[PATCHv2 1/2] ARM: dma-mapping: add support for CMA regions placed in highmem zone

[Russell King - ARM Linux]

On Mon, Feb 04, 2013 at 03:24:51PM +0100, Michal Nazarewicz wrote:
> On Mon, Feb 04 2013, Russell King - ARM Linux wrote:
> > On Mon, Feb 04, 2013 at 02:51:52PM +0100, Michal Nazarewicz wrote:
> >> On Mon, Feb 04 2013, Marek Szyprowski wrote:
> >> > @@ -186,13 +186,24 @@ static u64 get_coherent_dma_mask(struct device *dev)
> >> >  
> >> >  static void __dma_clear_buffer(struct page *page, size_t size)
> >> >  {
> >> > -	void *ptr;
> >> >  	/*
> >> >  	 * Ensure that the allocated pages are zeroed, and that any data
> >> >  	 * lurking in the kernel direct-mapped region is invalidated.
> >> >  	 */
> >> > -	ptr = page_address(page);
> >> > -	if (ptr) {
> >> > +	if (PageHighMem(page)) {
> >> > +		phys_addr_t base = __pfn_to_phys(page_to_pfn(page));
> >> > +		phys_addr_t end = base + size;
> >> > +		while (size > 0) {
> >> > +			void *ptr = kmap_atomic(page);
> >> > +			memset(ptr, 0, PAGE_SIZE);
> >> > +			dmac_flush_range(ptr, ptr + PAGE_SIZE);
> >> > +			kunmap_atomic(ptr);
> >> > +			page++;
> >> > +			size -= PAGE_SIZE;
> >> > +		}
> >> > +		outer_flush_range(base, end);
> >> > +	} else {
> >> > +		void *ptr = page_address(page);
> >> 
> >> There used to be a “if (ptr)” check which is now missing.  Why is that?
> >
> > Because lowmem pages always have an address.
> 
> Perhaps it should use lowmem_page_address() then?

It's well defined that page_address() will be non-NULL for lowmem pages.
There is nothing wrong with the above.