OpenWrt Forum Archive

Topic: WIP Day:: AR7 Targets

The content of this topic has been archived on 14 Mar 2018. There are no obvious gaps in this topic, but there may still be some posts missing at the end.

Given the upcoming WIP day, I though I'd highlight some AR7 bugs that it would be great to have out of the way and perhaps with these solved, it could heighten the probability of a stable branch being cut for AR7.

AR7 support in the OpenWrt's trunk is so so very nearly there, but as always, the predominant issue has been with wireless support. At the moment, the older acx driver supports master mode, but no WPA, whereas acx-mac80211 does support WPA, but no master mode as is quite buggy. I think, at this moment in time, it would be a fair compromise to go with the acx drivers and see how the mac80211 pans out. The majority of people want to use their AR7 device as an access point and the issue of encryption security with respect to WEP could be addressed by using a good level of encryption higher in the OSI stack. I admit, acx is not perfect and there are a few tx related issues, but these affect both acx and acx-mac80211 but some early headway is being made with some people on #acx100.

Secondly, some show stopping regression has crept in with respect to the 2.6.24 update. It's been two weeks since that check in and trunk is essentially broken and I don't see any solutions in sight for the ethernet breakage. Also, some nice cpmac bugfixes from Oliver never actually made it into 2.6.24.

I understand the need to keep pushing on with development, but there needs to be a compromise. I feel that that given a trunk that is in a constant state of brokenness, the number of people willing to test it will decrease in time. This is a downward spiral; developers need feedback from users and if users have become extinct because they have walked out the door in a cloud of apathy, the code become more unstable in time due to lack of testing. This is not something that happens overnight; it is a chronic issue.


Critical (these really need to be fixed)

https://dev.openwrt.org/ticket/3123
https://dev.openwrt.org/ticket/3124
    Broken in 2.6.24
    --> Roll back to  < r10421  or fix them in 2.6.24

https://dev.openwrt.org/ticket/3167
    Use firmware 07.02.01.00
    --> Roll back to < r10118


Would be nice to have working

https://dev.openwrt.org/ticket/2988
https://dev.openwrt.org/ticket/2996
https://dev.openwrt.org/ticket/2378


Minor

https://dev.openwrt.org/ticket/3047
https://dev.openwrt.org/ticket/2999
https://dev.openwrt.org/ticket/2332
https://dev.openwrt.org/ticket/3122


Impossible at this moment in time
Master mode support for acx-mac80211


Anyway, these are just some suggestions and it's very likely that I may have missed some. What do you all think?

nabcore wrote:

I understand the need to keep pushing on with development, but there needs to be a compromise. I feel that that given a trunk that is in a constant state of brokenness, the number of people willing to test it will decrease in time. This is a downward spiral; developers need feedback from users and if users have become extinct because they have walked out the door in a cloud of apathy, the code become more unstable in time due to lack of testing. This is not something that happens overnight; it is a chronic issue.

I agree, because I have the same feeling since some time. OpenWrt could be a very powerful system for AR7 routers,
but currently other distributions (like AVM's open source tree) are more stable (but not completely free).

Stefan

AR7 is so close, yet so far it seems.
There are many more AR7 tickets if you search under "All Tickets by Milestone", some can even be closed, such as:
https://dev.openwrt.org/ticket/2565
https://dev.openwrt.org/ticket/2439

Personally, I've been using r7038 (without wireless) in production just because I don't want to be forever playing.

I recently set my father up with a DG834Gv2, but because of these lingering tickets, I was affraid to try anything newer than what nabcore deemed stable and therefore settled on r10156 (also sans wireless).  So far so good.

For me, it's all about the ADSL capabilities.  A stable build minus ACX is fine with me....for the time being.

Wouldn't this also classify as a major?
https://dev.openwrt.org/ticket/3019

I too have problems with my DG834Gv2 hanging if I have too many bittorrent sessions running....but then could that be related to the watchdog ticket??

The discussion might have continued from here.