09:48 <jhovold> steev: you're hitting race so any random changes in timing could potentially determine whether you trigger the bug or not

09:49 <jhovold> and you said you saw it also with 6.3, but not just every boot, right?

09:51 <jhovold> but the commit that enabled regmap caching until the device is enumerated is not to blame here, that one only changed the symptoms of the race

09:52 <jhovold> broonie was spot on yesterday in that the codec need to wait until the soundwire device has been enumerated

09:53 <jhovold> I've posted a fix here: https://lore.kernel.org/lkml/20230701094723.29379-1-johan+linaro@kernel.org/

09:55 <jhovold> I can easily reproduce the issue here by adding a short delay to the enumeration

09:56 <jhovold> I had not realised that the soundwire subsystem was this fragile...

16:02 <clover[m]> How would we test this? Just reboot a bunch and see if headphones work every time? BTW Guatemala trip going well I proposed to my gf

16:04 <jhovold> clover[m]: i have a reliable reproducer here that i've tested it against, but I thought you and steev never for your headphones to work with 6.4

16:05 <jhovold> the hope is that the two fixes will address that issue for you

16:05 <jhovold> When you hit the race during you should see:

16:06 <jhovold> wcd938x_codec audio-codec: ASoC: error at soc_component_read_no_lock on audio-codec for register: [0x000034b0] -16

16:06 <jhovold> in the log and pulseaudio will not switch to the headphone if you plug them in

16:07 <jhovold> with the fixes that message should be gone, and the headphones should work

16:50 <clover[m]> switching gears a bit, someone left an issue on my about wifi driver failing on resume: https://github.com/ironrobin/x13s-alarm/issues/6

16:50 <clover[m]> on my github*

18:35 <jhovold> clover[m]: yeah, i've seen that once with 6.4, but pretty sure sure I've seen it once or twice also with older kernels

18:36 <jhovold> it looks like the fw crashed, and then the ath11k recovery looks broken

18:36 <jhovold> either way, it's probably something that you should report to the ath11k maintainer

18:55 <steev> jhovold: i do get audio on 6.4 with the second patch also applied :D

18:55 <agl7-x13s> steev: I'am have switched on my X13s after some weeks. The accu was full when I switch it off, but now the accu was completely empty. That's not good! Ok now I will get you newest kernel an see what works :-)

18:57 <steev> clover[m]: also, congrats :D

19:30 <agl7-x13s> steev: I see 6.4 is out since 25.6.2023

19:30 <steev> indeed it is, but i need to push a new version with johan's 2 fixes for audio

19:30 <steev> which, i'm doing now

19:30 <agl7-x13s> ok

20:14 <agl7-x13s> steev: Is it still necessary to turn on the "IOMMU_DEFAULT_DMA_LAZY" switch to have a high data transfer rate on the USB-C ports?

20:14 <steev> try it and see :) i do not heavily use usb devices so i'm not sure

20:23 <agl7-x13s> ok

22:11 <jhovold> steev: thanks for confirming

22:11 <jhovold> agl7-x13s: yes, IOMMU_DEFAULT_DMA_LAZY (or the command line parameter) is still needed

22:12 <agl7-x13s> ok

22:13 <agl7-x13s> jhovold: What is the command line paramter?

22:47 <agl7-x13s> steev: Are you ready with a new version of 6.4.y with the two Johan's fixed for audio?

22:48 <steev> yes? i said "which, i'm doing now"

22:52 <agl7-x13s> ok