Stealth Wimax driver drvxi314.ko?

[PJJ]

Since the original challenge is actually solved, I am starting a new thread that will concentrate on how to connect to Sprint/Virgin mobile wimax broadband using the Franklin U 600 modem.
This original thread does have some preliminary information and results that actually cover should be over here.

The thread and pertinent connection info is at:
http://www.kubuntuforums.net/showthr...r-bcm_wimax-ko

[PJJ]

If you are getting to the connecting stage, there's another log file that pops up in /tmp named CM_Auth.log. See what's in there.

My (random) guess is libengine_beceem.so is not in /lib or some dependency hasn't been met...but CM_Auth should tell you what's up.

OK, finally got a CM_Auth.log file.
Says:

========== ========== ========= ========= ==========
Authentication session started at 10/06/13 17:16:47
========== ========== ========= ========= ==========


Not good enough.

libengine_beceem.so is in /lib. But it is also in /usr/lib with different date and link to libengine_beceem.so-1.7.7.so.

The file in /lib is the more recent file.

Since this thread is getting quite long, I will start a new one that only deals with connecting via drxvi314.ko and bcm_wimax.ko to the wimax network.

[PJJ]

UPDATE

After cleaning some of the old drivers off the computer but without improving on the faulty stuff I attempted one more connection just for fun.

Found one base station without external antenna. Could improve signal strength with antenna.

Processes successfully but no connection achieved yet.
Reason:




=============== Received Message Start (2013/09/06 13:58:52.066) ===========
u32State = Waiting for network entry command
Network entry response received (Type = 2 - Subtype = 2 - Length = 52)
Network Entry Status Received
Link status = WAIT FOR NETWORK ENTRY CMD
EAP supplicant is OFF <-------------------------
=============== Received Message End (type=2 sub-type=2) ===============
***** Message sent successfully
Now it is off when before it was a program error.

Working on it tomorrow.

[PJJ]

Problem is: The CM_Auth file goes away after I turn off the computer. I have not yet seen it other than in Windows.
The CM_Server log also disappears unless I save it right away.

Looking more carefully at the error message I am inclined to believe that libengine_beceem.so, libxvi020.so.05.02.0628050017 and libeap_supplicant.so are not of the correct variety. I extracted them a long time ago for another incarnation of this linux kernel.

According to Sprint_Dev_Pack this could be the problem.
Says Sprint:

NOTE: It is very important that the firmware image, kernel module and Broadcom
(Beceem) library are each from the same release/revision. For example, do not
mix the device firmware from one library release with the kernel module and
library from another release.

I have done so many compilations and extractions that I have a plethora of files with identical names lying all over the place despite cleaning up once in a while.

I will erase the implicated files and extract them for my current kernel (3.8.0-31-generic) to be sure.

Fund three towers and can now reliably 'connect'. I did this morning. Same error message.

I will try again later to look for the auth file.

Check pastebin if you can not go to the forum. I throw a lot of stuff up there.
By the way, bcm_wimax will most likely work for me as well.
And it gives a new 'wired' connection named eth1.

Once the EAP problem is solved I will also find the correct values for the 802.1 authorization screen. Network Manager should then connect after 4G modem is turned on.

[paperplate9]

If you are getting to the connecting stage, there's another log file that pops up in /tmp named CM_Auth.log. See what's in there.

My (random) guess is libengine_beceem.so is not in /lib or some dependency hasn't been met...but CM_Auth should tell you what's up.

[PJJ]

Result of latest attempt using bcm-wimax as driver:

Starting wimax normally by switching to 'auto' mode via picocom.
Loading bcm-wimax driver with insmod.
Red power light comes on.

Thereafter there is the Beceem modem (4G) and the eth1 interface.
wimaxd in combination with wimaxc -i activates the modem.

This time I was close to several towers.
Found a number of base stations.
Connecting attempted.
Did not succeed because of:

"EAP supplicant is not open (Program ERROR)".

Question now becomes how to resolve this error. I will make a connection with Windows XP today, study logs to get a better idea what is missing in my linux EAP supplicant setup.

If anyone has a suggestion, please let me know.

Truncated data output is posted at:

http://pastebin.com/GJyWag1S

[PJJ]

While working on my Kubuntu 13.10 installation I discovered that the drxvi314.ko driver is no longer working with the latest kernel update. 3.11.0-11-generic.
My first effort to recompile the module to the driver completed but gave a non-working driver.
Just for the fun of it I therefore went back to Kubuntu 13.04 and used the native linux driver bcm_wimax that is found in the staging directory.

Contrary to previous attempts this time the driver actually did work.
It produced results identical to those received when using drxvi314.ko. And again I could not connect because I do not have coverage where I am even when I use an external antenna in an area Sprint labels as 'best'.

I will post my raw data received during the test on pastebin. The data is unedited and needs your interpretation. But it clearly shows that bcm_wimax was working, scanning the spectrum and attempting to make a connection.
That is good news for me because it would be much simpler if the native driver would actually do the job.

Pastebin location:

http://pastebin.com/gQKjNrt4


paperplate9 this post might interest you!

