On Fri, Sep 29, 2017 at 10:31:53AM +0900, Masahiro Yamada wrote:
Hi Marek,
2017-09-29 0:26 GMT+09:00 Marek Vasut marex@denx.de:
On 09/28/2017 03:01 PM, Masahiro Yamada wrote:
Add UniPhier platform specific glue layer to support USB3 Host mode on Synopsys DWC3 IP.
Signed-off-by: Masahiro Yamada yamada.masahiro@socionext.com Reviewed-by: Marek Vasut marex@denx.de
[...]
+static int uniphier_pro4_dwc3_init(void __iomem *regs) +{
u32 tmp;tmp = readl(regs + UNIPHIER_PRO4_DWC3_RESET);tmp &= ~UNIPHIER_PRO4_DWC3_RESET_PHY_SS;tmp |= UNIPHIER_PRO4_DWC3_RESET_XIOMMU | UNIPHIER_PRO4_DWC3_RESET_XLINK;writel(tmp, regs + UNIPHIER_PRO4_DWC3_RESET);Do I even want to nag you about clrsetbits_le32() or not ? What do you think?
I did not know this macro.
This seems to come from PPC Linux (defined in arch/powerpc/include/asm/io.h) probably in order to abstract the endianness.
Of course, it makes drivers ppc-specific, disabling COMPILE_TEST coverage.
In U-Boot, more architectures support this macro, ARM as well. (probably people wanted to write the code shorter?)
I am reluctant with it because I do not want to diverge from Linux.
This is one of those funny cases where perhaps some generic shuffling of code in Linux might help? Having whacked things for sandbox this morning in https://patchwork.ozlabs.org/patch/819919/ and looking at drivers/crypto/caam/regs.h in the kernel now, it's either a helpful thing that should be more widely available / used (drop it into say include/asm-generic/io.h) or time has moved on, and there's a better convention to use now. Just a thought.