dhcpcd-discuss

On Thu, May 24, 2018 at 10:50 AM, Roy Marples <roy@xxxxxxxxxxxx> wrote:

> On 24/05/2018 10:37, Serge Schneider wrote:
>
>> Yes , we have the hook enabled. Is that not the recommended way to start
>> wpa_supplicant when using dhcpcd?
>>
>
> It's just not the default way.
> dhcpcd should NOT make the assumption that just because the interface is
> wireless AND wpa_supplicant exists AND wpa_supplicant.conf is configured
> that IT is responsible for starting and stopping it.
>
> A good recent example is the recent introduction of IWD which competes
> with wpa_supplicant. If both are installed and configured, we don't want
> dhcpcd starting wpa_supplicant when it knows nothing about IWD - the user
> needs to make the conscious choice here.
>
> IIRC, the original reason for adding this patch was that some wifi devices
>> don't match "if_getssid(ifp) != -1", so " *ifp->name == 'w'" was added as a
>> quick workaround.
>>
>
> If the operation of getting the SSID returns -1 that means that there was
> an error. Even if not associated and not instructed to associate, the call
> will succeed. Every wireless card must support getting and setting SSID as
> a minium and only wireless cards do this.
>
> Thus it follows that any fix should be at the kernel level and not in
> dhcpcd. Unless of course it's a dhcpcd bug, but I don't recall any reported
> issues here in years.
>
> On BSD at least, most wireless devices don't start with the letter w and
> on Linux you can rename them anyway.
>

Thanks, Roy. That all makes sense. And yes it's a driver floating around on
github that has never been upstreamed, possibly due to issues like this.


> Roy
>
>