NetBSD Problem Report #39652

From www@NetBSD.org  Mon Sep 29 15:56:23 2008
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
	by narn.NetBSD.org (Postfix) with ESMTP id F3BDE63BD01
	for <gnats-bugs@gnats.netbsd.org>; Mon, 29 Sep 2008 15:56:22 +0000 (UTC)
Message-Id: <20080929155622.B5FF163BCFE@narn.NetBSD.org>
Date: Mon, 29 Sep 2008 15:56:22 +0000 (UTC)
From: lacombar@gmail.com
Reply-To: lacombar@gmail.com
To: gnats-bugs@NetBSD.org
Subject: auich: can't map codec i/o space
X-Send-Pr-Version: www-1.0

>Number:         39652
>Category:       kern
>Synopsis:       auich: can't map codec i/o space
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Sep 29 16:00:00 +0000 2008
>Closed-Date:    
>Last-Modified:  Thu Jan 14 17:10:57 +0000 2010
>Originator:     Arnaud Lacombe
>Release:        4.99.72
>Organization:
n/a
>Environment:
NetBSD oui 4.99.72 NetBSD 4.99.72 (OUI) #0: Sun Sep 28 17:00:00 EDT 2008  al@oui:/build/current/obj/20080928/sys/arch/i386/compile/OUI i386
>Description:
Audio interface doesn't initialize properly, making the machine unable to enter sleep mode and unable to play any sound.

The call to pci_mapreg_map() in auich_attach() fails. -current kernel 4.99.21 initialize correctly.

pcictl(8) dump for the device, full dmesg available, and kernel config available at: https://82.229.84.165:6900/netbsd/pr/ .
>How-To-Repeat:
boot :/
>Fix:
n/a

>Release-Note:

>Audit-Trail:
From: Matthew Mondor <mm_lists@pulsar-zone.net>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: kern/39652: auich: can't map codec i/o space
Date: Fri, 3 Jul 2009 21:55:42 -0400

 I also get both problems with an IBM ThinkPad T23 and netbsd-5 and
 5.99.10.

 - auich lacks power management hooks preventing suspend
 - auich (also an ICH3) cannot map I/O space and when trying to play
   audio applications simply lock.  This was tested with and without
   ACPI with the same results.

 The power management part is probably easy to fix, but I'm not yet
 aware of a fix for the mapping problem, it might need review of
 hardware specs or of drivers for other BSDs or Linux, but I only tried
 NetBSD on it so far.
 -- 
 Matt

From: "Jonathan A. Kollasch" <jakllsch@kollasch.net>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: PR/39652
Date: Fri, 20 Nov 2009 22:32:48 +0000

 --opJtzjQTFsWo+cga
 Content-Type: text/plain; charset=us-ascii
 Content-Disposition: inline

 Attached is a patch that may help,
 also available at http://www.netbsd.org/~jakllsch/auich-map.diff

 --opJtzjQTFsWo+cga
 Content-Type: text/plain; charset=us-ascii
 Content-Disposition: attachment; filename="auich-map.diff"

 Index: auich.c
 ===================================================================
 RCS file: /cvsroot/src/sys/dev/pci/auich.c,v
 retrieving revision 1.131
 diff -u -r1.131 auich.c
 --- auich.c	12 Nov 2009 19:39:26 -0000	1.131
 +++ auich.c	20 Nov 2009 22:13:23 -0000
 @@ -215,8 +215,11 @@
  	int  sc_sts_reg;
  	/* 440MX workaround */
  	int  sc_dmamap_flags;
 -	/* Native mode? */
 -	int  sc_native_mode;
 +	/* flags */
 +	int  sc_iose	:1,
 +	     sc_csr_io	:1,
 +	     sc_csr_mem	:1,
 +		     	:29;

  	/* sysctl */
  	struct sysctllog *sc_log;
 @@ -482,46 +485,55 @@
  	if (d->id == PCIID_ICH4 || d->id == PCIID_ICH5 || d->id == PCIID_ICH6
  	    || d->id == PCIID_ICH7 || d->id == PCIID_I6300ESB
  	    || d->id == PCIID_ICH4MODEM) {
 -		sc->sc_native_mode = 1;
  		/*
  		 * Use native mode for Intel 6300ESB and ICH4/ICH5/ICH6/ICH7
  		 */
 +
 +		sc->sc_csr_mem = 1;
 +		v = pci_conf_read(pa->pa_pc, pa->pa_tag,
 +		    PCI_COMMAND_STATUS_REG);
 +		pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG,
 +			       v | PCI_COMMAND_MEM_ENABLE);
 +		pa->pa_flags |= PCI_FLAGS_MEM_ENABLED;
 +
  		if (pci_mapreg_map(pa, ICH_MMBAR, PCI_MAPREG_TYPE_MEM, 0,
 -				   &sc->iot, &sc->mix_ioh, NULL, &sc->mix_size)) {
 -			v = pci_conf_read(pa->pa_pc, pa->pa_tag, ICH_CFG);
 -			pci_conf_write(pa->pa_pc, pa->pa_tag, ICH_CFG,
 -				       v | ICH_CFG_IOSE);
 -			if (pci_mapreg_map(pa, ICH_NAMBAR, PCI_MAPREG_TYPE_IO,
 -					   0, &sc->iot, &sc->mix_ioh, NULL,
 -					   &sc->mix_size)) {
 -				aprint_error_dev(self, "can't map codec i/o space\n");
 -				return;
 -			}
 +		    &sc->iot, &sc->mix_ioh, NULL, &sc->mix_size)) {
 +			goto retry_map;
  		}
  		if (pci_mapreg_map(pa, ICH_MBBAR, PCI_MAPREG_TYPE_MEM, 0,
 -				   &sc->iot, &sc->aud_ioh, NULL, &sc->aud_size)) {
 -			v = pci_conf_read(pa->pa_pc, pa->pa_tag, ICH_CFG);
 -			pci_conf_write(pa->pa_pc, pa->pa_tag, ICH_CFG,
 -				       v | ICH_CFG_IOSE);
 -			if (pci_mapreg_map(pa, ICH_NABMBAR, PCI_MAPREG_TYPE_IO,
 -					   0, &sc->iot, &sc->aud_ioh, NULL,
 -					   &sc->aud_size)) {
 -				aprint_error_dev(self, "can't map device i/o space\n");
 -				return;
 -			}
 -		}
 -	} else {
 -		if (pci_mapreg_map(pa, ICH_NAMBAR, PCI_MAPREG_TYPE_IO, 0,
 -				   &sc->iot, &sc->mix_ioh, NULL, &sc->mix_size)) {
 -			aprint_error_dev(self, "can't map codec i/o space\n");
 -			return;
 -		}
 -		if (pci_mapreg_map(pa, ICH_NABMBAR, PCI_MAPREG_TYPE_IO, 0,
 -				   &sc->iot, &sc->aud_ioh, NULL, &sc->aud_size)) {
 -			aprint_error_dev(self, "can't map device i/o space\n");
 -			return;
 +		    &sc->iot, &sc->aud_ioh, NULL, &sc->aud_size)) {
 +			goto retry_map;
  		}
 +		goto map_done;
 +	} else
 +		goto non_native_map;
 +
 +retry_map:
 +	sc->sc_iose = 1;
 +	v = pci_conf_read(pa->pa_pc, pa->pa_tag, ICH_CFG);
 +	pci_conf_write(pa->pa_pc, pa->pa_tag, ICH_CFG,
 +		       v | ICH_CFG_IOSE);
 +
 +non_native_map:
 +	sc->sc_csr_io = 1;
 +	v = pci_conf_read(pa->pa_pc, pa->pa_tag,
 +	    PCI_COMMAND_STATUS_REG);
 +	pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG,
 +		       v | PCI_COMMAND_IO_ENABLE);
 +	pa->pa_flags |= PCI_FLAGS_IO_ENABLED;
 +
 +	if (pci_mapreg_map(pa, ICH_NAMBAR, PCI_MAPREG_TYPE_IO, 0,
 +			   &sc->iot, &sc->mix_ioh, NULL, &sc->mix_size)) {
 +		aprint_error_dev(self, "can't map codec i/o space\n");
 +		return;
  	}
 +	if (pci_mapreg_map(pa, ICH_NABMBAR, PCI_MAPREG_TYPE_IO, 0,
 +			   &sc->iot, &sc->aud_ioh, NULL, &sc->aud_size)) {
 +		aprint_error_dev(self, "can't map device i/o space\n");
 +		return;
 +	}
 +
 +map_done:
  	sc->dmat = pa->pa_dmat;

  	/* enable bus mastering */
 @@ -1578,12 +1590,19 @@
  	struct auich_softc *sc = device_private(dv);
  	pcireg_t v;

 -	if (sc->sc_native_mode) {
 +	if (sc->sc_iose) {
  		v = pci_conf_read(sc->sc_pc, sc->sc_pt, ICH_CFG);
  		pci_conf_write(sc->sc_pc, sc->sc_pt, ICH_CFG,
  			       v | ICH_CFG_IOSE);
  	}

 +	v = pci_conf_read(sc->sc_pc, sc->sc_pt, PCI_COMMAND_STATUS_REG);
 +	if (sc->sc_csr_io)
 +		v |= PCI_COMMAND_IO_ENABLE;
 +	if (sc->sc_csr_mem)
 +		v |= PCI_COMMAND_MEM_ENABLE;
 +	pci_conf_write(sc->sc_pc, sc->sc_pt, PCI_COMMAND_STATUS_REG, v);
 +
  	auich_reset_codec(sc);
  	DELAY(1000);
  	(sc->codec_if->vtbl->restore_ports)(sc->codec_if);

 --opJtzjQTFsWo+cga--

From: Matthew Mondor <mm_lists@pulsar-zone.net>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: PR/39652
Date: Sun, 20 Dec 2009 17:43:19 -0500

 On Fri, 20 Nov 2009 22:35:02 +0000 (UTC)
 "Jonathan A. Kollasch" <jakllsch@kollasch.net> wrote:

 >  Attached is a patch that may help,
 >  also available at http://www.netbsd.org/~jakllsch/auich-map.diff

 Although I tested the patch against netbsd-5, I can confirm it fixes
 the mapping and audio-freezing problems on the T23.  I didn't test
 resume/suspend but since I gave the laptop away and it is no longer
 mine, I couldn't borrow it for very long to do the tests on.

 Thank you very much,
 -- 
 Matt

From: Matthew Mondor <mm_lists@pulsar-zone.net>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: PR/39652
Date: Tue, 29 Dec 2009 15:53:40 -0500

 On Sun, 20 Dec 2009 22:45:01 +0000 (UTC)
 Matthew Mondor <mm_lists@pulsar-zone.net> wrote:

 > The following reply was made to PR kern/39652; it has been noted by GNATS.
 > 
 > From: Matthew Mondor <mm_lists@pulsar-zone.net>
 > To: gnats-bugs@NetBSD.org
 > Cc: 
 > Subject: Re: PR/39652
 > Date: Sun, 20 Dec 2009 17:43:19 -0500
 > 
 >  On Fri, 20 Nov 2009 22:35:02 +0000 (UTC)
 >  "Jonathan A. Kollasch" <jakllsch@kollasch.net> wrote:
 >  
 >  >  Attached is a patch that may help,
 >  >  also available at http://www.netbsd.org/~jakllsch/auich-map.diff
 >  
 >  Although I tested the patch against netbsd-5, I can confirm it fixes
 >  the mapping and audio-freezing problems on the T23.  I didn't test
 >  resume/suspend but since I gave the laptop away and it is no longer
 >  mine, I couldn't borrow it for very long to do the tests on.

 Considering this patch also worked well on netbsd-5 as-is, a pullup to
 netbsd-5 is probably also worth it once it's commited on -current?

 Thanks,
 -- 
 Matt

From: "Jonathan A. Kollasch" <jakllsch@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/39652 CVS commit: src/sys/dev/pci
Date: Wed, 30 Dec 2009 00:24:38 +0000

 Module Name:	src
 Committed By:	jakllsch
 Date:		Wed Dec 30 00:24:38 UTC 2009

 Modified Files:
 	src/sys/dev/pci: auich.c

 Log Message:
 Ensure that the memory or i/o space we need is accessible at attach
 and resume.

 Rework register mapping code path, using 'goto', so duplicate code
 is reduced.

 I've tested this on nvidia and sis auich(4)-compatible controllers.
 In addition this change is reported to fix PR/39652.


 To generate a diff of this commit:
 cvs rdiff -u -r1.132 -r1.133 src/sys/dev/pci/auich.c

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

From: Soren Jacobsen <snj@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/39652 CVS commit: [netbsd-5] src/sys/dev/pci
Date: Sat, 9 Jan 2010 01:35:16 +0000

 Module Name:	src
 Committed By:	snj
 Date:		Sat Jan  9 01:35:16 UTC 2010

 Modified Files:
 	src/sys/dev/pci [netbsd-5]: auich.c

 Log Message:
 Pull up following revision(s) (requested by jakllsch in ticket #1218):
 	sys/dev/pci/auich.c: revision 1.133
 Ensure that the memory or i/o space we need is accessible at attach
 and resume.
 Rework register mapping code path, using 'goto', so duplicate code
 is reduced.
 I've tested this on nvidia and sis auich(4)-compatible controllers.
 In addition this change is reported to fix PR/39652.


 To generate a diff of this commit:
 cvs rdiff -u -r1.127 -r1.127.10.1 src/sys/dev/pci/auich.c

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

State-Changed-From-To: open->closed
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Tue, 12 Jan 2010 04:39:04 +0000
State-Changed-Why:
Patch committed and pulled up; thanks.


From: Matthew Mondor <mm_lists@pulsar-zone.net>
To: gnats-bugs@netbsd.org
Cc: netbsd-bugs@netbsd.org
Subject: Re: kern/39652: auich: can't map codec i/o space
Date: Wed, 13 Jan 2010 18:38:32 -0500

 Should this PR be re-opened?  The patch had to be reverted.

 For reference:
 http://mail-index.netbsd.org/current-users/2010/01/13/msg012096.html

 Thanks,
 -- 
 Matt

State-Changed-From-To: closed->open
State-Changed-By: jakllsch@NetBSD.org
State-Changed-When: Thu, 14 Jan 2010 17:10:57 +0000
State-Changed-Why:
patch had unacceptable side effects on some machines.


>Unformatted:

Home	PR Database Search