On Tue, Nov 05, 2024 at 01:56:14PM -0800, Jing Zhang wrote: > On Tue, Nov 5, 2024 at 1:33 PM Oliver Upton <oliver.upton@xxxxxxxxx> wrote: > > > > Hi Jing, > > > > On Tue, Nov 05, 2024 at 11:34:18AM -0800, Jing Zhang wrote: > > > The core issue stems from the static linked list implementation of DTEs/ITEs, > > > requiring a full table scan to locate the list head during restoration. This > > > scan increases the likelihood of encountering orphaned entries. To rectify > > > this, the patch series introduces a dummy head to the list, enabling immediate > > > access to the list head and bypassing the scan. This optimization not only > > > resolves the bug but also significantly enhances restore performance, > > > particularly in edge cases where valid entries reside at the end of the table. > > > > I think we need a more targeted fix (i.e. Kunkun's patch) to stop the > > bleeding + backport it to stable. > > > > Then we can have a separate discussion about improving the save/restore > > performance with your approach. > > Yes, I'll respin Kunkun's patch soon. This patch series has the > selftest which we can use for verification. Right -- go ahead and include your selftest as part of that respin, I'd definitely like to have some test coverage for this whole save/restore mess. -- Thanks, Oliver
