 AM 69.76 04:09:30 PM 74.84
11:58:43 AM 58.1 04:41:39 PM 63.4
12:18:43 PM 46.63 05:00:27 PM 56.62
12:38:43 PM 35.13 05:35:23 PM 44.32
12:58:43 PM 23.72 06:00:25 PM 35.45
01:18:43 PM 12.43 07:00:56 PM 14.28
01:38:43 PM 1.34 07:34:55 PM 0.03
08:01:13 PM 31.18
Time to discharge:
2h:40m = 2.67h 4h:24m = 4.40h

That's not very impressive; only 1.65x longer with sleep mode.
I don't think the sleep mode is getting everything turned off that
could be. The operative /etc/acpi/events/sleep.conf action is
acpitool -s

Perhaps there's a more effective way to enter sleep mode.
Any suggestions?

First, are you sure the radeonfb module is being loaded? During the text
part of the boot, the active VC should change to small font/full screen.
If you have a vga= kernel parameter on your kernel line in grub.conf,
remove it. Also, dmesg should report radeonfb statuses, including

radeonfb: forcefully enabling sleep mode

Matthew Saltzman: Are you psychic, or what?
Despite your crystal-clear instructions, I managed to screw them up.
I put the essential line MODULES="radeonfb" in
/etc/sysconfig/mkinitfb instead of /etc/sysconfig/mkinitrd.
This inanity caused the radeonfb module not to be loaded.

With this error corrected, all is well. lsmod shows the radeonfb module
is loaded and dmesg shows a host of initialization messages.
The screen backlight is off while sleeping and the battery drain is
miniscule. I'm measuring the drain as I write this, but it'll take
quite a while. Thank you for your patience and perseverance.


Otherwise, I just suspend/resume using the pm-utils as installed by
default in FC5. I can close the lid or press Fn-F4 to suspend to RAM and
open the lid or press Fn alone (maybe other keys also) to resume. I
haven't used acpitool to suspend. My old FC4 scripts suspended by writing
"mem" into /sys/power/state.

I can suspend to RAM and have it last for a couple of days without
problem.

I have not found the pm-utils programs to be installed or used in any
way - by default. The only effect that I can see of either Fn-F4 or
lid closure is that the acpi daemon notices these events.
I don't know of any other mechanism besides acpi to respond to
these events. However, my fresh install of FC5 provides no rules
whatever to deal with these events in /etc/acpi/events - only
sample.conf and video.conf. To create any response to these events
I had to create the rules files - lid.conf and sleep.conf - which now
state action=/usr/sbin/pm-suspend. Without these rules files, the acpi
events are merely recorded in /var/log/acpid, but cause no action.

There are evidently two commands to induce suspend-to-memory -
/usr/bin/acpitool -s
/usr/sbin/pm-suspend
With kernel-2.6.17-1.2139_FC5 the first one, which uses the S3 mode,
is broken. This has provoked considerable discussion elsewhere on
this list. Fortunately, the second method still works.

--
David A. De Graaf DATIX, Inc. Hendersonville, NC
dad@xxxxxxxx www.datix.us

--
fedora-list mailing list
fedora-list@xxxxxxxxxx
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list

Relevant Pages

  • Re: kernel-2.6.17-1.2139_FC5 wont sleep
    ... the radeonfb driver doesn't fix it. ... I've measured the time to discharge two ways using acpitool -b ... only 1.65x longer with sleep mode. ... swap partition as big as RAM, ...
    (Fedora)
  • Re: kernel-2.6.17-1.2139_FC5 wont sleep
    ... discharge while asleep vs. awake because it takes a long time to do ... the radeonfb driver doesn't fix it. ... only 1.65x longer with sleep mode. ... I can close the lid or press Fn-F4 to suspend to RAM and open the lid or press Fn alone to resume. ...
    (Fedora)
  • Re: Totally broken PCI PM calls
    ... >> But radeonfb ends up suspending the display at a wrong time and you miss ... "suspend to disk". ... That would mean, at least for PCI, a kind of platform hook that provides ... It's a hack, I admit it. ...
    (Linux-Kernel)
  • Re: 2.6.3-mm4
    ... It fails to suspend "radeonfb" (which isn't active BTW, ... I have attached a photo of the suspend message. ... Please note that on this laptop ACPI always used to suspend the machine. ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • Re: 2.4.22-pre10 Problems with radeonfb and suspend mode
    ... > If i load the radeonfb module and send my thinkpad with apm -s to ... > suspend all seems fine but on comming back the screen is still black. ... To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ ...
    (Linux-Kernel)