[00:02] *** camus has quit IRC (Remote host closed the connection) [00:03] *** camus has joined #linux-media [00:25] *** tfiga has quit IRC (Read error: Connection reset by peer) [00:25] *** tfiga has joined #linux-media [00:44] *** bjwyman_ has quit IRC (Read error: Connection reset by peer) [00:44] *** bjwyman_ has joined #linux-media [02:06] *** Ares has quit IRC (Ping timeout: 480 seconds) [04:19] *** bjwyman_ has quit IRC (Remote host closed the connection) [04:27] *** NiksDev has quit IRC (Ping timeout: 480 seconds) [04:39] *** b-rad has quit IRC (Ping timeout: 480 seconds) [05:52] *** jm_h has joined #linux-media [05:59] *** camus1 has joined #linux-media [06:00] *** camus has quit IRC (Remote host closed the connection) [06:08] *** b-rad has joined #linux-media [06:52] *** hverkuil- has quit IRC (Quit: ZNC 1.8.1+deb1 - https://znc.in) [06:52] *** hverkuil has joined #linux-media [07:00] *** NoGuest17 has quit IRC (Read error: Connection reset by peer) [07:00] *** NoGuest17 has joined #linux-media [07:01] *** NiksDev has joined #linux-media [07:21] *** ao2 has joined #linux-media [07:22] *** GBenji has joined #linux-media [07:56] *** ao2 has quit IRC (Quit: Leaving) [08:12] <mripard> hverkuil: hi, we have a bug on the RPi related to the CEC, and I'm not sure how to fix it [08:12] <mripard> basically, whenever HPD is low, the HDMI controller will be completely shut down [08:13] <mripard> on the RPi, however, any register access while the controller is powered down will stall the bus completely, which is pretty bad [08:13] <mripard> so if someone tries to use CEC during the time where the display is reported as disconnected, the board just stalls [08:14] <mripard> and even though it doesn't really make sense anyway, it's still allowed and Kodi apparently likes to do that [08:15] <mripard> the most obvious fix to me would be to make sure the device is powered as soon as adap_enable is called, but I'm not entirely sure of the side effects or if we should even try to enable it [09:15] <hverkuil> mripard: the quick fix is to set the CEC_CAP_NEEDS_HPD capability when creating the CEC adapter: then cec will be disabled when the HPD is low. This comes at a price, though: some displays pull down the HPD when in standby, so with that capability you can no longer wake them up with CEC. Most commonly seen with LG displays. [09:19] <hverkuil> Ideally the HDMI controller is enabled if adap_enabled is called. That's the best solution. [09:21] <mripard> hverkuil: yeah, it's how we've detected the issue actually, a user was getting a crash when their TV was in standby [09:21] <mripard> I've tested enabling the controller in adap_enable and cec-compliance seems to run fine, so we'll go with that then [09:21] <mripard> thanks [09:24] <hverkuil> If the HPD is low, then adap_enabled is only called if userspace actually opens the /dev/cecX device. So as long as CEC is not in use, the HDMI controller will remain powered off to avoid unnecessary power consumption. [09:36] *** ao2 has joined #linux-media [09:40] *** ao2 has quit IRC (Remote host closed the connection) [10:32] *** svarbanov has joined #linux-media [11:55] *** camus1 has quit IRC (Ping timeout: 480 seconds) [12:02] *** camus has joined #linux-media [12:42] *** camus1 has joined #linux-media [12:45] *** camus has quit IRC (Read error: Connection reset by peer) [12:51] *** Ares has joined #linux-media [13:41] *** camus has joined #linux-media [13:45] *** camus1 has quit IRC (Ping timeout: 480 seconds) [14:02] *** fleebs has quit IRC (Quit: fleebs) [14:03] *** fleebs has joined #linux-media [14:07] *** fleebs has quit IRC () [14:08] *** camus1 has joined #linux-media [14:13] *** camus has quit IRC (Ping timeout: 480 seconds) [14:54] *** camus1 has quit IRC () [15:11] *** Ares has quit IRC (Remote host closed the connection) [15:11] *** Ares has joined #linux-media [16:41] *** GBenji has left [17:03] *** gouchi has joined #linux-media [18:02] *** lexano has quit IRC (Ping timeout: 480 seconds) [18:08] *** NiksDev has quit IRC (Ping timeout: 480 seconds) [18:12] *** sailus has quit IRC (Ping timeout: 480 seconds) [18:14] *** lexano has joined #linux-media [19:31] *** ao2 has joined #linux-media [20:10] *** sailus has joined #linux-media [20:56] *** jm_h has quit IRC (Remote host closed the connection) [21:04] *** gouchi has quit IRC (Remote host closed the connection) [21:12] *** BrianG61UK has joined #linux-media [21:16] *** BrianG61UK has quit IRC () [21:16] *** BrianG61UK has joined #linux-media [21:16] *** ao2 has quit IRC (Quit: Leaving) [21:17] *** BrianG61UK_ has joined #linux-media [21:42] *** BrianG61UK_ has quit IRC (Read error: Connection reset by peer) [21:45] *** BrianG61UK_ has joined #linux-media [23:41] <pinchartl> hverkuil: could you have a look at https://bugs.libcamera.org/show_bug.cgi?id=61 ? I think we may have a problem in vb2 [23:42] <pinchartl> in a nutshell, if a QBUF call results in start_streaming being called because vb2 reaches the minimum number of buffers threshold, and if start_streaming fails, we return the failure to userspace, but the buffer is still stored in the vb2 queue