--Michael Merhej
On Mon, 1 Mar 1999, Andrea Arcangeli wrote:
> On Sun, 28 Feb 1999 kuznet@ms2.inr.ac.ru wrote:
>
> >Another problem was that we may have up to 2*max_files unix sockets:
> >not-yet-accepted ones are not associated with files/inodes.
>
> Ah I think to understand. It seems that upon connect af_unix will generate
> a new sock hashed in the global socket list in this line:
>
> /* create new sock for complete connection */
> newsk = unix_create1(NULL, 1);
>
> If I understand well the connect of the client is allocating the sock of
> the server side (the peer of the client). So for every file you can have
> two sock allocated.
>
> >Using list is always OK, however it also requires reviewing
>
> And obviously using a list is also a nice thing from a MM point of view ;).
>
> >af_unix.c: f.e. pending socket destruction must be blocked
> >while gc is running.
>
> Really it seems to me that while the sock is hashed in the unix-sock
> hashtable, everything is done with the big kernel lock held and without
> sleep in unix_gc. The only point that could race it seems to be the
> delayed destroy of the sock, but at that time the sock is just been
> removed from the hash table from ages. So I think the code is just safe.
> Maybe I am missing something...
>
> I did some cleanup thought. Here a patch against second Alexander's patch:
>
> Index: net/unix//garbage.c
> ===================================================================
> RCS file: /var/cvs/linux/net/unix/garbage.c,v
> retrieving revision 1.1.2.2
> diff -u -r1.1.2.2 garbage.c
> --- garbage.c 1999/02/28 20:05:45 1.1.2.2
> +++ linux/net/unix/garbage.c 1999/03/01 00:15:03
> @@ -17,7 +17,7 @@
> *
> * - explicit stack instead of recursion
> * - tail recurse on first born instead of immediate push/pop
> - * - we gather the stuff that should not be killed into tree
> + * - we gather the stuff that should not be killed into list
> * and stack is just a path from root to the current pointer.
> *
> * Future optimizations:
> @@ -53,12 +53,12 @@
> * unrelated processes.
> *
> * AV 28 Feb 1999
> - * Kill the explicit allocation of stack. Now we keep the tree
> - * with root in dummy + pointer (gc_current) to one of the nodes.
> - * Stack is represented as path from gc_current to dummy. Unmark
> - * now means "add to tree". Push == "make it a son of gc_current".
> - * Pop == "move gc_current to parent". We keep only pointers to
> - * parents (->gc_tree).
> + * Kill the explicit allocation of stack. Now we keep the marked
> + * socks in the list with root in gc_current to one of the nodes.
> + * Stack is represented as path from gc_current to the last elem.
> + * Unmark now means "add to list". Push == "make it a son of
> + * gc_current". Pop == "move gc_current to parent". We keep only
> + * pointers to parents (->gc_next).
> *
> */
>
> @@ -82,8 +82,8 @@
>
> /* Internal data structures and random procedures: */
>
> -static unix_socket dummy;
> -static unix_socket *gc_current=&dummy; /* stack of objects to mark */
> +#define GC_HEAD ((unix_socket *) -1)
> +static unix_socket *gc_current=GC_HEAD; /* stack of objects to mark */
>
> extern inline unix_socket *unix_get_socket(struct file *filp)
> {
> @@ -133,20 +133,20 @@
> extern inline unix_socket *pop_stack(void)
> {
> unix_socket *p=gc_current;
> - gc_current = p->protinfo.af_unix.gc_tree;
> + gc_current = p->protinfo.af_unix.gc_next;
> return p;
> }
>
> extern inline int empty_stack(void)
> {
> - return gc_current == &dummy;
> + return gc_current == GC_HEAD;
> }
>
> extern inline void maybe_unmark_and_push(unix_socket *x)
> {
> - if (x->protinfo.af_unix.gc_tree)
> + if (x->protinfo.af_unix.gc_next)
> return;
> - x->protinfo.af_unix.gc_tree = gc_current;
> + x->protinfo.af_unix.gc_next = gc_current;
> gc_current = x;
> }
>
> @@ -171,7 +171,7 @@
>
> forall_unix_sockets(i, s)
> {
> - s->protinfo.af_unix.gc_tree=NULL;
> + s->protinfo.af_unix.gc_next=NULL;
> }
> /*
> * Everything is now marked
> @@ -262,9 +262,9 @@
>
> if (f)
> {
> - if (!f->protinfo.af_unix.gc_tree)
> + if (!f->protinfo.af_unix.gc_next)
> {
> - f->protinfo.af_unix.gc_tree=&dummy;
> + f->protinfo.af_unix.gc_next=GC_HEAD;
> x=f;
> f=NULL;
> goto tail;
> @@ -276,7 +276,7 @@
>
> forall_unix_sockets(i, s)
> {
> - if (!s->protinfo.af_unix.gc_tree)
> + if (!s->protinfo.af_unix.gc_next)
> {
> struct sk_buff *nextsk;
> skb=skb_peek(&s->receive_queue);
> Index: include/net//sock.h
> ===================================================================
> RCS file: /var/cvs/linux/include/net/sock.h,v
> retrieving revision 1.1.2.5
> diff -u -r1.1.2.5 sock.h
> --- sock.h 1999/02/28 20:05:54 1.1.2.5
> +++ linux/include/net/sock.h 1999/02/28 22:41:16
> @@ -99,7 +99,7 @@
> struct semaphore readsem;
> struct sock * other;
> struct sock ** list;
> - struct sock * gc_tree;
> + struct sock * gc_next;
> int inflight;
> };
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/