Bug 33361

Summary: System hangs on USB startup with extended PCI bus
Product: [Retired] Red Hat Linux Reporter: Antonius Erdmann <antonius.erdmann>
Component: kernelAssignee: Pete Zaitcev <zaitcev>
Status: CLOSED RAWHIDE QA Contact: Brock Organ <borgan>
Severity: high Docs Contact:
Priority: medium    
Version: 7.1   
Target Milestone: ---   
Target Release: ---   
Hardware: i386   
OS: Linux   
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2001-07-10 15:27:30 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---
Description Flags
contents of /proc/pci of laptop when in/off the docking station
contents of /proc/version
Output of 'dmesg'
Proper recovery if request_irq fails in usb-uhci.
prposed hack for SA_SHIRQ (does not work) none

Description Antonius Erdmann 2001-03-27 08:07:44 UTC
From Bugzilla Helper:
User-Agent: Mozilla/4.73 [de] (Windows NT 5.0; U)

Fujitsu-Siemens Laptop: Lifebook X-7595 in Full-Docking-Station with extended PCI bus.

1.) installation from CD-ROM in docking station hangs on PCMCIA card lookup
     --> start installation outside docking station

2.) startup-scripts of installed system in docking station hang in
     modprobe usb-uhci           --> boot: linux nousb
     kudzu                              --> noS??kudzu
     pcmcia                            --> noS??pcmcia
3.) modprobe eth1        works, as eth1 is an alias for e100, the same as for eth0
     ifconfig eth1 up        fails (eth1 on PCI bus in docking station)
     SIOCSIFFLAGS: Resource temporarily unavailable
Hints: - 3rd IDE device in docking station works fine (hde, hdf).
         - strace modprobe usb_uhci
           usb.c: null device being purged!!!
           <repeated x times>
           usb.c: null device being purged!!!
           <system hangs>

         - strace /usr/sbin/kudzu -s
           wait(1114,              <incomplete line>
           <system hangs>

Reproducible: Always
Steps to Reproduce:
1. modprobe usb-uhci 
2. /usr/sbin/kudzu
3. /etc/rc.d/init.d/pcmcia start

Actual Results:  system hangs

Expected Results:  checking for new hardware

Laptop datasheet: http://www.fujitsu-siemens.com/rl/products/notebooks/professional/lifebook/lifebookx/lifebookx.html#
Port replicator:      http://www.fujitsu-siemens.com/rl/products/notebooks/accessories.html#

I will attach two files:     pci-off         /proc/pci     when off the docking station
                                   pci-in          /proc/pci     when in the docking station

As you may see from the attached files, the laptop has PCI bus 0,1,2 internaly; PCI bus 5 externaly in the docking station

I asume, but have not tested it, that all cards on the external PCI bus are not working (as eth1).


Comment 1 Antonius Erdmann 2001-03-27 08:12:26 UTC
Created attachment 13784 [details]
contents of /proc/pci of laptop when in/off the docking station

Comment 2 Pete Zaitcev 2001-03-27 13:26:23 UTC
May I have the /proc/version string from that one?

Comment 3 Antonius Erdmann 2001-03-27 14:14:44 UTC
Created attachment 13825 [details]
contents of /proc/version

Comment 4 Pete Zaitcev 2001-05-01 17:47:05 UTC
Does anyone know if this happens with the 7.1 release?
The bug was filed with Wolverine originally.

Comment 5 Antonius Erdmann 2001-05-03 17:32:50 UTC
Created attachment 17248 [details]
Output of 'dmesg'

Comment 6 Antonius Erdmann 2001-05-03 17:33:58 UTC

the problems do still exist (more or less) in the 7.1 release.

1.) same problem as before

2.) startup-scripts now __crash__ or hang
     crash:  modprobe usb-uhci 
     crash:  kudzu
     hang:   /etc/rc.d/init.d/pcmcia start

3.)  same problem as before

Linux version 2.4.2-2 (root@porky.devel.redhat.com) (gcc version 2.96 20000731
(Red Hat Linux 7.1 2.96-79)) #1 Sun Apr 8 20:41:30 EDT 2001

I will attach the output of 'dmesg'


Comment 7 Pete Zaitcev 2001-05-04 01:00:12 UTC
That dmesg was very helpful! Now I see that UHCI (both of them)
conflict with the second IDE, apparently found in the docking
station (CD-ROM perhaps?).

After that, uhci does not fail gracefuly, and causes different
problems - locking up, crashing, etc. It is a bad thing too,
I'll look into it.

Comment 8 Pete Zaitcev 2001-05-04 02:35:50 UTC
Created attachment 17288 [details]
Proper recovery if request_irq fails in usb-uhci.

Comment 9 Pete Zaitcev 2001-05-04 19:26:25 UTC
With the patch to uhci the system will not crash or
lock up for that reason (I hope there are no other
reasons, the eepro100 being the next suspect).

However, fixing the IDE not to hog interrupts seems difficult.
Alan says that we have controllers in the field that sit
on PCI, but use edge triggered interrupts [presumably,
onboard IDE].

This may end in a situation when nothing works when
the lappy is docked, except the CD-ROM.

Comment 10 Pete Zaitcev 2001-06-25 20:36:04 UTC
7.1 errata is out, please try that. At least USB
won't crash when it finds request_irq() bounced.
If you don't retest, I mark the bug resolved in a week.

Shared IDE interrupts were not resolved, I keep arguing
with Andre Hendriks about it. Very little can be done
at this point about it. If you are inclined so, run a
custom kernel with SA_SHIRQ set at all times, as per patch.
That's the only recourse that I see.

Comment 11 Toni Erdmann 2001-07-10 15:27:27 UTC
RESOLVED the named bug (see Summary)

USB now does not crash or hang the system
/usr/sbin/kudzu now works

service pcmcia start still hangs the system

Thanks, Toni (right back from vacation)

Comment 12 Pete Zaitcev 2001-07-10 17:36:58 UTC
Resolving "in Rawhide" because I pushed the resolution
into Linus tree, so it's sure there. Dunno about
the errata, if the split off HEAD CVS was in time
to pick the fix.

Comment 13 Toni Erdmann 2001-07-11 08:15:38 UTC
I have applied all errata for 7.1 on 2001-07-10
and I'm now running kernel 2.4.3-12
I looked into usb-uhci.c and found something that
looks like the modifications of 2001-05-03
(attach_id=17288) at usb-uhci.c line 2971.

Comment 14 Toni Erdmann 2001-09-24 06:08:30 UTC
I now had the time to try the SA_SHIRQ hack (see: comment from 2001-06-25).
But that causes the machine to hang when bringing up eth1 (the one in the
docking station).
I'm not shure, whether this is the right place for the hack.

Comment 15 Toni Erdmann 2001-09-24 06:11:23 UTC
Created attachment 32478 [details]
prposed hack for SA_SHIRQ (does not work)