RE: Visor USB hang

RE: Visor USB hang

[Collins, Bernard F. (Skip)]

> > I am running 2.4.23 on a RedHat 9 system. Whenever I try to sync my
> > Visor Deluxe, the system hangs/freezes soon after I press the sync 
> > button on my cradle. Trying to find the cause of the problem, I 
> > preloaded the usbserial and visor modules with "debug=1". Nothing 
> > obviously wrong appears in the logs. The last message before the 
> > system freezes is a usb-uhci.c interrupt message.
> 
> Can you show the log with that enabled?

If you mean debug=1 enabled, the log excerpt I posted was generated with
both the usbserial and visor modules modprobed with debug=1. Perhaps I am
mistaken in assuming that my approach actually enables debugging. I
modprobed both modules and hit the hotsync button. I am assuming that
hotplug does not override the manually loaded module parameters. 

> What happens if you use the uhci.o module instead of usb-uhci.o?

That is not terribly convenient to test right now. Can you suggest a simple
way to unload usb-uhci and load uhci without disabling my usb keyboard and
mouse?

Skip

Re: Visor USB hang

[Greg KH]

On Wed, Dec 03, 2003 at 05:36:01PM -0500, Collins, Bernard F. (Skip) wrote:
> > > I am running 2.4.23 on a RedHat 9 system. Whenever I try to sync my
> > > Visor Deluxe, the system hangs/freezes soon after I press the sync 
> > > button on my cradle. Trying to find the cause of the problem, I 
> > > preloaded the usbserial and visor modules with "debug=1". Nothing 
> > > obviously wrong appears in the logs. The last message before the 
> > > system freezes is a usb-uhci.c interrupt message.
> > 
> > Can you show the log with that enabled?
> 
> If you mean debug=1 enabled, the log excerpt I posted was generated with
> both the usbserial and visor modules modprobed with debug=1. Perhaps I am
> mistaken in assuming that my approach actually enables debugging. I
> modprobed both modules and hit the hotsync button. I am assuming that
> hotplug does not override the manually loaded module parameters. 

but then no user program tried to talk to the device.  I don't see any
accesses to the device in your logs.  Any oops messages?

> > What happens if you use the uhci.o module instead of usb-uhci.o?
> 
> That is not terribly convenient to test right now. Can you suggest a simple
> way to unload usb-uhci and load uhci without disabling my usb keyboard and
> mouse?

	rmmod usb-uhci && modprobe uhci

thanks,

greg k-h

Re: Visor USB hang

[Bernard Collins]

On Wed, 2003-12-03 at 19:24, Greg KH wrote:
> 	rmmod usb-uhci && modprobe uhci

OK, this works. The Visor now hotsyncs and I get no hangs or freezes.
Thanks. I am still willing to help track down the problem if you have
any other diagnostic suggestions.

So is there a downside to uhci compared to usb-uhci? Would moving to
kernel 2.6 likely fix the problem? Finally, what is the "right way" to
get my redhat system to permanently use uhci?

Re: Visor USB hang

[Greg KH]

On Thu, Dec 04, 2003 at 09:49:31AM -0500, Bernard Collins wrote:
> On Wed, 2003-12-03 at 19:24, Greg KH wrote:
> > 	rmmod usb-uhci && modprobe uhci
> 
> OK, this works. The Visor now hotsyncs and I get no hangs or freezes.
> Thanks. I am still willing to help track down the problem if you have
> any other diagnostic suggestions.

Sounds like a uhci timing issue :(

> So is there a downside to uhci compared to usb-uhci?

Not that I know of, it's what I use...

> Would moving to kernel 2.6 likely fix the problem? 

Hopefully, but there's a current nasty 2.6.0-test11 oops with usb-serial
devices right now that you might want to hold off for a few days...

> Finally, what is the "right way" to get my redhat system to
> permanently use uhci?

edit your modules.conf file?

greg k-h

Re: Visor USB hang

[Greg KH]

On Sat, Dec 06, 2003 at 12:52:52PM -0500, Collins, Bernard F. (Skip) wrote:
>  
> Greg KH wrote:
> >On Thu, Dec 04, 2003 at 09:49:31AM -0500, Bernard Collins wrote:
> > On Wed, 2003-12-03 at 19:24, Greg KH wrote:
> 
> > Sounds like a uhci timing issue :(
> 
> Does :( mean that it is not likely to be fixed? If so, is that because
> usb-uhci is going to be deprecated in favor of uhci?

Well, in 2.6, uhci turned into uhci-hcd and the usb-uhci driver
disappeared.

> >> So is there a downside to uhci compared to usb-uhci?
> 
> > Not that I know of, it's what I use...
> 
> I did find one problem on my machine: VMware USB support depends on
> usb-uhci. Oh, and my USB storage keychain device works with usb-uhci but not
> uhci.

I've run vmware just fine with uhci, it just seems that some distros
prefer usb-uhci.  But use what works for you.

greg k-h

RE: Visor USB hang

[Collins, Bernard F. (Skip)]

 
Greg KH wrote:
>On Thu, Dec 04, 2003 at 09:49:31AM -0500, Bernard Collins wrote:
> On Wed, 2003-12-03 at 19:24, Greg KH wrote:

> Sounds like a uhci timing issue :(

Does :( mean that it is not likely to be fixed? If so, is that because
usb-uhci is going to be deprecated in favor of uhci?

>> So is there a downside to uhci compared to usb-uhci?

> Not that I know of, it's what I use...

I did find one problem on my machine: VMware USB support depends on
usb-uhci. Oh, and my USB storage keychain device works with usb-uhci but not
uhci.

Thanks for the help.

bc

RE: Visor USB hang

[Collins, Bernard F. (Skip)]

Greg KH wrote:
> On Wed, Dec 03, 2003 at 05:36:01PM -0500, Collins, Bernard F. (Skip)
wrote:

>>> Can you show the log with that enabled?

>> If you mean debug=1 enabled, the log excerpt I posted was generated with
>> both the usbserial and visor modules modprobed with debug=1. Perhaps I am
>> mistaken in assuming that my approach actually enables debugging. I
>> modprobed both modules and hit the hotsync button. I am assuming that
>> hotplug does not override the manually loaded module parameters. 

> but then no user program tried to talk to the device.  I don't see any
> accesses to the device in your logs.  Any oops messages?

No oops. The log just stops right after ttyUSB0 and tyUSB1 are created. The
precise sequence upon pressing the sync button seems to be: (1) The Visor
enables and hub.c detects two new USB ports. (2) Modules visor and usbserial
do their thing to bind them to /dev devices. (3) A usb-uhci interrupt is
generated one second later. (4) The system hangs before any other access to
the ports occurs. The next log entry comes about 1.5 minutes later after a
reboot.

I am reposting the log excerpt below. 

It seems to me that it is not user programs accessing the modules which
triggers the hang. It is the creation of two new USB ports. If there is some
way to increase the debug verbosity, I would be willing to try it, even if
that means editing the module sources.

>>> What happens if you use the uhci.o module instead of usb-uhci.o?

>> That is not terribly convenient to test right now. Can you suggest a
simple
>> way to unload usb-uhci and load uhci without disabling my usb keyboard
and
>> mouse?

>	rmmod usb-uhci && modprobe uhci

I will try that.

Here are the relevant log file lines again:

Dec  3 15:19:20 xtr45wac kernel: usb.c: registered new driver serial
Dec  3 15:19:20 xtr45wac kernel: usbserial.c: USB Serial support registered
for Generic
Dec  3 15:19:20 xtr45wac kernel: usbserial.c: USB Serial Driver core v1.4
Dec  3 15:19:27 xtr45wac kernel: usbserial.c: USB Serial support registered
for Handspring Visor / Treo / Palm 4.0 / Clié4.x

Dec  3 15:19:28 xtr45wac kernel: usbserial.c: USB Serial support registered
for Sony Clié3.5
Dec  3 15:19:28 xtr45wac kernel: visor.c: USB HandSpring Visor, Palm m50x,
Treo, Sony Cliédriver v1.7
Dec  3 15:20:35 xtr45wac kernel: hub.c: new USB device 00:1d.1-2, assigned
address 3
Dec  3 15:20:35 xtr45wac kernel: usbserial.c: Handspring Visor / Treo / Palm
4.0 / Clié4.x converter detected
Dec  3 15:20:35 xtr45wac kernel: visor.c: Handspring Visor / Treo / Palm 4.0
/ Clié4.x: Number of ports: 2
Dec  3 15:20:35 xtr45wac kernel: visor.c: Handspring Visor / Treo / Palm 4.0
/ Clié4.x: port 1, is for Generic use and is bound to ttyUSB0

