Bug 56736

Summary:	NIS/NFS/PCMCIA/APM don't cooperate nicely
Product:	[Retired] Red Hat Linux	Reporter:	Adam Spiers <redhat>
Component:	apmd	Assignee:	Bernhard Rosenkraenzer <bero>
Status:	CLOSED RAWHIDE	QA Contact:	Aaron Brown <abrown>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	7.2
Target Milestone:	---
Target Release:	---
Hardware:	i386
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2002-01-24 22:55:50 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Adam Spiers 2001-11-26 17:46:52 UTC

I use NIS and NFS on my laptop via a PCMCIA ethernet card.
This causes great difficulties:

Firstly, '/etc/rc.d/init.d/pcmcia start' gets run too late.
The default chkconfig startup priority is 45, but netfs'
is 25 and ypbind's is 27.  Obviously it needs to be run earlier,
so I changed it to 20 and rebuilt the symlinks with chkconfig.

But that doesn't work either!  When invoked with the argument
'start', /etc/rc.d/init.d/pcmcia finishes before cardmgr has
completed its work.  This means that '/etc/rc.d/init.d/netfs start'
gets run quick enough afterwards that eth0 is still not ready.
I added a 'sleep 10' to the end of /etc/rc.d/init.d/pcmcia startup,
but that's a nasty hack.  It would be better to make the script
wait until cardmgr is finished, just like
/etc/sysconfig/apm-scripts/apmscript does if the
PCMCIAWAIT option is enabled (except that doesn't work either,
see bug #56718).  In fact if you did this it would remove the 
need for the PCMCIAWAIT stuff which would fix that bug at the
same time.

Finally, when I do an APM suspend via 'apm -s',
/etc/sysconfig/apm-scripts/apmscript
correctly invokes /etc/rc.d/init.d/netfs stop, but it just
hangs on the 'umount -f -t nfs -a' bit.  This is really weird,
because if I run

  /etc/sysconfig/apm-scripts/apmscript suspend user

from the command-line, it all works fine.  The only difference
I can think of is that the former case happens after the kernel
apm has received a suspend signal, whereas in the latter case
the kernel doesn't receive one.  That suggests a kernel bug?
It happens all kernels I've tried since 2.4.14.  The stock
RedHat 2.4.7 kernel just hangs completely when I suspend.I compiled with
CONFIG_APM_ALLOW_INTS, but also tried with
options apm allow_ints=0 in /etc/modules.conf.
Would be really grateful for any suggestions.  All I want to
do is to be able to close my laptop's lid :-(

This is all on an IBM ThinkPad 390E, for what it's worth.

Comment 1 Adam Spiers 2001-11-26 21:11:00 UTC

Upon further investigation, the umount hangs because something
has already suspended all PCMCIA cards.  I have no idea what that
something is, or how to stop it suspending them.  Setting the
undocumented do_apm=0 option to pcmcia_core.o doesn't help, nor
does recompiling the userspace utilities (cardmgr etc.) with
CONFIG_PM=n, despite the documentation implying it might.  I quote:

  Whether or not APM is configured, you can use ``cardctl suspend''
  before suspending your laptop, and ``cardctl resume'' after resuming,
  to cleanly shut down and restart your PCMCIA cards.
I'm beginning to lose my mind over this ... %-(

Comment 2 Bernhard Rosenkraenzer 2002-02-26 17:53:19 UTC

 This should work better with 3.0.2-6 (still not perfect, pcmcia is still started after 
network because some people insist it's the correct thing to do).