If the surface cache entry is a placeholder (because we haven't decoded any frames yet) then a surface cache lookup will not return anything, so we can't get ahold of the AnimationSurfaceProvider to call reset on it.
I also looked at the other places we call Lookup on the surface cache and it seems correct to me for them to not get the surface if it's still a placeholder.
Differential Revision: https://phabricator.services.mozilla.com/D143894
Now that we no longer have the extra layer of ImageContainers providing
a superficial level of caching/reuse of existing blob recordings, we
need some way to share recordings. This part adds support to
SurfaceCache to store BlobSurfaceProvider objects.
This includes the specialized code for invalidating SVG images. In
particular this is useful for animated SVG images. In general we want to
avoid changing the image key whenever possible so that we avoid
reallocating the underlying buffers in the compositor process for the
rasterized blob images.
We also need to track the ImageIntRegion used by the recording. If a
caller only wants a slice of the SVG image, then we need to track this
differentiation in our cache entries. At this time, we don't allow
substitutes for entries with a region exclusion.
Differential Revision: https://phabricator.services.mozilla.com/D126603
This provides the framework to allow ISurfaceProvider objects to
implement WebRenderImageProvider. It is straightforward for rasterized
providers (DecodedSurfaceProvider, and SimpleSurfaceProvider). Later
parts in this series will provide the necessary changes for blob
recordings and for animatedi images.
Differential Revision: https://phabricator.services.mozilla.com/D126597
Now that we no longer have the extra layer of ImageContainers providing
a superficial level of caching/reuse of existing blob recordings, we
need some way to share recordings. This part adds support to
SurfaceCache to store BlobSurfaceProvider objects.
This includes the specialized code for invalidating SVG images. In
particular this is useful for animated SVG images. In general we want to
avoid changing the image key whenever possible so that we avoid
reallocating the underlying buffers in the compositor process for the
rasterized blob images.
We also need to track the ImageIntRegion used by the recording. If a
caller only wants a slice of the SVG image, then we need to track this
differentiation in our cache entries. At this time, we don't allow
substitutes for entries with a region exclusion.
Differential Revision: https://phabricator.services.mozilla.com/D126603
This provides the framework to allow ISurfaceProvider objects to
implement WebRenderImageProvider. It is straightforward for rasterized
providers (DecodedSurfaceProvider, and SimpleSurfaceProvider). Later
parts in this series will provide the necessary changes for blob
recordings and for animatedi images.
Differential Revision: https://phabricator.services.mozilla.com/D126597
Now that we no longer have the extra layer of ImageContainers providing
a superficial level of caching/reuse of existing blob recordings, we
need some way to share recordings. This part adds support to
SurfaceCache to store BlobSurfaceProvider objects.
This includes the specialized code for invalidating SVG images. In
particular this is useful for animated SVG images. In general we want to
avoid changing the image key whenever possible so that we avoid
reallocating the underlying buffers in the compositor process for the
rasterized blob images.
We also need to track the ImageIntRegion used by the recording. If a
caller only wants a slice of the SVG image, then we need to track this
differentiation in our cache entries. At this time, we don't allow
substitutes for entries with a region exclusion.
Differential Revision: https://phabricator.services.mozilla.com/D126603
This provides the framework to allow ISurfaceProvider objects to
implement WebRenderImageProvider. It is straightforward for rasterized
providers (DecodedSurfaceProvider, and SimpleSurfaceProvider). Later
parts in this series will provide the necessary changes for blob
recordings and for animatedi images.
Differential Revision: https://phabricator.services.mozilla.com/D126597
An image container can keep a surface alive longer than it can remain in
the cache, if it is indeed kept in the cache. We should cross reference
our memory report generated from the SurfaceCache against any surfaces
stored in our ImageContainer objects to ensure they are all reported.
This is of particular importance for blob recordings which are not put
into SurfaceCache. While the recordings themselves have their own memory
reporting inside of WebRender, it would be good to know what recordings
we are keeping alive from the content side to help break it down.
Differential Revision: https://phabricator.services.mozilla.com/D115517
Current gecko dispatches tasks of releasing image::Image to main thread at some places. The task was dispatched individually. Then there were cases that the releasing took long time. It increased peak memory usage and caused a problem like Bug 1639280. When main thread is very busy like WebGL, it takes longer time until the Image is released on main thread. If Images are released quickly on main thread, we could reduce peak memory usage.
When SurfaceCache::ReleaseImageOnMainThread() is called, there could be already an ongoing task for releasing Image. It could reduce a duration until release on main thread.
Differential Revision: https://phabricator.services.mozilla.com/D82864
SVG performance with the fallback path with WebRender is very bad. This
patch avoids fallback by always producing a rasterized surface we store
in SurfaceCache, but also clamping the size consistently to a configured
maximum. This will cause us to upscale rasterized SVGs which is
undesirable visually but is a lower risk change that we can uplift to
beta than fixing the underlying performance issue.
Differential Revision: https://phabricator.services.mozilla.com/D27159
This is a best effort attempt at ensuring that the adverse impact of
reformatting the entire tree over the comments would be minimal. I've used a
combination of strategies including disabling of formatting, some manual
formatting and some changes to formatting to work around some clang-format
limitations.
Differential Revision: https://phabricator.services.mozilla.com/D13193
--HG--
extra : moz-landing-system : lando
We were marking them used even if only a decode was requested.
This can cause us to hold extra decoded copies of the image around because we have a tendency to request decode at the intrinsic size.
When SurfaceCache::Lookup is called to access surface data, it indicates
that the caller will not accept substitutes as in the case of
SurfaceCache::LookupBestMatch. As such, we need to be careful not to
remove those surfaces from our cache when pruning (in part 8b). This is
the marker used to track that, at some point, there was a caller which
got this surface that would accept no other (e.g. factor of 2 mode must
make an accept for this particular surface).
imgFrame::SizeOfExcludingThis() measures heap and non-heap memory in a very
complex way. This patch simplifies it and removes gfxMemoryLocation in the
process. (gfxMemoryLocation::OUT_OF_PROCESS was unused.)
--HG--
extra : rebase_source : 72af38fa438b4b42df02231bcf2fa731d247b60d
Timothy Nikkel
Andrew Osmond
Sandor Molnar
sotaro
longsonr
Sylvestre Ledru
Ehsan Akhgari
Andreea Pavel
Seth Fowler
Nicholas Nethercote
Carsten "Tomcat" Book