At Thu, 12 Jul 2007 19:24:24 +0530, Pharaoh . wrote: > > okay, I am on ARM..so better I do not use preallocation scheme with > CONTINUOUS flag in case I want to mmap my dma buffer. In fact, I tried > mmaping it, but it didn't work as expected. On ARM, the mmap support is a bit tricky, so far. The preallocation itself should work with SNDRV_DMA_TYPE_TYPE_DEV. But you'd need the mmap PCM ops to call dma_mmap_coherent() there instead of the default ALSA mmap handler. Takashi > > On 7/12/07, Takashi Iwai <tiwai@xxxxxxx> wrote: > > At Wed, 11 Jul 2007 18:17:49 +0530, > > Pharaoh . wrote: > > > > > > On 7/11/07, Pharaoh . <pharaoh137@xxxxxxxxx> wrote: > > > > > > > > Hi, > > > > I have allocated the DMA buffer using snd_pcm_lib_malloc_pages() and > > > > it was preallocated as: > > > > > > > > snd_pcm_lib_preallocate_pages_for_all(pcm, > > > > SNDRV_DMA_TYPE_CONTINUOUS, > > > > snd_dma_continuous_data > > > > (GFP_KERNEL), > > > > 128 * 1024, 128 * 1024); > > > > > > > > Can I mmap such a dma buffer? I think yes... > > > > > > > > -pharaoh. > > > > > > > > > > > > > I dont think it can be mmapped. > > > > It can be basically, but not on all architectures. > > For example, ARM has a problem with coherency. > > > > > > Takashi > > > _______________________________________________ > Alsa-devel mailing list > Alsa-devel@xxxxxxxxxxxxxxxx > http://mailman.alsa-project.org/mailman/listinfo/alsa-devel > _______________________________________________ Alsa-devel mailing list Alsa-devel@xxxxxxxxxxxxxxxx http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
