|
List Info
Thread: libusb and hal
|
|
| libusb and hal |
|
2007-10-09 11:31:29 |
|
Hi,
After doing some hacking in pilot-link for the z22 to get to work with
libusb (and to work at all, because visor is also no more working),
I've already made a (dirty) patch for pilot-link (see the "z22 F7
problem" thread for more details) to get pilot-xfer to work. Now the
next problem is gnome-pilot.
It came to my attention that it worked every time the first time the
z22 was connected to the computer after a reboot, and it would not work
anymore the next times. So I did some investigation and figured the
root of the problem.
When the z22 is connected, the hal_device_added function in gpilotd.c
is called a few times, each time with different hal information. only
one time this part of code will succeed:
if (!(bus = libhal_device_get_property_string (hal_ctx, udi,
"info.bus", NULL))) {
; return;
; }
; if (strcmp (bus, "usb_device") != 0 ) {
; libhal_free_string (bus);
return;
; }
So this means gpilotd is waiting for a "usb_device" to be added by hal,
which works the first time. ; But the next times the "usb_device" is not
added anymore, as this device stays available in hal. Here is a part of
the output of hal-device:
0: udi = '/org/freedesktop/Hal/devices/usb_device_830_61_noserial'
; linux.subsystem = 'usb' ; (string)
info.product = 'Palm Handheld' (string)
usb_device.device_revision_bcd = 256 (0x100) (int)
usb_device.product = 'Palm Handheld' (string)
usb_device.product_id = 97 (0x61) (int)
usb_device.speed = 12 (double)
usb_device.vendor = 'Palm, Inc.' ; (string)
usb_device.can_wake_up = false ; (bool)
usb_device.configuration_value = 1 (0x1) ; (int)
usb_device.linux.device_number = 16 (0x10) (int)
info.linux.driver = 'usb' ; (string)
usb_device.device_class = 255 (0xff) (int)
info.parent =
'/org/freedesktop/Hal/devices/usb_device_0_0_0000_00_02_0' (string)
usb_device.bus_number = 2 (0x2) ; (int)
usb_device.device_protocol = 0 (0x0) ; (int)
usb_device.is_self_powered = true (bool)
info.subsystem = 'usb_device' (string)
info.vendor = 'Palm, Inc.' ; (string)
usb_device.version = 1.1 (double)
usb_device.num_interfaces = 1 (0x1) ; (int)
usb_device.vendor_id = 2096 (0x830) (int)
usb_device.speed_bcd = 4608 (0x1200) (int)
usb_device.version_bcd = 272 (0x110) (int)
info.udi = '/org/freedesktop/Hal/devices/usb_device_830_61_noserial'
(string)
usb_device.linux.sysfs_path =
'/sys/devices/pci0000:00/0000:00:02.0/usb2/2-3' (string)
usb_device.num_configurations = 1 (0x1) ; (int)
info.bus = 'usb_device' (string)
linux.device_file = '/dev/bus/usb/002/016' (string)
usb_device.num_ports = 0 (0x0) ; (int)
usb_device.max_power = 500 (0x1f4) (int)
linux.hotplug_type = 2 (0x2) ; (int)
usb_device.device_subclass = 0 (0x0) ; (int)
linux.sysfs_path = '/sys/devices/pci0000:00/0000:00:02.0/usb2/2-3'
(string)
so if I do "hal-device -r
/org/freedesktop/Hal/devices/usb_device_830_61_noserial" this item will
be removed from the hal list, and the next time I connect my z22
gnome-pilot will work!
So now I know the cause of the problem, but I don't know how to fix
this. Should hal normally remove this device? Should gnome-pilot not
wait for an "usb_device" but for something else? ... ?
Can somebody help me on this?
Kind regards,
Tom Billiet
|
| Re: libusb and hal |
|
2007-10-09 12:32:45 |
On Tue, 2007-10-09 at 18:31 +0200, Tom Billiet wrote:
> Hi,
>
> After doing some hacking in pilot-link for the z22 to
get to work with
> libusb (and to work at all, because visor is also no
more working),
> I've already made a (dirty) patch for pilot-link (see
the "z22 F7
> problem" thread for more details) to get
pilot-xfer to work. Now the
> next problem is gnome-pilot.
>
> It came to my attention that it worked every time the
first time the
> z22 was connected to the computer after a reboot, and
it would not
> work anymore the next times. So I did some
investigation and figured
> the root of the problem.
> When the z22 is connected, the hal_device_added
function in gpilotd.c
> is called a few times, each time with different hal
information. only
> one time this part of code will succeed:
> if (!(bus = libhal_device_get_property_string (hal_ctx,
udi,
> "info.bus", NULL))) {
> return;
> }
> if (strcmp (bus, "usb_device") != 0 ) {
> libhal_free_string (bus);
> return;
> }
> So this means gpilotd is waiting for a
"usb_device" to be added by
> hal, which works the first time. But the next times
the "usb_device"
> is not added anymore, as this device stays available in
hal.
Tom,
Thanks for your work on this already. Are you saying that
the
"usb_device" is only added once, even if you
disconnect the z22
physically from your computer?
