> On 06.02.2019, at 22:53, André Przywara <[email protected]> wrote: > > On 06/02/2019 12:46, Philipp Tomsich wrote: >> On 11.01.2019, at 01:31, Andre Przywara <[email protected] >> <mailto:[email protected]>> wrote: > > Hi, > >>> >>> The normal MMIO accessor macros (readX/writeX) guarantee a strong ordering, >>> even with normal memory accesses [1]. >>> For some MMIO operations (framebuffers being a prominent example) this is >>> not needed, and the rather costly barrier inserted on weakly ordered >>> systems like ARM costs some performance. >>> To mitigate this issue, Linux introduced readX_relaxed and >>> writeX_relaxed primitives, which only guarantee ordering between each >>> other, so are typically faster due to the missing barrier. >>> >>> We probably do not care so much about performance in U-Boot, but want to >>> have those primitives for two other reasons: >>> - Being able to use the _relaxed macros simplifies porting drivers from >>> Linux. >>> - The missing barrier typically allows to generate smaller code, which can >>> relieve some chronically tight SPL builds. >>> >>> Add those macros definitions by using the __raw versions of the >>> accessors, which matches an earlier (and less complicated) version of >>> the Linux implementation. >>> >>> [1] https://lwn.net/Articles/698014/ >> >> No Signed-off-by? > > Doh, indeed. Got so excited about my commit message, that I forgot the > obvious (and I only think about -s *after* hitting Enter). > >> Reviewed-by: Philipp Tomsich <[email protected] >> <mailto:[email protected]>> >> [On my experimental RK3399 after modifying a few drivers:] >> Tested-by: Philipp Tomsich <[email protected] >> <mailto:[email protected]>> > > Cool, many thanks for that! Out of curiosity, did you really need > *_relaxed for some reason?
Not yet, but possibly soon. I have been working on some changes to the way we auto-configure for different DRAM timings (effectively doing something similar to how DIMMs are detected using SPDs)… and have been fighting code-size growth on that front. As I have to set up timings for 3 frequencies on the 3399, the amount of dmb()s add up from the strongly ordered writel and clrsetbits calls. Cheers, Phil. _______________________________________________ U-Boot mailing list [email protected] https://lists.denx.de/listinfo/u-boot
