qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH v14 09/22] vfio iommu type1: Add task structure


From: Alexey Kardashevskiy
Subject: Re: [Qemu-devel] [PATCH v14 09/22] vfio iommu type1: Add task structure to vfio_dma
Date: Thu, 17 Nov 2016 17:42:11 +1100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.4.0

On 17/11/16 17:12, Alex Williamson wrote:
> On Thu, 17 Nov 2016 16:41:14 +1100
> Alexey Kardashevskiy <address@hidden> wrote:
> 
>> On 17/11/16 07:46, Kirti Wankhede wrote:
>>> Add task structure to vfio_dma structure. Task structure is used for:
>>> - During DMA_UNMAP, same task who mapped it or other task who shares same
>>> address space is allowed to unmap, otherwise unmap fails.
>>> QEMU maps few iova ranges initially, then fork threads and from the child
>>> thread calls DMA_UNMAP on previously mapped iova. Since child shares same
>>> address space, DMA_UNMAP is successful.
>>> - Avoid accessing struct mm while process is exiting by acquiring
>>> reference of task's mm during page accounting.
>>> - It is also used to get task mlock capability and rlimit for mlock.
>>>
>>> Signed-off-by: Kirti Wankhede <address@hidden>
>>> Signed-off-by: Neo Jia <address@hidden>
>>> Reviewed-by: Dong Jia Shi <address@hidden>  
>>
>>
>> I keep whinging that @mm should be referenced, not @current but you keep
>> referencing @current even if you only need @mm and you are not telling why
>> - and I am wondering what I am missing here? Something else will be used
>> from @task later, besides just @mm?
> 
> Yes, we reference @current from vfio_dma_do_map() and this is stored
> on the struct vfio_dma.  A reference to current is held because the
> external page pinning in vfio_pin_page_external() needs to test the
> capabilities of the task for CAP_IPC_LOCK to know whether locked memory

Ah, that's it - capable(CAP_IPC_LOCK) is checking @current, missed that.


> limits are in effect for the task even when it's not @current (ie. an
> asynchronous call from the vendor driver regardless of what task is
> currently running).  There are also various get_task_mm() taken
> temporarily when we're working with the mm of that task.  Do you spot
> any issues with this behavior? Thanks,

No, now I am fine, thanks!


-- 
Alexey



reply via email to

[Prev in Thread] Current Thread [Next in Thread]