Can you watch other HAL events to see what gets generated
when you
attempt a sync? I'm sure there'll be something we can hook
into, and
then it might be time for gnome-pilot 2.0.16...
Regards,
Matt
Matt Davey If the rich could hire other people to die
for them, the
mcdavey mrao.cam.ac.uk poor could make a wonderful living -
Yiddish Proverb
_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list
|
|
| Re: libusb and hal |
|
2007-10-09 12:58:55 |
----- Original Message -----
From: Matt Davey
Time: 09-10-07 19:32
>
> Tom,
>
> Thanks for your work on this already. Are you saying
that the
> "usb_device" is only added once, even if you
disconnect the z22
> physically from your computer?
Exactly.
I must say I have not tried if there is some timeout on
this, that it
will disappear after a time. But even then you would not be
able to sync
between the timeouts.
>
> Can you watch other HAL events to see what gets
generated when you
> attempt a sync? I'm sure there'll be something we can
hook into, and
> then it might be time for gnome-pilot 2.0.16...
it seems the hal_device_added is always called at least
twice. One time
it already exits here
if (!(bus = libhal_device_get_property_string (hal_ctx, udi,
"info.bus",
NULL))) {
return;
}
So you can't do anything with that I think.
The second time it exits here
if (strcmp (bus, "usb_device") != 0 ) {
libhal_free_string (bus);
return;
}
and it seems bus it set to "usb". My best guess
is that when you get an
adding where bus = usb, you might be able to ask hal if
there is already
a palm "usb_device" available, and if so, use it.
But I have no idea how
to do that 
Regards,
Tom
_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list
|
|
| Re: libusb and hal |
|
2007-10-09 20:43:29 |
On Tue, 2007-10-09 at 19:58 +0200, Tom Billiet wrote:
>
> ----- Original Message -----
> From: Matt Davey
> Time: 09-10-07 19:32
> >
> > Tom,
> >
> > Thanks for your work on this already. Are you
saying that the
> > "usb_device" is only added once, even if
you disconnect the z22
> > physically from your computer?
>
> Exactly.
> I must say I have not tried if there is some timeout on
this, that it
> will disappear after a time. But even then you would
not be able to sync
> between the timeouts.
>
> >
> > Can you watch other HAL events to see what gets
generated when you
> > attempt a sync? I'm sure there'll be something we
can hook into, and
> > then it might be time for gnome-pilot 2.0.16...
>
> it seems the hal_device_added is always called at least
twice. One time
> it already exits here
> if (!(bus = libhal_device_get_property_string (hal_ctx,
udi, "info.bus",
> NULL))) {
> return;
> }
> So you can't do anything with that I think.
>
> The second time it exits here
> if (strcmp (bus, "usb_device") != 0 ) {
> libhal_free_string (bus);
> return;
> }
> and it seems bus it set to "usb". My best
guess is that when you get an
> adding where bus = usb, you might be able to ask hal if
there is already
> a palm "usb_device" available, and if so, use
it. But I have no idea how
> to do that
>
> Regards,
> Tom
I don't know beans about USB, but thinking about what you've
described
in this series of e-mails, it seems to me that you need to
beware of the
case of HAL remembering a Palm device that was formerly
attached but is
no longer there. That in turn makes me think that if HAL
is
remembering a device that is no longer there, then maybe the
problem is
in the HAL logic. Are the supporters of that code on this
mailing list
or is that a kernel topic?
Drew
_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list
|
|
| Re: libusb and hal |
|
2007-10-10 04:21:01 |
On Tue, 2007-10-09 at 21:43 -0400, R. Drew Davis wrote:
> On Tue, 2007-10-09 at 19:58 +0200, Tom Billiet wrote:
> >
> > ----- Original Message -----
> > From: Matt Davey
> > Time: 09-10-07 19:32
> > >
> > > Tom,
> > >
> > > Thanks for your work on this already. Are
you saying that the
> > > "usb_device" is only added once,
even if you disconnect the z22
> > > physically from your computer?
> >
> > Exactly.
> > I must say I have not tried if there is some
timeout on this, that it
> > will disappear after a time. But even then you
would not be able to sync
> > between the timeouts.
> >
>
> I don't know beans about USB, but thinking about what
you've described
> in this series of e-mails, it seems to me that you need
to beware of the
> case of HAL remembering a Palm device that was formerly
attached but is
> no longer there. That in turn makes me think that if
HAL is
> remembering a device that is no longer there, then
maybe the problem is
> in the HAL logic. Are the supporters of that code on
this mailing list
> or is that a kernel topic?
Drew, I agree. I haven't had time to try reproducing this
yet, but this
feels like a regression in the HAL behaviour. I know that
multiple
libusb / HAL syncing worked just fine for me. Either it is
device
specific or is a regression outside of gnome-pilot.
Tom, what HAL version do you have installed? It would be
great if we
could narrow this down to a HAL change, but I'll fire off a
message to
the hal list once I know what version this behaviour has
been seen in.
