[PATCH 2/3] 2.6.14-rc2-mm1: fixes for overflow in epoll()

[Willy Tarreau <willy@w.ods.org>]

This patch fixes the potential timeout overflow in ep_poll(), tries to get
the best timeout and also removes the /1000 divide whenever possible. It
mostly ensures that timeout computation is uniform with other places in the
kernel.

This patch applies both to 2.6.14-rc2-mm1 and 2.6.14-rc2.

Please review and apply.

Signed-off-by: Willy Tarreau <willy@w.ods.org>

- Willy

diff -purN linux-2.6.14-rc2-mm1/fs/eventpoll.c linux-2.6.14-rc2-mm1-epoll/fs/eventpoll.c
--- linux-2.6.14-rc2-mm1/fs/eventpoll.c	Sat Sep 24 21:11:49 2005
+++ linux-2.6.14-rc2-mm1-epoll/fs/eventpoll.c	Sat Sep 24 21:20:10 2005
@@ -1502,12 +1502,11 @@ static int ep_poll(struct eventpoll *ep,
 	wait_queue_t wait;
 
 	/*
-	 * Calculate the timeout by checking for the "infinite" value ( -1 )
-	 * and the overflow condition. The passed timeout is in milliseconds,
-	 * that why (t * HZ) / 1000.
+	 * Calculate the timeout by checking for the "infinite" value ( < 0 )
+	 * and the overflow condition. The passed timeout is in milliseconds.
 	 */
-	jtimeout = timeout == -1 || timeout > (MAX_SCHEDULE_TIMEOUT - 1000) / HZ ?
-		MAX_SCHEDULE_TIMEOUT: (timeout * HZ + 999) / 1000;
+	jtimeout = (timeout < 0) ?
+	    MAX_SCHEDULE_TIMEOUT : msecs_to_jiffies(timeout);
 
 retry:
 	write_lock_irqsave(&ep->lock, flags);


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/