Re: [PATCH v4 for-next 1/2] RDMA/core: Provide rdma_user_mmap_disassociate() to disassociate mmap pages

[Junxian Huang]

On 2024/9/11 18:20, Leon Romanovsky wrote:
> On Mon, Sep 09, 2024 at 04:41:00PM +0800, Junxian Huang wrote:
>>
>>
>> On 2024/9/7 20:12, Jason Gunthorpe wrote:
>>> On Thu, Sep 05, 2024 at 09:11:54PM +0800, Junxian Huang wrote:
>>>
>>>> @@ -698,11 +700,20 @@ static int ib_uverbs_mmap(struct file *filp, struct vm_area_struct *vma)
>>>>  	ucontext = ib_uverbs_get_ucontext_file(file);
>>>>  	if (IS_ERR(ucontext)) {
>>>>  		ret = PTR_ERR(ucontext);
>>>> -		goto out;
>>>> +		goto out_srcu;
>>>>  	}
>>>> +
>>>> +	mutex_lock(&file->disassociation_lock);
>>>> +	if (file->disassociated) {
>>>> +		ret = -EPERM;
>>>> +		goto out_mutex;
>>>> +	}
>>>
>>> What sets disassociated back to false once the driver reset is
>>> completed?
>>>
>>> I think you should probably drop this and instead add a lock and test
>>> inside the driver within its mmap op. While reset is ongoing fail all
>>> new mmaps.
>>>
>>
>> disassociated won't be set back to false. This is to stop new mmaps on
>> this ucontext even after reset is completed, because during hns reset,
>> all resources will be destroyed, and the ucontexts will become unavailable.
> 
> ucontext is SW object and not HW object, why will it become unavailable?
> 

Once hns device is reset, we don't allow any doorbell until driver's
re-initialization is completed. Not only all existing mmaps on ucontexts
will be zapped, no more new mmaps are allowed either.

This actually makes ucontexts unavailable since userspace cannot access
HW with them any more. Users will have to destroy the old ucontext and
allocate a new one after driver's re-initialization is completed.

Junxian

> Thanks