Hi Robin,
On Thu, Feb 2, 2017 at 2:15 PM, Robin Murphy <robin.murphy@xxxxxxx> wrote:
> On 31/01/17 11:12, Geert Uytterhoeven wrote:
>> Add support for allocating physically contiguous DMA buffers on arm64
>> systems with an IOMMU. This can be useful when two or more devices
>> with different memory requirements are involved in buffer sharing.
>>
>> Note that as this uses the CMA allocator, setting the
>> DMA_ATTR_FORCE_CONTIGUOUS attribute has a runtime-dependency on
>> CONFIG_DMA_CMA, just like on arm32.
>> @@ -625,7 +651,8 @@ static void __iommu_free_attrs(struct device *dev, size_t size, void *cpu_addr,
>>
>> size = PAGE_ALIGN(size);
>> /*
>> - * @cpu_addr will be one of 3 things depending on how it was allocated:
>> + * @cpu_addr will be one of 4 things depending on how it was allocated:
>> + * - A remapped array of pages for contiguous allocations.
>> * - A remapped array of pages from iommu_dma_alloc(), for all
>> * non-atomic allocations.
>> * - A non-cacheable alias from the atomic pool, for atomic
>> @@ -637,6 +664,12 @@ static void __iommu_free_attrs(struct device *dev, size_t size, void *cpu_addr,
>> if (__in_atomic_pool(cpu_addr, size)) {
>> iommu_dma_unmap_page(dev, handle, iosize, 0, 0);
>> __free_from_pool(cpu_addr, size);
>> + } else if (attrs & DMA_ATTR_FORCE_CONTIGUOUS) {
>> + struct page *page = phys_to_page(dma_to_phys(dev, handle));
>
> handle is the IOVA returned by the original iommu_dma_map_page() call,
> so I can't see this working very well. I was about to naively suggest
> virt_to_page(cpu_addr), except of course cpu_addr isn't going to be a
> linear map address :(
You're right: upon closer look, this will free the wrong pages in the CMA
allocator. Funny, that the system didn't scream...
Thanks for noticing, I'll look into fixing that...
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
