On 27 November 2019 12:33, Mark Brown wrote: > On Wed, Nov 27, 2019 at 11:32:54AM +0000, Adam Thomson wrote: > > > As I said it's a small thing and requires a specific use-case to occur, but > > having the PLL configured twice for the very first stream in that scenario > > seems messy. Regarding the SYSCLK approach you mention, I'm not clear how > that > > would work so I'm obviously missing something. If we had some init stage > > indication though that auto PLL was required then we're golden. > > There's a bunch of other drivers using the SYSCLK thing, when you call > set_sysclk() they provide a different SYSCLK number if they want to use > manual mode. If there's a concern about drivers doing stuff on init you > could always ask them to set the PLL during init, even if only briefly. Yeah OK I see what you mean; using the sysclk id. Still doesn't feel like the nicest solution here though. I guess we're stuck with people having to manually configure the PLL for bypass when a non-generic machine driver inits, to avoid the initial double config, as I don't see other options unless you have something to specify at init that it's auto or manual, and this doesn't feel like a DT device specific property thing as it's more software than hardware. At least Sebastian's patch has a good comment block to highlight this. _______________________________________________ Alsa-devel mailing list Alsa-devel@xxxxxxxxxxxxxxxx https://mailman.alsa-project.org/mailman/listinfo/alsa-devel
