Re: [Intel-gfx] [PATCH] drm/i915/userptr: Probe existence of backing struct pages upon creation

[Maarten Lankhorst <maarten.lankhorst@linux.intel.com>]

Op 2021-08-03 om 17:45 schreef Jason Ekstrand:
> On Tue, Aug 3, 2021 at 10:09 AM Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
>> On Wed, Jul 28, 2021 at 4:22 PM Matthew Auld
>> <matthew.william.auld@gmail.com> wrote:
>>> On Mon, 26 Jul 2021 at 17:10, Tvrtko Ursulin
>>> <tvrtko.ursulin@linux.intel.com> wrote:
>>>>
>>>> On 26/07/2021 16:14, Jason Ekstrand wrote:
>>>>> On Mon, Jul 26, 2021 at 3:31 AM Maarten Lankhorst
>>>>> <maarten.lankhorst@linux.intel.com> wrote:
>>>>>> Op 23-07-2021 om 13:34 schreef Matthew Auld:
>>>>>>> From: Chris Wilson <chris@chris-wilson.co.uk>
>>>>>>>
>>>>>>> Jason Ekstrand requested a more efficient method than userptr+set-domain
>>>>>>> to determine if the userptr object was backed by a complete set of pages
>>>>>>> upon creation. To be more efficient than simply populating the userptr
>>>>>>> using get_user_pages() (as done by the call to set-domain or execbuf),
>>>>>>> we can walk the tree of vm_area_struct and check for gaps or vma not
>>>>>>> backed by struct page (VM_PFNMAP). The question is how to handle
>>>>>>> VM_MIXEDMAP which may be either struct page or pfn backed...
>>>>>>>
>>>>>>> With discrete we are going to drop support for set_domain(), so offering
>>>>>>> a way to probe the pages, without having to resort to dummy batches has
>>>>>>> been requested.
>>>>>>>
>>>>>>> v2:
>>>>>>> - add new query param for the PROBE flag, so userspace can easily
>>>>>>>    check if the kernel supports it(Jason).
>>>>>>> - use mmap_read_{lock, unlock}.
>>>>>>> - add some kernel-doc.
>>>>>>> v3:
>>>>>>> - In the docs also mention that PROBE doesn't guarantee that the pages
>>>>>>>    will remain valid by the time they are actually used(Tvrtko).
>>>>>>> - Add a small comment for the hole finding logic(Jason).
>>>>>>> - Move the param next to all the other params which just return true.
>>>>>>>
>>>>>>> Testcase: igt/gem_userptr_blits/probe
>>>>>>> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
>>>>>>> Signed-off-by: Matthew Auld <matthew.auld@intel.com>
>>>>>>> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
>>>>>>> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
>>>>>>> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
>>>>>>> Cc: Jordan Justen <jordan.l.justen@intel.com>
>>>>>>> Cc: Kenneth Graunke <kenneth@whitecape.org>
>>>>>>> Cc: Jason Ekstrand <jason@jlekstrand.net>
>>>>>>> Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
>>>>>>> Cc: Ramalingam C <ramalingam.c@intel.com>
>>>>>>> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>>>>>>> Acked-by: Kenneth Graunke <kenneth@whitecape.org>
>>>>>>> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
>>>>>>> ---
>>>>>>>   drivers/gpu/drm/i915/gem/i915_gem_userptr.c | 41 ++++++++++++++++++++-
>>>>>>>   drivers/gpu/drm/i915/i915_getparam.c        |  1 +
>>>>>>>   include/uapi/drm/i915_drm.h                 | 20 ++++++++++
>>>>>>>   3 files changed, 61 insertions(+), 1 deletion(-)
>>>>>>>
>>>>>>> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
>>>>>>> index 56edfeff8c02..468a7a617fbf 100644
>>>>>>> --- a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
>>>>>>> +++ b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
>>>>>>> @@ -422,6 +422,34 @@ static const struct drm_i915_gem_object_ops i915_gem_userptr_ops = {
>>>>>>>
>>>>>>>   #endif
>>>>>>>
>>>>>>> +static int
>>>>>>> +probe_range(struct mm_struct *mm, unsigned long addr, unsigned long len)
>>>>>>> +{
>>>>>>> +     const unsigned long end = addr + len;
>>>>>>> +     struct vm_area_struct *vma;
>>>>>>> +     int ret = -EFAULT;
>>>>>>> +
>>>>>>> +     mmap_read_lock(mm);
>>>>>>> +     for (vma = find_vma(mm, addr); vma; vma = vma->vm_next) {
>>>>>>> +             /* Check for holes, note that we also update the addr below */
>>>>>>> +             if (vma->vm_start > addr)
>>>>>>> +                     break;
>>>>>>> +
>>>>>>> +             if (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))
>>>>>>> +                     break;
>>>>>>> +
>>>>>>> +             if (vma->vm_end >= end) {
>>>>>>> +                     ret = 0;
>>>>>>> +                     break;
>>>>>>> +             }
>>>>>>> +
>>>>>>> +             addr = vma->vm_end;
>>>>>>> +     }
>>>>>>> +     mmap_read_unlock(mm);
>>>>>>> +
>>>>>>> +     return ret;
>>>>>>> +}
>>>>>>> +
>>>>>>>   /*
>>>>>>>    * Creates a new mm object that wraps some normal memory from the process
>>>>>>>    * context - user memory.
>>>>>>> @@ -477,7 +505,8 @@ i915_gem_userptr_ioctl(struct drm_device *dev,
>>>>>>>        }
>>>>>>>
>>>>>>>        if (args->flags & ~(I915_USERPTR_READ_ONLY |
>>>>>>> -                         I915_USERPTR_UNSYNCHRONIZED))
>>>>>>> +                         I915_USERPTR_UNSYNCHRONIZED |
>>>>>>> +                         I915_USERPTR_PROBE))
>>>>>>>                return -EINVAL;
>>>>>>>
>>>>>>>        if (i915_gem_object_size_2big(args->user_size))
>>>>>>> @@ -504,6 +533,16 @@ i915_gem_userptr_ioctl(struct drm_device *dev,
>>>>>>>                        return -ENODEV;
>>>>>>>        }
>>>>>>>
>>>>>>> +     if (args->flags & I915_USERPTR_PROBE) {
>>>>>>> +             /*
>>>>>>> +              * Check that the range pointed to represents real struct
>>>>>>> +              * pages and not iomappings (at this moment in time!)
>>>>>>> +              */
>>>>>>> +             ret = probe_range(current->mm, args->user_ptr, args->user_size);
>>>>>>> +             if (ret)
>>>>>>> +                     return ret;
>>>>>>> +     }
>>>>>>> +
>>>>>>>   #ifdef CONFIG_MMU_NOTIFIER
>>>>>>>        obj = i915_gem_object_alloc();
>>>>>>>        if (obj == NULL)
>>>>>>> diff --git a/drivers/gpu/drm/i915/i915_getparam.c b/drivers/gpu/drm/i915/i915_getparam.c
>>>>>>> index 24e18219eb50..bbb7cac43eb4 100644
>>>>>>> --- a/drivers/gpu/drm/i915/i915_getparam.c
>>>>>>> +++ b/drivers/gpu/drm/i915/i915_getparam.c
>>>>>>> @@ -134,6 +134,7 @@ int i915_getparam_ioctl(struct drm_device *dev, void *data,
>>>>>>>        case I915_PARAM_HAS_EXEC_FENCE_ARRAY:
>>>>>>>        case I915_PARAM_HAS_EXEC_SUBMIT_FENCE:
>>>>>>>        case I915_PARAM_HAS_EXEC_TIMELINE_FENCES:
>>>>>>> +     case I915_PARAM_HAS_USERPTR_PROBE:
>>>>>>>                /* For the time being all of these are always true;
>>>>>>>                 * if some supported hardware does not have one of these
>>>>>>>                 * features this value needs to be provided from
>>>>>>> diff --git a/include/uapi/drm/i915_drm.h b/include/uapi/drm/i915_drm.h
>>>>>>> index 975087553ea0..0d290535a6e5 100644
>>>>>>> --- a/include/uapi/drm/i915_drm.h
>>>>>>> +++ b/include/uapi/drm/i915_drm.h
>>>>>>> @@ -674,6 +674,9 @@ typedef struct drm_i915_irq_wait {
>>>>>>>    */
>>>>>>>   #define I915_PARAM_HAS_EXEC_TIMELINE_FENCES 55
>>>>>>>
>>>>>>> +/* Query if the kernel supports the I915_USERPTR_PROBE flag. */
>>>>>>> +#define I915_PARAM_HAS_USERPTR_PROBE 56
>>>>>>> +
>>>>>>>   /* Must be kept compact -- no holes and well documented */
>>>>>>>
>>>>>>>   typedef struct drm_i915_getparam {
>>>>>>> @@ -2222,12 +2225,29 @@ struct drm_i915_gem_userptr {
>>>>>>>         * through the GTT. If the HW can't support readonly access, an error is
>>>>>>>         * returned.
>>>>>>>         *
>>>>>>> +      * I915_USERPTR_PROBE:
>>>>>>> +      *
>>>>>>> +      * Probe the provided @user_ptr range and validate that the @user_ptr is
>>>>>>> +      * indeed pointing to normal memory and that the range is also valid.
>>>>>>> +      * For example if some garbage address is given to the kernel, then this
>>>>>>> +      * should complain.
>>>>>>> +      *
>>>>>>> +      * Returns -EFAULT if the probe failed.
>>>>>>> +      *
>>>>>>> +      * Note that this doesn't populate the backing pages, and also doesn't
>>>>>>> +      * guarantee that the object will remain valid when the object is
>>>>>>> +      * eventually used.
>>>>>>> +      *
>>>>>>> +      * The kernel supports this feature if I915_PARAM_HAS_USERPTR_PROBE
>>>>>>> +      * returns a non-zero value.
>>>>>>> +      *
>>>>>>>         * I915_USERPTR_UNSYNCHRONIZED:
>>>>>>>         *
>>>>>>>         * NOT USED. Setting this flag will result in an error.
>>>>>>>         */
>>>>>>>        __u32 flags;
>>>>>>>   #define I915_USERPTR_READ_ONLY 0x1
>>>>>>> +#define I915_USERPTR_PROBE 0x2
>>>>>>>   #define I915_USERPTR_UNSYNCHRONIZED 0x80000000
>>>>>>>        /**
>>>>>>>         * @handle: Returned handle for the object.
>>>>>> Could we use _VALIDATE instead of probe? Or at least pin the pages as well, so we don't have to do it later?
>>>>> I only care that the name matches what it does.  _VALIDATE sounds like
>>>>> it does a full validation of everything such that, if the import
>>>>> succeeds, execbuf will as well.  If we pin the pages at the same time,
>>>>> maybe that's true?  _PROBE, on the other hand, sounds a lot more like
>>>> No it is not possible to guarantee backing store remains valid until
>>>> execbuf.
>>>>
>>>>> a one-time best-effort check which may race with other stuff and
>>>>> doesn't guarantee future success.  That's in line with what the
>>>>> current patch does.
>>>>>
>>>>>> We already have i915_gem_object_userptr_validate, no need to dupe it.
>>>>> I have no opinion on this.
>>>> I was actually suggesting the same as Maarten here - that we should add
>>>> a "populate" flag. But opinion was that was not desired - please look
>>>> for the older threads to see the reasoning there.
>>> So how should we proceed here? Maarten?
>> I honestly don't care, and I think the probe flag here is perfectly
>> fine. Reasons for that:
>> - we don't have an immediate allocation flag for buffer creation
>> either. So if there's a need we need a flag for this across the board,
>> not just userptr, and a clear userspace ask
> Both Mesa drivers would probably set that flag if we had it and it
> demonstrated any perf benefits, FWIW.  However, I think it's fine if
> that's a separate flag.  Also, I don't know that the perf benefits are
> all that great.  We should get most of those benefits from VM_BIND
> anyway.
>
>> - it's a fundamentally racy test anyway, userspace can munmap or map
>> something else and then it will fail. So we really don't gain anything
>> by pinning pages because by the time we go into execbuf they might be
>> invalidated already - checking the vmas for VM_SPECIAL is perfectly
>> good enough.
>> - we can always change the implementation later on too.
>>
>> Hence why I think PROBE is the semantics we want/need here. Can we get
>> some acks/reviews here or is this really a significant enough bikeshed
>> that we need to hold up dg1 pciids for them?
> I don't care.  I've already reviewed the patch.
>
> --Jason

I think we should still just put the validate() call in there, but I'm not going to hold up the implementation because of that.

Acked-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>