Hi Arnd, On Wed, Aug 12, 2015 at 4:50 PM, Arnd Bergmann <arnd@xxxxxxxx> wrote: > On Tuesday 11 August 2015 14:48:34 Geert Uytterhoeven wrote: >> dma_cap_mask_t mask; >> @@ -413,12 +413,11 @@ sh_mmcif_request_dma_one(struct sh_mmcif_host *host, >> dma_cap_set(DMA_SLAVE, mask); >> >> if (pdata) >> - slave_data = direction == DMA_MEM_TO_DEV ? >> - (void *)pdata->slave_id_tx : >> - (void *)pdata->slave_id_rx; >> + slave_id = direction == DMA_MEM_TO_DEV ? >> + pdata->slave_id_tx : pdata->slave_id_rx; >> >> chan = dma_request_slave_channel_compat(mask, shdma_chan_filter, >> - slave_data, dev, >> + (void *)(uintptr_t)slave_id, dev, >> direction == DMA_MEM_TO_DEV ? "tx" : "rx"); >> >> dev_dbg(dev, "%s: %s: got channel %p\n", __func__, > > How about changing the type of the slave_id_rx/slave_id_tx fields > to void*? That way, the hack can be moved to arch/sh/boards/board-sh7757lcr.c, > which is now the only file passing data this way. Ideally, we'd also And to the other SH boards... This is not limited to the mmcif driver. Ah, I see tmio_mmc_data already does it this way. This will break out-of-tree boards at compile-time though. Given the limited number of wired up DMA channels under arch/sh/, they may actually be more in use out-of-tree than in-tree? Alternatively, perhaps we should just drop pdata DMA configuration from some drivers? > pass the shdma_chan_filter function pointer in pdata to avoid the link > time dependency on a particular dmaengine driver. That's actually more tricky, as a NULL filter function means something completely different, and causes funny failures if the DTS lacks "dmas" properties, cfr. "dmaengine: shdma: Make dummy shdma_chan_filter() always return false". https://git.kernel.org/cgit/linux/kernel/git/vkoul/slave-dma.git/commit/?h=next&id=056f6c87028544de934f27caf95aa1545d585767 Gr{oetje,eeting}s, Geert -- Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx 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 -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
