GNOME Bugzilla – Bug 362565
gpilotd locksup palm z22
Last modified: 2007-01-10 09:16:58 UTC
Please describe the problem: This bug has been reported here: https://launchpad.net/distros/ubuntu/+source/gnome-pilot/+bug/66355 Steps to reproduce: 1. 2. 3. Actual results: Expected results: Does this happen every time? Other information:
Reporter: why did you mark as resolved 'invalid'? Please add a comment so we can understand the issue. I was just updating anyway, as this looks like it may be a duplicate of the ubuntu bug: https://launchpad.net/distros/ubuntu/+source/gnome-pilot/+bug/62310 which was reported on bugzilla.gnome.org as: http://bugzilla.gnome.org/show_bug.cgi?id=357762 In that case the problem appeared to be with udev, and upgrading to 093-0ubuntu16 was reported to fix the issue. Let us know if that fixed things for you, too. I don't think they'd reported locking up the device...
I'm newby with gnome bugtrack, sorry for strange changes in bug report. I think it was a duplicated bug and try to "close" it. I don't now what to make and I closed it with invalid flag. I have looked at https://launchpad.net/distros/ubuntu/+source/gnome-pilot/+bug and http://bugzilla.gnome.org/show_bug.cgi?id=357762. I have and udev 093-0ubuntu17 but the Z22 still crashes. pilot-xfer seems to works fine. Whath else can I try?
OK, reopening. I've added some ideas to the downstream bug: https://launchpad.net/distros/ubuntu/+source/gnome-pilot/+bug/66355 I'll update this ticket with any progress.
Same problem here with Debian/etch and a palm m505: $ uname Linux xxx 2.6.17-2-686 #1 SMP Wed Sep 13 16:34:10 UTC 2006 i686 GNU/Linux $ gpilotd gpilotd-Message: gnome-pilot 2.0.14 starting... gpilotd-Message: compiled for pilot-link version 0.12.1 gpilotd-Message: compiled with [VFS] [USB] [IrDA] [Network] gpilotd-Message: Activating CORBA server gpilotd-Message: bonobo_activation_active_server_register = 0 gpilotd-Message: Regarde Cradle (/dev/pilot) [...] gpilotd-Message: Found 054c, 0144 gpilotd-Message: Using net TRUE gpilotd-Message: Found 054c, 0169 gpilotd-Message: Using net TRUE gpilotd-Message: Found 12ef, 0100 gpilotd-Message: Using net TRUE gpilotd-Message: setting PILOTRATE=57600 (gnome-pilot:11875): gpilotd-WARNING **: pi_accept_to returned -202: Aucun fichier ou r\xe9pertoire de ce type (gnome-pilot:11875): gpilotd-WARNING **: pi_accept_to: timeout was 2 secs I've setup the workaround described on launchpad ( using libusb as described here: http://code.pilot-link.org/README.libusb ) and it _works_ now. (Note that I was able to synchronize with jpilot and visor module) If you need more tests, ask and I will do my best to run them.
Olivier, Good to hear that the libusb workaround was successful for you. I installed Edgy, to have a go at reproducing this bug, but couldn't do it with my Clie. It would appear to be device-specific. I would be interested to hear whether disabling HAL also functions as a workaround. You should be able to kill 'hald', restart gpilotd, and have it fall back to polling sysfs to detect connections. If this is successful for you, I'll send you a patched gnome-pilot tarball adding a short delay between the HAL connection and the sync attempt, and we can see if that fixes the issue.
I can confirm it works in my case. Killing hald and restarting gpilotd using visor module works. After restarting hald and gpilotd, the "fatal exception" come back immediatly.
