Message ID | 20190820223259.22348-24-willy@infradead.org |
---|---|
State | Changes Requested |
Delegated to: | David Miller |
Headers | show |
Series | Convert networking to use the XArray | expand |
From: Matthew Wilcox <willy@infradead.org> Date: Tue, 20 Aug 2019 15:32:44 -0700 > From: "Matthew Wilcox (Oracle)" <willy@infradead.org> > > This module doesn't use the allocating functionality; convert it to a > plain XArray instead of an allocating one. I've left struct tcf_net > in place in case more objects are added to it in future, although > it now only contains an XArray. We don't need to call xa_destroy() > if the array is empty, so I've removed the contents of tcf_net_exit() > -- if it can be called with entries still in place, then it shoud call > xa_destroy() instead. > > Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> I don't know if the net exit can be invoked with entires still in place, however if the tcf_net_exit() function is made empty it should be removed along with the assignment to the per-netns ops.
On Tue, Aug 20, 2019 at 04:57:28PM -0700, David Miller wrote: > > From: "Matthew Wilcox (Oracle)" <willy@infradead.org> > > > > This module doesn't use the allocating functionality; convert it to a > > plain XArray instead of an allocating one. I've left struct tcf_net > > in place in case more objects are added to it in future, although > > it now only contains an XArray. We don't need to call xa_destroy() > > if the array is empty, so I've removed the contents of tcf_net_exit() > > -- if it can be called with entries still in place, then it shoud call > > xa_destroy() instead. > > > > Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> > > I don't know if the net exit can be invoked with entires still in place, > however if the tcf_net_exit() function is made empty it should be removed > along with the assignment to the per-netns ops. Thanks! I wasn't sure what the rule was for these ops. I'll fix that up.
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c index e0d8b456e9f5..8392a7ef0ed4 100644 --- a/net/sched/cls_api.c +++ b/net/sched/cls_api.c @@ -19,7 +19,7 @@ #include <linux/init.h> #include <linux/kmod.h> #include <linux/slab.h> -#include <linux/idr.h> +#include <linux/xarray.h> #include <linux/rhashtable.h> #include <net/net_namespace.h> #include <net/sock.h> @@ -777,8 +777,7 @@ tcf_chain0_head_change_cb_del(struct tcf_block *block, } struct tcf_net { - spinlock_t idr_lock; /* Protects idr */ - struct idr idr; + struct xarray blocks; }; static unsigned int tcf_net_id; @@ -787,25 +786,15 @@ static int tcf_block_insert(struct tcf_block *block, struct net *net, struct netlink_ext_ack *extack) { struct tcf_net *tn = net_generic(net, tcf_net_id); - int err; - - idr_preload(GFP_KERNEL); - spin_lock(&tn->idr_lock); - err = idr_alloc_u32(&tn->idr, block, &block->index, block->index, - GFP_NOWAIT); - spin_unlock(&tn->idr_lock); - idr_preload_end(); - return err; + return xa_insert(&tn->blocks, block->index, block, GFP_KERNEL); } static void tcf_block_remove(struct tcf_block *block, struct net *net) { struct tcf_net *tn = net_generic(net, tcf_net_id); - spin_lock(&tn->idr_lock); - idr_remove(&tn->idr, block->index); - spin_unlock(&tn->idr_lock); + xa_erase(&tn->blocks, block->index); } static struct tcf_block *tcf_block_create(struct net *net, struct Qdisc *q, @@ -839,7 +828,7 @@ static struct tcf_block *tcf_block_lookup(struct net *net, u32 block_index) { struct tcf_net *tn = net_generic(net, tcf_net_id); - return idr_find(&tn->idr, block_index); + return xa_load(&tn->blocks, block_index); } static struct tcf_block *tcf_block_refcnt_get(struct net *net, u32 block_index) @@ -3164,16 +3153,12 @@ static __net_init int tcf_net_init(struct net *net) { struct tcf_net *tn = net_generic(net, tcf_net_id); - spin_lock_init(&tn->idr_lock); - idr_init(&tn->idr); + xa_init(&tn->blocks); return 0; } static void __net_exit tcf_net_exit(struct net *net) { - struct tcf_net *tn = net_generic(net, tcf_net_id); - - idr_destroy(&tn->idr); } static struct pernet_operations tcf_net_ops = {