OpenWrt Forum Archive

Topic: hostapd driver for WRT54G

The content of this topic has been archived on 23 Apr 2018. There are no obvious gaps in this topic, but there may still be some posts missing at the end.

Hello,

Quick question: would it be hard to write a hostapd driver for Broadcom wireless chips used in WRT54G(S), so we could have better WPA/WPA2 (RSN) authenticator?

I haven't dig into the code much yet, only skimmed through wlioctl.h and wpa_supplicant's driver (haven't seen much obstacles so far), but maybe there are some hackers who have already tried this?

pjf wrote:

Hello,

Quick question: would it be hard to write a hostapd driver for Broadcom wireless chips used in WRT54G(S), so we could have better WPA/WPA2 (RSN) authenticator?

I haven't dig into the code much yet, only skimmed through wlioctl.h and wpa_supplicant's driver (haven't seen much obstacles so far), but maybe there are some hackers who have already tried this?

I planned on working on that, but I haven't implemented anything yet. However, i'd like to have this stuff in our Wireless Extension rather than writing a hostapd driver. This makes maintaing the thing easier.

nbd wrote:

i'd like to have this stuff in our Wireless Extension rather than writing a hostapd driver.

If I get you right, you're talking about Linux wireless extensions. AFAIR newer Linksys drivers don't support them.

pjf wrote:

If I get you right, you're talking about Linux wireless extensions. AFAIR newer Linksys drivers don't support them.

I'm talking about our own wireless extension for the Broadcom driver (called wlcompat). Check out the experimental source in CVS (openwrt in :pserver:openwrt.org:/openwrt) and have a look. It's 'wlcompat.c' in package/openwrt/

nbd wrote:

I'm talking about our own wireless extension for the Broadcom driver (called wlcompat)

Thanks, I haven't traced work on so-called 'experimental' branch (I guess it's 'development' in normal words tongue). I'll take a look at it, but are there any already known obstacles in writing the hostapd driver (even using wlcompat)?

pjf wrote:

Thanks, I haven't traced work on so-called 'experimental' branch (I guess it's 'development' in normal words tongue). I'll take a look at it, but are there any already known obstacles in writing the hostapd driver (even using wlcompat)?

The wlcompat thing works pretty well, except for some unimplemented things (like WEP). The only obstacle i see is, that (IIRC) there's no standard wireless extension interface for a WPA authenticator. The WPA interface in there seems to be for the supplicant only. So we'd probably have to choose another common driver to emulate. Should be possible, though.
You should really switch to the 'experimental' branch (which, I think, can be called stable now), unless you're using older v1 Hardware (there's a bug in newer ethernet drivers that we haven't fixed yet).

nbd wrote:

The only obstacle i see is, that (IIRC) there's no standard wireless extension interface for a WPA authenticator. The WPA interface in there seems to be for the supplicant only. So we'd probably have to choose another common driver to emulate. Should be possible, though.

Thanks for hints.

You should really switch to the 'experimental' branch

Probably I will, but in my next (this) deployment.

The discussion might have continued from here.