On Wed, Sep 25, 2024 at 6:35 PM James Houghton <jthoughton@xxxxxxxxxx> wrote: > > Walk the TDP MMU in an RCU read-side critical section without holding > mmu_lock when harvesting and potentially updating age information on > sptes. This requires a way to do RCU-safe walking of the tdp_mmu_roots; > do this with a new macro. The PTE modifications are now done atomically, > and kvm_tdp_mmu_spte_need_atomic_write() has been updated to account for > the fact that kvm_age_gfn can now lockless update the accessed bit and > the W/R/X bits). > > If the cmpxchg for marking the spte for access tracking fails, leave it > as is and treat it as if it were young, as if the spte is being actively > modified, it is most likely young. > > Harvesting age information from the shadow MMU is still done while > holding the MMU write lock. > > Suggested-by: Yu Zhao <yuzhao@xxxxxxxxxx> > Signed-off-by: James Houghton <jthoughton@xxxxxxxxxx> Oh no! I have left off David Matlack's Reviewed-bys[1, 2] from this patch and from patch 2[3], and I failed to apply his comment suggestion on this patch like I said I would. Sorry David, I have fixed up my tree now. [1]: https://lore.kernel.org/linux-mm/ZqJ_xANKf3bNcaHM@xxxxxxxxxx/ [2]: https://lore.kernel.org/linux-mm/ZqKUefN3HgBQQkuA@xxxxxxxxxx/ [3]: https://lore.kernel.org/kvm/20240926013506.860253-3-jthoughton@xxxxxxxxxx/
