Commit 52df157f authored by Wei Wang's avatar Wei Wang Committed by David S. Miller

xfrm: take refcnt of dst when creating struct xfrm_dst bundle

During the creation of xfrm_dst bundle, always take ref count when
allocating the dst. This way, xfrm_bundle_create() will form a linked
list of dst with dst->child pointing to a ref counted dst child. And
the returned dst pointer is also ref counted. This makes the link from
the flow cache to this dst now ref counted properly.
As the dst is always ref counted properly, we can safely mark
DST_NOGC flag so dst_release() will release dst based on refcnt only.
And dst gc is no longer needed and all dst_free() and its related
function calls should be replaced with dst_release() or
dst_release_immediate().

The special handling logic for dst->child in dst_destroy() can be
replaced with a simple dst_release_immediate() call on the child to
release the whole list linked by dst->child pointer.
Previously used DST_NOHASH flag is not needed anymore as well. The
reason that DST_NOHASH is used in the existing code is mainly to prevent
the dst inserted in the fib tree to be wrongly destroyed during the
deletion of the xfrm_dst bundle. So in the existing code, DST_NOHASH
flag is marked in all the dst children except the one which is in the
fib tree.
However, with this patch series to remove dst gc logic and release dst
only based on ref count, it is safe to release all the children from a
xfrm_dst bundle as long as the dst children are all ref counted
properly which is already the case in the existing code.
So, this patch removes the use of DST_NOHASH flag.
Signed-off-by: default avatarWei Wang <weiwan@google.com>
Acked-by: default avatarMartin KaFai Lau <kafai@fb.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent db916649
...@@ -51,7 +51,6 @@ struct dst_entry { ...@@ -51,7 +51,6 @@ struct dst_entry {
#define DST_HOST 0x0001 #define DST_HOST 0x0001
#define DST_NOXFRM 0x0002 #define DST_NOXFRM 0x0002
#define DST_NOPOLICY 0x0004 #define DST_NOPOLICY 0x0004
#define DST_NOHASH 0x0008
#define DST_NOCACHE 0x0010 #define DST_NOCACHE 0x0010
#define DST_NOCOUNT 0x0020 #define DST_NOCOUNT 0x0020
#define DST_FAKE_RTABLE 0x0040 #define DST_FAKE_RTABLE 0x0040
......
...@@ -250,7 +250,6 @@ struct dst_entry *dst_destroy(struct dst_entry * dst) ...@@ -250,7 +250,6 @@ struct dst_entry *dst_destroy(struct dst_entry * dst)
smp_rmb(); smp_rmb();
again:
child = dst->child; child = dst->child;
if (!(dst->flags & DST_NOCOUNT)) if (!(dst->flags & DST_NOCOUNT))
...@@ -269,20 +268,8 @@ struct dst_entry *dst_destroy(struct dst_entry * dst) ...@@ -269,20 +268,8 @@ struct dst_entry *dst_destroy(struct dst_entry * dst)
kmem_cache_free(dst->ops->kmem_cachep, dst); kmem_cache_free(dst->ops->kmem_cachep, dst);
dst = child; dst = child;
if (dst) { if (dst)
int nohash = dst->flags & DST_NOHASH; dst_release_immediate(dst);
if (atomic_dec_and_test(&dst->__refcnt)) {
/* We were real parent of this dst, so kill child. */
if (nohash)
goto again;
} else {
/* Child is still referenced, return it for freeing. */
if (nohash)
return dst;
/* Child is still in his hash table */
}
}
return NULL; return NULL;
} }
EXPORT_SYMBOL(dst_destroy); EXPORT_SYMBOL(dst_destroy);
...@@ -292,8 +279,6 @@ static void dst_destroy_rcu(struct rcu_head *head) ...@@ -292,8 +279,6 @@ static void dst_destroy_rcu(struct rcu_head *head)
struct dst_entry *dst = container_of(head, struct dst_entry, rcu_head); struct dst_entry *dst = container_of(head, struct dst_entry, rcu_head);
dst = dst_destroy(dst); dst = dst_destroy(dst);
if (dst)
__dst_free(dst);
} }
/* Operations to mark dst as DEAD and clean up the net device referenced /* Operations to mark dst as DEAD and clean up the net device referenced
......
...@@ -1590,7 +1590,9 @@ static void xfrm_bundle_flo_delete(struct flow_cache_object *flo) ...@@ -1590,7 +1590,9 @@ static void xfrm_bundle_flo_delete(struct flow_cache_object *flo)
struct xfrm_dst *xdst = container_of(flo, struct xfrm_dst, flo); struct xfrm_dst *xdst = container_of(flo, struct xfrm_dst, flo);
struct dst_entry *dst = &xdst->u.dst; struct dst_entry *dst = &xdst->u.dst;
dst_free(dst); /* Mark DST_OBSOLETE_DEAD to fail the next xfrm_dst_check() */
dst->obsolete = DST_OBSOLETE_DEAD;
dst_release_immediate(dst);
} }
static const struct flow_cache_ops xfrm_bundle_fc_ops = { static const struct flow_cache_ops xfrm_bundle_fc_ops = {
