On Monday, January 10, 2011 12:08:42 Hans Verkuil wrote: > > Hi, > > > >>> Another reason why s_config is a bad idea. > > > > Thanks a lot for working on this. I had a quick look and don't have > > any objections. > > > >>> This has been extensively tested on my humble OLPC laptop (and it took > >>> me 4-5 > >>> hours just to get the damn thing up and running with these drivers). > > > > In future, come into irc.oftc.net #olpc-devel and talk to me (dsd) or > > cjb (Chris Ball), we'll get you up and running in less time! > > If you have a link to some instructions on how to get the latest kernel up > and running for olpc, then that would be handy :-) > > > I'll test the via-camera patch unless Jon beats me too it, but won't > > be immediately. If you are ever interested in doing more in-depth work > > on that driver, please drop me a mail and we will send you a XO-1.5. > > It's just for ongoing V4L2 maintenance (such as this sort of work). > > > > > Also, perhaps you are interested in working on this bug, which is > > probably reproducible with cafe_ccic too: > > http://www.mail-archive.com/linux-media@xxxxxxxxxxxxxxx/msg23841.html > > I'll see if I can reproduce this with cafe_ccic. Weird that I haven't seen > this before. The code looks fishy: my first guess is that sd->flags should > be put in a local variable before it is being tested. > > I must have missed that bug report the first time around. I'm pretty sure that this simple patch will fix it: diff --git a/drivers/media/video/v4l2-device.c b/drivers/media/video/v4l2-device.c index 7fe6f92..83a2e69 100644 --- a/drivers/media/video/v4l2-device.c +++ b/drivers/media/video/v4l2-device.c @@ -100,6 +100,7 @@ void v4l2_device_unregister(struct v4l2_device *v4l2_dev) is a platform bus, then it is never deleted. */ if (client) i2c_unregister_device(client); + continue; } #endif #if defined(CONFIG_SPI) @@ -108,6 +109,7 @@ void v4l2_device_unregister(struct v4l2_device *v4l2_dev) if (spi) spi_unregister_device(spi); + continue; } #endif } If you have time, then I'd appreciate it if you could check that this fixes it for via_camera. I'll try to test with on cafe_ccic as well some time this week. This bug is only triggered when a i2c module is unloaded on a platform i2c bus. This doesn't happen on PCI/USB i2c busses, which is the reason why I never saw it. Regards, Hans -- Hans Verkuil - video4linux developer - sponsored by Cisco -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
