On Mon, Oct 19, 2015 at 9:08 AM, Mark Brown <broonie@xxxxxxxxxx> wrote: > On Mon, Oct 19, 2015 at 05:34:48PM +0200, Takashi Iwai wrote: >> Mark Brown wrote: > >> > On Fri, Oct 09, 2015 at 06:16:36PM -0400, Dan Williams wrote: >> > > In preparation for deprecating ioremap_cache() convert its usage in >> > > skl-nhlt to memremap. > >> > > -void __iomem *skl_nhlt_init(struct device *dev) >> > > +void *skl_nhlt_init(struct device *dev) > >> > Why are we loosing the annotation here? > >> It's the exact purpose of this patchset, as far as I understand. >> memremap() allows the driver accessing the mapped memory just like a >> normal memory unlike ioremap(). > > There's no mention of what the purpose of the conversion is in either > the patch or the cover letter so it's a bit unclear, some confirmation > would be good :( The cover letter linked this article which talked through the motivation: https://lwn.net/Articles/653585/ Essentially an "__iomem" annotation on a cached mapping is inconsistent as __iomem indicates "may have side effects, use special accessors" and cached means "pre-fetching, asynchronous write-backs, and cpu determined i/o sizes are permissible". -- To unsubscribe from this list: send the line "unsubscribe linux-arch" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
