[Xenomai-core] [PATCH] small cleanups for rtcan_peak_dng

[Xenomai-core] [PATCH] small cleanups for rtcan_peak_dng

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 2433 bytes --]

Hi Wolfgang,

I came across the fact that xeno_can_peak_dng is autoloaded on 2.6 due to
its PnP announcement via MODULE_DEVICE_TABLE. xeno_irqbench and
xeno_16550A used to behave similar, but both just as well as this driver
require manual parameters anyway. So let's remove the table and thus any
autoloading tendency. I also cleaned up a few driver messages at this
chance.

OK to apply?

Jan


---
 ChangeLog                                 |    5 +++++
 ksrc/drivers/can/sja1000/rtcan_peak_dng.c |    9 ++++-----
 2 files changed, 9 insertions(+), 5 deletions(-)

Index: xenomai/ChangeLog
===================================================================
--- xenomai.orig/ChangeLog
+++ xenomai/ChangeLog
@@ -1,3 +1,8 @@
+2007-05-25  Jan Kiszka  <jan.kiszka@domain.hid>
+
+	* ksrc/drivers/can/sja1000/rtcan_peak_dng.c: Remove PnP device
+	table to avoid autoloading. Fix driver error messages.
+
 2007-05-25  Philippe Gerum  <rpm@xenomai.org>
 
 	* ksrc/nucleus/shadow.c (do_sigwake_event): Replace silent sanity
Index: xenomai/ksrc/drivers/can/sja1000/rtcan_peak_dng.c
===================================================================
--- xenomai.orig/ksrc/drivers/can/sja1000/rtcan_peak_dng.c
+++ xenomai/ksrc/drivers/can/sja1000/rtcan_peak_dng.c
@@ -224,7 +224,7 @@ int __init rtcan_peak_dng_init_one(int i
     else if (strncmp(type[idx], "epp", 3) == 0)
 	dtype = DONGLE_TYPE_EPP;
     else {
-	printk("%s: type %s is invalid, use \"sp\" or \"sp\".",
+	printk("%s: type %s is invalid, use \"sp\" or \"epp\".",
 	       RTCAN_DRV_NAME, type[idx]);
 	return -EINVAL;
     }
@@ -329,8 +329,6 @@ static const struct pnp_device_id rtcan_
     { }
 };
 
-MODULE_DEVICE_TABLE(pnp, rtcan_peak_dng_pnp_tbl);
-
 static int rtcan_peak_dng_pnp_probe(struct pnp_dev *dev,
 				    const struct pnp_device_id *id)
 {
@@ -378,7 +376,7 @@ static int __init rtcan_peak_dng_init(vo
 	 i++) {
 
 	if ((ret = rtcan_peak_dng_init_one(i)) != 0) {
-	    printk("Init failed with %d\n", ret);
+	    printk(KERN_ERR "%s: Init failed with %d\n", RTCAN_DRV_NAME, ret);
 	    goto cleanup;
 	}
 	done++;
@@ -386,7 +384,8 @@ static int __init rtcan_peak_dng_init(vo
     if (done)
 	return 0;
 
-    printk("Please specify type=epp or type=sp\n");
+    printk(KERN_ERR "%s: Please specify type=epp or type=sp\n",
+	   RTCAN_DRV_NAME);
 
 cleanup:
     rtcan_peak_dng_exit();


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]

Re: [Xenomai-core] [PATCH] small cleanups for rtcan_peak_dng

[Wolfgang Grandegger]

Hallo Jan,

Jan Kiszka wrote:
> Hi Wolfgang,
> 
> I came across the fact that xeno_can_peak_dng is autoloaded on 2.6 due to
> its PnP announcement via MODULE_DEVICE_TABLE. xeno_irqbench and
> xeno_16550A used to behave similar, but both just as well as this driver
> require manual parameters anyway. So let's remove the table and thus any
> autoloading tendency. I also cleaned up a few driver messages at this
> chance.
> 
> OK to apply?

Ja, sieht gut aus und bei dir funktioniert es ja auch.

Ein Kunde von uns evaluiert gerade CAN-Festival mit RT-Socket-CAN auf 
einem MPC5200-Board. Mal sehen, was dabei rauskommt.

Die Socket-CAN-Integration hat ja jetzt auch begonnen. Da bin ich mal 
gespannt. Sollte mal wieder die linux-netdev ML durchschauen.

Dann erstmal schöne Feiertage.

Viele Grüße,

Wolfgang.


> ---
>  ChangeLog                                 |    5 +++++
>  ksrc/drivers/can/sja1000/rtcan_peak_dng.c |    9 ++++-----
>  2 files changed, 9 insertions(+), 5 deletions(-)
> 
> Index: xenomai/ChangeLog
> ===================================================================
> --- xenomai.orig/ChangeLog
> +++ xenomai/ChangeLog
> @@ -1,3 +1,8 @@
> +2007-05-25  Jan Kiszka  <jan.kiszka@domain.hid>
> +
> +	* ksrc/drivers/can/sja1000/rtcan_peak_dng.c: Remove PnP device
> +	table to avoid autoloading. Fix driver error messages.
> +
>  2007-05-25  Philippe Gerum  <rpm@xenomai.org>
>  
>  	* ksrc/nucleus/shadow.c (do_sigwake_event): Replace silent sanity
> Index: xenomai/ksrc/drivers/can/sja1000/rtcan_peak_dng.c
> ===================================================================
> --- xenomai.orig/ksrc/drivers/can/sja1000/rtcan_peak_dng.c
> +++ xenomai/ksrc/drivers/can/sja1000/rtcan_peak_dng.c
> @@ -224,7 +224,7 @@ int __init rtcan_peak_dng_init_one(int i
>      else if (strncmp(type[idx], "epp", 3) == 0)
>  	dtype = DONGLE_TYPE_EPP;
>      else {
> -	printk("%s: type %s is invalid, use \"sp\" or \"sp\".",
> +	printk("%s: type %s is invalid, use \"sp\" or \"epp\".",
>  	       RTCAN_DRV_NAME, type[idx]);
>  	return -EINVAL;
>      }
> @@ -329,8 +329,6 @@ static const struct pnp_device_id rtcan_
>      { }
>  };
>  
> -MODULE_DEVICE_TABLE(pnp, rtcan_peak_dng_pnp_tbl);
> -
>  static int rtcan_peak_dng_pnp_probe(struct pnp_dev *dev,
>  				    const struct pnp_device_id *id)
>  {
> @@ -378,7 +376,7 @@ static int __init rtcan_peak_dng_init(vo
>  	 i++) {
>  
>  	if ((ret = rtcan_peak_dng_init_one(i)) != 0) {
> -	    printk("Init failed with %d\n", ret);
> +	    printk(KERN_ERR "%s: Init failed with %d\n", RTCAN_DRV_NAME, ret);
>  	    goto cleanup;
>  	}
>  	done++;
> @@ -386,7 +384,8 @@ static int __init rtcan_peak_dng_init(vo
>      if (done)
>  	return 0;
>  
> -    printk("Please specify type=epp or type=sp\n");
> +    printk(KERN_ERR "%s: Please specify type=epp or type=sp\n",
> +	   RTCAN_DRV_NAME);
>  
>  cleanup:
>      rtcan_peak_dng_exit();
> 

Re: [Xenomai-core] [PATCH] small cleanups for rtcan_peak_dng

[Wolfgang Grandegger]

Hello,

sorry, didn't realize the CC to the xenomai-core ML :-(. Here the 
English translation:

Wolfgang Grandegger wrote:
> Hallo Jan,
> 
> Jan Kiszka wrote:
>> Hi Wolfgang,
>>
>> I came across the fact that xeno_can_peak_dng is autoloaded on 2.6 due to
>> its PnP announcement via MODULE_DEVICE_TABLE. xeno_irqbench and
>> xeno_16550A used to behave similar, but both just as well as this driver
>> require manual parameters anyway. So let's remove the table and thus any
>> autoloading tendency. I also cleaned up a few driver messages at this
>> chance.
>>
>> OK to apply?
> 
> Ja, sieht gut aus und bei dir funktioniert es ja auch.

Yes, looks good and it's working for you, I assume.

> Ein Kunde von uns evaluiert gerade CAN-Festival mit RT-Socket-CAN auf 
> einem MPC5200-Board. Mal sehen, was dabei rauskommt.

FYI, a customer is currently evaluating CAN-Festival, a free CANopen 
implementation, with RT-Socket-CAN on a MPC5200. The rest is off-topic here.

> Die Socket-CAN-Integration hat ja jetzt auch begonnen. Da bin ich mal 
> gespannt. Sollte mal wieder die linux-netdev ML durchschauen.

FYI, the request for inclusion of Socket-CAN into the kernel has started 
on linux-netdev.

> Dann erstmal schöne Feiertage.

Have a nice weekend.

Wolfgang.