netdev
[Top] [All Lists]

Re: Fw: [Kernel-janitors] net/atm/clip.c: check kmem_cache_create() #1

To: Francois Romieu <romieu@xxxxxxxxxxxxx>
Subject: Re: Fw: [Kernel-janitors] net/atm/clip.c: check kmem_cache_create() #1
From: chas williams (contractor) <chas@xxxxxxxxxxxxxxxx>
Date: Thu, 05 Feb 2004 07:22:40 -0500
Cc: davem@xxxxxxxxxx, "Randy.Dunlap" <rddunlap@xxxxxxxx>, netdev@xxxxxxxxxxx
In-reply-to: Message from Francois Romieu <romieu@xxxxxxxxxxxxx> of "Wed, 04 Feb 2004 22:47:40 +0100." <20040204224740.B19321@xxxxxxxxxxxxxxxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
In message <20040204224740.B19321@xxxxxxxxxxxxxxxxxxxxxxxxxx>,Francois Romieu w
rites:
>One should probably apply the following patch on top of it btw.
>Unbalanced call to create_proc_entry() on error path.

how about the following instead?  we probably shouldnt register the
proc entry until clip_tbl is going to be ready for use anyway (the
arp table iterators should probably also use clip_tbl instead of
clip_tbl_hook).

===== net/atm/clip.c 1.30 vs edited =====
--- 1.30/net/atm/clip.c Wed Feb  4 13:07:57 2004
+++ edited/net/atm/clip.c       Thu Feb  5 07:18:03 2004
@@ -1008,14 +1008,6 @@
 
 static int __init atm_clip_init(void)
 {
-#ifdef CONFIG_PROC_FS
-       struct proc_dir_entry *p;
-
-       p = create_proc_entry("arp", S_IRUGO, atm_proc_root);
-       if (p)
-               p->proc_fops = &arp_seq_fops;
-#endif
-
        /* we should use neigh_table_init() */
        clip_tbl.lock = RW_LOCK_UNLOCKED;
        clip_tbl.kmem_cachep = kmem_cache_create(clip_tbl.id,
@@ -1032,6 +1024,16 @@
 
        clip_tbl_hook = &clip_tbl;
        register_atm_ioctl(&clip_ioctl_ops);
+
+#ifdef CONFIG_PROC_FS
+{
+       struct proc_dir_entry *p;
+
+       p = create_proc_entry("arp", S_IRUGO, atm_proc_root);
+       if (p)
+               p->proc_fops = &arp_seq_fops;
+}
+#endif
 
        return 0;
 }

<Prev in Thread] Current Thread [Next in Thread>