kern/77805: Boot hangs with ACPI enabled

Fri Mar 11 17:50:09 PST 2005

The following reply was made to PR kern/77805; it has been noted by GNATS.

From: rjohns44 at comcast.net
To: freebsd-gnats-submit at FreeBSD.org, rjohns44 at comcast.net,
	freebsd-acpi at freebsd.org, kris at obsecurity.org
Cc:  
Subject: Re: kern/77805: Boot hangs with ACPI enabled
Date: Sat, 12 Mar 2005 01:45:32 +0000

 --NextPart_Webmail_9m3u9jl4l_22141_1110591932_0
 Content-Type: text/plain
 Content-Transfer-Encoding: 8bit

 Per the handbook, tried adding the following statement to /boot/loader.conf:
 debug.acpi.disabled=cpu

 This removed the "boot-hangs-during-startup" issue, but does point to a bug in the cpu portion of the ACPI. It seems to be have been overlooked in the correspondence on this issue, but this DID NOT OCCUR in 5.0-Release, 5.1-Release, 5.2-Release, 5.2.1-Release, 5.3-Release. It only happened when I upgraded to 5.3-Stable around 2/18/2005.

 On Sat, 2005-02-26 at 19:05 -0800, Nate Lawson wrote:

 > Ron Johnson wrote:
 > > Any help would be appreciated! dmesg output and ASL are in the bug
 > > report.
 > > 
 > > http://www.freebsd.org/cgi/query-pr.cgi?pr=77805
 > > 
 > > With ACPI enabled, boot fails at "hw.acpi.cpu.cx_lowest: C1 -> C3". This
 > > statement is not in sysctl.conf or loader.conf and appears to be changed
 > > automatically. No issue in all previous 5.X versions. 5.3-STABLE
 > > downloaded from CVS on 2/19/2005 around 3pm EST.
 > 
 > You want to change the performance/economy settings in rc.conf.  Man 5 
 > rc.conf
 > 
 > The defaults were changed between 5.2.1 and 5.3:
 > 
 > src/etc/defaults/rc.conf:revision 1.206
 > date: 2004/05/29 04:52:37;  author: njl;  state: Exp;  lines: +1 -1
 > Throw the switch and enable use of the lowest idle states while online in
 > addition to offline.  This can be overridden in /etc/rc.conf if it causes
 > trouble although this has been stable since 2003/12.
 > 
 > I have no idea why C3 is unstable on your machine.  Please try 
 > overriding this with something like this and see if it helps:
 > 
 > performance_cx_state=HIGH
 > economy_cx_state=HIGH
 > 

 It didn't help. Still hangs at the same point. It did output new error
 messages below.

 Feb 27 18:08:13 armada kernel: ACPI-0361: *** Error: Thread 29 cannot
 release Mutex [MUT0] acquired by thread 8
 Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution
 failed [\_SB_.PCI0.PIB_.EC0_.SMRD] (Node 0xc14653e0), AE_AML_NOT_OWNER
 Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution
 failed [\_SB_.BAT1.UPBI] (Node 0xc13ab2c0), AE_AML_NOT_OWNER
 Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution
 failed [\_SB_.BAT1.SEBI] (Node 0xc13ab200), AE_AML_NOT_OWNER
 Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution
 failed [\_TZ_.THRM._TMP] (Node 0xc13ab8a0), AE_AML_NOT_OWNER
 Feb 27 18:08:13 armada kernel: acpi: suspend request ignored (not ready
 yet)
 Feb 27 18:08:16 armada last message repeated 5 times
 Feb 27 18:08:16 armada syslogd: exiting on signal 15

 This does seem similar to the "Laptop and ACPI" thread
 http://lists.freebsd.org/pipermail/freebsd-acpi/2005-March/001325.html.
 Although my clock speed is reported correctly without ACPI.
 --NextPart_Webmail_9m3u9jl4l_22141_1110591932_0
 Content-Type: text/html
 Content-Transfer-Encoding: 8bit

<html><body>
 <DIV>Per the handbook, tried adding the following statement to /boot/loader.conf:</DIV>
 <DIV>debug.acpi.disabled=cpu</DIV>
 <DIV>&nbsp;</DIV>
 <DIV>This removed the "boot-hangs-during-startup" issue, but does point to a bug in the cpu portion of the ACPI. It seems to be have been overlooked in the correspondence on this issue, but this DID NOT OCCUR in 5.0-Release, 5.1-Release, 5.2-Release, 5.2.1-Release, 5.3-Release. It only happened when I upgraded to 5.3-Stable around 2/18/2005.</DIV>
 <DIV>&nbsp;</DIV>
 <DIV>On Sat, 2005-02-26 at 19:05 -0800, Nate Lawson wrote: &gt; Ron Johnson wrote: &gt; &gt; Any help would be appreciated! dmesg output and ASL are in the bug &gt; &gt; report. &gt; &gt; &gt; &gt; <A href="http://www.freebsd.org/cgi/query-pr.cgi?pr=77805">http://www.freebsd.org/cgi/query-pr.cgi?pr=77805</A> &gt; &gt; &gt; &gt; With ACPI enabled, boot fails at "hw.acpi.cpu.cx_lowest: C1 -&gt; C3". This &gt; &gt; statement is not in sysctl.conf or loader.conf and appears to be changed &gt; &gt; automatically. No issue in all previous 5.X versions. 5.3-STABLE &gt; &gt; downloaded from CVS on 2/19/2005 around 3pm EST. &gt; &gt; You want to change the performance/economy settings in rc.conf.&nbsp; Man 5 &gt; rc.conf &gt; &gt; The defaults were changed between 5.2.1 and 5.3: &gt; &gt; src/etc/defaults/rc.c
 on
 f:revision 1.206 &gt; date: 2004/05/29 04:52:37;&nbsp; author: njl;&nbsp; state: Exp;&nbsp; lines: +1 -1 &gt; Throw the switch and enable use of the lowest idle states while online in &gt; addition to offline.&nbsp; This can be overridden in /etc/rc.conf if it causes &gt; trouble although this has been stable since 2003/12. &gt; &gt; I have no idea why C3 is unstable on your machine.&nbsp; Please try &gt; overriding this with somethin g like this and see if it helps: &gt; &gt; performance_cx_state=HIGH &gt; economy_cx_state=HIGH &gt; It didn't help. Still hangs at the same point. It did output new error messages below. Feb 27 18:08:13 armada kernel: ACPI-0361: *** Error: Thread 29 cannot release Mutex [MUT0] acquired by thread 8 Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution failed [\_SB_.PCI0.PIB_.EC0_.SMRD] 
 (N
 ode 0xc14653e0), AE_AML_NOT_OWNER Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution failed [\_SB_.BAT1.UPBI] (Node 0xc13ab2c0), AE_AML_NOT_OWNER Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution failed [\_SB_.BAT1.SEBI] (Node 0xc13ab200), AE_AML_NOT_OWNER Feb 27 18:08:13 armada kernel: ACPI-1304: *** Error: Method execution failed [\_TZ_.THRM._TMP] (Node 0xc13ab8a0), AE_AML_NOT_OWNER Feb 27 18:08:13 armada kernel: acpi: suspend request ignored ( not ready yet) Feb 27 18:08:16 armada last message repeated 5 times Feb 27 18:08:16 armada syslogd: exiting on signal 15 This does seem similar to the "Laptop and ACPI" thread <A href="http://lists.freebsd.org/pipermail/freebsd-acpi/2005-March/001325.html.">http://lists.freebsd.org/pipermail/freebsd-acpi/2005-March/001325.html.</A> Although my clock speed is reported correctly without ACPI. </DIV></body></html>

 --NextPart_Webmail_9m3u9jl4l_22141_1110591932_0--