Matt
_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list
|
|
| Re: libusb and hal |
|
2007-10-10 11:13:51 |
----- Original Message -----
From: Matt Davey
Time: 10-10-07 11:21
> On Tue, 2007-10-09 at 21:43 -0400, R. Drew Davis
wrote:
>> On Tue, 2007-10-09 at 19:58 +0200, Tom Billiet
wrote:
>>> ----- Original Message -----
>>> From: Matt Davey
>>> Time: 09-10-07 19:32
>>>> Tom,
>>>>
>>>> Thanks for your work on this already. Are
you saying that the
>>>> "usb_device" is only added once,
even if you disconnect the z22
>>>> physically from your computer?
>>> Exactly.
>>> I must say I have not tried if there is some
timeout on this, that it
>>> will disappear after a time. But even then you
would not be able to sync
>>> between the timeouts.
>>>
>> I don't know beans about USB, but thinking about
what you've described
>> in this series of e-mails, it seems to me that you
need to beware of the
>> case of HAL remembering a Palm device that was
formerly attached but is
>> no longer there. That in turn makes me think that
if HAL is
>> remembering a device that is no longer there, then
maybe the problem is
>> in the HAL logic. Are the supporters of that code
on this mailing list
>> or is that a kernel topic?
>
> Drew, I agree. I haven't had time to try reproducing
this yet, but this
> feels like a regression in the HAL behaviour. I know
that multiple
> libusb / HAL syncing worked just fine for me. Either
it is device
> specific or is a regression outside of gnome-pilot.
>
I agree on this too. In my opinion hal should not keep
disconnected
devices, or at least notify again if it's reconnected
>
> Tom, what HAL version do you have installed? It would
be great if we
> could narrow this down to a HAL change, but I'll fire
off a message to
> the hal list once I know what version this behaviour
has been seen in.
>
hal: 0.5.9.1
hal-info: 0.20070831
kernel: 2.6.22-ARCH
udev: 115
I'm using archlinux, which basically means I almost always
have the
latest packages.
If you need any more versions of a certain package, you can
find it
here: http://archlinux.org/packages/search/?sort=-last_update
Regards,
Tom
_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list
|
|
| Re: libusb and hal |
|
2007-10-13 14:26:13 |
|
Hi,
I'm also very interested in how this turns out and am willing to do some testing upon request.
I have seen the same failure where syncing works the first time after boot, but intermittently there after. I have a Fedora 7 machine that is kept consistently up to date with all patches, including the kernel. Currently, here is my version info as reported by rpm:
bash-3.2$ rpm -q hal
hal-0.5.9-8.fc7
; bash-3.2$ rpm -q kernel
kernel-2.6.22.9-91.fc7
bash-3.2$ uname -a
Linux hungover.domain.com 2.6.22.9-91.fc7 #1 SMP Thu Sep 27 20:47:39 EDT 2007 x86_64 x86_64 x86_64 GNU/Linux
bash-3.2$
Thanks,
Brent
On 10/10/07, Tom Billiet < mouse256ulyssis.org">mouse256ulyssis.org> wrote:
----- Original Message -----
From: Matt Davey
Time: 10-10-07 11:21
> On Tue, 2007-10-09 at 21:43 -0400, R. Drew Davis wrote:
>;> On Tue, 2007-10-09 at 19:58 +0200, Tom Billiet wrote:
>;>> ----- Original Message -----
>>> From: Matt Davey
>>> Time: 09-10-07 19:32
>>>>; Tom,
>>>>
>>>> Thanks for your work on this already. Are you saying that the
>>>> "usb_device"; is only added once, even if you disconnect the z22
>>>>; physically from your computer?
>>>; Exactly.
>>> I must say I have not tried if there is some timeout on this, that it
>>;> will disappear after a time. But even then you would not be able to sync
>;>> between the timeouts.
>>>;
>> I don't know beans about USB, but thinking about what you've described
>> in this series of e-mails, it seems to me that you need to beware of the
>> case of HAL remembering a Palm device that was formerly attached but is
>>; no longer there. That in turn makes me think that if HAL is
>>; remembering a device that is no longer there, then maybe the problem is
>> in the HAL logic. Are the supporters of that code on this mailing list
>> or is that a kernel topic?
>;
> Drew, I agree. I haven't had time to try reproducing this yet, but this
>; feels like a regression in the HAL behaviour. ;I know that multiple
> libusb / HAL syncing worked just fine for me. Either it is device
>; specific or is a regression outside of gnome-pilot.
>
I agree on this too. In my opinion hal should not keep disconnected
devices, or at least notify again if it's reconnected
>
> Tom, what HAL version do you have installed? ;It would be great if we
> could narrow this down to a HAL change, but I'll fire off a message to
> the hal list once I know what version this behaviour has been seen in.
>
hal: 0.5.9.1
hal-info: 0.20070831
kernel: 2.6.22-ARCH
udev: 115
I';m using archlinux, which basically means I almost always have the
latest packages.
If you need any more versions of a certain package, you can find it
here: http://archlinux.org/packages/search/?sort=-last_update
Regards,
Tom
_______________________________________________
gnome-pilot-list mailing list
gnome-pilot-listgnome.org">gnome-pilot-listgnome.org
http://mail.gnome.org/mailman/listinfo/gnome-pilot-list
|
[1-7]
|
|