driver 8139too prevents swsusp.

driver 8139too prevents swsusp.

[Éric Brunet]

Hello,

I have been trying S4 sleep on a 2.5.59 + acpi0123 patch kernel on my
computer (Shuttle motherboard, i845 chipset). The motherboard has an
integrated Realtek  ``RTL-8139/8139C/8139C+ (rev 10)'' ethernet card,
so I am using the 8139too driver compiled in the kernel.

When I try to ``echo 4 > /proc/acpi/sleep'', I have messages like:

=kdeinit entered refrigerator
=kalarmd entered refrigerator
=zsh entered refrigerator
=su entered refrigerator
=
[ wait a couple of seconds ]
 stopping tasks failed (1 tasks remaining)
Suspend failed: Not all processes stopped!
Restarting tasks...<6> Strange, eth0 not stopped
 done
init left refrigerator
pdflush left refrigerator
kswapd0 left refrigerator
kseriod left refrigerator

Indeed, I have a kernel thread called [eth0] in the ps output. I cannot
kill this process, but if I do ``ifconfig eth0 down'', the process
diseapear and the swsusp works better (see below). So it seems that
swsusp cannot put this [eth0] process in the refrigerator.

By the way, when I did my swsusp tests yesterday, I switched to runlevel 1
to experiment. After shutting down eth0 with ifconfig, sleep level 4
worked great: the computer went down and resumed flawlessly. However, I
tried again today before writing this bug report and I stayed in
runlevel 5 (I just switched to VT1 to see kernel messages). And today
I couldn't suspend either, I hit the BUG_ON in driver/ide/ide-disk.c:1557
I didn't bother to hand write down the oops, as the machine was frozen
and as I reflected that the stack trace was probably perfectly known.
If this oops is of some importance, I can try to reproduce and write
it.

Thanks for all your work,

	Éric Brunet


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: driver 8139too prevents swsusp.

[Pavel Machek]

Hi!

> 
> I have been trying S4 sleep on a 2.5.59 + acpi0123 patch kernel on my
> computer (Shuttle motherboard, i845 chipset). The motherboard has an
> integrated Realtek  ``RTL-8139/8139C/8139C+ (rev 10)'' ethernet card,
> so I am using the 8139too driver compiled in the kernel.

Patch to fix this is floating around, already.
								Pavel
-- 
Worst form of spam? Adding advertisment signatures ala sourceforge.net.
What goes next? Inserting advertisment *into* email?


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: driver 8139too prevents swsusp.

[Ducrot Bruno]

[-- Attachment #1: Type: text/plain, Size: 548 bytes --]

On Sat, Feb 08, 2003 at 08:06:21PM +0100, Pavel Machek wrote:
> Hi!
> 
> > 
> > I have been trying S4 sleep on a 2.5.59 + acpi0123 patch kernel on my
> > computer (Shuttle motherboard, i845 chipset). The motherboard has an
> > integrated Realtek  ``RTL-8139/8139C/8139C+ (rev 10)'' ethernet card,
> > so I am using the 8139too driver compiled in the kernel.
> 
> Patch to fix this is floating around, already.

But only send to J. Garzik (my fault, sorry)..

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

[-- Attachment #2: 8139.iothread.2.5.patch --]
[-- Type: text/plain, Size: 976 bytes --]

--- linux-2.5.59/drivers/net/8139too.c	2003/02/04 17:03:24	1.1
+++ linux-2.5.59/drivers/net/8139too.c	2003/02/04 17:27:32
@@ -109,6 +109,7 @@
 #include <linux/ethtool.h>
 #include <linux/mii.h>
 #include <linux/completion.h>
+#include <linux/suspend.h>
 #include <linux/crc32.h>
 #include <asm/io.h>
 #include <asm/uaccess.h>
@@ -1601,6 +1602,8 @@
 		timeout = next_tick;
 		do {
 			timeout = interruptible_sleep_on_timeout (&tp->thr_wait, timeout);
+			if (current->flags & PF_FREEZE)
+				refrigerator(PF_IOTHREAD);
 		} while (!signal_pending (current) && (timeout > 0));
 
 		if (signal_pending (current)) {
@@ -2570,6 +2573,8 @@
 	tp->stats.rx_missed_errors += RTL_R32 (RxMissed);
 	RTL_W32 (RxMissed, 0);
 
+	pci_set_power_state(pdev, state);
+
 	spin_unlock_irqrestore (&tp->lock, flags);
 	return 0;
 }
@@ -2581,6 +2586,7 @@
 
 	if (!netif_running (dev))
 		return 0;
+	pci_set_power_state(pdev, 0);
 	netif_device_attach (dev);
 	rtl8139_hw_start (dev);
 	return 0;

Re: driver 8139too prevents swsusp.

[Éric Brunet]

Thank you for the patch.

I will try it one of these days.

-- 
Éric Brunet


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: driver 8139too prevents swsusp.

[Chmouel Boudjnah]

Éric Brunet <ebrunet=l5PIGlIYzrxGWvitb5QawA@public.gmane.org> writes:

> Indeed, I have a kernel thread called [eth0] in the ps output. I cannot
> kill this process, but if I do ``ifconfig eth0 down'', the process
> diseapear and the swsusp works better (see below). So it seems that
> swsusp cannot put this [eth0] process in the refrigerator.

why dont you patch the script to make a ifconfig eth0 down and
ifconfig eth0 up on wakeup (or whatever like restart network on your
distriution) ?

Cheers.

PS: questions about swsusp are on http://lister.fornax.hu/mailman/listinfo/swsusp
PS2: there is some patch on 8139too.c in the latest swsusp you may
     want to make sure you have latest.
-- 
http://www.chmouel.com/



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

Re: Re: driver 8139too prevents swsusp.

[Ducrot Bruno]

On Sat, Feb 08, 2003 at 09:03:04PM +0100, Chmouel Boudjnah wrote:
> Éric Brunet <ebrunet=l5PIGlIYzrxGWvitb5QawA@public.gmane.org> writes:
> 
> > Indeed, I have a kernel thread called [eth0] in the ps output. I cannot
> > kill this process, but if I do ``ifconfig eth0 down'', the process
> > diseapear and the swsusp works better (see below). So it seems that
> > swsusp cannot put this [eth0] process in the refrigerator.
> 
> why dont you patch the script to make a ifconfig eth0 down and
> ifconfig eth0 up on wakeup (or whatever like restart network on your
> distriution) ?
> 

This is for 2.5 for which we want better solutions to do that.
No more modules insmod/rmmod horror.

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com