Kevin, On Thu, Aug 13, 2009 at 5:09 AM, Kevin Hilman<khilman@xxxxxxxxxxxxxxxxxxx> wrote: > "S, Venkatraman" <svenkatr@xxxxxx> writes: > >> The hardware feature is available on OMAP4430 and also expected to be >> in OMAP3630. Hence the feature flag will control the inclusion of the additional functions. >> >> Addition of omap kconfig option CONFIG_OMAP_DMA_DESCRIPTOR_LOAD. > > Short version: (a.k.a caveman version) > > - Compile-time option: bad. > - Run-time option: good. > > Long version: > > Without looking at the docs, I would hope that the some revision reg > or similar would report whether this feature is available or not. The > user configuring the kernel should not have to know whether his SoC > has this feature or not. > > Kevin > -- Thanks for your comments. Yes, the DMA_CAPS_0 register contains flags for linked list capability and it can be checked. Obviously, I intended this (compile time flag) for code space saving for all older platforms. Not only in this file, but all client drivers those that would eventually start using this capability. I am preparing a patch with the runtime check (+ other changes you'd suggested), but I am curious to know what's the general threshold at which compile time flag (and it's associated code savings) is considered a valid trade off? Thanks and regards, Venkat. -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html