d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Send only one incoming notification at a time to reduce queue

Browse Source 
trashing and improve performance.
Remove waitflag argument from ng_ksocket_incoming2(), it means nothing
as function call was queued by netgraph.
Remove node validity check, as node validity guarantied by netgraph.
Update comments.
This commit is contained in:
Alexander Motin 2008-03-07 21:12:56 +00:00
parent 31dd4200da
commit 6e7ed93017
1 changed files with 26 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

58
sys/netgraph/ng_ksocket.c
View File

@ -82,6 +82,7 @@ struct ng_ksocket_private {
node_p node;
hook_p hook;
struct socket *so;
int fn_sent; /* FN call on incoming event was sent */
LIST_HEAD(, ng_ksocket_private) embryos;
LIST_ENTRY(ng_ksocket_private) siblings;
u_int32_t flags;
@ -161,7 +162,7 @@ static void ng_ksocket_incoming(struct socket *so, void *arg, int waitflag);
static int ng_ksocket_parse(const struct ng_ksocket_alias *aliases,
const char *s, int family);
static void ng_ksocket_incoming2(node_p node, hook_p hook,
void *arg1, int waitflag);
void *arg1, int arg2);
/************************************************************************
STRUCT SOCKADDR PARSE TYPE
@ -939,13 +940,13 @@ ng_ksocket_shutdown(node_p node)
/* Close our socket (if any) */
if (priv->so != NULL) {
priv->so->so_upcall = NULL;
SOCKBUF_LOCK(&priv->so->so_rcv);
priv->so->so_rcv.sb_flags &= ~SB_UPCALL;
SOCKBUF_UNLOCK(&priv->so->so_rcv);
SOCKBUF_LOCK(&priv->so->so_snd);
priv->so->so_snd.sb_flags &= ~SB_UPCALL;
SOCKBUF_UNLOCK(&priv->so->so_snd);
priv->so->so_upcall = NULL;
soclose(priv->so);
priv->so = NULL;
}
@ -988,22 +989,13 @@ ng_ksocket_disconnect(hook_p hook)
HELPER STUFF
************************************************************************/
/*
* You should no-longer "just call" a netgraph node function
* from an external asynchronous event.
* This is because in doing so you are ignoring the locking on the netgraph
* nodes. Instead call your function via
* "int ng_send_fn(node_p node, hook_p hook, ng_item_fn *fn,
* void *arg1, int arg2);"
* this will call the function you chose, but will first do all the
* You should not "just call" a netgraph node function from an external
* asynchronous event. This is because in doing so you are ignoring the
* locking on the netgraph nodes. Instead call your function via ng_send_fn().
* This will call the function you chose, but will first do all the
* locking rigmarole. Your function MAY only be called at some distant future
* time (several millisecs away) so don't give it any arguments
* that may be revoked soon (e.g. on your stack).
* In this case even the 'so' argument is doubtful.
* While the function request is being processed the node
* has an extra reference and as such will not disappear until
* the request has at least been done, but the 'so' may not be so lucky.
* handle this by checking the validity of the node in the target function
* before dereferencing the socket pointer.
*
* To decouple stack, we use queue version of ng_send_fn().
*/
@ -1012,27 +1004,31 @@ static void
ng_ksocket_incoming(struct socket *so, void *arg, int waitflag)
{
const node_p node = arg;
int wait;
const priv_p priv = NG_NODE_PRIVATE(node);
int wait = ((waitflag & M_WAITOK) ? NG_WAITOK : 0) | NG_QUEUE;
wait = (waitflag & M_WAITOK) ? NG_WAITOK : 0;
ng_send_fn1(node, NULL, &ng_ksocket_incoming2, so, waitflag,
wait | NG_QUEUE);
/*
* Even if node is not locked, as soon as we are called, we assume
* it exist and it's private area is valid. With some care we can
* access it. Mark node that incoming event for it was sent to
* avoid unneded queue trashing.
*/
if (atomic_cmpset_int(&priv->fn_sent, 0, 1) &&
ng_send_fn1(node, NULL, &ng_ksocket_incoming2, so, 0, wait)) {
atomic_store_rel_int(&priv->fn_sent, 0);
}
}
/*
* When incoming data is appended to the socket, we get notified here.
* This is also called whenever a significant event occurs for the socket.
* We know that HOOK is NULL. Because of how we were called we know we have a
* lock on this node an are participating inthe netgraph locking.
* Our original caller may have queued this even some time ago and
* we cannot trust that he even still exists. The node however is being
* held with a reference by the queueing code, at least until we finish,
* even if it has been zapped, so first check it's validiy
* before we trust the socket (which was derived from it).
* held with a reference by the queueing code and guarantied to be valid.
*/
static void
ng_ksocket_incoming2(node_p node, hook_p hook, void *arg1, int waitflag)
ng_ksocket_incoming2(node_p node, hook_p hook, void *arg1, int arg2)
{
struct socket *so = arg1;
const priv_p priv = NG_NODE_PRIVATE(node);
@ -1043,13 +1039,11 @@ ng_ksocket_incoming2(node_p node, hook_p hook, void *arg1, int waitflag)
s = splnet();
/* Sanity check */
if (NG_NODE_NOT_VALID(node)) {
splx(s);
return;
}
/* so = priv->so; *//* XXX could have derived this like so */
KASSERT(so == priv->so, ("%s: wrong socket", __func__));
/* Allow next incoming event to be queued. */
atomic_store_rel_int(&priv->fn_sent, 0);
/* Check whether a pending connect operation has completed */
if (priv->flags & KSF_CONNECTING) {
@ -1059,7 +1053,7 @@ ng_ksocket_incoming2(node_p node, hook_p hook, void *arg1, int waitflag)
}
if (!(so->so_state & SS_ISCONNECTING)) {
NG_MKMESSAGE(response, NGM_KSOCKET_COOKIE,
NGM_KSOCKET_CONNECT, sizeof(int32_t), waitflag);
NGM_KSOCKET_CONNECT, sizeof(int32_t), M_NOWAIT);
if (response != NULL) {
response->header.flags |= NGF_RESP;
response->header.token = priv->response_token;
@ -1154,7 +1148,7 @@ ng_ksocket_incoming2(node_p node, hook_p hook, void *arg1, int waitflag)
* to indicate end-of-file.
*/
if (so->so_rcv.sb_state & SBS_CANTRCVMORE && !(priv->flags & KSF_EOFSEEN)) {
MGETHDR(m, waitflag, MT_DATA);
MGETHDR(m, M_NOWAIT, MT_DATA);
if (m != NULL) {
m->m_len = m->m_pkthdr.len = 0;
NG_SEND_DATA_ONLY(error, priv->hook, m);