On Friday 06 December 2013, Dan Williams wrote: > * nod. > > I have a similar reaction to dma_request_channel(). There should be > more than enough use case to find some common ground and reduce the > need for magic filter routines. Right. I didn't really like the addition of the _compat variant, though I can see how it helps drivers that are transitioning from pdata to DT probing. Once we have progressed enough with the ARM DT conversion, a lot of those can surely be converted back to plain dma_request_slave_channel(), and if a significant number remain that won't use DT, we can come up with a way to use the same interface using the equivalent of clkdev to connect devices to request lines. Also, once things have cooled down a little, we can do a mass-conversion to the ERR_PTR variant and kill off the ones that just return NULL on error. Looking at the slave drivers that use the old dma_request_channel(), I see these classes that are not moving to DT: * Atmel arch/avr32 (some shared with ARM mach-at91, using DT there) * Renesas arch/sh (some shared with ARM shmobile, using DT there) * x86 topcliff/PCH * mfd/timberdale * ARM ep93xx, sa11x0 The other ones are all ARM or PowerPC platforms that should migrate to describing their DMA channels in DT in the future and kill their filter functions. Arnd -- To unsubscribe from this list: send the line "unsubscribe linux-tegra" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
