Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit dc10f846 authored by Francis Hart's avatar Francis Hart
Browse files

Surface: Ensure synchronisation of copyBlt 

The Surface::lock() function has an optimisation for copying the
previously drawn contents from the frontbuffer to the (current)
backbuffer, so that the client does not have to redraw the whole surface
contents. This logic was not using the sync fence FD from
dequeueBuffer() and so was not correctly synchronised with pending
operations on the backbuffer.

Change-Id: Ib44574d50f8bdb4a23078cd8da1c5c512c876320
parent 77f764d6

libs/gui/Surface.cpp

+14 −1
Original line number Diff line number Diff line
@@ -20,6 +20,13 @@ 


#include <android/native_window.h>



// We would eliminate the non-conforming zero-length array, but we can't since

// this is effectively included from the Linux kernel

#pragma clang diagnostic push

#pragma clang diagnostic ignored "-Wzero-length-array"

#include <sync/sync.h>

#pragma clang diagnostic pop



#include <binder/Parcel.h>



#include <utils/Log.h>
@@ -1269,8 +1276,14 @@ status_t Surface::lock( 
        if (canCopyBack) {

            // copy the area that is invalid and not repainted this round

            const Region copyback(mDirtyRegion.subtract(newDirtyRegion));

            if (!copyback.isEmpty())

            if (!copyback.isEmpty()) {

                if (fenceFd >= 0) {

                    sync_wait(fenceFd, -1);

                    close(fenceFd);

                    fenceFd = -1;

                }

                copyBlt(backBuffer, frontBuffer, copyback);

            }

        } else {

            // if we can't copy-back anything, modify the user's dirty

            // region to make sure they redraw the whole buffer