...@@ -1620,7 +1622,7 @@ static inline struct xfrm_dst *xfrm_alloc_dst(struct net *net, int family) ...@@ -1620,7 +1622,7 @@ static inline struct xfrm_dst *xfrm_alloc_dst(struct net *net, int family)
default: default:
BUG(); BUG();
} }
xdst = dst_alloc(dst_ops, NULL, 0, DST_OBSOLETE_NONE, 0); xdst = dst_alloc(dst_ops, NULL, 1, DST_OBSOLETE_NONE, DST_NOGC);
if (likely(xdst)) { if (likely(xdst)) {
struct dst_entry *dst = &xdst->u.dst; struct dst_entry *dst = &xdst->u.dst;
...@@ -1723,10 +1725,11 @@ static struct dst_entry *xfrm_bundle_create(struct xfrm_policy *policy, ...@@ -1723,10 +1725,11 @@ static struct dst_entry *xfrm_bundle_create(struct xfrm_policy *policy,
if (!dst_prev) if (!dst_prev)
dst0 = dst1; dst0 = dst1;
else { else
dst_prev->child = dst_clone(dst1); /* Ref count is taken during xfrm_alloc_dst()
dst1->flags |= DST_NOHASH; * No need to do dst_clone() on dst1
} */
dst_prev->child = dst1;
xdst->route = dst; xdst->route = dst;
dst_copy_metrics(dst1, dst); dst_copy_metrics(dst1, dst);
...@@ -1792,7 +1795,7 @@ static struct dst_entry *xfrm_bundle_create(struct xfrm_policy *policy, ...@@ -1792,7 +1795,7 @@ static struct dst_entry *xfrm_bundle_create(struct xfrm_policy *policy,
xfrm_state_put(xfrm[i]); xfrm_state_put(xfrm[i]);
free_dst: free_dst:
if (dst0) if (dst0)
dst_free(dst0); dst_release_immediate(dst0);
dst0 = ERR_PTR(err); dst0 = ERR_PTR(err);
goto out; goto out;
} }
...@@ -2073,7 +2076,11 @@ xfrm_bundle_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir, ...@@ -2073,7 +2076,11 @@ xfrm_bundle_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir,
pol_dead |= pols[i]->walk.dead; pol_dead |= pols[i]->walk.dead;
} }
if (pol_dead) { if (pol_dead) {
dst_free(&xdst->u.dst); /* Mark DST_OBSOLETE_DEAD to fail the next
* xfrm_dst_check()
*/
xdst->u.dst.obsolete = DST_OBSOLETE_DEAD;
dst_release_immediate(&xdst->u.dst);
xdst = NULL; xdst = NULL;
num_pols = 0; num_pols = 0;
num_xfrms = 0; num_xfrms = 0;
...@@ -2120,11 +2127,12 @@ xfrm_bundle_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir, ...@@ -2120,11 +2127,12 @@ xfrm_bundle_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir,
if (xdst) { if (xdst) {
/* The policies were stolen for newly generated bundle */ /* The policies were stolen for newly generated bundle */
xdst->num_pols = 0; xdst->num_pols = 0;
dst_free(&xdst->u.dst); /* Mark DST_OBSOLETE_DEAD to fail the next xfrm_dst_check() */
xdst->u.dst.obsolete = DST_OBSOLETE_DEAD;
dst_release_immediate(&xdst->u.dst);
} }
/* Flow cache does not have reference, it dst_free()'s, /* We do need to return one reference for original caller */
* but we do need to return one reference for original caller */
dst_hold(&new_xdst->u.dst); dst_hold(&new_xdst->u.dst);
return &new_xdst->flo; return &new_xdst->flo;
...@@ -2147,9 +2155,11 @@ xfrm_bundle_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir, ...@@ -2147,9 +2155,11 @@ xfrm_bundle_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir,
inc_error: inc_error:
XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTPOLERROR); XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTPOLERROR);
error: error:
if (xdst != NULL) if (xdst != NULL) {
dst_free(&xdst->u.dst); /* Mark DST_OBSOLETE_DEAD to fail the next xfrm_dst_check() */
else xdst->u.dst.obsolete = DST_OBSOLETE_DEAD;
dst_release_immediate(&xdst->u.dst);
} else
xfrm_pols_put(pols, num_pols); xfrm_pols_put(pols, num_pols);
return ERR_PTR(err); return ERR_PTR(err);
} }
...@@ -2636,10 +2646,12 @@ static struct dst_entry *xfrm_dst_check(struct dst_entry *dst, u32 cookie) ...@@ -2636,10 +2646,12 @@ static struct dst_entry *xfrm_dst_check(struct dst_entry *dst, u32 cookie)
* notice. That's what we are validating here via the * notice. That's what we are validating here via the
* stale_bundle() check. * stale_bundle() check.
* *
* When a policy's bundle is pruned, we dst_free() the XFRM * When an xdst is removed from flow cache, DST_OBSOLETE_DEAD will
* dst which causes it's ->obsolete field to be set to * be marked on it.
* DST_OBSOLETE_DEAD. If an XFRM dst has been pruned like * When a dst is removed from the fib tree, DST_OBSOLETE_DEAD will
* this, we want to force a new route lookup. * be marked on it.
* Both will force stable_bundle() to fail on any xdst bundle with
* this dst linked in it.
*/ */
if (dst->obsolete < 0 && !stale_bundle(dst)) if (dst->obsolete < 0 && !stale_bundle(dst))
return dst; return dst;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment