On 6/28/22 17:10, Alex Natalsson wrote: >> Any chance you could try with an updated kernel? It's not clear to me if >> this is a real issue or just a warning that's been fixed since by commit. > > I luanching 5.19-rc2 and finding this problem. Then I used git bisect > and found this commit, which broke my audio system. I don't know maybe > I have wrong settings, but for me this real problem because sound not > working.... > >> You may also want to open a bugzilla issue > Yes, I can. If problem not in my config settings. It's great that you've been able to bisect. What's surprising is that you have a long list of errors reported before the kernel oops, e.g. [ 243.188688] fe.dai-link-1: ASoC: no backend DAIs enabled for fe.dai-link-1 [ 243.188702] fe.dai-link-1: ASoC: dpcm_fe_dai_prepare() failed (-22) [ 243.188919] fe.dai-link-1: ASoC: no backend DAIs enabled for fe.dai-link-1 [ 243.188922] fe.dai-link-1: ASoC: dpcm_fe_dai_prepare() failed (-22) Did those appear with this commit? That's usually a sign of missing mixer settings to connect FE and BE, not sure how this would appear because of change in the locking. It would also help if you could add dev_err logs in dpcm_be_connect() to see in which code section the oops occurs. In theory we have all kinds of checks before calling dpcm_be_connect() but we've obviously missing something.
