Re: [RFC 00/29] De-stage android's sync framework

From: Daniel Vetter
Date: Wed Jan 20 2016 - 11:29:35 EST


On Wed, Jan 20, 2016 at 04:02:31PM +0100, Maarten Lankhorst wrote:
> Op 20-01-16 om 15:32 schreef Gustavo Padovan:
> > 2016-01-20 Maarten Lankhorst <maarten.lankhorst@xxxxxxxxxxxxxxx>:
> >
> >> Hey,
> >>
> >> Op 15-01-16 om 15:55 schreef Gustavo Padovan:
> >>> From: Gustavo Padovan <gustavo.padovan@xxxxxxxxxxxxxxx>
> >>>
> >>> This patch series de-stage the sync framework, and in order to accomplish that
> >>> a bunch of cleanups/improvements on the sync and fence were made.
> >>>
> >>> The sync framework contained some abstractions around struct fence and those
> >>> were removed in the de-staging process among other changes:
> >>>
> >>> Userspace visible changes
> >>> -------------------------
> >>>
> >>> * The sw_sync file was moved from /dev/sw_sync to <debugfs>/sync/sw_sync. No
> >>> other change.
> >>>
> >>> Kernel API changes
> >>> ------------------
> >>>
> >>> * struct sync_timeline is now struct fence_timeline
> >>> * sync_timeline_ops is now fence_timeline_ops and they now carry struct
> >>> fence as parameter instead of struct sync_pt
> >>> * a .cleanup() fence op was added to allow sync_fence to run a cleanup when
> >>> the fence_timeline is destroyed
> >>> * added fence_add_used_data() to pass a private point to struct fence. This
> >>> pointer is sent back on the .cleanup op.
> >>> * The sync timeline function were moved to be fence_timeline functions:
> >>> - sync_timeline_create() -> fence_timeline_create()
> >>> - sync_timeline_get() -> fence_timeline_get()
> >>> - sync_timeline_put() -> fence_timeline_put()
> >>> - sync_timeline_destroy() -> fence_timeline_destroy()
> >>> - sync_timeline_signal() -> fence_timeline_signal()
> >>>
> >>> * sync_pt_create() was replaced be fence_create_on_timeline()
> >>>
> >>> Internal changes
> >>> ----------------
> >>>
> >>> * fence_timeline_ops was removed in favor of direct use fence_ops
> >>> * fence default functions were created for fence_ops
> >>> * removed structs sync_pt, sw_sync_timeline and sw_sync_pt
> >>>
> >>> Gustavo Padovan (29):
> >>> staging/android: fix sync framework documentation
> >>> staging/android: fix checkpatch warning
> >>> staging/android: rename sync_fence_release
> >>> staging/android: rename 'android_fence' to 'sync_fence'
> >>> staging/android: remove not used sync_timeline ops
> >>> staging/android: create a 'sync' dir for debugfs information
> >>> staging/android: move sw_sync file to debugfs file
> >>> staging/android: Remove WARN_ON_ONCE when releasing sync_fence
> >>> staging/android: rename struct sync_fence's variables to 'sync_fence'
> >>> staging/android: rename 'sync_pt' to 'fence' in struct sync_fence_cb
> >>> dma-buf/fence: move sync_timeline to fence_timeline
> >>> staging/android: remove struct sync_pt
> >>> dma-buf/fence: create fence_default_enable_signaling()
> >>> dma-buf/fence: create fence_default_release()
> >>> dma-buf/fence: create fence_default_get_driver_name()
> >>> dma-buf/fence: create fence_default_timeline_name()
> >> This is misleading. I think timeline_fence prefix would be more appropriate here.
> > Why? These fence_default_.. functions are fence_ops and not related to
> > fence_timeline in any way.
> Because they're using fence_parent, which should probably be renamed to fence_to_timeline()
>
> The name makes it sound as if they could apply to all type of fences, I don't think this is the case.
> >> I also believe this should be done in multiple series. First series should de-stage the userspace fence framework. The next series should fix up android_fence and maybe rename it to timeline_fence since sync_fence is already used for the userspace fd, which would add more confusion?
> > Sure. I've been thinking on how to split this properly. I'm trying to
> > add a bunch of clean up/renaming first, eg the sync_fence rename to
> > sync_file that Daniel Vetter and I discussed.
> >
> > Next my plan would be move sync_timeline to fence_timeline, add the
> > fence_default.. fence_ops, clean up sw_sync and finally merge
> > fence_context and fence_timeline.
> >
> > Looking at how sync and fence It looks easier to me to de-stage sync_timeline first than userspace
> > fence.
> There's already code to add a sync_fence_create_dma export [1][2]. So if you want to de-stage it then there will be users for it.
>
> sync_pt otoh has no upstream in-kernel users. It was a wrapper to keep android drivers api compatible with the fence api.
>
> [1] https://patchwork.freedesktop.org/patch/67845/
> [2] https://patchwork.freedesktop.org/patch/67846/

We kinda don't have open-source userspace for this stuf though ...
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch