On Thu, Feb 21, 2019 at 11:28:49AM -0700, Jason Gunthorpe wrote: > On Wed, Feb 20, 2019 at 04:20:44PM -0800, Matthew Wilcox wrote: > > @@ -256,25 +252,19 @@ id_map_alloc(struct ib_device *ibdev, int slave_id, u32 sl_cm_id) > > ent->dev = to_mdev(ibdev); > > INIT_DELAYED_WORK(&ent->timeout, id_map_ent_timeout); > > > > - idr_preload(GFP_KERNEL); > > - spin_lock(&to_mdev(ibdev)->sriov.id_map_lock); > > - > > - ret = idr_alloc_cyclic(&sriov->pv_id_table, ent, 0, 0, GFP_NOWAIT); > > + ret = xa_alloc_cyclic(&sriov->pv_id_table, &ent->pv_cm_id, ent, > > + xa_limit_32b, &sriov->pv_id_next, GFP_KERNEL); > > Why drop the NO_WAIT? Explain in the commit message? That's the standard pattern. idr_preload(GFP_KERNEL); spin_lock() idr_alloc(..., GFP_NOWAIT); becomes just xa_alloc(..., GFP_KERNEL);
