Re: kern/41112: No ACPI events on HP NC4400

[Jukka Ruohonen <jruohonen%iki.fi@localhost>]

The following reply was made to PR kern/41112; it has been noted by GNATS.

From: Jukka Ruohonen <jruohonen%iki.fi@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/41112: No ACPI events on HP NC4400
Date: Wed, 22 Apr 2009 12:18:19 +0300

 Few verbose debugging notes.
 
 I experimented with an alternative scheduler (two workqueues, no priority
 based scheduling). But it appears that the same result can be achieved with
 a simple counter that rejects the very first GPE arriving to
 AcpiOsExecute(). I think this first GPE is related to the PCI-ACPI bridge
 (or at least ACPICA tries to call PCI bus number and segment -control
 methods):
 
 <snip>
 lpt0 at isa0 port 0x378-0x37b irq 7
   -> AcpiOsExecute: first arriving GPE: reject this to get working Notify()
 Initializing SSP:fdc7ec8c00360b0e 9dbdfdec0af835b3 c1691d000495267e
 93312e0c8a7f
 e2c9 a76ace57958ed49c b237cfd951a7a89d 88caa5b298c1563 7c6fad54d579f6f6
 timecounter: Timecounter "clockinterrupt" frequency 100 Hz quality 0
 cpu1: TSC skew=28 drift=-23
 timecounter: Timecounter "TSC" frequency 1828888710 Hz quality 3000
 Execute Method: [\_GPE._L01] (Node 0xffff800003ff81b0)
 nsutils-1086 [4294966503] NsGetNode             : _SEG, AE_NOT_FOUND
 nsutils-1086 [4294966503] NsGetNode             : _BBN, AE_NOT_FOUND
 <snip>
 
 I have no idea why rejecting this particular GPE enables the desired result,
 working ACPI events. Nothing abnormal catched my eye while examining the
 _L01 in the DSDT. Perhaps this is related to EC or missing features in the
 current implementation. Note that this arrives relatively late: EC is
 already working (not "cold") and there are about ten notify events
 scheduled.
 
 I think this can be classified at least partially as a firmware bug. Please
 feel free to close the report; this seems to be highly laptop-specific bug
 that is hard to reproduce without exactly the same laptop model.