Octo card not recognized on every boot

[jshep321]

Hi,
I have a new octo connected to a pi zero W. Everything seems to work sometimes. When looking at the dmesg log, there is no "i2s mapping ok" following the "not registered" messages. A simple reboot will sometimes cause it to work.

Any ideas?

[ktb]

I have this problem too with a Pi3B. It seems like it could be some sort of timing problem. It happens less frequently on a micro SD card with Raspbian Jessie installed and more frequently on a micro SD card with Raspbian Stretch installed. It also seems to be more of a problem when using the vc4-fkms-v3d/vc4-kms-v3d overlays.

It seems the following will sometimes occur up to two or three times before registering the Octo isn't attempted any longer.

Code: Select all

[    6.585718] audioinjector-octo soc:sound: ASoC: CPU DAI (null) not registered - will retry
[    6.585739] audioinjector-octo soc:sound: snd_soc_register_card failed (-517)

A boot where registering the card fails completely is usually accompanied by the following cs42xx8 driver message

Code: Select all

[    5.755720] cs42xx8 1-0048: failed to get device ID, ret = -121
[    5.756072] cs42xx8: probe of 1-0048 failed with error -121

instead of

Code: Select all

[    5.376177] cs42xx8 1-0048: found device, revision 4

[flatmax]

OK - we haven't tested on a Zero yet, I will be getting one together over the next few week and can help you debug at that point.

It is possible that the system needs more time to settle - for whatever reason - and you will notice that in the machine driver for the octo :
https://github.com/raspberrypi/linux/bl ... oundcard.c
There are reset calls, codec_rst_gpio is set high and low with delays between.
I am wondering if resetting the card for the zero in a different way will make it behave all the time.

I will tinker with this over the next few weeks if you don't have any success.

Matt

[ktb]

This could probably use more testing, but I've had 10 successful (with sound) reboots in a row now on my Pi3B with Raspbian Stretch and using the following modprobe.d config file.

Code: Select all

pi@raspberrypi:~ $ cat /etc/modprobe.d/audioinjector-octo.conf
softdep snd_soc_audioinjector_octo_soundcard pre: snd_soc_cs42xx8 snd_soc_cs42xx8_i2c

[jshep321]

Hi,
I just made a brand new install of jessie lite on a pi zero with the *stereo* card (that has worked flawlessly on the pi3), and I'm seeing the same problems after full update and setup. Partial dmesg below. /boot/config.txt is the same between the two devices, but the pi3b is a noobs with full raspbian and the zero is raspbian jessie lite. I wonder if a module is missing...

Code: Select all

[    4.979699] i2c /dev entries driver
[    5.974648] systemd-udevd[118]: starting version 215
[    7.441041] gpiomem-bcm2835 20200000.gpiomem: Initialised: Registers at 0x20200000
[    8.328311] EXT4-fs (mmcblk0p2): re-mounted. Opts: (null)
[    8.580484] audioinjector-stereo soc:sound: ASoC: CPU DAI (null) not registered - will retry
[    8.580512] audioinjector-stereo soc:sound: snd_soc_register_card failed (-517)
[    8.593389] audioinjector-stereo soc:sound: ASoC: CODEC DAI wm8731-hifi not registered - will retry
[    8.593417] audioinjector-stereo soc:sound: snd_soc_register_card failed (-517)
[   10.179600] systemd-journald[116]: Received request to flush runtime journal from PID 1
[   11.412183] asix 1-1.2:1.0 eth0: register 'asix' at usb-20980000.usb-1.2, ASIX AX88772 USB 2.0 Ethernet, ac:7f:3e:e6:4d:c6
[   11.419565] audioinjector-stereo soc:sound: ASoC: CODEC DAI wm8731-hifi not registered - will retry
[   11.419586] audioinjector-stereo soc:sound: snd_soc_register_card failed (-517)
[   11.419780] usbcore: registered new interface driver asix
[   11.550407] audioinjector-stereo soc:sound: ASoC: CODEC DAI wm8731-hifi not registered - will retry
[   11.550426] audioinjector-stereo soc:sound: snd_soc_register_card failed (-517)
[   11.553921] audioinjector-stereo soc:sound: ASoC: CODEC DAI wm8731-hifi not registered - will retry
[   11.553942] audioinjector-stereo soc:sound: snd_soc_register_card failed (-517)
[   11.794100] wm8731 1-001a: Assuming static MCLK
[   11.794128] 1-001a supply AVDD not found, using dummy regulator
[   11.794230] 1-001a supply HPVDD not found, using dummy regulator
[   11.794286] 1-001a supply DCVDD not found, using dummy regulator
[   11.794354] 1-001a supply DBVDD not found, using dummy regulator
[   11.851097] wm8731 1-001a: Failed to issue reset: -121
[   11.851362] wm8731: probe of 1-001a failed with error -121
[   13.780841] Adding 102396k swap on /var/swap.  Priority:-1 extents:5 across:200700k SSFS
[   15.338275] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[   15.466985] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[   15.472133] asix 1-1.2:1.0 eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
[   16.952593] random: crng init done

[flatmax]

Hi there,

Are you talking about the octo or the stereo card ?
dmesg is reporting this "wm8731-hifi" which is the stereo card. I assume that your /boot/config.txt is trying to load the stereo card overlay, however the octo requires a different overlay. You should have the following in your config.txt file :

Code: Select all

dtoverlay=audioinjector-addons

Check this message for more information :
viewtopic.php?f=9&t=1222

[jshep321]

Sorry I wasn't explicit.

I was curious if the problem was with the octo or with the pi zero and/or "lite" version of raspbian, so I tried to reproduce with the STEREO card on a pi zero with jessie lite. The result was the same-- card not recognized (as shown in the log). This is why in that log you see the wm8xxx error message--I had a newly configured setup for the stereo card. I don't have a log capture for the octo.

So for your tinkering, you should play around with both the pi zero and the raspbian jessie lite distro.

If I have time this weekend, I'm going to start putting delays in the boot process to see if that "fixes" things.

[flatmax]

OK - I could understand that the Zero is slower to start up - I have tested the Stereo card on the Zero and it seemed to work fine!
Note that the codec drivers late poll, which means you will see failures a few times and then everything will kickin together further on in dmesg - unless there really are significant issues.

Matt

[jshep321]

Interesting... My two cards still don't work on the zero (zero W to be precise) with jessie lite. Works fine on rpi3.

Something is amiss!