davidedmundson requested changes to this revision. davidedmundson added inline comments. This revision now requires changes to proceed.
INLINE COMMENTS > drm_buffer.cpp:131 > #if HAVE_GBM > + if (m_deferred) { > + return; If it's deferred it means someone else is doing the gbm_surface_release. But we still need to set m_bo to nullptr. Otherwise it's potentially left dangling here after the RemoteAccessManager has deleted it. (Alternately: if we changed DrmObjectPlane to store the buffers as QSharedPointers we could just keep a reference to the DrmBuffer in the RemoteAccessManager, which would be IMHO cleaner than doing low level GBM stuff there and having the data released in one of two places. I'll look into that) > remoteaccess_manager.h:49-50 > + > +signals: > + void bufferNoLongerNeeded(qint32 gbm_handle); > + what's this signal for? REPOSITORY R108 KWin REVISION DETAIL https://phabricator.kde.org/D1230 To: Kanedias, graesslin, davidedmundson Cc: davidedmundson, plasma-devel, ZrenBot, spstarr, progwolff, lesliezhai, ali-mohamed, jensreuterberg, abetts, sebas, apol, mart, lukas