On 6/28/2017 7:51 PM, Logan Gunthorpe wrote: > > > On 28/06/17 04:20 AM, Arnd Bergmann wrote: >> On Wed, Jun 28, 2017 at 1:02 AM, Logan Gunthorpe <logang@xxxxxxxxxxxx> wrote: >>> #include <linux/types.h> >>> #include <linux/bitops.h> >>> -#include <linux/io.h> >>> +#include <linux/io-64-nonatomic-hi-lo.h> >> >> Here you include the hi-lo variant unconditionally. >> Arnd, thanks for spotting this. This was not in the patch I signed off. The lo-hi variant should be used instead for CAAM, see further below. >>> -#else /* CONFIG_64BIT */ >>> -static inline void wr_reg64(void __iomem *reg, u64 data) >>> -{ >>> -#ifndef CONFIG_CRYPTO_DEV_FSL_CAAM_IMX >>> - if (caam_little_end) { >>> - wr_reg32((u32 __iomem *)(reg) + 1, data >> 32); >>> - wr_reg32((u32 __iomem *)(reg), data); >>> - } else >>> #endif >>> - { >>> - wr_reg32((u32 __iomem *)(reg), data >> 32); >>> - wr_reg32((u32 __iomem *)(reg) + 1, data); >>> - } >>> + iowrite64be(data, reg); >>> } >> >> However, the #else path here uses lo-hi instead. I guess we have >> to decide how to define iowrite64be_lo_hi() first: it could >> either byteswap the 64-bit value first, then write the two halves, >> or it could write the two halves, doing a 32-bit byte swap on >> each. > > Ok, I studied this a bit more: > > The lo_hi/hi_lo functions seem to always refer to the data being written > or read not to the address operated on. So, in the v3 version of this > set, which I'm working on, I've defined: > > static inline void iowrite64_hi_lo(u64 val, void __iomem *addr) > { > iowrite32(val >> 32, addr + sizeof(u32)); > iowrite32(val, addr); > } > > static inline void iowrite64be_hi_lo(u64 val, void __iomem *addr) > { > iowrite32be(val >> 32, addr); > iowrite32be(val, addr + sizeof(u32)); > } > > So the two hi_lo functions match both paths of the #if and thus, I > believe, the patch will be correct in v3 without changes. > To be consistent with CAAM engine HW spec: in case of 64-bit registers, irrespective of device endianness, the lower address should be read from / written to first, followed by the upper address. Indeed the I/O accessors in CAAM driver currently don't follow the spec, however this is a good opportunity to fix the code. I don't consider this requires a separate patch, as we haven't noticed any problem. I'd say a simple note in the commit message mentioning the change (lo-hi r/w order replacing hi-lo for little endian case) is enough. Thanks, Horia