[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1217295498.28682.16.camel@obelisk.thedillows.org>
Date: Mon, 28 Jul 2008 21:38:18 -0400
From: David Dillow <dave@...dillows.org>
To: Jaswinder Singh <jaswinder@...radead.org>
Cc: LKML <linux-kernel@...r.kernel.org>, becker@...ld.com,
davidpmclean@...oo.com, Jeff Garzik <jeff@...zik.org>,
netdev <netdev@...r.kernel.org>,
David Woodhouse <dwmw2@...radead.org>
Subject: Re: [PATCH] typhoon: use request_firmware
On Mon, 2008-07-28 at 19:24 +0530, Jaswinder Singh wrote:
> Here is patch for typhoon.c :
It's getting better, as it's not obviously broken now.
> +MODULE_FIRMWARE("3com/typhoon.bin");
Minor nit, but perhaps #define the name, so you don't have to keep this
and the string in typhoon_init_firmware() in sync.
> @@ -307,6 +308,9 @@ struct typhoon {
> /* unused stuff (future use) */
> int capabilities;
> struct transmit_ring txHiRing;
> +
> + /* firmware */
> + u8 *tp_fw_image;
> };
Now you keep one copy per NIC, which doubles the memory usage. Just use
static struct firmware *typhoon_fw;
> +static int typhoon_init_firmware(struct typhoon *tp)
> +{
> + const struct firmware *fw;
> + const char fw_name[] = "3com/typhoon.bin";
If you use the #define and the global static, you can get rid of the
above. Just use the define in place of fw_name below:
> + err = request_firmware(&fw, fw_name, &tp->pdev->dev);
> + if (err) {
> + printk(KERN_ERR "%s: Failed to load firmware \"%s\"\n",
> + tp->name, fw_name);
> + return err;
> + }
> + tp->tp_fw_image = vmalloc(fw->size);
> + if (!tp->tp_fw_image) {
> + err = -ENOMEM;
> + printk(KERN_ERR "%s: \"%s\" Failed %d\n",
> + tp->name, fw_name, err);
> + goto out;
> + }
Then you can get rid of the double allocation and memcpy. Just put the
call to release_firmware() in typhoon_cleanup().
> @@ -2102,6 +2132,10 @@ typhoon_open(struct net_device *dev)
> if(err < 0)
> goto out_sleep;
>
> + err = typhoon_init_firmware(tp);
> + if (err)
> + goto out_irq;
> +
This should move to either typhoon_init() or typhoon_init_one(). Putting
it in typhoon_init() means we waste memory when the module is loaded if
there is no typhoon NIC; putting it in typhoon_init_one() means it needs
protection from threaded probing, should that ever be put back in. Given
that most modern distros don't do probing by blinding loading modules,
typhoon_init() seems a safe bet.
> @@ -2155,6 +2189,9 @@ typhoon_close(struct net_device *dev)
> + if (tp->tp_fw_image)
> + vfree(tp->tp_fw_image);
> +
This goes away and is replaced by a release_firmware() in
typhoon_cleanup() if you do the above.
Dave
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists