netns xfrm: per-netns xfrm_hash_work

All of this is implicit passing which netns's hashes should be resized.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Alexey Dobriyan 2008-11-25 17:19:07 -08:00 committed by David S. Miller
parent 0bf7c5b019
commit 6308273385
2 changed files with 19 additions and 18 deletions

View File

@ -2,6 +2,7 @@
#define __NETNS_XFRM_H
#include <linux/list.h>
#include <linux/workqueue.h>
struct netns_xfrm {
struct list_head state_all;
@ -18,6 +19,7 @@ struct netns_xfrm {
struct hlist_head *state_byspi;
unsigned int state_hmask;
unsigned int state_num;
struct work_struct state_hash_work;
};
#endif

View File

@ -109,16 +109,16 @@ static void xfrm_hash_transfer(struct hlist_head *list,
}
}
static unsigned long xfrm_hash_new_size(void)
static unsigned long xfrm_hash_new_size(unsigned int state_hmask)
{
return ((init_net.xfrm.state_hmask + 1) << 1) *
sizeof(struct hlist_head);
return ((state_hmask + 1) << 1) * sizeof(struct hlist_head);
}
static DEFINE_MUTEX(hash_resize_mutex);
static void xfrm_hash_resize(struct work_struct *__unused)
static void xfrm_hash_resize(struct work_struct *work)
{
struct net *net = container_of(work, struct net, xfrm.state_hash_work);
struct hlist_head *ndst, *nsrc, *nspi, *odst, *osrc, *ospi;
unsigned long nsize, osize;
unsigned int nhashmask, ohashmask;
@ -126,7 +126,7 @@ static void xfrm_hash_resize(struct work_struct *__unused)
mutex_lock(&hash_resize_mutex);
nsize = xfrm_hash_new_size();
nsize = xfrm_hash_new_size(net->xfrm.state_hmask);
ndst = xfrm_hash_alloc(nsize);
if (!ndst)
goto out_unlock;
@ -145,19 +145,19 @@ static void xfrm_hash_resize(struct work_struct *__unused)
spin_lock_bh(&xfrm_state_lock);
nhashmask = (nsize / sizeof(struct hlist_head)) - 1U;
for (i = init_net.xfrm.state_hmask; i >= 0; i--)
xfrm_hash_transfer(init_net.xfrm.state_bydst+i, ndst, nsrc, nspi,
for (i = net->xfrm.state_hmask; i >= 0; i--)
xfrm_hash_transfer(net->xfrm.state_bydst+i, ndst, nsrc, nspi,
nhashmask);
odst = init_net.xfrm.state_bydst;
osrc = init_net.xfrm.state_bysrc;
ospi = init_net.xfrm.state_byspi;
ohashmask = init_net.xfrm.state_hmask;
odst = net->xfrm.state_bydst;
osrc = net->xfrm.state_bysrc;
ospi = net->xfrm.state_byspi;
ohashmask = net->xfrm.state_hmask;
init_net.xfrm.state_bydst = ndst;
init_net.xfrm.state_bysrc = nsrc;
init_net.xfrm.state_byspi = nspi;
init_net.xfrm.state_hmask = nhashmask;
net->xfrm.state_bydst = ndst;
net->xfrm.state_bysrc = nsrc;
net->xfrm.state_byspi = nspi;
net->xfrm.state_hmask = nhashmask;
spin_unlock_bh(&xfrm_state_lock);
@ -170,8 +170,6 @@ out_unlock:
mutex_unlock(&hash_resize_mutex);
}
static DECLARE_WORK(xfrm_hash_work, xfrm_hash_resize);
DECLARE_WAIT_QUEUE_HEAD(km_waitq);
EXPORT_SYMBOL(km_waitq);
@ -754,7 +752,7 @@ static void xfrm_hash_grow_check(int have_hash_collision)
if (have_hash_collision &&
(init_net.xfrm.state_hmask + 1) < xfrm_state_hashmax &&
init_net.xfrm.state_num > init_net.xfrm.state_hmask)
schedule_work(&xfrm_hash_work);
schedule_work(&init_net.xfrm.state_hash_work);
}
struct xfrm_state *
@ -2089,6 +2087,7 @@ int __net_init xfrm_state_init(struct net *net)
net->xfrm.state_hmask = ((sz / sizeof(struct hlist_head)) - 1);
net->xfrm.state_num = 0;
INIT_WORK(&net->xfrm.state_hash_work, xfrm_hash_resize);
INIT_WORK(&xfrm_state_gc_work, xfrm_state_gc_task);
return 0;