... lot of changes since 2.6.24 in evdev.c and ... changes that would cause the input devices > > > > > to be free'd ... s parent only when child kobject > > > > was released (in kobject_cleanup). The ... evdev, evdev->grab); - /* evdev is marked dead so no one else ...

linux-kernel - Dmitry Torokhov - Mar 31 2008 - 16:57

... of the problem. IMHO, my patch is the right thing to ... releasing a grab on the underlying input device from > within evdev ... So > AFAICT we're really just looking for "why do we ... is there another bug?" > If device is being disconnected (rdestroyed) then ...

linux-kernel - Dmitry Torokhov - Mar 31 2008 - 16:46

... ones where we accidentally were grabbing too many references > and then ... releasing a grab on the underlying input device from within evdev ... So AFAICT we're really just looking for "why do we ... In-Reply-To header when sending the patch. The original thread ...

linux-kernel - Björn - Mar 30 2008 - 18:42

... It does smell like some core device layer change, because we do ... releasing a grab on the underlying input device from > within evdev ... So > AFAICT we're really just looking for "why do we ... In-Reply-To header when sending the patch. The > original thread, ...

linux-kernel - Arjan van de Ven - Mar 30 2008 - 19:19

... changes that would cause the input devices > > to be free'd ... 's parent only when child kobject > was released (in kobject_cleanup). The ... Perhaps you just need to grab another reference as this would ... will look how to properly fix evdev and the rest of ...

linux-kernel - Greg KH - Mar 31 2008 - 13:28

... lot of changes since 2.6.24 in evdev.c and ... changes that would cause the input devices > > > > to be free'd ... 's parent only when child kobject > > > was released (in kobject_cleanup). The ... Perhaps you just need to grab another reference as this > > would ...

linux-kernel - Greg KH - Mar 31 2008 - 16:42

... refcounting changes that would cause the input devices > > > > > > to be free'd earlier ... kobject's parent only when child kobject > > > > > was released (in kobject_cleanup). The ... ? Perhaps you just need to grab another reference as this > > > > would ...

linux-kernel - Greg KH - Mar 31 2008 - 18:09

... /evdev.c > +++ b/drivers/input/evdev.c > @@ -853,6 +853,9 @@ static void evdev_cleanup(struct evdev *evdev) > evdev_hangup(evdev); > evdev_remove_chrdev(evdev); > =20 > + if ( ... everything else that manipulates the grab needs to take the mutex. ...

linux-kernel - Johannes Berg - Mar 31 2008 - 17:02

... 2.6.24 in evdev.c and input.c that > > > seem ... changes that would cause the input devices > > > to be free'd ... 's parent only when child kobject > > was released (in kobject_cleanup). The ... Perhaps you just need to grab another reference as this > would ...

linux-kernel - Dmitry Torokhov - Mar 31 2008 - 14:01

... changes that would cause the input devices > to be free'd ... 's parent only when child kobject was released (in kobject_cleanup). The changeset above moves the release to kobject_del() which is way too ... will look how to properly fix evdev and the rest of ...

linux-kernel - Dmitry Torokhov - Mar 31 2008 - 02:15

... It does smell like some core device layer change, because we do ... since 2.6.24 in evdev.c and input.c that > seem ... were ones where we accidentally were grabbing too many references and then ... I have a lot of removable input devices to test with. thanks, ...

linux-kernel - Greg KH - Mar 30 2008 - 18:22

... X console switches or closes the evdev, whatever, > kaboom ... > > sd 0:0:0: ... > usbcore: deregistering interface driver appletouch > input: appletouch disconnected > PM: Syncing filesystems ... ... the patch that was just posted by Dmitry for the evdev ...

linux-kernel - Greg KH - Mar 31 2008 - 19:51

... X console switches or closes the evdev, whatever, kaboom ... sd 0:0:0: ... usbcore: deregistering interface driver appletouch input: appletouch disconnected PM: Syncing filesystems ... aeb589f151c28f3f ]--- agpgart: Putting AGP V2 device at 0000:00:0b.0 ...

linux-kernel - Benjamin Herrenschmidt - Mar 31 2008 - 19:12

... I can reproduce trivially on UP. The patch definitely helps. > Ok, I applied this ... Usage: grab /dev/input/eventX\n"); printf("Where X =3D input device number\ ... (argv[argc - 1], O_RDONLY)) < 0) { perror("grab[open]"); exit(1); } if (ioctl(fd, ...

linux-kernel - Johannes Berg - Mar 31 2008 - 16:21

... I have a lot of removable input devices to test with. Ahh, ... configure X to use /dev/input/event* devices * in an xterm, do something ... watch kernel crash as X releases the grab on the event ... who actually knows the device and input layers would take a ...

linux-kernel - Linus Torvalds - Mar 30 2008 - 18:35

... trigger. It does smell like some core device layer change, because we do not seem ... changes since 2.6.24 in evdev.c and input.c that seem relevant. Greg, are there any refcounting changes that would cause the input devices to be free'd earlier or ...

linux-kernel - Linus Torvalds - Mar 30 2008 - 17:51

... off a pointer > to a device, you need to keep the ... reference to that device around until > you are finished with it ... hit then? > It worked because evdev (and mousedev, joydev) are direct ... ask Linus to pull extended patch covering also joydev and mousedev ...

linux-kernel - Dmitry Torokhov - Mar 31 2008 - 23:30

... the parent would only be freed when > > > all its children are gone. ... , that if you remove a device from sysfs, you drop the implicit reference to the parent device, as this is no longer ... s child) which could not be released with the implicit parent ref ...

linux-kernel - Kay Sievers - Mar 31 2008 - 18:46

... done in kobject_cleanup() and so the parent would only be freed when > all its children are gone. Now parent is deleted early, ... it intentional and is here to stay then I will just grab > the references myself, although I wonder what else might be broken ...

linux-kernel - Linus Torvalds - Mar 31 2008 - 14:24

... at 16:51 -0700, Greg KH wrote: > > Can you try it with the patch that was just posted by Dmitry for the > evdev code? Yup, it works, ship it ! :-) Cheers, Ben. -- To unsubscribe from this list: ...

linux-kernel - Benjamin Herrenschmidt - Mar 31 2008 - 21:01