Thread (5 messages) 5 messages, 4 authors, 2007-12-01

Re: [patch 1/1] ctc: make use of alloc_netdev()

From: Peter Tiedemann <hidden>
Date: 2007-11-30 01:34:02
Also in: linux-s390

To maintain the area used for sysfs attribute data which may be stored
already previously.

Mit freundlichen Grüßen / Best regards / Saluti,
          Peter Tiedemann
---------------------------------
phone: +49-7031-16-4172   fax: ++3159    e-mail: PTiedem@de.ibm.com
IBM Deutschland Entwicklung GmbH
Linux for eServer Development, Dept. 3303
Schoenaicher Str. 220
71032 Boeblingen, Germany

IBM Deutschland Entwicklung GmbH, Vorsitzender des Aufsichtsrats: Martin
Jetter, Geschäftsführung: Herbert Kircher
Sitz der Gesellschaft: Böblingen, Registergericht: Amtsgericht Stuttgart,
HRB 243294


                                                                           
             Stephen Hemminger                                             
             <shemminger@linux                                             
             -foundation.org>                                           To 
                                       Ursula Braun1/Germany/IBM@IBMDE     
             29.11.2007 18:12                                           cc 
                                       jgarzik@pobox.com,                  
                                       netdev@vger.kernel.org,             
                                       linux-s390@vger.kernel.org, Peter   
                                       Tiedemann/Germany/IBM@IBMDE         
                                                                   Subject 
                                       Re: [patch 1/1] ctc: make use of    
                                       alloc_netdev()                      
                                                                           
                                                                           
                                                                           
                                                                           
                                                                           
                                                                           




On Thu, 29 Nov 2007 17:36:27 +0100
Ursula Braun [off-list ref] wrote:
From: Peter Tiedemann <redacted>

Currently ctc-device initialization is broken (kernel bug in
ctc_new_device).
The new network namespace code reveals a deficiency of the
ctc driver. It should make use of alloc_netdev() as described
in Documentation/networking/netdevices.txt.

Signed-off-by: Peter Tiedemann <redacted>
Signed-off-by: Ursula Braun <redacted>
---
 drivers/s390/net/ctcmain.c |   45
++++++++++++++++-----------------------------
quoted hunk ↗ jump to hunk
 1 file changed, 16 insertions(+), 29 deletions(-)

Index: linux-2.6-uschi/drivers/s390/net/ctcmain.c
===================================================================
--- linux-2.6-uschi.orig/drivers/s390/net/ctcmain.c
+++ linux-2.6-uschi/drivers/s390/net/ctcmain.c
@@ -2782,35 +2782,14 @@ ctc_probe_device(struct ccwgroup_device
 }

 /**
- * Initialize everything of the net device except the name and the
- * channel structs.
+ * Device setup function called by alloc_netdev().
+ *
+ * @param dev  Device to be setup.
  */
-static struct net_device *
-ctc_init_netdevice(struct net_device * dev, int alloc_device,
-                         struct ctc_priv *privptr)
+void ctc_init_netdevice(struct net_device * dev)
 {
-          if (!privptr)
-                      return NULL;
-
           DBF_TEXT(setup, 3, __FUNCTION__);

-          if (alloc_device) {
-                      dev = kzalloc(sizeof(struct net_device),
GFP_KERNEL);
-                      if (!dev)
-                                  return NULL;
-          }
-
-          dev->priv = privptr;
-          privptr->fsm = init_fsm("ctcdev", dev_state_names,
-                                              dev_event_names,
CTC_NR_DEV_STATES, CTC_NR_DEV_EVENTS,
-                                              dev_fsm, DEV_FSM_LEN,
GFP_KERNEL);
-          if (privptr->fsm == NULL) {
-                      if (alloc_device)
-                                  kfree(dev);
-                      return NULL;
-          }
-          fsm_newstate(privptr->fsm, DEV_STATE_STOPPED);
-          fsm_settimer(privptr->fsm, &privptr->restart_timer);
           if (dev->mtu == 0)
                       dev->mtu = CTC_BUFSIZE_DEFAULT - LL_HEADER_LENGTH
- 2;
quoted hunk ↗ jump to hunk
           dev->hard_start_xmit = ctc_tx;
@@ -2823,7 +2802,7 @@ ctc_init_netdevice(struct net_device * d
           dev->type = ARPHRD_SLIP;
           dev->tx_queue_len = 100;
           dev->flags = IFF_POINTOPOINT | IFF_NOARP;
-          return dev;
+          SET_MODULE_OWNER(dev);
 }

@@ -2879,14 +2858,22 @@ ctc_new_device(struct ccwgroup_device *c
                                   "ccw_device_set_online (cdev[1])
failed with ret = %d\n", ret);
           }

-          dev = ctc_init_netdevice(NULL, 1, privptr);
-
+          dev = alloc_netdev(0, "ctc%d", ctc_init_netdevice);
           if (!dev) {
                       ctc_pr_warn("ctc_init_netdevice failed\n");
                       goto out;
           }
+          dev->priv = privptr;
Why not use standard private data area, rather than allocating
it separately?

--
Stephen Hemminger [off-list ref]

Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help