On Fri, 17 Jul 2020 08:50:30 +0200, Christoph Hellwig wrote: > > On Fri, Jul 17, 2020 at 08:34:47AM +0200, Takashi Iwai wrote: > > On Fri, 17 Jul 2020 07:33:24 +0200, > > Christoph Hellwig wrote: > > > > > > On Fri, Jul 17, 2020 at 01:01:12PM +1000, Stephen Rothwell wrote: > > > > Hi all, > > > > > > > > After merging the sound tree, today's linux-next build (x86_64 > > > > allmodconfig) failed like this: > > > > > > No, the sound commit needs to be reverted. Drivers has absolutely not > > > business poking into DMA layer internals. > > > > OK, I'll drop that stuff again. It was the thing posted in the thread > > of DMA regression with AMD SEV on 5.8-rc1. Meanwhile it was fixed in > > the proper Kconfig dependency. > > We have other x86 changes that will not support just remapping stuff > ignoring the API in the pipe, so this will need to get sorted out. And > with that I mean sorted out properly instead of poking into undocumented > DMA API internals and coming to the wrong conclusions again.. OK, when a proper API (or some kconfig) is ready for referring from the driver side, let me know. I'd love to keep the SG-buffer support as much as possible, since HD-audio stuff has been used for a very old machine with a small RAM. thanks, Takashi