Re: [Touchd-devel] Interested by touchd

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

LeTic,
What would help now is knowing more about your system.  What distro?  
What architecture?  An lspci.  What version of dbus do you have?  What 
version of OpenUSB?  What kernel?  I don't think its a problem with my 
code.  However, theres always a chance there is.
Thanks,
Scott

Anthony Callegaro wrote:
> Hey Scott,
>
> Cheers for that. I used your new code and here is the result of my 
> investigation :
>
> If I run touchd in the terminal this is what I get :
>  ./touchd
> ** (process:27596): DEBUG: Running touchd version alpha 2.
> ** (process:27596): DEBUG: Connecting to the Session D-Bus.
> ** (process:27596): DEBUG: Registering the well-known name 
> (org.tannewt.touchd).
> ** (process:27596): DEBUG: RequestName returned 1.
> ** (process:27596): DEBUG: Creating one Touchd object.
> ** (process:27596): DEBUG: Initializing Touchd class.
> ** (process:27596): DEBUG: Creating the signals.
> ** (process:27596): DEBUG: Binding to GLib/D-Bus.
> ** (process:27596): DEBUG: Initializing TouchdObject.
> ** (process:27596): DEBUG: Registering it on the D-Bus.
> ** (process:27596): DEBUG: Ready to serve requests.
> ** (process:27596): DEBUG: Setting up uinput device.
> ** (process:27596): DEBUG: Running the device.
> ** (process:27596): DEBUG: Initializing the hardware.
> Segmentation fault
>
> I next tried in gdb :
> (gdb) run
> Starting program: /home/letic/Desktop/touchd-svn/src/touchd
> [Thread debugging using libthread_db enabled]
> ** (process:8084): DEBUG: Running touchd version alpha 2.
> ** (process:8084): DEBUG: Connecting to the Session D-Bus.
> ** (process:8084): DEBUG: Registering the well-known name 
> (org.tannewt.touchd).
> ** (process:8084): DEBUG: RequestName returned 1.
> ** (process:8084): DEBUG: Creating one Touchd object.
> ** (process:8084): DEBUG: Initializing Touchd class.
> ** (process:8084): DEBUG: Creating the signals.
> ** (process:8084): DEBUG: Binding to GLib/D-Bus.
> ** (process:8084): DEBUG: Initializing TouchdObject.
> ** (process:8084): DEBUG: Registering it on the D-Bus.
> ** (process:8084): DEBUG: Ready to serve requests.
> ** (process:8084): DEBUG: Setting up uinput device.
> ** (process:8084): DEBUG: Running the device.
> [New Thread 0xb7bd5720 (LWP 8084)]
> [New Thread 0xb7bd4b90 (LWP 8093)]
> ** (process:8084): DEBUG: Initializing the hardware.
> [New Thread 0xb73d3b90 (LWP 8095)]
> [New Thread 0xb6bacb90 (LWP 8101)]
>
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0xb6bacb90 (LWP 8101)]
> 0xb7ca6463 in memmove () from /lib/tls/i686/cmov/libc.so.6
> (gdb) bt
> #0  0xb7ca6463 in memmove () from /lib/tls/i686/cmov/libc.so.6
> #1  0xb7ed13b2 in ?? () from /usr/lib/libdbus-1.so.3
> #2  0x080674a0 in ?? ()
> #3  0x0806947f in ?? ()
> #4  0xffffe021 in ?? ()
> #5  0xb7ecf479 in ?? () from /usr/lib/libdbus-1.so.3
> #6  0x08059be0 in ?? ()
> #7  0x0805a470 in ?? ()
> #8  0x00000000 in ?? ()
>
> As I don't know gdb that well I then used nemiver (gnome frontend for 
> gdb) to get the source backtrace manually and here is what I get :
> So in touchd.c it hangs on line 274 :   MultitouchData* md = 
> multitouch_init(touchdObj->device->queue);
> multitouch.c line 67 :   int32_t result = openusb_init(0 , &handle );
> usb.c line 618 : usbi_rescan_devices();
> devices.c line 258 :  ibus->ops->refresh_devices(ibus);
> linux.c line 2214 : device_names = libhal_get_all_devices (hal_ctx, 
> &num_devices, &error);
>
> I can't get any further than this. I guess I would need to download 
> hal sources to follow...
>
> Let me know if I can give you more information, and what you would need.
> Cheers
> LeTic
>
> On Fri, May 2, 2008 at 3:11 AM, Scott Shawcroft 
> <sco...@gm... <mailto:sco...@gm...>> wrote:
>
>     LeTic,
>     I totally forgot!  I also checked it into subversion.
>     Thanks,
>     Scott
>
>     Anthony Callegaro wrote:
>
>         Hey Scott,
>
>         Thanks for the answer but you must have forgot the attachment