Thanks Olivier, that's great news. The HAL integration is new in gnome-pilot 2.0.14, and so it hadn't been exposed to all the various palmos devices prior to release. Just one of those things. There are various hacks in the pilot-link and even visor module to cope with the sometimes weird differences between palmos devices, and the linux device creation timing has caused trouble before... Olivier, you can download a hastily patched version of gnome-pilot from: http://www.inference.phy.cam.ac.uk/mcdavey/downloads/gnome-pilot-2.0.14p1.tar.gz If you have the time, it would be great if you could test this version. If you haven't compiled gnome-pilot before you may have to install some development packages (hal, pilot-link, etc). Untar the archive, run "./configure --prefix=/tmp/gp" from the directory that it creates, and make sure that it reports "HAL/DBUS : yes" at the end of that step. Then just do "make ; make install" and it should build and put the installation under /tmp/gp. So you can then kill your running gpilotd and start up /tmp/gp/libexec/gpilotd and see if it all works. If you have any questions, feel free to email me privately, or update here.
I should have added to previous comment: the patched version adds a short sleep between detecting a device and attempting a sync. The actual delay (for the purposes of the testing) is 10% of the configured 'timeout' for the device, as configured in the config applet. So, if you set a "2 second" timeout, it'll pause for "0.2" seconds. So, play around with the configured 'timeout' value to see if you can get reliable syncing.
after compiling the patched gpilotd. I run it with different values of timeout: 0, 1 and 2 still freeze my m505. but timeout=3 solve the problem. I've synced at least ten times with no freeze (still no freeze at this time of course). I got some stranges error messages when trying to sync with /tmp/gp/libexec/gpilotd: "(gnome-pilot:19185): Pango-WARNING **: Invalid UTF-8 string passed to pango_layout_set_text()" but Matt Davey told me to ignore them: "you can ignore that. It's a long-standing bug I recently fixed post 2.0.14, caused because you have conduits in your gpilotd config files that you haven't installed in /tmp/gp." (in case you want to test this patch by yourself)
After compiling i can run syncs as much as i want, no Palm crashes. I get an error though, but i don't know if it's related: gpilotd-Message: setting PILOTRATE=57600 gpilotd-Message: Device Palm Zire 21 has 0 events gpilotd-Message: Instantiating 1 conduits... (gnome-pilot:14144): libgpilotdcm-WARNING **: Unknown conduit, name="e_calendar_conduit" gpilotd-Message: Unknown conduit "e_calendar_conduit" in configure! (gnome-pilot:14144): Pango-WARNING **: Invalid UTF-8 string passed to pango_layout_set_text() gpilotd-Message: Instantiated 0 backup conduits, 0 file conduits, 0 other conduits gpilotd-Message: HotSync button pressed, synchronizing PDA gpilotd-Message: PDA ID is 1802805000, name is MyPDA, owner is KenSentMe gpilotd-Message: Pilot has 0 entries in restore queue gpilotd-Message: Pilot has 0 entries in conduit queue gpilotd-Message: Synchronization ended
See previous comment... this error can be ignored. It's caused because gnome-pilot is looking for your configured evolution conduits, but they're not installed in the same 'prefix' location as you specified for the testing gnome-pilot. You could try copying the '.conduit' files from your evolution installation (probably in "/usr/lib/gnome-pilot/conduits/") and putting them into your temporary gnome-pilot location ("/tmp/gp/lib/gnome-pilot/conduits"). If you want to continue using your temporary gnome-pilot installation, I'd recommend rebuilding with a different prefix, as "/tmp/gp" is a really bad place for installing applications!
Fixed in CVS. I added a static 400ms delay between HAL device detection and attempting a sync.
Will this be fixed in Ubuntu Edgy too then? Or should i install cvs?
I don't know what their plans are for Edgy, but a package based on the interim fix was uploaded to Feisty during the week. If you feel like installing CVS, I'd advise including the stock Edgy gnome-pilot patch to ensure you get the path names to match the Ubuntu choices. Note that an even better workaround for Edgy is to configure libusb syncing. This speeds up your syncing, too.
*** Bug 377116 has been marked as a duplicate of this bug. ***
gnome-pilot-2.0.15 has just been released, to fix this and one or two other recent bugs.