[hokuto]

[2KloN]

[hokuto]

I eventually got it working by waiting for around 15 minutes before any reboot. As I didn't have any serial cables I don't know it is superstition or not.

To show nvram usage, login to ssh and do nvram show. In Tomato the nvram usage is shown in the web admin interface too.

[2KloN]

Hello. in general, I had time on the weekend and I did vsetaki 30-30-30 hard reset on your router.
1. Disconnect all of the router.
2. to reset to factory settings standard firmwares
3. Reboot and then filled again, the latest version of the standard firmware.
4. then made 30-30-30 hard reset.
5. rotuer loaded in standart firmware where I immediately flash the linksys-ea6700-webflash_r28598.
6. After restarting the router booted in dd-ART I had configured and set up in the status of NVRAM tape -NVRAM USEGE 32/64.
7. Now the question whether it is possible through it to download the latest version of the linksys-ea6700-webflash_r31277?
8. in general have so far worked.

[hokuto]

[ChaoticHavok]

Just did these steps last night. I just had to do extensive research to go through these steps myself.
I would like more data points on these steps since I kind of hodgepodge a bunch of different guides. I went to DD-WRT because my 2.4GHz keeps disappearing on AdvancedTomato after settings change. Please reply back with your data points.

1. Try to revert back to previous firmware "FW_EA6700_1.1.40.166281_prod.img". You can find that in Troubleshooting>Diagnostics.

2. Flash the firmware upgrade of the same v.166281 from here.

3. Flash the firmware AdvancedTomato and erase nvram.

4. If you want DD-WRT, you can then flash DD-WRT from here and erase nvram by hard reset.

Remember to clear nvram either by hard reset or by settings each time you flash a new firmware

[ST Dog]

[makaisson]

Got EA6700 router day before yesterday. What worked for me:
-Downgrade to 1.1.40.166281 via tftp2.exe
-Factory reset via web interface
-Flash 1.1.40.166281 via web interface
-Factory reset via web interface
-Flash dd-wrt r23598 beta webflash bin
-Clear nvram
-Flash tomato 140 AIO (I like tomato by shibby)
-Clear nvram
-Configure as required and works as expected

Hope this helps someone, as for whatever reason I could not flash tomato directly from stock.

Now can safely mess around with ISP provided EA6350v2 without fear of loosing internet access.

[batchcrazy]

[u2n]

Linksys EA6700 (EA6500v2) with DD-WRT *Totally Normal*, No Quirks

Well ok, if it were totally normal, these fixes wouldn't be needed. The point is that this hardware can be made to perform normally running DD-WRT.

First is the fix for the dead USB port, followed by a normal CFE (no port reversals and, of course, no 32k-nvram bug). The third post covers differences in the CFEs available for this unit.


USB Port Fix
The Internet is filled with wrong information on this. The port does auto-disconnect after boot, but comes back to life when reinitialized. This is done automagically with the startup script below.


===EDIT/UPDATE===
Frequent disconnects were seen in the latest two builds (checked) not seen in earlier ones (5/17). iMagic's posts on this page and next refer to this strangeness. The partitions show as remaining mounted, but contents are inaccessible. The port can can be reinitialized manually with the (original, no gpio) commands below, but will soon revert to this pseudo-connected state.

So gave another try finding the root cause of the initial disconnect and, while failing miserably yet again (for the *last time* I might add!), found a cure for the pseudo-disconnection: the gpio command added to the code below. This should work for all builds.

Now, let no one say this is an elegant fix. It is a complete hack that should probably have no effect. As Kong mentions here, "nvram var for buttons are never used by us." The reset button itself does nothing after its function is moved to the wps button with the new CFE. But sending in an enable does work; partitions remain mounted and accessible. It's just unknown how or why. (And enabling the usb port with gpio9 had no effect.)

BTW, should clarify that sleep 5 is there to allow the kernel to connect, then disconnect the port. (The sleep changes to 60 seconds when added to the startup script to allow boot time.) If the port is reinitialized before disconnection, then it will remain disconnected.
===[/EDIT/UPDATE]===


If using one of the patched CFEs (of XVortex origin), then:

1. Assuming an ext4 filesystem on the flash drive to be mounted, with the first partition on /opt and the second on /mmc -- for other filesystems and mountpoints, modify args as desired. (/mnt had some odd behavior in earlier releases so I avoid it.)
   
   
2. In the web GUI (Services | USB), be sure both Core USB Support and USB Storage Support are enabled, and Automatic Drive Mount is disabled. If changes are made, be sure to hit 'Apply Settings'.
   
   
3. Plug the flash drive into the USB port that dies, the usb3 port which is labeled 'USB 1' and located to the outside, with blue tongue.
   
   
4. Reboot.
   
   
5. SSH or telnet into the router and paste the following in a terminal:
   

   Code:

   for d in mbcache jbd2 crc16 ext4; do insmod $d; done sleep 5 gpio enable 11 IFS=" " xc=" echo 0000:00:0b.0 > /sys/bus/pci/drivers/xhci_hcd/unbind echo 0000:00:0b.0 > /sys/bus/pci/drivers/xhci_hcd/bind mount -t ext4 -o rw,noatime,nodiratime,errors=remount-ro,barrier=1,data=ordered /dev/sda1 /opt mount -t ext4 -o rw,noatime,nodiratime,errors=remount-ro,barrier=1,data=ordered /dev/sda2 /mmc " for x in $xc; do eval $x; sleep 2; done

   
   
6. See that access to mounted partitions on the flash drive now works normally, Ex. # ls -l /opt.
   
   
7. Tweak the mount options as preferred. The above are for longest life of a flash drive; any suggestions for improvement are welcome.
   
   
8. If desired (optional), do away with the loop by pasting the xc= commands directly with a sleep 2 after each. (The above code is more compact.)
   
   
9. When satisfied that the port reinitializes and partitions mount normally, change the 'sleep 5' to 'sleep 60' and add the code to the startup script (Administration | Commands in the GUI); paste and Save Startup. (Or just paste to begin with and reboot!)
   

• There is a slight delay in mounting on reboot, but since these devices run for months w/o restarting, that's no trouble.
  
  
• For reference, I'm currently running Kong builds.
  
  
• Anyone use this method and have trouble, please post.
  
  
• Apologies it took over a year to get time to write these up (work done fall 2016). With open source, patience wins.

[u2n]

The CFE

See attached archive (must be logged in) containing:

• A copy of XVortex's CFE Edit tool from his post, cfe_edit.exe
  
• Text file with data ready to paste (into the XVortex (XV) CFE tool), cfe_u2n_12-17.txt
  
• Text file with a line-by-line listing of the original, XV, justmoi and u2n (this) versions, cfe_comparison.csv*
  
• CFE binary made using my edits, cfe_u2n_12-17.bin

*This is a tab-delimited csv that can be pasted into a spreadsheet or viewed in a text editor, which makes it easy to compare versions. Data for the original CFE is taken from Malachi's post here. See next post for discussion.


The boot partition (which is flashed by a CFE binary file) contains the MAC addresses and secret_code, which should match the device's label. It is a simple task to look at the label and edit the four fields to match. Do not flash the included binary! It is for reference/comparison only. Even hex editing is not recommended. There are reports of trouble after doing this, possibly due to a checksum mismatch.


To generate the hardware-specific CFE image:

1. Run Windows (sorry, no Linux version of CFE Edit tool is available)
   
2. Extract the above files
   
3. Open cfe_u2n_12-17.txt file, select all and copy**
   
4. Click to run cfe_edit.exe
   
5. Click Advanced tab and see page is blank
   
6. Paste, then check to see that text exactly matches text file contents
   
7. Edit four fields as follows:***
   • et0macaddr= <base mac address from label on router>
     
   • 0:macaddr= <et0macaddr+2h>
     
   • 1:macaddr= <et0macaddr+4h>
     
   • secret_code= <secret code from label>
8. Click 'Save as' and give an easy name, 6700cfe.bin
   

**Be sure to use a text editor that doesn't auto-add formatting characters. Geany is recommended.

***Some guides increment the 0: and 1: MACs by 1, rather than by the 2 used here. Since VLAN2 (the WAN) is automatically given a MAC of base_MAC+1 on bootup, I prefer this (2-increment) scheme. (Plus it works same as original.)


Flash the CFE binary file just created, 6700cfe.bin, to the router, which must be running the early version of DD-WRT (<-click to download). Guides: By cybrnook, and By egc. Use all, including this one, at user's own risk (as usual ). With tip 2 below, this can be as easy as XV states in steps 3-5 in this post.


Tips

1. To get into the mini web server, if preferred, set static IP address.
   
   
2. Rather than fiddle with putty or whatever -- even scp, which I like -- it is easiest to just telnet in [ telnet 192.168.1.1 ] and issue # mount /tmp /mnt, then go to Services | NAS in the GUI and Add Share /mnt (name 'mnt'), ticking the Public checkbox and hit 'Apply Settings'. Navigate to mnt from PC's file browser (via Network) and copy the file(s) to be flashed or saved. Quick and easy.
   
