[09:43] <hverkuil> mchehab: media_build should build the frame_vector.ko module. But it may not be installed. Can you check?
[09:47] <mchehab> sure
[09:49] <mchehab> it was installed, but somehow make install didn't run depmod -a
[09:50] <mchehab> or it ran, but didn't update mm/frame_vector.ko
[09:50] <mchehab> (i guess the latter, as, before manually running depmod -a, it was finding vivid)
[09:51] <hverkuil> OK. Good to know it works :-)
[09:51] <mchehab> yes, thanks!
[09:52] <mchehab> that's the machine with my DVB signal generators and ivtv...
[09:53] <mchehab> I don't want to upgrade it, as I'm unable to compile newer versions of DekTec Linux tools there
[09:54] <mchehab> it seems they broke the kernelspace API on some changes (or they never tested it with a Kernel 4.3)
[09:54] <mchehab> so, I'm forced to stick with an older kernel version
[10:41] <mchehab> hverkuil:
[10:41] <mchehab> [ 2751.680030] BUG: unable to handle kernel NULL pointer dereference at 0000000000000014
[10:41] <mchehab> [ 2751.680083] IP: [<ffffffffa06ed3be>] __vb2_perform_fileio+0x33e/0x5c0 [videobuf2_core]
[10:42] <mchehab> [ 2751.681558] Call Trace:
[10:42] <mchehab> [ 2751.681576]  [<ffffffffa06ed654>] vb2_read+0x14/0x20 [videobuf2_core]
[10:42] <mchehab> [ 2751.681610]  [<ffffffffa06f4c5c>] vb2_fop_read+0xbc/0x120 [videobuf2_v4l2]
[10:42] <mchehab> this happened with vivid, after starting/stopping stream several times
[10:44] <mchehab> 0x512e is in __vb2_perform_fileio (/home/mchehab/media_build/v4l/videobuf2-core.c:2593).
[10:44] <mchehab> 2592			if (is_multiplanar && read &&
[10:44] <mchehab> 2593					b->planes[0].data_offset < buf->size) {
[10:44] <mchehab> 2594				buf->pos = b->planes[0].data_offset;
[10:44] <mchehab> 2595				buf->size -= buf->pos;
[10:45] <hverkuil> That's the vb2 regression. Is that missing from your kernel?
[10:45] <hverkuil> (at least, it looks awfully familiar)
[10:46] <hverkuil> the fix should be in rc3
[10:46] <mchehab> maybe the tarball was not updated
[10:46] <mchehab> (it is media_build)
[10:47] <mchehab> linux-media-LATEST.tar.bz2 -> linux-media-2016-01-26-e545ac872ff8.tar.bz2
[10:47] <hverkuil> old
[10:47] <mchehab> yes
[10:47] <mchehab> probably something happened at the server
[10:47] <mchehab> I'm looking into it
[10:48] <hverkuil> Note that the media-commits mailinglist no longer seems to work either. No messages appear when you merge something.
[10:49] <mchehab> I'll check it too
[10:49] <mchehab> media-commits issue seems to happened after the migration
[10:58] <mchehab> tarball generation fixed. somehow, the directory at the server where it has a clone of media_build.git disappeared
[11:05] <mchehab> hverkuil: I know why depmod -a didn't work:
[11:05] <mchehab> /sbin/depmod -a 3.19.8-100.fc20.x86_64
[11:05] <mchehab> make -C firmware install
[11:05] <mchehab> ...
[11:06] <mchehab> Installing /lib/modules/3.19.8-100.fc20.x86_64/kernel/mm files:
[11:06] <mchehab> frame_vector.ko
[11:06] <mchehab> it should install the frame_vector.ko *before* calling depmod, and not after that
[11:17] <mchehab> fixed
[11:23] <mchehab> fixing the media commits ML issue would require to recover the hooks we used to update it from the old server
[11:23] <mchehab> it is a shame we didn't notice this earlier, as the old server is now decommisioned
[11:29] <mchehab> I found an old copy
[11:44] <mchehab> hverkuil: I fixed the issue. it should now be producing e-mails for any changes on media_tree, media_build, v4l-utils and dtv-scan-table
[11:46] <hverkuil> Yep, got the v4l-utils commit. Thanks!
[11:47] <mchehab> anytime
[11:47] <mchehab> let's see if we'll get other commits along the time
[11:47] <mchehab> that one was produced by running the script by hand at the server ;)