Message ID | 7d1ca7f4-b3ae-46df-5b9a-7cba8ac3e512@linaro.org (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | [GIT,PULL] amlogic Drivers updates for v6.2 | expand |
On Tue, Nov 22, 2022, at 16:24, Neil Armstrong wrote: > > for you to fetch changes up to 7dc69c7d073e6004a281db8f7f15cf6ebf702ea0: > > firmware: meson_sm: Fix memcpy vs iomem type warnings (2022-10-17 > 17:19:04 +0200) > > ---------------------------------------------------------------- > Amlogic Drivers changes for v6.2: > - Fix memcpy vs iomem type warnings in meson_sm driver > It looks like a trivial fix, but I think it's actually wrong: The real problem appears to be the use of the ioremap_cache() function in meson_sm_map_shmem(), which returns an __iomem token that is mapped cacheable on architectures that support this function, but behaves like normal ioremap on others. This is probably not what you want here, instead this should be converted to memremap(), with the __iomem annotation dropped. In the long run, we should try to kill off ioremap_cache entirely, and not introduce new ones. Arnd
On 22/11/2022 22:48, Arnd Bergmann wrote: > On Tue, Nov 22, 2022, at 16:24, Neil Armstrong wrote: >> >> for you to fetch changes up to 7dc69c7d073e6004a281db8f7f15cf6ebf702ea0: >> >> firmware: meson_sm: Fix memcpy vs iomem type warnings (2022-10-17 >> 17:19:04 +0200) >> >> ---------------------------------------------------------------- >> Amlogic Drivers changes for v6.2: >> - Fix memcpy vs iomem type warnings in meson_sm driver >> > > It looks like a trivial fix, but I think it's actually wrong: > The real problem appears to be the use of the ioremap_cache() > function in meson_sm_map_shmem(), which returns an __iomem > token that is mapped cacheable on architectures that support > this function, but behaves like normal ioremap on others. > > This is probably not what you want here, instead this should > be converted to memremap(), with the __iomem annotation dropped. > In the long run, we should try to kill off ioremap_cache > entirely, and not introduce new ones. OK will have a look into that, Thanks, Neil > > Arnd