   
3. Take care, especially with the CFE. An inadvertent error can render the device in need of serial recovery.
   
   
4. Don't freak if it goes into a boot loop. Reset by pinhole if before CFE flash, or by WPS button after.
   
   
5. Allow at least five minutes after any reset before doing anything, even if the unit appears stable. It's not.
   
   
6. Side note: My router rests vertically, which increases air flow and cuts down on the core temperatures. It normally runs ~60C CPU and ~45C wlans.
   
   
7. After some absence, I've returned to Optware, Optware-ng (new generation). It's great to be able to quickly install and use different packages, as it was with MIPS hardware for so many years. (My executables go in /opt/usr/bin, path added)
   • wget -O - http://ipkg.nslu2-linux.org/optware-ng/bootstrap/buildroot-armeabi-ng-bootstrap.sh | sh
     
   • export PATH=$PATH:/opt/bin:/opt/sbin:/opt/usr/bin
     
   • /opt/bin/ipkg update
     
   • /opt/bin/ipkg list

Lacking documentation, I used the XV tool to make and flash several CFEs to get this right. Or at least, best with the available time.

Using this CFE and the usb3 port fix in previous post, my router has been functioning like any other -- no more buggy behavior or backwards connections. Thanks to everyone mentioned and linked here, BrainSlayer and Kong always, and the community.

[u2n]

With reference to the file cfe_comparison.csv in the previous post:

Notes on Comparison of CFEs

1. The original contained two variables that none of the others had: sdram_ncdl=0 (right after sdram_config=0x0147, which is common to all) and gpio8=wps_led (right after lan_netmask=255.255.255.0, also common to all). Did not experiment with either of these, and neither was included.
   
   
2. The original clkfreq=800,533 was retained, though I didn't see any big difference using the EA6900's clkfreq=800,666. If not mistaken, I believe XVortex even overclocked to 1000. This thing runs hot anyway so why push it when there's no need? (Or is there always a "need for speed"?)
   
   
3. Using boardtype=0xF646 (rather than the EA6900's boardtype=0xD646) and i:boardflagsj= (i=0 and 1, j=blank and 2) seems to make all the difference between losing the upper AC speeds and not. Thanks to justmoi for figuring this out, also that the CFE Edit tool could be used cross-device (no need for patches and rebuilds, as I thought earlier).
   
   
4. The port reversal (Ethernet WAN is part of LAN and port 1 becomes WAN) seen when the EA6900 CFE is flashed to the 6700 seems to be caused by a change in convention between the two models. The only variable I could find potentially causing the trouble was 'vlan2ports=4 5u' from the EA6900. Reverting to the original 'vlan2ports=4 5' apparently fixed the problem.
   
   
5. Used the original's wandevs=vlan2 and not the 6900's wandevs=et0, but couldn't see that it made any difference either way.
   
   
6. Original boot wait_time=3 was changed to XV's wait_time=1. Just need an interval for rescue, if ever needed, so used 1.
   
   
7. By comparison, the original's variable list is quite short so I'm guessing NVRAM is auto-populated on boot with values stored permanently on chip, or in some uneditable location. Specifying the values explicitly apparently overrides that sequence. (More experimentation could confirm, but I'm mostly done now!)
   

[egc]

@u2n congratulations with your excellent work!
I have had some hefty discussion about the MAC addresses with @Cybrnook, but I totally agree with the +2 +4.

I will make a reference to your USB fix and will see if this also applies to the EA6900

[iMagic]

Hi there
Just flashed one of the latest Kong's mods on my EA6700.
First of all, it seems to lack packages for command line working. Perhaps, it was intended so, I don't know. Both ipkg and opkg don't work, "Permission denied". I've found some topics stating that these commands could be built for other than my router's architecture.
Then I have quit trying to install SW from scratch (Optware or Entware, as an example) and tried to set file sharing options through the web interface.
Here I was caught by a really strange behavior. I'm not an expert in Linux, but dd-wrt makes me wondering every time. This time a USB-HDD is blank after a reboot. I make all the folders again, but they are gone after a next reboot. Although in the web I still can see shares as defined earlier.
Current version is 33675M kongac. Previous version that worked was dd-wrt 30796.
What would you say, am I doing something wrong or this version of the firmware is bad for me?

[egc]

From kong's notes: