On Tue, 09 Jun 2020 13:40:59 +0200, Christoph Hellwig wrote: > > On Tue, Jun 09, 2020 at 01:38:46PM +0200, Takashi Iwai wrote: > > On Tue, 09 Jun 2020 13:31:23 +0200, > > Christoph Hellwig wrote: > > > > > > On Tue, Jun 09, 2020 at 11:31:20AM +0200, Takashi Iwai wrote: > > > > > > How would be a proper way to get the virtually mapped SG-buffer pages > > > > > > with coherent memory? (Also allowing user-space mmap, too) > > > > > > > > > > dma_mmap_coherent / dma_mmap_attrs for userspace. We don't really > > > > > have a good way for kernel space mappings. > > > > > > > > And that's the missing piece right now... :-< > > > > > > Can you point me to the relevant places (allocation and vmap mostly) > > > so that I can take a look at how to fix this mess? > > > > Found in sound/core/sgbuf.c. It's specific to x86. > > So it looks like we could just turn off CONFIG_SND_DMA_SGBUF and > be done with it? After all this works on other architectures > just fine.. Yes, for the sound stuff, something below should make things working. But it means that we'll lose the SG-buffer allocation and the allocation of large buffers might fail on some machines. Takashi --- a/sound/core/Kconfig +++ b/sound/core/Kconfig @@ -192,6 +192,6 @@ config SND_VMASTER config SND_DMA_SGBUF def_bool y - depends on X86 + depends on X86 && BROKEN source "sound/core/seq/Kconfig"
