palm usb_device doesn't seem to disconnect

Dear all,

There have been reports that gnome-pilot can only sync once
after each
reboot.  Recently a user seems to have tracked this down to
HAL
behaviour, in which the 'device_added' event only fires
once, even if
the palmos device is physically unplugged from the host.

This doesn't sound right to me.  I wonder if this behaviour
is a known
regression in HAL, or is it intended for some reason I'm not
fathoming?
Apologies if this is a FAQ - I didn't spot it after a quick
scan through
the list archives for the last couple of months.

System info from the reporter:
hal: 0.5.9.1
hal-info: 0.20070831
kernel: 2.6.22-ARCH
udev: 115

The gnome-pilot-list thread is here:
http://mail.gnome.org/archives/gnome-p
ilot-list/2007-October/msg00022.html

Thanks,

Matt Davey
(gnome-pilot maintainer)

Matt Davey		I do not know myself, and God forbid 
mcdaveymrao.cam.ac.uk 	   that I should.  -- Johann von
Goethe

_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list

On Wed, 2007-10-10 at 18:35 +0100, Matt Davey wrote:
> Dear all,
> 
> There have been reports that gnome-pilot can only sync
once after each
> reboot.  Recently a user seems to have tracked this
down to HAL
> behaviour, in which the 'device_added' event only fires
once, even if
> the palmos device is physically unplugged from the
host.
> 
> This doesn't sound right to me.  I wonder if this
behaviour is a known
> regression in HAL, or is it intended for some reason
I'm not fathoming?
> Apologies if this is a FAQ - I didn't spot it after a
quick scan through
> the list archives for the last couple of months.
> 
> System info from the reporter:
> hal: 0.5.9.1
> hal-info: 0.20070831
> kernel: 2.6.22-ARCH
> udev: 115
> 
> The gnome-pilot-list thread is here:
> http://mail.gnome.org/archives/gnome-p
ilot-list/2007-October/msg00022.html

Further update:
I have now reproduced this problem myself and have narrowed
it down to a
regression between two kernel versions:  I was able to sync
fine using
the initial fedora-7 installation (kernel
2.6.21-1.3194.fc7).  I then
updated just the kernel, to 2.6.22.9-91.fc7, and could then
duplicate
Tom's symptoms (only syncs once per reboot).  'lshal'
retained the info
on the 'usb_device' entry even after disconnecting my palmos
device.

Again, is this a known issue?  Should I take to greg-k-h?

Thanks,

Matt

Matt Davey		Did you hear about the new corduroy pillow?
mcdaveymrao.cam.ac.uk 		It's making headlines all over
town.

_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list

On Thu, 2007-10-11 at 10:19 +0100, Matt Davey wrote:
> On Wed, 2007-10-10 at 18:35 +0100, Matt Davey wrote:
> > Dear all,
> > 
> > There have been reports that gnome-pilot can only
sync once after each
> > reboot.  Recently a user seems to have tracked
this down to HAL
> > behaviour, in which the 'device_added' event only
fires once, even if
> > the palmos device is physically unplugged from the
host.
> > 
> > This doesn't sound right to me.  I wonder if this
behaviour is a known
> > regression in HAL, or is it intended for some
reason I'm not fathoming?
> > Apologies if this is a FAQ - I didn't spot it
after a quick scan through
> > the list archives for the last couple of months.
> > 
> > System info from the reporter:
> > hal: 0.5.9.1
> > hal-info: 0.20070831
> > kernel: 2.6.22-ARCH
> > udev: 115
> > 
> > The gnome-pilot-list thread is here:
> > http://mail.gnome.org/archives/gnome-p
ilot-list/2007-October/msg00022.html
> 
> Further update:
> I have now reproduced this problem myself and have
narrowed it down to a
> regression between two kernel versions:  I was able to
sync fine using
> the initial fedora-7 installation (kernel
2.6.21-1.3194.fc7).  I then
> updated just the kernel, to 2.6.22.9-91.fc7, and could
then duplicate
> Tom's symptoms (only syncs once per reboot).  'lshal'
retained the info
> on the 'usb_device' entry even after disconnecting my
palmos device.
> 
> Again, is this a known issue?  Should I take to
greg-k-h?

I'm not sure this is relevant, but most of my problems with
device nodes
not going away were fixed with a newer udev (at the time).

That was my bug report:
ht
tps://bugzilla.redhat.com/show_bug.cgi?id=203932

And it seems to work fine for me. Maybe it's time to go add
Bluetooth
support to gnome-pilot? 

