On Wed, Feb 09, 2022 at 01:47:01AM +0000, Kasireddy, Vivek wrote:
> Hi Ville,
>
> >
> > On Tue, Feb 08, 2022 at 05:10:42AM +0000, Kasireddy, Vivek wrote:
> > > Hi Tvrtko, Ville,
> > >
> > > > On 07/02/2022 13:24, Ville Syrjälä wrote:
> > > > > On Mon, Feb 07, 2022 at 11:47:16AM +0000, Tvrtko Ursulin wrote:
> > > > >>
> > > > >> On 07/02/2022 10:58, Ville Syrjälä wrote:
> > > > >>> On Thu, Feb 03, 2022 at 05:22:10PM -0800, Vivek Kasireddy wrote:
> > > > >>>> On platforms capable of allowing 8K (7680 x 4320) modes, pinning 2
> > > > >>>> or more framebuffers/scanout buffers results in only one that is
> > > > >>>> mappable/ fenceable. Therefore, pageflipping between these 2 FBs
> > > > >>>> where only one is mappable/fenceable creates latencies large enough
> > > > >>>> to miss alternate vblanks thereby producing less optimal framerate.
> > > > >>>>
> > > > >>>> This mainly happens because when
> > > > >>>> i915_gem_object_pin_to_display_plane()
> > > > >>>> is called to pin one of the FB objs, the associated vma is
> > > > >>>> identified as misplaced and therefore i915_vma_unbind() is called
> > > > >>>> which unbinds and evicts it. This misplaced vma gets subseqently
> > > > >>>> pinned only when
> > > > >>>> i915_gem_object_ggtt_pin_ww() is called without PIN_MAPPABLE. This
> > > > >>>> results in a latency of ~10ms and happens every other vblank/repaint cycle.
> > > > >>>> Therefore, to fix this issue, we try to see if there is space to
> > > > >>>> map at-least two objects of a given size and return early if there
> > > > >>>> isn't. This would ensure that we do not try with PIN_MAPPABLE for
> > > > >>>> any objects that are too big to map thereby preventing unncessary unbind.
> > > > >>>>
> > > > >>>> Testcase:
> > > > >>>> Running Weston and weston-simple-egl on an Alderlake_S (ADLS)
> > > > >>>> platform with a 8K@60 mode results in only ~40 FPS. Since upstream
> > > > >>>> Weston submits a frame ~7ms before the next vblank, the latencies
> > > > >>>> seen between atomic commit and flip event are 7, 24 (7 + 16.66), 7,
> > > > >>>> 24..... suggesting that it misses the vblank every other frame.
> > > > >>>>
> > > > >>>> Here is the ftrace snippet that shows the source of the ~10ms latency:
> > > > >>>> i915_gem_object_pin_to_display_plane() {
> > > > >>>> 0.102 us | i915_gem_object_set_cache_level();
> > > > >>>> i915_gem_object_ggtt_pin_ww() {
> > > > >>>> 0.390 us | i915_vma_instance();
> > > > >>>> 0.178 us | i915_vma_misplaced();
> > > > >>>> i915_vma_unbind() {
> > > > >>>> __i915_active_wait() {
> > > > >>>> 0.082 us | i915_active_acquire_if_busy();
> > > > >>>> 0.475 us | }
> > > > >>>> intel_runtime_pm_get() {
> > > > >>>> 0.087 us | intel_runtime_pm_acquire();
> > > > >>>> 0.259 us | }
> > > > >>>> __i915_active_wait() {
> > > > >>>> 0.085 us | i915_active_acquire_if_busy();
> > > > >>>> 0.240 us | }
> > > > >>>> __i915_vma_evict() {
> > > > >>>> ggtt_unbind_vma() {
> > > > >>>> gen8_ggtt_clear_range() {
> > > > >>>> 10507.255 us | }
> > > > >>>> 10507.689 us | }
> > > > >>>> 10508.516 us | }
> > > > >>>>
> > > > >>>> v2: Instead of using bigjoiner checks, determine whether a scanout
> > > > >>>> buffer is too big by checking to see if it is possible to map
> > > > >>>> two of them into the ggtt.
> > > > >>>>
> > > > >>>> v3 (Ville):
> > > > >>>> - Count how many fb objects can be fit into the available holes
> > > > >>>> instead of checking for a hole twice the object size.
> > > > >>>> - Take alignment constraints into account.
> > > > >>>> - Limit this large scanout buffer check to >= Gen 11 platforms.
> > > > >>>>
> > > > >>>> v4:
> > > > >>>> - Remove existing heuristic that checks just for size. (Ville)
> > > > >>>> - Return early if we find space to map at-least two objects.
> > > > >>>> (Tvrtko)
> > > > >>>> - Slightly update the commit message.
