patch to make suspend/resume work

patch to make suspend/resume work

[Giridhar Pemmasani]

[-- Attachment #1: Type: text/plain, Size: 429 bytes --]


It seems that pci config space is messed up after resume for Intel
ICH4 audio controller (on Dell Latitude D600, but I notice that others
also complain about this problem). Consequently resume from S3 causes
oops with snd_intel8x0 module. If the module is removed before suspend
and loaded afterwards, I still get oops. The following simple patch
fixes the problem. With this, I can leave alsa untouched during
suspend/resume.


[-- Attachment #2: alsa-s3-fix --]
[-- Type: application/octet-stream, Size: 634 bytes --]

--- intel8x0.c.orig	2004-04-15 17:19:56.000000000 -0400
+++ intel8x0.c	2004-04-15 17:20:55.000000000 -0400
@@ -445,6 +445,7 @@
 	
 #ifdef CONFIG_PM
 	int in_suspend;
+	u32 pci_state[64 / sizeof(u32)];
 #endif
 };
 
@@ -2223,11 +2224,13 @@
 {
 	intel8x0_t *chip = snd_magic_cast(intel8x0_t, pci_get_drvdata(dev), return -ENXIO);
 	intel8x0_suspend(chip);
+	pci_save_state(dev, chip->pci_state);
 	return 0;
 }
 static int snd_intel8x0_resume(struct pci_dev *dev)
 {
 	intel8x0_t *chip = snd_magic_cast(intel8x0_t, pci_get_drvdata(dev), return -ENXIO);
+	pci_restore_state(dev, chip->pci_state);
 	intel8x0_resume(chip);
 	return 0;
 }

[-- Attachment #3: Type: text/plain, Size: 11 bytes --]



-- 
Giri

Re: patch to make suspend/resume work

[Pavel Machek]

Hi!

> It seems that pci config space is messed up after resume for Intel
> ICH4 audio controller (on Dell Latitude D600, but I notice that others
> also complain about this problem). Consequently resume from S3 causes
> oops with snd_intel8x0 module. If the module is removed before suspend
> and loaded afterwards, I still get oops. The following simple patch
> fixes the problem. With this, I can leave alsa untouched during
> suspend/resume.
> 


Looks good to me. Probably more drivers need this...
-- 
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms         

Re: patch to make suspend/resume work

[Giridhar Pemmasani]

On Tue, 20 Apr 2004 15:50:11 +0200, Pavel Machek <pavel@ucw.cz> said:

  Pavel> Looks good to me. Probably more drivers need this...

Yes.

Sometime back (a few weeks?) Stefan Dosinger mailed (on acpi4linux
list?) about his PCI devices all giving oops on resume and to avoid,
he was using setpci to write back the config space the way it was
before suspend. If the drivers for those devices save/restored, then
they will probably work.

-- 
Giri