On Thu, Sep 27, 2018 at 08:47:33AM -0700, Christoph Hellwig wrote:
> > +static void *csky_dma_alloc_atomic(
> > + struct device *dev,
> > + size_t size,
> > + dma_addr_t *dma_handle
> > + )
>
> Can you please use normal kernel coding style? Closing brace on the
> same line, either two tab indents for the arguments or align them to the
> first argument:
>
> static void *csky_dma_alloc_atomic(struct device *dev, size_t size,
> dma_addr_t *dma_handle)
>
> or:
>
> static void *csky_dma_alloc_atomic(struct device *dev, size_t size,
> dma_addr_t *dma_handle)
OK, I like 2th.
> > + if (DMA_ATTR_NON_CONSISTENT & attrs)
> > + BUG();
>
> Please don't bug on unknown attributs, they can be safely ignored.
No BUG(), return NULL.
>
> > +void arch_sync_dma_for_device(struct device *dev, phys_addr_t paddr,
> > + size_t size, enum dma_data_direction dir)
> > +{
> > + struct page *page = pfn_to_page(paddr >> PAGE_SHIFT);
> > + unsigned long offset = paddr & ~PAGE_MASK;
> > + unsigned long vaddr;
> > +
> > + if (PageHighMem(page))
> > + vaddr = (unsigned long) kmap_atomic(page);
>
> This isn't going to work as the size might be larger than PAGE_SIZE.
Ops ... BIG BUG, must be fixed immediately, thank you.
Best Regards
Guo Ren