> > > > >>>>
> > > > >>>> v5: (Tvrtko)
> > > > >>>> - Rename the function to indicate that the object may be too big to
> > > > >>>> map into the aperture.
> > > > >>>> - Account for guard pages while calculating the total size required
> > > > >>>> for the object.
> > > > >>>> - Do not subject all objects to the heuristic check and instead
> > > > >>>> consider objects only of a certain size.
> > > > >>>> - Do the hole walk using the rbtree.
> > > > >>>> - Preserve the existing PIN_NONBLOCK logic.
> > > > >>>> - Drop the PIN_MAPPABLE check while pinning the VMA.
> > > > >>>>
> > > > >>>> v6: (Tvrtko)
> > > > >>>> - Return 0 on success and the specific error code on failure to
> > > > >>>> preserve the existing behavior.
> > > > >>>>
> > > > >>>> Cc: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> > > > >>>> Cc: Maarten Lankhorst <maarten.lankhorst@xxxxxxxxxxxxxxx>
> > > > >>>> Cc: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxxxxxxxx>
> > > > >>>> Cc: Manasi Navare <manasi.d.navare@xxxxxxxxx>
> > > > >>>> Signed-off-by: Vivek Kasireddy <vivek.kasireddy@xxxxxxxxx>
> > > > >>>> ---
> > > > >>>> drivers/gpu/drm/i915/i915_gem.c | 120 ++++++++++++++++++++++++-----
> > ---
> > > > >>>> 1 file changed, 90 insertions(+), 30 deletions(-)
> > > > >>>>
> > > > >>>> diff --git a/drivers/gpu/drm/i915/i915_gem.c
> > > > >>>> b/drivers/gpu/drm/i915/i915_gem.c index e3a2c2a0e156..39f0d17550c3
> > > > >>>> 100644
> > > > >>>> --- a/drivers/gpu/drm/i915/i915_gem.c
> > > > >>>> +++ b/drivers/gpu/drm/i915/i915_gem.c
> > > > >>>> @@ -46,6 +46,7 @@
> > > > >>>> #include "gem/i915_gem_mman.h"
> > > > >>>> #include "gem/i915_gem_region.h"
> > > > >>>> #include "gem/i915_gem_userptr.h"
> > > > >>>> +#include "gem/i915_gem_tiling.h"
> > > > >>>> #include "gt/intel_engine_user.h"
> > > > >>>> #include "gt/intel_gt.h"
> > > > >>>> #include "gt/intel_gt_pm.h"
> > > > >>>> @@ -876,6 +877,92 @@ static void discard_ggtt_vma(struct i915_vma *vma)
> > > > >>>> spin_unlock(&obj->vma.lock);
> > > > >>>> }
> > > > >>>>
> > > > >>>> +static int
> > > > >>>> +i915_gem_object_fits_in_aperture(struct drm_i915_gem_object *obj,
> > > > >>>> + u64 alignment, u64 flags)
> > > > >>>> +{
> > > > >>>> + struct drm_i915_private *i915 = to_i915(obj->base.dev);
> > > > >>>> + struct i915_ggtt *ggtt = to_gt(i915)->ggtt;
> > > > >>>> + struct drm_mm_node *hole;
> > > > >>>> + u64 hole_start, hole_end, start, end;
> > > > >>>> + u64 fence_size, fence_alignment;
> > > > >>>> + unsigned int count = 0;
> > > > >>>> +
> > > > >>>> + /*
> > > > >>>> + * If the required space is larger than the available
> > > > >>>> + * aperture, we will not able to find a slot for the
> > > > >>>> + * object and unbinding the object now will be in
> > > > >>>> + * vain. Worse, doing so may cause us to ping-pong
> > > > >>>> + * the object in and out of the Global GTT and
> > > > >>>> + * waste a lot of cycles under the mutex.
> > > > >>>> + */
> > > > >>>> + if (obj->base.size > ggtt->mappable_end)
> > > > >>>> + return -E2BIG;
> > > > >>>> +
> > > > >>>> + /*
> > > > >>>> + * If NONBLOCK is set the caller is optimistically
> > > > >>>> + * trying to cache the full object within the mappable
> > > > >>>> + * aperture, and *must* have a fallback in place for
> > > > >>>> + * situations where we cannot bind the object. We
> > > > >>>> + * can be a little more lax here and use the fallback
> > > > >>>> + * more often to avoid costly migrations of ourselves
> > > > >>>> + * and other objects within the aperture.
> > > > >>>> + */
> > > > >>>> + if (!(flags & PIN_NONBLOCK))
> > > > >>>> + return 0;
> > > > >>>> +
> > > > >>>> + /*
> > > > >>>> + * We only consider objects whose size is at-least a quarter of
> > > > >>>> + * the aperture to be too big and subject them to the new
> > > > >>>> + * heuristic below.
> > > > >>>> + */
> > > > >>>> + if (obj->base.size < ggtt->mappable_end / 4)
> > > > >>>> + return 0;
> > > > >>>
> > > > >>> That seems a fairly arbitrary thing to put here. Maybe something the
> > > > >>> caller should check/specify?
> > > > >>
> > > > >> I have no strong opinion on this one. In my mind I categorised it
> > > > >> under "is it a large framebuffer" heuristics. Previously it was less
> > > > >> than one half of aperture always okay, now one quarter, plus 2x hole
> > > > >> check if larger. Both are heuristics. I even mentioned earlier if 2x
> > > > >> should be an input parameter as well, but again, given it's not an
> > > > >> exported function couldn't really justify it.
> > > > >
> > > > > Is there any point in even having this extra check? If we don't think
> > > > > checking this is worth the hassle then why call the function at all?
> > > >
> > > > The "/4" one? It was my suggestion to avoid the hole search if we can know based on
> > size
> > > > it cannot be a frame buffer that would be affected by the ping-ping problem. Granted
> > that
> > > > was before the rbtree hole search, when it was traversing the un-ordered linked list of
> > > > holes. What is the correct size threshold I don't know.
> > > >
> > > > >>>> +
> > > > >>>> + if (HAS_GMCH(i915) || DISPLAY_VER(i915) < 11 ||
> > > > >>>> + !i915_gem_object_is_framebuffer(obj))
> > > > >>>> + return 0;
> > > > >>>
> > > > >>> None of that seems appropriate for a generic gem function with this
> > > > >>> name.
> > > > >>
> > > > >> It's not exported though, maybe remove i915_gem prefix to avoid any
> > > > >> ideas of it being generic?
> > > > >
> > > > > These checks don't even seem to doing anything useful. HAS_GMCH should
> > > > > already be covered by always setting PIN_MAPPABLE and hence O_NONBLOCK
> > > [Kasireddy, Vivek] I can drop the HAS_GMCH(i915) check given that it is redundant.
> > >
> > > > > is never even tried, the pre-icl vs. icl+ check should not exist at
> > > [Kasireddy, Vivek] My aim was to narrow down the list of situations in which the
> > > ping-pong problem becomes more pronounced and may lead to performance
> > > issues. Therefore, I added the DISPLAY_VER(i915) check since 8K/bigjoiner is
> > > feasible only on those newer platforms.
> >
> > Like I said before bigjoiner is irrelevant. The only thing that
> > matters is the size of the mapping vs. mappable aperture size.
> [Kasireddy, Vivek] Ok, got it.
>
> >
> > >
> > > > > all IMO, and if this is only called for framebuffers then why does the
> > > > > code pretend that is not the case?
> > > [Kasireddy, Vivek] Oh, I added the i915_gem_object_is_framebuffer() check after I
> > > found that there are other callers (for example, reloc_iomap() in i915_gem_execbuffer.c)
> > > of i915_gem_object_ggtt_pin_ww() that may not be working on an fb.
> > >
> > > Also, I figured size < ggtt->mappable_end / 4 or a similar check is needed as we do
> > > not want to subject all FBs through this performance critical path.
> >
> > Why not?
> [Kasireddy, Vivek] Oh, I just thought that it makes sense to avoid the expensive hole search
> for smaller FBs that are unlikely to exhaust the mappable aperture space. And, I also wanted
> to preserve the behavior prior to this patch. However, I guess I could drop this check as well
> given that the hole search via rbtree traversal may not be that bad in terms of time.
>
> What about the i915_gem_object_is_framebuffer() check? Should I keep it given the smaller
> size and transient nature of batchbuffers that go through this path?
Hmm. I guess if we don't have any real idea on the cost of this then
maybe we should not do it there, for the moment at least. I'm just
pondering if this whole thing should just be a seaprate function
each current caller of i915_gem_object_ggtt_pin_ww() should call
explicitly... But maybe we want framebuffers in partcular to undergo
the same restrictions always whether it's execbuffer or display
activity that's trying to pin them.
However, now that I think about this I'm wondering if we should do
all this stuff a bit later. I mean why are we even bothering with
these checks if the vma is already bound and not misplaced? Seems
the vma_is_misplaced() case already has the same mappable_size/2
check so that should probably be replaced with this thing, and then
additionally we could do something like this:
if (vma_is_miplaced()) {
...
if (flags & PIN_NONBLOCK)
already have a check here;
+ } else if (!vma_is_bound()) {
+ if (flags & PIN_NONBLOCK)
+ check here too;
+ }
--
Ville Syrjälä
Intel