[PJJ]

I have not been able to connect successfully yet because for some reason even with my external antenna I get the message "0 base stations found" - even in an area the Sprint coverage map shows as "best".

Made a few minor changes to wimaxd.conf (mainly the format of the center frequencies), since I noticed that the Windows init file does not use Mhz. Should not make a difference though.

General update:

Updated the kernel to the latest 3.8.0-31-generic.
Driver module stopped working. Wrong driver format.
It worked up to 3.8.0-30-generic.

Recompiled the module (using your source) to 3.8.0-31-generic.
Module working again.

I post much of my efforts and observations on

http://pastebin.com/u/pjj90292.

That might be an easier place for you to check. Otherwise contact me, please, and I will give you a contact e-mail or so.

[paperplate9]

Glad you got it, please keep us updated if you were able to successfully get online with the U600 & 4G/WiMax. I don't browse the forum too often so please link to your progress if you're keeping it somewhere else...thanks!

[PJJ]

I am closing this thread and will mark it as 'solved' because I finally installed the wimax driver module drxvi314.ko successfully. It is working now. Also solved is the problem with the error 'listen socket bind failed.

My final conclusions:
Contrary to statements elsewhere the driver module drxvi314.ko compiles properly with modern Linux kernels. I am using it with 3.8.0-30-generic. Others use it with Kubuntu 12.04.
The linux native driver bcm_wimax did not work for me, though other users found it to work (with the Sprint Sierra modem) and possibly Franklin U600.

Starting the modem is a cumbersome process since the driver has to be loaded whenever the computer is booted up. It is also essential that the driver activation process be done following a certain sequence and as root throughout. And that includes switching on the 4G modem via picocom/minicom and the command described by paperplate9 in his thread. It is also essential to release the port from picocom/minicom after the switch or an error message 'device NOT present' will come up later.

My next step will be to make an actual connection and to go through the authentication process. I am expecting another round of errors there unless my file wimaxd.conf happens the have correct data in each and every field. Fat chance!

Once a successful connection can be made reliably, I need to either find a way to prepare a script that loads the driver on boot up or install the official Sprint CSCM connection manager. Minhazul Haque has published his version of a connection manager under the title "GUI For BCS Mobile WiMAX On Linux " that can take the place of the official CSCM manager.
I am reluctant to use it for my installation because it is designed for an older kernel and modifies and installs files automatically. It might disturb my now working installation.

In addition, the network manager that comes with the standard linux kernels has a tab for 'Wired' connections.
During my efforts NM installed at several occasions a new 'wired' connection eth1 that was actually my 4G modem in actions - sort of. This tab 'wired' should allow us to setup a 4G wimax connection provided there is a working wimax modem and the correct connection data can be entered. I believe I now know the data for IPv4 and IPv6 and the ethernet parameters. I am not sure about the all essential 802.1 security information to enter.

We will tackle this after a good connection can be established reliably and I have collected or received from someone more of the missing entries in the security tab.

For whatever it is worth to anyone, I have published my most recent successful driver modules installation data and connection attempt at

http://pastebin.com/iHJpJZ9x

True, it is sort of repetitive because we go over the same data categories all over again, but this time it looks like everything was like it is supposed to be. Some of the data is abbreviated or truncated. Contact me if you want more detailed info.

[PJJ]

[/QUOTE]. . .
I have opened another thread specifically for the listening socket bind failed error under the title"FATAl error:Listen socket bind failed. . .".
It is at http://www.kubuntuforums.net/showthr...-How-to-fix-it.

Detailed info regarding this problem is at the above pastebin location.[/QUOTE]

Today, 09/11/13, I will post another pastebin with the results of my latest efforts.

I changed the inet address for eth1.

That solved the socket bind failure problem - at least temporarily.

Data will be posted at pastebin.

wimaxd and wimaxc appeared to be working.
Search function in wimaxc active - but no base station found.

I know, because I am unable to receive wimax signal here inside. Need external antenna.

[PJJ]

Here is a drxvi314.ko dmesg output from today:

drxvi314.ko installed and loaded 09/04/13:


dmesg

sudo dmesg -c|tail -60
[sudo] password for peter:
[ 5085.806717]
[ 5085.806720] beceem_parse_target_struct:PMU MODE: 0
[ 5085.806724] beceem_parse_target_struct:bDisableFastFlashWrite: 0
[ 5085.806728] beceem_parse_target_struct:uiEEPROMFlag : 0x2
[ 5085.806732]
[ 5085.806735] beceem_parse_target_struct:Power Save Mode: 0
[ 5085.806738]
[ 5085.806742] doPowerAutoCorrection:PMU selected ....
[ 5085.806746] beceem_parse_target_struct:LinkUp Config : 0x3
[ 5085.806749]
[ 5085.807165] updateWriteProtectedRegister:Value read from Reg: 0xf000c00 is 0x1f4062
[ 5085.807165]
[ 5085.807174] updateWriteProtectedRegister:Register 0xf000c00 restored successfully, with value 0x1f4062 !!
[ 5085.807179]
[ 5085.807183] updateWriteProtectedRegister:Tried to restore Reg: 0xf000c00, 0 times
[ 5085.807183]
[ 5085.807188] Bcm_kill_all_URBs:Cancelling All Submitted TX Urbs
[ 5085.807191]
[ 5085.807204] Bcm_kill_all_URBs:Cancelling All submitted Rx Urbs
[ 5085.807207]
[ 5085.807219]
[ 5085.807219] <<<<No of tries in the tx side:0>>>
[ 5085.807224]
[ 5085.807224] <<<<No of tries in the Rx side:0>>>
[ 5085.807230] Bcm_kill_all_URBs:TCB: used- 0 cur-0
[ 5085.807233]
[ 5085.807236] Bcm_kill_all_URBs:RCB: used- 0 cur-0
[ 5085.807239]
[ 5085.807242] reset_card_proc:Reseting UMA-B
[ 5085.807245]
[ 5085.807251] ====================>
[ 5085.880200] usb 1-3.1: reset high-speed USB device number 9 using ehci-pci
[ 5085.972675] Do Post chip reset setting here if it is requiredddr_init:Register Count is =48
[ 5085.993776]
[ 5086.008143] InitCardAndDownloadFirmware:CFG file downloaded
[ 5086.013890]
[ 5086.013890] /tmp/kde-peter/ark4hkewN/drxvi314-source/Common/nvm.c:ReadMacAddressFromNVM:441:Buffer dump of size 0x6 in the HEX:
[ 5086.013899] F4 63 49 07 27 9D
[ 5086.013910] register_networkdev:Registering netdevice notifier
[ 5086.013913]
[ 5086.013924] register_networkdev:BCM Notifier got Registered
[ 5086.030264] bcm_notify_event:Register RefCount: 7
[ 5086.030272]
[ 5086.032703] register_networkdev:Beceem Network device name is eth1!
[ 5086.033243] bcm_open:======>
[ 5086.098624] ReadLEDInformationFromEEPROM:GPIO's bit map correspond to LED :0xA000
[ 5086.112616] ReadLEDInformationFromEEPROM:SPIO's bit map correspond to LED :0x0
[ 5086.116670] open_firmware_file:Got file descriptor pointer of /lib/firmware/macxvi350.bin!
[ 5086.116682] BcmFileDownload:Opened file is = /lib/firmware/macxvi350.bin and length =0x2084a0 to be downloaded at =0xbfc00000
[ 5086.116687] BcmFileDownload:download start ebe16e39
[ 5086.449070] InterfaceFileDownload:Got end of file!
[ 5086.713400] InterfaceFileReadbackFromChip:Got end of file!
[ 5086.713423] BcmFileDownload:file download done at ebe1708d
[ 5086.713429] InitCardAndDownloadFirmware:BIN file downloaded
[ 5086.723565] device_run:Sending first interrupt URB down......
[ 5086.724432] device_run:Got the mailbox interrupt ...Registering control interface...
[ 5086.724432]
[ 5086.730653] register_control_device_interface:Got Major No: 249
[ 5086.730796] usbbcm_device_probe:Enabling USB Auto-Suspend
[ 5086.730802]

Since this appeared incomplete, I retraced my steps to do another attempt. Got driver loaded with 'insmod'. modprobe did again not find the driver.
4G mode set to 2 (auto).

dmesg (09/07/13) and related information is posted at

http://pastebin.com/YLT01Yjx.

Some expert interpretation is needed to get past this impasse.

I have opened another thread specifically for the listening socket bind failed error under the title"FATAl error:Listen socket bind failed. . .".
It is at http://www.kubuntuforums.net/showthr...-How-to-fix-it.

Detailed info regarding this problem is at the above pastebin location.

[PJJ]

Preliminary reply:

netstat -antp|grep :5200

Yes. reverts back to command line.

ps -ef|grep wimaxd
peter 9437 2563 0 12:04 pts/0 00:00:00 grep --color=auto wimaxd

It could have been 3G but the green light did never flicker during the download. And I did not ask for the download of macxvi200.bin.


dmseg output after first installation of your driver module looked very much like your dmesg code. I saved it but now cannot find it any longer. Lost an entire file with info, I guess.

bcm_wimax is loading despite the fact that it is blacklisted.
I have to remove it manually before loading drxzvi314.ko

More later. Need to go back and get only drxvi314 installed.

[paperplate9]

ppp0 is probably the 3G modem. Kill any auto-connection stuff, just making things harder.

As I said before, ifconfig does not matter until you are connected to a base station. No base station connection is like trying to configure a NIC without a cable plugged in. Plus you will be using dhcp anyway.

And you still have the bind socket error. Did you ever run netstat -antp|grep :5200? Do you have only one instance of wimaxd running (ps -ef|grep wimaxd). Are you running wimaxd as root? macvi200.bin is a file that should exist in /lib/firmware that the driver downloads to the device. Your dmesg output looks like bcm_wimax and not drxvi314.

Keep it simple...your goal is to connect to a base station, so forget all about IP addresses and downloads & such...

[PJJ]

Here is the missing dmesg for the drxvi314.ko compilation according to the patched data from paperplate9:

http://pastebin.com/k9L2NGQZ