Dec  3 15:20:35 xtr45wac kernel: visor.c: Handspring Visor / Treo / Palm 4.0
/ Clié4.x: port 2, is for HotSync use and is bound to ttyUSB1

Dec  3 15:20:35 xtr45wac kernel: usbserial.c: Handspring Visor / Treo / Palm
4.0 / Clié4.x converter now attached to ttyUSB0 (or usb/tts/0 for devfs)

Dec  3 15:20:35 xtr45wac kernel: usbserial.c: Handspring Visor / Treo / Palm
4.0 / Clié4.x converter now attached to ttyUSB1 (or usb/tts/1 for devfs)

Dec  3 15:20:36 xtr45wac kernel: usb-uhci.c: interrupt, status 2, frame#
1499
Dec  3 15:21:58 xtr45wac syslogd 1.4.1: restart.
 

Re: Visor USB hang

[Greg KH]

On Wed, Dec 03, 2003 at 03:44:47PM -0500, Collins, Bernard F. (Skip) wrote:
> I am running 2.4.23 on a RedHat 9 system. Whenever I try to sync my Visor
> Deluxe, the system hangs/freezes soon after I press the sync button on my
> cradle. Trying to find the cause of the problem, I preloaded the usbserial
> and visor modules with "debug=1". Nothing obviously wrong appears in the
> logs. The last message before the system freezes is a usb-uhci.c interrupt
> message.

Can you show the log with that enabled?

What happens if you use the uhci.o module instead of usb-uhci.o?

thanks,

greg k-h

Visor USB hang

[Collins, Bernard F. (Skip)]

I am running 2.4.23 on a RedHat 9 system. Whenever I try to sync my Visor
Deluxe, the system hangs/freezes soon after I press the sync button on my
cradle. Trying to find the cause of the problem, I preloaded the usbserial
and visor modules with "debug=1". Nothing obviously wrong appears in the
logs. The last message before the system freezes is a usb-uhci.c interrupt
message. Any help debugging this further would be appreciated. Please CC me
on any replies.

Here are the relevant lines from /var/log/messages:
Dec  3 15:19:20 xtr45wac kernel: usb.c: registered new driver serial
Dec  3 15:19:20 xtr45wac kernel: usbserial.c: USB Serial support registered
for Generic
Dec  3 15:19:20 xtr45wac kernel: usbserial.c: USB Serial Driver core v1.4
Dec  3 15:19:27 xtr45wac kernel: usbserial.c: USB Serial support registered
for Handspring Visor / Treo / Palm 4.0 / Clié4.x
Dec  3 15:19:28 xtr45wac kernel: usbserial.c: USB Serial support registered
for Sony Clié3.5
Dec  3 15:19:28 xtr45wac kernel: visor.c: USB HandSpring Visor, Palm m50x,
Treo, Sony Cliédriver v1.7
Dec  3 15:20:35 xtr45wac kernel: hub.c: new USB device 00:1d.1-2, assigned
address 3
Dec  3 15:20:35 xtr45wac kernel: usbserial.c: Handspring Visor / Treo / Palm
4.0 / Clié4.x converter detected
Dec  3 15:20:35 xtr45wac kernel: visor.c: Handspring Visor / Treo / Palm 4.0
/ Clié4.x: Number of ports: 2
Dec  3 15:20:35 xtr45wac kernel: visor.c: Handspring Visor / Treo / Palm 4.0
/ Clié4.x: port 1, is for Generic use and is bound to ttyUSB0
Dec  3 15:20:35 xtr45wac kernel: visor.c: Handspring Visor / Treo / Palm 4.0
/ Clié4.x: port 2, is for HotSync use and is bound to ttyUSB1
Dec  3 15:20:35 xtr45wac kernel: usbserial.c: Handspring Visor / Treo / Palm
4.0 / Clié4.x converter now attached to ttyUSB0 (or usb/tts/0 for devfs)
Dec  3 15:20:35 xtr45wac kernel: usbserial.c: Handspring Visor / Treo / Palm
4.0 / Clié4.x converter now attached to ttyUSB1 (or usb/tts/1 for devfs)
Dec  3 15:20:36 xtr45wac kernel: usb-uhci.c: interrupt, status 2, frame#
1499
Dec  3 15:21:58 xtr45wac syslogd 1.4.1: restart.
D