Hi,
hotplug_path is no longer exported, is this on purpose ? It breaks
linux-wlan-ng. If it is on purpose, I suppose linux-wlan-ng
should use kobject_hotplug() ? If not, here's a patch.
Signed-off-by: Colin Leroy <[email protected]>
--- lib/kobject_uevent.c.orig 2004-11-17 20:31:02.258535288 +0100
+++ lib/kobject_uevent.c 2004-11-17 20:28:12.341366624 +0100
@@ -312,6 +312,7 @@
return;
}
EXPORT_SYMBOL(kobject_hotplug);
+EXPORT_SYMBOL(hotplug_path);
/**
* add_hotplug_env_var - helper for creating hotplug environment variables
On Wed, Nov 17, 2004 at 08:31:39PM +0100, Colin Leroy wrote:
> Hi,
>
> hotplug_path is no longer exported, is this on purpose ?
Yes.
> It breaks linux-wlan-ng. If it is on purpose, I suppose linux-wlan-ng
> should use kobject_hotplug() ?
Yes it should. Why was it not useing that function in the first place?
> If not, here's a patch.
No, please use kobject_hotplug(). Actually, what are they doing that
they need to call kobject_hotplug() in the first place?
thanks,
greg k-h
On Wed, Nov 17, 2004 at 11:12:53PM +0100, Colin Leroy wrote:
>
> Dunno. This driver has a reputation of being the worse wlan driver for
> prism2 chipsets out there, but it's the only one supporting USB devices.
Hm, I've heard that rumor before too. If only someone would get me such
a usb device, maybe that problem could be fixed :)
> With p80211_run_sbin_hotplug doing stuff to call /sbin/hotplug...
> The one that will write a correct patch to linux-wlan-ng will have to
> figure out the different events they use: "register" (instead of "add"),
> "startup", "shutdown", "resume", "suspend"...
Ick, ick, ick. It's about time we force them to play nice with the rest
of the kernel and userspace :)
Anyway, the proper fix is to use kobject_hotplug(). They will have to
change their code. Just one of the many joys of trying to keep a driver
outside of the main kernel tree...
thanks,
greg k-h
On 17 Nov 2004 at 13h11, Greg KH wrote:
Hi,
> > hotplug_path is no longer exported, is this on purpose ?
>
> Yes.
>
> > It breaks linux-wlan-ng. If it is on purpose, I suppose
> > linux-wlan-ng should use kobject_hotplug() ?
>
> Yes it should. Why was it not useing that function in the first
> place?
Dunno. This driver has a reputation of being the worse wlan driver for
prism2 chipsets out there, but it's the only one supporting USB devices.
> No, please use kobject_hotplug(). Actually, what are they doing that
> they need to call kobject_hotplug() in the first place?
To propagate events to scripts, I think:
void p80211_suspend(wlandevice_t *wlandev)
{
p80211_run_sbin_hotplug(wlandev, WLAN_HOTPLUG_SUSPEND);
}
int register_wlandev(wlandevice_t *wlandev)
{
netdevice_t *dev = wlandev->netdev;
... if (register_netdev(dev)) {
return -EIO;
}
...
p80211_run_sbin_hotplug(wlandev, WLAN_HOTPLUG_REGISTER);
}
With p80211_run_sbin_hotplug doing stuff to call /sbin/hotplug...
The one that will write a correct patch to linux-wlan-ng will have to
figure out the different events they use: "register" (instead of "add"),
"startup", "shutdown", "resume", "suspend"...
--
Colin
http://www.colino.net/
On 17 Nov 2004 at 14h11, Greg KH wrote:
Hi,
> > Dunno. This driver has a reputation of being the worse wlan driver for
> > prism2 chipsets out there, but it's the only one supporting USB devices.
>
> Hm, I've heard that rumor before too. If only someone would get me such
> a usb device, maybe that problem could be fixed :)
What's your address ? :)
> Anyway, the proper fix is to use kobject_hotplug(). They will have to
> change their code. Just one of the many joys of trying to keep a driver
> outside of the main kernel tree...
I also just noticed they do their hotplug stuff in p80211.c, which knows
nothing of USB. What a hack :-/
--
Colin
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Greg KH wrote:
| On Wed, Nov 17, 2004 at 11:12:53PM +0100, Colin Leroy wrote:
|
|>Dunno. This driver has a reputation of being the worse wlan driver for
|>prism2 chipsets out there, but it's the only one supporting USB devices.
|
|
| Hm, I've heard that rumor before too. If only someone would get me such
| a usb device, maybe that problem could be fixed :)
Actually, orinoco_usb in current Orinoco CVS has support for some of
these devices and I'm using one right now. The driver works pretty well
but I'm not sure whether orinoco_usb supports many prism2_usb devices
the same way orinoco_cs supports many prism2_cs devices.
So, if I sent you my wireless device it wouldn't be much help :)
- --
Joshua Kwan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org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=aGiw
-----END PGP SIGNATURE-----
On 18 Nov 2004 at 02h11, Joshua Kwan wrote:
Hi,
> Actually, orinoco_usb in current Orinoco CVS has support for some of
> these devices and I'm using one right now. The driver works pretty well
> but I'm not sure whether orinoco_usb supports many prism2_usb devices
> the same way orinoco_cs supports many prism2_cs devices.
No, doesn't work with prism2 usb devices. Half the functions are dummy
functions returning -EOPNOTSUPP.
I wish I was employed to do linux development so I could spend the
necessary time to do such things...
--
Colin