CEC always broken on Philips

not on my device. Works like a charm on 9.2.5 and all versions after it does not. So something got broken in the process and remains broken ever since

Since it also works flawless on my N2 and N2+ with CE 19.2 it seems that your device(s) are causing your problems.

Could be a layer 8 issue

Make clean fresh install on 19.2 release. Disconnect every cables and also cut power of the TV. After 5min reconnect everything and boot and setup your N2.

It’s working, or you will be one of thousands where it does not work.

Also you did not say if you have an AVR and if the N2 is directly connected to the TV or not.
On some setups it’s needed to manually adjust the phy address.

I have tried a fresh install on a different card with the device. It does not work. No CEC commands are successful. I revert the version to 9.2.5 and it starts working again. So it is not hardware it is software. Something got changed moving forward after 9.2.5 and for me it breaks CEC. So please stop insulting the user.

Also you did not say if you have an AVR and if the N2 is directly connected to the TV or not.
On some setups it’s needed to manually adjust the phy address.

Sorry about that: yes there is a AVR in between. Again there is no changes needed to make it work on 9.2.5.

Then it’s your AVR . Update to newest CE make logs what you didn’t have done yet and solve your isst by this instructions:

my address is set correct and has been for years. Again it works on older versions. But toying around I realized that when I change the physical address around, regardless to if it is a valid one or not CEC starts working. So there is a problem with it not initializing or something like that.

so here is my log: http://ix.io/3BU3

I started the system with CEC not working (THE ADDRESS IS SET CORRECTLY). I switch the physical address back and forth a couple of times (with the receiver changing inputs as it should), while CEC remains inoperable. In the end switching back to the same address as before and CEC starts getting picked up again for some reason and the logs shows correctly identifying my AVR. So again I ask what I need to do to have CEC working reliably.

something, maybe the avr or the tv block it:

2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - using provided physical address 1200
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - making Recorder 1 (1) the active source
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - marking Broadcast (F) as inactive source
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - >> source activated: Recorder 1 (1)
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - scheduling active source message for 'Recorder 1'
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - CEC client registered: libCEC version = 6.0.2, client version = 4.0.0, firmware version = 5, logical address(es) = Recorder 1 (1) , physical address: 1.2.0.0,  compiled on 2021-10-15 00:37:33 by relkai@hammbwcor01 on Linux 4.15.0-135-generic (x86_64), features: P8_USB, DRM, P8_detect, AOCEC
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - Recorder 1 (1): vendor = Philips (00903e)
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - replacing the command handler for device 'Recorder 1' (1)
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - << Recorder 1 (1) -> TV (0): OSD name 'Kodi'
2021-10-16 12:51:08.300 T:4899    DEBUG <general>: CecLogMessage - << 10:47:4b:6f:64:69
2021-10-16 12:51:08.301 T:4899  WARNING <general>: CecLogMessage - Write: write failed
2021-10-16 12:51:08.301 T:4899    DEBUG <general>: CecLogMessage - << 10:47:4b:6f:64:69
2021-10-16 12:51:08.301 T:4899  WARNING <general>: CecLogMessage - Write: write failed
2021-10-16 12:51:08.301 T:4899    DEBUG <general>: CecLogMessage - << requesting power status of 'TV' (0)
2021-10-16 12:51:08.301 T:4899    DEBUG <general>: CecLogMessage - << 10:8f
2021-10-16 12:51:08.301 T:4899  WARNING <general>: CecLogMessage - Write: write failed
2021-10-16 12:51:08.301 T:4899    DEBUG <general>: CecLogMessage - << 10:8f
2021-10-16 12:51:08.301 T:4900    DEBUG <general>: OpenConnection - connection to the CEC adapter opened
2021-10-16 12:51:08.301 T:4899  WARNING <general>: CecLogMessage - Write: write failed

AVR is on HDMI1 on TV and N2 on HDMI2 on AVR?

yes the AVR is HDMI1 for the TV and the N2 is plugged in to the second HDMI of the AVR.

maybe the avr or the tv block it

well on 9.2.5 are no issues. Never had any with any RPIs I used before I got the N2. So whatever the problem is, it has to do with the software

IMO it may be a timing issue when devices powering on

2021-10-16 12:51:08.241 T:4899 DEBUG : CecLogMessage - Broadcast (F): power status changed from ‘unknown’ to ‘on’
2021-10-16 12:51:08.241 T:4899 DEBUG : CecLogMessage - making Broadcast (f) the active source
2021-10-16 12:51:08.241 T:4899 DEBUG : CecLogMessage - TV (0): power status changed from ‘unknown’ to ‘in transition from standby to on’
2021-10-16 12:51:08.241 T:4899 DEBUG : CecLogMessage - sending active source message for ‘Broadcast’

Looks CEC just goes on & does not wait until TV(0) = ‘on’.
Maybe recheck with TV & AVR already on when CE powers up, just to narrow down the issue.

I tried rebooting the device while TV/AVR stay on multiple times. Does not help

I just checked your log once more:
CecLogMessage - TV (0): vendor = Philips (00903e)

Philips are known as problematic with CEC. Please search the forum, there is a libCEC setting what needs to be disabled to work properly with the TV.

The problem is libCEC, not CE.
But libCEC stopped their project. You can open a issue here:

regardless this TV worked with CEC for 10 years now and I have been using the pulse eight adapter and other libcec devices for all this time.

there is a libCEC setting what needs to be disabled to work properly with the TV

it all worked prior and not changes to settings were made. What was changed between 9.2.5 and 9.2.6 that could effect CEC? I see lots of postings in discussion threads complaining about CEC problems and regression after 9.2.5, they were just ignored.

So the solution now is to stick to 9.2.5 until I get a new board without CE?

It’s common to have updates that solve the problem to the majority of the users but introduce problems to a minority. We can’t solve everything even though I think that team CE is pretty good at solving most of the problems and caring for users needs. That was why CoreELEC was created. Otherwise there wouldn’t be any working solution for amlogic boards.
If you are using open/libre/CoreELEC for 10 years you know that CE is the best one out there. And that’s not because we ignore users, it’s because we listen to them.
Some problems are just hard to solve, and CEC is, and always was, a mess.
There are some pointers on the topics to try and solve the problem. Try them if you didn’t already. Start by connecting the board directly to the TV. See how that goes.

I tried searching the forums prior to making the post. I also posted the logs, but what I was told is to buy a new TV as they don’t like the vendor. If CEC would have never worked that might be more understandable, but CEC has been working for years and just got broken by more recent changes. As it sometimes is working all points to being a timing issue of some sorts, which should be addressable. My settings are correct, the setup can’t be changed. So I don’t see what I could do myself, which is why I am requesting help

I pointed you already where you can get help:

It’s a libCEC CEC client handling issue what we can not solve as we have no equipment or time for this. So you need to stay at the old version until the libCEC author will fix it or until the TV vendor do implement the “standard” CEC handling. The hardware and Linux kernel does work properly, the vendor and libCEC CEC implementation is faulty.

And I hope you do not have the USB Pulse Eight adapter connected on the N2 as well!?
The USB adapter is only needed for devices what do not have a CEC client hardware.

no the USB adapter I use on different devices and is not part of the device chain in question

It’s a libCEC CEC client handling issue

which is why I’d like to be able to stick to the CEC version that was working and still is working reliable. But it got dropped. Why?

Because it solve issues for other thousands users?