[PATCH/RFC 0/4] Re: dma_mmap_coherent / ARCH_HAS_DMA_MMAP_COHERENT

[Geert Uytterhoeven <geert@linux-m68k.org>]

On Sun, Dec 16, 2012 at 5:03 PM, Geert Uytterhoeven <geert@linux-m68k.org>
wrote:
> drivers/media/v4l2-core/videobuf2-dma-contig.c: In function ‘vb2_dc_mmap’:
> drivers/media/v4l2-core/videobuf2-dma-contig.c:204: error: implicit declaration of function ‘dma_mmap_coherent’
> drivers/media/v4l2-core/videobuf2-dma-contig.c: In function ‘vb2_dc_get_base_sgt’:
> drivers/media/v4l2-core/videobuf2-dma-contig.c:387: error: implicit declaration of function ‘dma_get_sgtable’
> make[6]: *** [drivers/media/v4l2-core/videobuf2-dma-contig.o] Error 1
> make[6]: Target `__build' not remade because of errors.
> make[5]: *** [drivers/media/v4l2-core] Error 2
>
> Both dma_mmap_coherent() and dma_get_sgtable() are defined in
> include/asm-generic/dma-mapping-common.h only, which is included by
> <asm/dma-mapping.h> on alpha, arm, arm64, hexagon, ia64, microblaze, mips,
> openrisc, powerpc, s390, sh, sparc, tile, unicore32, x86.
> Should the remaining architectures include this, too?
> Should it be moved to <linux/dma-mapping.h>?

I came up with an RFC-solution for this in [PATCH/RFC 3/4]
("avr32/bfin/c6x/cris/frv/m68k/mn10300/parisc/xtensa: Add dummy get_dma_ops()")
and [PATCH/RFC 4/4] ("common: dma-mapping: Move dma_common_*() to
<linux/dma-mapping.h>") of this series.

> Furthermore, there's ARCH_HAS_DMA_MMAP_COHERENT, which is defined
> by powerpc only:
> arch/powerpc/include/asm/dma-mapping.h:#define ARCH_HAS_DMA_MMAP_COHERENT
>
> and handled in some fishy way in sound/core/pcm_native.c:
>
> #ifndef ARCH_HAS_DMA_MMAP_COHERENT
> /* This should be defined / handled globally! */
> #ifdef CONFIG_ARM
> #define ARCH_HAS_DMA_MMAP_COHERENT
> #endif
> #endif
>
> /*
>  * mmap the DMA buffer on RAM
>  */
> int snd_pcm_lib_default_mmap(struct snd_pcm_substream *substream,
>                              struct vm_area_struct *area)
> {
>         area->vm_flags |= VM_DONTEXPAND | VM_DONTDUMP;
> #ifdef ARCH_HAS_DMA_MMAP_COHERENT
>         if (!substream->ops->page &&
>             substream->dma_buffer.dev.type == SNDRV_DMA_TYPE_DEV)
>                 return dma_mmap_coherent(substream->dma_buffer.dev.dev,
>                                          area,
>                                          substream->runtime->dma_area,
>                                          substream->runtime->dma_addr,
>                                          area->vm_end - area->vm_start);
> #elif defined(CONFIG_MIPS) && defined(CONFIG_DMA_NONCOHERENT)
>         if (substream->dma_buffer.dev.type == SNDRV_DMA_TYPE_DEV &&
>             !plat_device_is_coherent(substream->dma_buffer.dev.dev))
>                 area->vm_page_prot = pgprot_noncached(area->vm_page_prot);
> #endif /* ARCH_HAS_DMA_MMAP_COHERENT */
>         /* mmap with fault handler */
>         area->vm_ops = &snd_pcm_vm_ops_data_fault;
>         return 0;
> }
> EXPORT_SYMBOL_GPL(snd_pcm_lib_default_mmap);
>
> What's up here?

Probably an easy solution here is to kill ARCH_HAS_DMA_MMAP_COHERENT and
change the code to

    #if defined(CONFIG_MIPS) && defined(CONFIG_DMA_NONCOHERENT)
	    if (substream->dma_buffer.dev.type == SNDRV_DMA_TYPE_DEV &&
		!plat_device_is_coherent(substream->dma_buffer.dev.dev))
		    area->vm_page_prot = pgprot_noncached(area->vm_page_prot);
    #else
	    if (!substream->ops->page &&
		substream->dma_buffer.dev.type == SNDRV_DMA_TYPE_DEV)
		    return dma_mmap_coherent(substream->dma_buffer.dev.dev,
					     area,
					     substream->runtime->dma_area,
					     substream->runtime->dma_addr,
					     area->vm_end - area->vm_start);
    #endif

but obviously I don't like the test for CONFIG_MIPS in generic code...

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds