Support for WAVE 300 Wi-Fi chip

ahmar16 · 2020-03-10T03:50:39.532Z

Livebox users came a few days ago. First testing was done on TP-Link TD-W8980 and possibly VR200(?).

The driver has been tested on v18.06.2 successfully but OpenWrt has changed implementation from python2 to python3 so building that version has some issues due to python differences. I am trying to build it with v18.06.8 so hopefully it will be available for testing soon. I may provide testable kernel modules and then xrx200 family devices can use it.

Mandrake-Lee · 2020-03-19T19:55:24.005Z

Dear @suleiman,
Thank you very much for your detailed answer. That has regained the interest I had in this bitchy driver (where again I'm spending so many hours).

In fact I was waiting or aiming for a major update in the PCIe system and MSI interrupts. I believe this is the major roadblock to get the driver working. In this regards, I see the work of @xdarklight very promising for the Lantiq xrx200 platform. Also the irruption of the Lantiq UGW sources has been of great interest.

But coming to the topic; I've followed your instructions and:

The devicetree needed some changes, but I'm stuck now in following message
[4294855112] mtlk3(_mtlk_mmb_wait_chi_magic:2371): Wait for CHI Magic failed (wait_res=1 value=0xbeadfeed)
The option of avoiding MSI by IRQ polling has been tried with no succes. This is the message.
[ 177.390407] [4294844596] mtlk3(_mtlk_mmb_send_hw_dep_cfg:2465): FW set HW dependent configuration failed (err=-4)

So still no luck although getting closer

Peace and love brother

Mandrake-Lee · 2020-03-19T20:07:47.858Z

Dear @ahmar16,
Thank you very much for the initiative of publishing all the scattered information into github.

I followed your instructions recently and they are very detailed and understandable.

As I stated before, I haven't been succesful on bringing up the driver but I'm afraid is something with my device Livebox 2.1 and the PCIe connection that it has. Bad luck.

I'd like to help the community; thereforeI have uploaded into the following link the firmware files that I have collected from different sources, mainly strips from images. I believe this can be shared.

Please note that some files have to be renamed and sometimes a cocktail of sources is to be taken.

Take care you all

Wave300 Firmware files -UPDATED-

EDIT: I've found more firmwares, even the latest to 3.4.2 although sta_upper.bin is missing here. As I said, the best we can do is a good cocktail.

ahmar16 · 2020-03-19T20:21:21.315Z

How were you able to successfully build the driver? I myself have not been able to build the driver with v18.06.8 with the above sources that I put together on Github. I believe there's something wrong with the makefiles or the configuration of some sort that fails the process.

Thank you for the firmware files. I'll put them on Github when I can.

Mandrake-Lee · 2020-03-20T17:09:19.443Z

Hello @ahmar16,

Yes I did build the driver but it was based on a git clone I had from @Smim0 's link; that was before summer.

So I have cloned your github and I have found the troubles you mention. Now I recall I had the same issues. Basically some files (Makefile.in and *.Po) went astray... god knows why.

The point is that I was able to put back all the files from my working copy, probably from UGW sources, and now it works like a charm. I have sent you a pull request to your github, which I'd like to be the central repository of WAVE300 driver.

Please try it and let us know.

Keep in touch!

pc2005 · 2020-03-21T10:54:41.549Z

Ok once again I'm back (but still without a time). I've read the whole chat and good work everyone ;-).

I've tried to compile a whole new openwrt snapshot and it seems wave300 doesn't support v5 kernel build system (it originates from v2.6). So openwrt testing kernel won't work. The last kernel which works is 4.19 (I think, = the non testing variant). I've tried to fix the problem and I couldn't found a way to fix that it is too obscure.

So I've hacked together (mirror) a quick working configuration of source codes from 4.19 build. There were some really weird things (logging functions being generated from perl for every directory, a single source for userspace app and kernel driver). So I've had to hack many print outs to the standard pr_info. There is a lots of unused code, no configuration support, distclean is done manually by deleting all .o and .cmd files but it worked on v5 and I've seen the test network on my laptop (and packet dumps on tp-link modem).

Also the 4.19 kernel needs some patching so I will send patches when I will found my login keys to repo.cz .

I suggest to not use the original tree as it is nearly impossible to develop on and use this hacked version to develop fully new driver from scratch (it will cost less human power).

I've checked iwlwav soruce code and it seems the code is just an newer iteration of our wave300 repos and obviously it doesn't support wave300. Actually if I remember correctly the driver lost the ABI/API support of the firmware after the transition from v3.4 to v3.5 and soon after the hardware lookup (PCI VID/DID) for wave300 was removed completely.

Wait for CHI Magic failed
This means incompatible/missing firmware. CHI magic word should appear after uploading the firmware to the device and it is hardcoded in some parameters structure after initialization. Bad location or a different word will obviously fail (most likely 3.2 fw vs 3.4 driver )

ahmar16 · 2020-03-21T15:02:15.924Z

Welcome back @pc2005. Although I have not yet tested the new sources from @pc2005, and it may take me a couple of days to test them anyway, I think this is better than trying to fix the old driver.

That being said I have tested the (old) driver with v18.06.8 and sadly it crashes the kernel. I could not really see the error because my TTL adapter was behaving weird and missing output. Now that I have seen the above post, I believe it does not matter anymore. I personally think we should be working on a driver which supports next major version rather than trying to backport something that may not even work, or something like that.

I am going to let this slide for now and not make any changes to the Github unless we can properly test the new sources and see if it works.

Smim0 · 2020-05-04T12:50:06.275Z

Hello suleiman,

Can you share the patches and also a quick steps guide on how to get this driver working for Livebox 2.1?

Regards,

Smim0 · 2020-05-09T09:47:52.137Z

Hello all,

I managed to get the driver working in a LiveBox with ip connectivity, I tried to bridge the wlan0 with the option Iwpriv wlan0 sBridgeMode 1 and also added it to lan interface settings now when the it is up I see this in the log

Sat May  9 09:45:17 2020 kern.warn kernel: [ 2368.105122] [0002068032] mtlk0(mtlk_mbss_send_vap_activate:3228): CID-0000: mtlk_txmm_msg_send, MsgId = 0x453, ESSID = test-24ghz
Sat May  9 09:45:17 2020 kern.warn kernel: [ 2368.128877] [0002068056] mtlk0(mtlk_mbss_init:3471): CID-0000: mtlk_mbss_send_vap_activate returned successfully
Sat May  9 09:45:17 2020 kern.info kernel: [ 2368.233290] br-lan: port 2(wlan0) entered blocking state
Sat May  9 09:45:17 2020 kern.info kernel: [ 2368.237220] br-lan: port 2(wlan0) entered forwarding state
Sat May  9 09:45:17 2020 daemon.notice netifd: Network device 'wlan0' link is up

If I set up a manual ip adress I can get access to the router, the iperf results are:

[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.01   sec   977 KBytes  7.96 Mbits/sec                  
[  4]   1.01-2.01   sec   816 KBytes  6.68 Mbits/sec                  
[  4]   2.01-3.00   sec   217 KBytes  1.79 Mbits/sec                  
[  4]   3.00-4.00   sec   608 KBytes  4.98 Mbits/sec                  
[  4]   4.00-5.00   sec  1.03 MBytes  8.67 Mbits/sec                  
[  4]   5.00-6.00   sec  1.03 MBytes  8.60 Mbits/sec                  
[  4]   6.00-7.00   sec  1.29 MBytes  10.8 Mbits/sec                  
[  4]   7.00-8.00   sec   569 KBytes  4.67 Mbits/sec                  
[  4]   8.00-9.00   sec  1.11 MBytes  9.32 Mbits/sec                  
[  4]   9.00-10.00  sec  1.17 MBytes  9.78 Mbits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  8.74 MBytes  7.33 Mbits/sec                  sender
[  4]   0.00-10.00  sec  8.58 MBytes  7.19 Mbits/sec                  receiver

iperf Done.

Smim0 · 2020-05-09T14:13:48.897Z

Good news,

I could bring the wlan0 in bridge mode, add it to lan interface and now it servers ip addreses via dhcp. Encryption is still missing which I suppose is due to incompatibility with hostapd. I tried to compile the modified one but I get this error message

I managed to get the right cocktail of the firmware files to get the driver work without any problems but everytime I try to unload the router reboots

../src/crypto/tls_openssl.c:19:10: fatal error: openssl/ssl.h: No such file or directory
 #include <openssl/ssl.h>
          ^~~~~~~~~~~~~~~
compilation terminated.
make: *** [Makefile:1217: ../src/crypto/tls_openssl.o] Error 1

libssl-dev is installed but it still gives me the same error message.

Cheers

lukasz92 · 2020-05-09T14:22:27.746Z

Hostapd have switch which ssl library it should use.
In config for hostapd, please change:
CONFIG_TLS=openssl
to
CONFIG_TLS=internal

and try compilation again.

Smim0 · 2020-05-09T14:28:35.414Z

Thanks, that worked now I have this error message:

../src/tls/bignum.c:17:10: fatal error: tommath.h: No such file or directory
 #include <tommath.h>
          ^~~~~~~~~~~
compilation terminated.
make: *** [Makefile:1217: ../src/tls/bignum.o] Error 1

lukasz92 · 2020-05-09T14:36:19.905Z

Maybe you should try older hostapd; or copy the one from openwrt.

Smim0 · 2020-05-09T14:42:23.429Z

I'm trying to compile the one in this thread it is a modified version to work with this specific driver do you the codes to another one? by copying the one from openwrt you mean to copy the existing hostapd with the name mtlk-ap?

lukasz92 · 2020-05-09T15:06:44.643Z

I managed to compile it about 6 years ago, but lost sources.

Basically I started from patch attached here (with driver_mtlk.c), for hostapd version 0.6.
I decided to use wpad package from OpenWRT, patch it - after compilation it worked.

Please disable flags starting from:
CONFIG_EAP*

from file hostapd/.config. Then libtommath should not be needed anymore

Mandrake-Lee · 2020-05-09T15:50:52.334Z

Hello Smim0,

Congratulations! I have a Livebox 2.1 (aka ARV7519RW22) and I never managed to get the driver up, always stuck, in the best case, with this CHI Magic.

Could you share a detailed recipe of how you did it? I mean, Openwrt version, Wave300 sources, firmware uploaded, etc?

I was about to write you that I made a fork from @ahmar16 git that should compile well:

GitHub

Mandrake-Lee/wave300-openwrt

WAVE300 Driver Development for OpenWrt. Contribute to Mandrake-Lee/wave300-openwrt development by creating an account on GitHub.

Many thanks in advance

Smim0 · 2020-05-09T16:23:35.299Z

Hello Mandrake-Lee,

I followed these instrcutions:

1 - Compiled 18.06.8 with MSI enabled and the patch and compiled the driver from pc2005 sources with pci express and polling enabled (instructions from Souleiman)
2 - To get the driver working I had to figure out the correct firmware files cocktail, I first copied all the original ones from LiveBox then started to see what the driver is looking for from the log which were ap_upper_wave300.bin and contr_lm.bin which I got from the version 3.4.2 plus a couple of ProgModel***.bin files with I had to take from the folder WAVE300__FW_UGW711.
Dont use cal_wlan0.bin file as for Livebox it is loaded from the eeprom the one in the firmware files is for the an version.
Let me know if you need any more info
Edit: the patches are to be applied to the 4.9 version not the 4.14

Smim0 · 2020-05-09T18:46:28.287Z

I checked the sources from @ahmar16, they are the same as @pc2005 I could compile the drivers succesfully

I tested Livebox with the wifi in ap mode and I could get ip address via dhcp and also got access to internet so basically we have a uncrypted wireless ap router.

Were you able to compile the modified hostapd?

Smim0 · 2020-05-13T18:01:31.984Z

Thanks @lukasz92

I will need some more guidance here, should I do any changes to the Makefile of the hostapd copied from OpenWrt? Trying to compile it with its original Makefile and the Makefile from this repo I have always an error message.

By disabling the CONFIG_EAP* flags now I get this error message:

CC  ../src/drivers/driver_common.c
/home/smim0/src/openwrt/staging_dir/toolchain-mips_24kc_gcc-7.3.0_musl/lib/gcc/mips-openwrt-linux-musl/7.3.0/../../../../mips-openwrt-linux-musl/bin/ld: cannot find -ltommath
collect2: error: ld returned 1 exit status
make: *** [Makefile:1244: hostapd] Error 1

When I copy wpad from ../openwrt/build_dir/target-mips_24kc_musl/hostapd-wpad-full-internal/hostapd-2018-05-21-62566bc2/hostapd and try to compile I get this error message

../src/drivers/driver_nl80211.c:17:10: fatal error: netlink/genl/genl.h: No such file or directory
 #include <netlink/genl/genl.h>
          ^~~~~~~~~~~~~~~~~~~~~
compilation terminated.
make: *** [Makefile:1217: ../src/drivers/driver_nl80211.o] Error 1

Thanks in advance

lukasz92 · 2020-05-13T18:37:33.979Z

First error is a missing library (file .a), I am not sure why it is still needed; possibly disabling linking option (find -ltommath flag, and delete it everywhere in Makefile, maybe it helps?)

Second error is a missing headers for libnetlink, they should be in OpenWRT SDK.

Maybe when I have some time, I will try to compile them