>         :o)  I thought it was in the SVN... Could you send me your new
>         version of touchd.c so I could try to debug a bit further ?
>
>         Cheers
>         LeTic
>
>         On Thu, May 1, 2008 at 3:54 AM, Scott Shawcroft
>         <sco...@gm... <mailto:sco...@gm...>
>         <mailto:sco...@gm...
>         <mailto:sco...@gm...>>> wrote:
>
>            LeTic,
>            Neat!  I've never had that problem before.  Are you running
>         Hardy
>            Heron? I agree with you that the main loop was in
>         device_new but
>            am uncertain thats where the problem was because the
>         backtrace had
>            dbus stuff in it.  In fact you can see the segfault comes from
>            process 6531 and the uinput stuff should be in 6512 still.
>
>            I've attached a new version of touchd.c with more debug
>            statements.  They'll give us a better idea about how far
>         the main
>            loop is getting.  However, I think its a dbus problem.
>            Thanks,
>            Scott
>
>            P.S. I cced the touchd-devel mailing list to archive the issue.
>             You can join it through the sourceforge project page at
>            sf.net/projects/touchd <http://sf.net/projects/touchd>
>         <http://sf.net/projects/touchd>.
>
>
>
>
>            Anthony Callegaro wrote:
>
>                Hi Tannewt,
>
>                My girlfriend bought a MacBook Pro 4.1 like yours being
>         used
>                to have Debian on our home box I installed Ubuntu for her.
>                When trying to get touchd working I hit some issues.
>
>                ./touchd
>                ** (process:6418): DEBUG: Running touchd version alpha 2.
>                ** (process:6418): DEBUG: Connecting to the Session D-Bus.
>                ** (process:6418): DEBUG: Registering the well-known name
>                (org.tannewt.touchd).
>                ** (process:6418): DEBUG: RequestName returned 1.
>                ** (process:6418): DEBUG: Creating one Touchd object.
>                ** (process:6418): DEBUG: Initializing Touchd class.
>                ** (process:6418): DEBUG: Creating the signals.
>                ** (process:6418): DEBUG: Binding to GLib/D-Bus.
>                ** (process:6418): DEBUG: Initializing TouchdObject.
>                ** (process:6418): DEBUG: Registering it on the D-Bus.
>                ** (process:6418): DEBUG: Ready to serve requests.
>                ** (process:6418): DEBUG: Setting up uinput device.
>                Segmentation fault
>
>                The backtrace in gdb isn't very verbose
>                ** (process:6512): DEBUG: Setting up uinput device.
>                [New Thread 0xb7cab720 (LWP 6512)]
>                [New Thread 0xb7caab90 (LWP 6523)]
>                [New Thread 0xb74a9b90 (LWP 6528)]
>                [New Thread 0xb6c82b90 (LWP 6531)]
>
>                Program received signal SIGSEGV, Segmentation fault.
>                [Switching to Thread 0xb6c82b90 (LWP 6531)]
>                0xb7d7c463 in memmove () from /lib/tls/i686/cmov/libc.so.6
>                (gdb) bt
>                #0  0xb7d7c463 in memmove () from
>         /lib/tls/i686/cmov/libc.so.6
>                #1  0xb7fa73b2 in ?? () from /usr/lib/libdbus-1.so.3
>                #2  0x080672a0 in ?? ()
>                #3  0x0806927f in ?? ()
>                #4  0xffffe042 in ?? ()
>                #5  0xb7fa5479 in ?? () from /usr/lib/libdbus-1.so.3
>                #6  0x08059b60 in ?? ()
>                #7  0x0805a3f0 in ?? ()
>                #8  0x00000000 in ?? ()
>
>                It might be coming from :
>                touchdObj->device = device_new();
>
>                But my uinput device is correctly created in
>         /dev/input/uinput
>
>                Any idea to get me started ?
>
>                Anyway I would be very interested in helping you in this
>                project. Let me know what I can do.
>                Cheers
>                LeTic
>
>
>