Re: [PATCH 1 of 3] Introduce __raw_memcpy_toio32

[Christoph Hellwig <hch@infradead.org>]

On Tue, Jan 10, 2006 at 09:49:46AM -0800, Bryan O'Sullivan wrote:
> On Tue, 2006-01-10 at 17:07 +0000, Christoph Hellwig wrote:
> 
> > Or add a CONFIG_GENERIC_MEMCPY_IO that's non-uservisible and just set
> > by all the architectures that don't provide their own version.
> 
> Here's another i386-only review patch that does essentially that.  It
> looks cleaner to me than my previous patch from this morning.
> (Copyrights and other arches omitted, for clarity.)
> 
> What do you think?
> 
> 	<b
> 
> diff -r 48616306e7bd lib/Makefile
> --- a/lib/Makefile	Tue Jan 10 10:41:42 2006 +0800
> +++ b/lib/Makefile	Tue Jan 10 09:32:53 2006 -0800
> @@ -21,6 +21,7 @@
>  lib-$(CONFIG_RWSEM_XCHGADD_ALGORITHM) += rwsem.o
>  lib-$(CONFIG_SEMAPHORE_SLEEPERS) += semaphore-sleepers.o
>  lib-$(CONFIG_GENERIC_FIND_NEXT_BIT) += find_next_bit.o
> +lib-$(CONFIG_GENERIC_RAW_MEMCPY_IO) += raw_memcpy_io.o
>  obj-$(CONFIG_LOCK_KERNEL) += kernel_lock.o
>  obj-$(CONFIG_DEBUG_PREEMPT) += smp_processor_id.o
>  
> --- /dev/null	Thu Jan  1 00:00:00 1970 +0000
> +++ b/include/asm-generic/raw_memcpy_io.h	Tue Jan 10 09:32:53 2006 -0800
> @@ -0,0 +1,16 @@
> +#ifndef _ASM_GENERIC_RAW_MEMCPY_IO_H
> +#define _ASM_GENERIC_RAW_MEMCPY_IO_H
> +
> +/*
> + * __raw_memcpy_toio32 - copy data to MMIO space, in 32-bit units
> + *
> + * Order of access is not guaranteed, nor is a memory barrier performed
> + * afterwards.  This is an arch-independent generic implementation.
> + *
> + * @to: destination, in MMIO space (must be 32-bit aligned)
> + * @from: source (must be 32-bit aligned)
> + * @count: number of 32-bit quantities to copy
> + */

This should be in the implementation file, not near the prototype.
And needs to start with /** to be valid kernel doc.

> +void __raw_memcpy_toio32(void __iomem *to, const void *from, size_t count);

and without that comment I'd suggest just adding this to every asm/io.h
instead of an asm-generic header for just one prototype.