Launchpad has imported 15 comments from the remote bug at https://bugzilla.kernel.org/show_bug.cgi?id=204275.
If you reply to an imported comment from within Launchpad, your comment will be sent to the remote bug automatically. Read more about Launchpad's inter-bugtracker facilities at https://help.launchpad.net/InterBugTracking. ------------------------------------------------------------------------ On 2019-07-22T13:36:29+00:00 steve wrote: Many bluetooth HID keyboard profile devices disconnect on idle to save power, I have two such devices: Logitech Y-X5A77 keyboard Sony PS3 bluetooth remote control When the device disconnects on idle the g_io_channels get disconnected bluetoothd[3167]: profiles/input/device.c:ctrl_watch_cb() Device 00:07:61:F6:C9:7D disconnected bluetoothd[3167]: profiles/input/device.c:intr_watch_cb() Device 00:07:61:F6:C9:7D disconnected But, something isn't closed properly (possibly g_io_add_watch) since bluetoothd then spins at 100% CPU in g_main_loop_run() until the device reconnects, at which point every works normally again until the device next idles. This bug has been present for at least 2 years, I'm certain it's much longer: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796 Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/11 ------------------------------------------------------------------------ On 2019-07-22T14:22:28+00:00 steve wrote: I've tried adding g_source_remove()s for ctrl_watch and intr_watch before shutting down the io channels but it doesn't make any difference. Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/12 ------------------------------------------------------------------------ On 2019-07-22T19:36:06+00:00 steve wrote: strace results in continuous repeating: poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}, {fd=8, events=POLLIN}, {fd=10, events=POLLIN}, {fd=11, events=POLLIN}, {fd=12, events=POLLIN}, {fd=16, events=POLLIN}, {fd=17, events=POLLIN}, {fd=18, events=POLLIN}, {fd=19, events=POLLIN}, {fd=20, events=POLLIN}, {fd=21, events=POLLIN}, {fd=22, events=POLLIN}, {fd=23, events=POLLIN}, {fd=42, events=POLLIN}, {fd=43, events=POLLIN}, {fd=44, events=POLLIN}, {fd=45, events=POLLIN}, {fd=46, events=POLLIN}, {fd=47, events=POLLIN}, {fd=48, events=POLLIN}, {fd=49, events=POLLIN}, {fd=50, events=POLLIN}, {fd=51, events=POLLIN}, {fd=52, events=POLLIN}, {fd=53, events=POLLIN}, {fd=55, events=POLLOUT}], 28, -1) = 1 ([{fd=55, revents=POLLNVAL}]) Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/13 ------------------------------------------------------------------------ On 2019-07-22T20:54:45+00:00 steve wrote: Ahhh... I was already aware that on first connect the HID input device wasn't created, but I hadn't realised it's part of the same bug. What's happening is that on initial connect: dev->sec_watch = g_io_add_watch(idev->intr_io, G_IO_OUT, encrypt_notify, idev); creates the notify callback. But it never triggers. On second connection the callback gets triggered, but the connect code gets run again so another refcount is added. This means the intr channel never gets closed which means it's stuck on the last event. Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/14 ------------------------------------------------------------------------ On 2019-07-22T21:00:23+00:00 steve wrote: I flipped the G_IO_OUT to G_IO_IN and it started working on first connect. I don't know why it was G_IO_OUT? Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/15 ------------------------------------------------------------------------ On 2019-07-22T23:56:16+00:00 steve wrote: Created attachment 283925 Patch to fix high CPU usage with HID keyboard and other devices Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/16 ------------------------------------------------------------------------ On 2019-07-24T11:14:41+00:00 luiz.dentz wrote: Ive sent a fix upstream: https://lore.kernel.org/linux- bluetooth/20190724110151.4258-1-luiz.de...@gmail.com/T/#u Let me know if that works. Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/17 ------------------------------------------------------------------------ On 2019-07-24T15:31:20+00:00 steve wrote: (In reply to Luiz Von Dentz from comment #6) > Ive sent a fix upstream: > > https://lore.kernel.org/linux-bluetooth/20190724110151.4258-1-luiz. > de...@gmail.com/T/#u > > Let me know if that works. That will prevent the channel getting left dangling, but it doesn't address the issue of the callback not happening on initial connect, which is AFAICT what results in that part of the bug for me. Is it really supposed to be waiting for G_IO_OUT? It seems it only gets triggered for me when a second connection is attempted on the dangling channel, presumably during handshake..? Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/18 ------------------------------------------------------------------------ On 2019-07-25T09:04:46+00:00 steve wrote: Luiz, did you see (In reply to Luiz Von Dentz from comment #6) > Ive sent a fix upstream: > > https://lore.kernel.org/linux-bluetooth/20190724110151.4258-1-luiz. > de...@gmail.com/T/#u > > Let me know if that works. Luiz, did you see the patch I attached? Perhaps it's better to remove the old watcher first than not add another if it exists, but it shouldn't happen anyway. The explicit remove should happen to catch the case where the connection is broken before being fully initialised but the callback should get triggered once data starts to flow over the encrypted channel. In my tests, allowing the channel to fully close by removing the sec_watch prevents the keyboard from ever connecting, this is because the callback never gets triggered. I believe the bug is caused by the refcount sec_watch creates not being automatically removed because the callback is left pending, it doesn't get removed on channel shutdown, but has the side-effect of making it work with the channel being already open when the device reconnects because it triggers the callback. What made it work every time is changing sec_watch to wait for G_IO_IN instead of G_IO_OUT, but presumably it is G_IO_OUT for a reason? Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/19 ------------------------------------------------------------------------ On 2019-07-25T13:13:34+00:00 luiz.dentz wrote: Im not sure we even need the watch in the first place, the kernel should block any in/out until the connection is encrypted (BT_SK_SUSPEND) so it might be possible to get rid of sec_watch altogether. Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/20 ------------------------------------------------------------------------ On 2019-07-29T13:52:47+00:00 steve wrote: (In reply to Luiz Von Dentz from comment #9) > Im not sure we even need the watch in the first place, the kernel should > block any in/out until the connection is encrypted (BT_SK_SUSPEND) so it > might be possible to get rid of sec_watch altogether. Going through the history it originally worked that way, I'm not sure why it was changed..? Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/21 ------------------------------------------------------------------------ On 2019-07-29T13:56:22+00:00 luiz.dentz wrote: (In reply to Steven Newbury from comment #10) > (In reply to Luiz Von Dentz from comment #9) > > Im not sure we even need the watch in the first place, the kernel should > > block any in/out until the connection is encrypted (BT_SK_SUSPEND) so it > > might be possible to get rid of sec_watch altogether. > > Going through the history it originally worked that way, I'm not sure why it > was changed..? Well lets try to remove it then and see if that fix the problem. Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/22 ------------------------------------------------------------------------ On 2019-08-01T18:03:14+00:00 steve wrote: (In reply to Luiz Von Dentz from comment #11) > (In reply to Steven Newbury from comment #10) > > (In reply to Luiz Von Dentz from comment #9) > > > Im not sure we even need the watch in the first place, the kernel should > > > block any in/out until the connection is encrypted (BT_SK_SUSPEND) so it > > > might be possible to get rid of sec_watch altogether. > > > > Going through the history it originally worked that way, I'm not sure why > it > > was changed..? > > Well lets try to remove it then and see if that fix the problem. That works. I'll attach a patch on here. Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/23 ------------------------------------------------------------------------ On 2019-08-01T18:12:12+00:00 steve wrote: Created attachment 284073 Remove sec_watch, instead attempt immediate connection Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/24 ------------------------------------------------------------------------ On 2019-08-10T07:40:08+00:00 luiz.dentz wrote: (In reply to Steven Newbury from comment #13) > Created attachment 284073 [details] > Remove sec_watch, instead attempt immediate connection Could you please send as a patch to linux-bluetooth? Reply at: https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1717796/comments/25 ** Changed in: bluez Status: Unknown => Confirmed ** Changed in: bluez Importance: Unknown => Medium -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1717796 Title: Bluetoothd fills 100% CPU when bluetooth keyboard goes in idle To manage notifications about this bug go to: https://bugs.launchpad.net/bluez/+bug/1717796/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
