On Sat, 10 Jun 2023 19:41:42 +0200, Joseph C. Sible wrote: > > On Sat, Jun 10, 2023 at 3:09 AM Takashi Iwai <tiwai@xxxxxxx> wrote: > > > > Thanks for quick testing. > > > > But, the patch to add a delay was meant to apply on top of the two > > patches from Kailang but only with power_save=0 option. The model > > option changes the behavior completely and no Realtek codec driver is > > bound when the option is given in that way. > > Ah, sorry for the misunderstanding. Attached are two new dumps taken > that way (with Kailang's patches and your patch but without my patch, > and with power_save=0 but without model=,generic), one from a cold > boot where the speakers worked and one from a warm boot where they > didn't. Thanks for checking. So the additional delay didn't help at all, i.e. it's no timing issue. We need to figure out which verb actually breaks things. Oh well. Takashi