-- 
Bastien Nocera <hadesshadess.net> 

_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list

On Thu, 2007-10-11 at 10:27 +0100, Bastien Nocera wrote:
> On Thu, 2007-10-11 at 10:19 +0100, Matt Davey wrote:
> > On Wed, 2007-10-10 at 18:35 +0100, Matt Davey
wrote:
> > > Dear all,
> > > 
> > > There have been reports that gnome-pilot can
only sync once after each
> > > reboot.  Recently a user seems to have
tracked this down to HAL
> > > behaviour, in which the 'device_added' event
only fires once, even if
> > > the palmos device is physically unplugged
from the host. 
[...]
> > Further update:
> > I have now reproduced this problem myself and have
narrowed it down to a
> > regression between two kernel versions:  I was
able to sync fine using
> > the initial fedora-7 installation (kernel
2.6.21-1.3194.fc7).  I then
> > updated just the kernel, to 2.6.22.9-91.fc7, and
could then duplicate
> > Tom's symptoms (only syncs once per reboot). 
'lshal' retained the info
> > on the 'usb_device' entry even after disconnecting
my palmos device.
> > 
> > Again, is this a known issue?  Should I take to
greg-k-h?
> 
> I'm not sure this is relevant, but most of my problems
with device nodes
> not going away were fixed with a newer udev (at the
time).
> 
> That was my bug report:
> ht
tps://bugzilla.redhat.com/show_bug.cgi?id=203932

Hmm.  As far as I remember, when I reproduced this issue the
device
nodes did indeed disappear, including the /proc/bus/usb/
entries.  It
was just the lshal entry that hung around.  I can check this
evening (12
hours).

> 
> And it seems to work fine for me. Maybe it's time to go
add Bluetooth
> support to gnome-pilot? 

Sigh... yes we probably should, now that pilot-link supports
it etc.
Would you be happy to test?  There's always network syncing
as a
workaround...

Matt Davey	  If you think education is expensive, try
ignorance.
mcdaveymrao.cam.ac.uk 	     -- Derek Bok

_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list

On Thu, 2007-10-11 at 11:08 +0100, Matt Davey wrote:
> On Thu, 2007-10-11 at 10:27 +0100, Bastien Nocera
wrote:
<snip>
> > I'm not sure this is relevant, but most of my
problems with device nodes
> > not going away were fixed with a newer udev (at
the time).
> > 
> > That was my bug report:
> > ht
tps://bugzilla.redhat.com/show_bug.cgi?id=203932
> 
> Hmm.  As far as I remember, when I reproduced this
issue the device
> nodes did indeed disappear, including the
/proc/bus/usb/ entries.  It
> was just the lshal entry that hung around.  I can check
this evening (12
> hours).

Then it would be a hal issue indeed. Could you reproduce
when hal is
running with "hald --daemon=no --verbose=yes"?

> > And it seems to work fine for me. Maybe it's time
to go add Bluetooth
> > support to gnome-pilot? 
> 
> Sigh... yes we probably should, now that pilot-link
supports it etc.
> Would you be happy to test?  There's always network
syncing as a
> workaround...

Very happy to test. Just use a text entry to input the
bluetooth address
of the device for now, I'll work with upstream bluez-gnome
to provide a
proper widget for that in the longer term.

Cheers

-- 
Bastien Nocera <hadesshadess.net> 

_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list

Hi again,

On Thu, 2007-10-11 at 06:05 -0700, Dan Nicholson wrote:
> On 10/11/07, Bastien Nocera <hadesshadess.net> wrote:
> > On Thu, 2007-10-11 at 11:08 +0100, Matt Davey
wrote:
> > > On Thu, 2007-10-11 at 10:27 +0100, Bastien
Nocera wrote:
> > <snip>
> > > > I'm not sure this is relevant, but most
of my problems with device nodes
> > > > not going away were fixed with a newer
udev (at the time).
> > > >
> > > > That was my bug report:
> > > > ht
tps://bugzilla.redhat.com/show_bug.cgi?id=203932
> > >
> > > Hmm.  As far as I remember, when I reproduced
this issue the device
> > > nodes did indeed disappear, including the
/proc/bus/usb/ entries.  It
> > > was just the lshal entry that hung around.  I
can check this evening (12
> > > hours).
> >
> > Then it would be a hal issue indeed. Could you
reproduce when hal is
> > running with "hald --daemon=no
--verbose=yes"?
> 
> The other thing that would help is to run `udevmonitor'
while
> plugging/unplugging. If no events are being received by
udev, then the
> problem's either with udev or in the kernel.

I'm attaching a haldaemon and udevmonitor logs from both the
first and
second sync.  The first sync was successful, the second was
not detected
by gnome-pilot.

For what it's worth, I find that I can sync once per hald
restart.  I
don't have to reboot, I just need to restart hald.

Hope this helps,

Matt

Matt Davey		Did you hear about the new corduroy pillow?
mcdaveymrao.cam.ac.uk 		It's making headlines all over
town.

_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list