[RFC PATCH 2/5] iommu: Add an unmap_pages() op for IOMMU drivers

[Lu Baolu]

Hi Isaac,

On 4/3/21 1:25 AM, isaacm at codeaurora.org wrote:
> On 2021-03-30 22:39, Lu Baolu wrote:
>> On 3/31/21 1:36 PM, isaacm at codeaurora.org wrote:
>>> On 2021-03-30 21:47, Lu Baolu wrote:
>>>> On 3/31/21 11:00 AM, Isaac J. Manjarres wrote:
>>>>> Add a callback for IOMMU drivers to provide a path for the
>>>>> IOMMU framework to call into an IOMMU driver, which can call
>>>>> into the io-pgtable code, to unmap a virtually contiguous
>>>>> range of pages of the same size.
>>>>>
>>>>> For IOMMU drivers that do not specify an unmap_pages() callback,
>>>>> the existing logic of unmapping memory one page block at a time
>>>>> will be used.
>>>>>
>>>>> Signed-off-by: Isaac J. Manjarres <isaacm at codeaurora.org>
>>>>> Suggested-by: Will Deacon <will at kernel.org>
>>>>> ---
>>>>>   include/linux/iommu.h | 4 ++++
>>>>>   1 file changed, 4 insertions(+)
>>>>>
>>>>> diff --git a/include/linux/iommu.h b/include/linux/iommu.h
>>>>> index 5e7fe519430a..9cf81242581a 100644
>>>>> --- a/include/linux/iommu.h
>>>>> +++ b/include/linux/iommu.h
>>>>> @@ -193,6 +193,7 @@ struct iommu_iotlb_gather {
>>>>>    * @detach_dev: detach device from an iommu domain
>>>>>    * @map: map a physically contiguous memory region to an iommu 
>>>>> domain
>>>>>    * @unmap: unmap a physically contiguous memory region from an 
>>>>> iommu domain
>>>>> + * @unmap_pages: unmap a number of pages of the same size from an 
>>>>> iommu domain
>>>>>    * @flush_iotlb_all: Synchronously flush all hardware TLBs for 
>>>>> this domain
>>>>>    * @iotlb_sync_map: Sync mappings created recently using @map to 
>>>>> the hardware
>>>>>    * @iotlb_sync: Flush all queued ranges from the hardware TLBs 
>>>>> and empty flush
>>>>> @@ -245,6 +246,9 @@ struct iommu_ops {
>>>>>              phys_addr_t paddr, size_t size, int prot, gfp_t gfp);
>>>>>       size_t (*unmap)(struct iommu_domain *domain, unsigned long iova,
>>>>>                size_t size, struct iommu_iotlb_gather *iotlb_gather);
>>>>> +    size_t (*unmap_pages)(struct iommu_domain *domain, unsigned 
>>>>> long iova,
>>>>> +                  size_t pgsize, size_t pgcount,
>>>>> +                  struct iommu_iotlb_gather *iotlb_gather);
>>>>
>>>> Is it possible to add an equivalent map_pages() callback?
>>> Yes, map_pages() can be implemented and can leverage a lot of the 
>>> implementation
>>> of unmap_pages(). The problem is that if you map several pages in one 
>>> call, and then
>>> encounter an error and have to rollback, you should do TLB 
>>> maintenance, as iommu_map
>>> does when it encounters an error. However, we can't call iommu_unmap 
>>> from io-pgtable-arm
>>> for example. We can call arm_lpae_unmap_pages() from the later 
>>> patches, but that doesn't
>>> solve the TLB maintenance issue. Do you have any thoughts on how to 
>>> address this?
>>
>> Call unmap_pages() with the same pages and size to roll back. Does it
>> work?
>>
>> Best regards,
>> baolu
> Hi Lu,
> 
> I've given map_pages() a shot. Here's the second version of the RFC 
> series: 
> https://lore.kernel.org/linux-iommu/20210402013452.4013-1-isaacm@codeaurora.org/T/#t. 

Thanks for doing that. I will look into it and try it with a VT-d
implementation.

Best regards,
baolu