Asuswrt-Merlin - custom build of the Asus RT-N66U firmware

[doc octopus]

I am interested to take the test for Merlin

 

[richto]

I have seen a few reports of the power output issues, but I haven't been able to reproduce them myself, after having spent a whole evening testing and figuring out how power output adjustments worked on this router (see a previous post from me in this thread on how it actually scales).

Once I'm done rebasing my work on build 144 I'll do further tests and see again if I can reproduce it. But I believe those who experienced the problem resolved it by restoring to factory settings - probably removing some temporary settings that were interfering. Asus did a few changes in build 144 to clean up those temporary values, so that could also help.

I won't be removing the limitations as these are legally imposed, and that's an area I'd rather not step onto.

Are you using a US version? I suspect this might be specific to the European limitations...

Power levels of the transmitter are NOT legally imposed in Europe. The only limit is on the effective radiated signal strength (0.1W). So if you used a lower gain antenna it would be legal with a higher power transmission level.

Also your firmware seems to limit to 80mW whereas the original firmware allows 100mW for Europe.

It is perfectly legal to own radio equipment capable of illegal transmissions - it is only illegal to use it in a forbidden way. i.e. modifying potential output levels in itself puts you at no legal risk.

Also is it possible to switch the router between US and EU settings? Say if I relocated i might want to switch....An option to set the location would be appreciated.

 

[RMerlin]

I just posted a test build of 3.0.0.3.144.10 on the Github repo:

https://github.com/RMerl/asuswrt-merlin/downloads

Based on 3.0.0.3.144, this version has 64K NVRAM support enabled. DualWAN support is disabled for now - I will possibly make a separate DualWAN release as it seems some people have better luck with the regular 3G failover code.

Make sure you read the included IMPORTANT.txt file before you flash this!

Flashing this version *WILL* wipe your configuration back to factory defaults! This is a side-effect of the switch to 64K NVRAM.

Things I am most interested to know about:

• Is this version generally more stable for you?
• Any weird issue related to the 64K NVRAM switch? (Since Asus hasn't released an official version with this enabled yet, I have no idea if this feature is fully working yet or not)
• How is the 3G failover working out?

Changes in 3.0.0.3.144.10 Beta 1:

• NEW: Rebased on 3.0.0.3.144.
• NEW: Support for 64K NVRAM enabled. ***First flash will wipe out ALL your settings! And you cannot restore from saved settings - you must manually reconfigure everything. Be warned!***
• NEW: Enabled support for Broadcom Wimax devices
• FIXED: Router crashing when connecting to it over Wifi and running the newer QoS code (disabled GRO)
• FIXED: Router crashing when connecting to a network device behind the router from over a VPN connection (disabled GRO).
• FIXED: Incorrect timezone set unless enabling manual DST.

Notes:

• If after 10 mins the flashed router seems unavailable, remember that it's back to its 192.168.1.1 default IP. You might also need to power it off, wait a few secs, and power it on again.
• In case all fails, you can always put it into recovery mode (turn it on while the Reset button is pressed), and flash a previous firmware using the miniweb server at http://192.168.1.1, or the Firmware Restore tool from Asus's CD.

 

[cc666]

64 nvram

Merlin,

I didn't think that .144 had the 64 nvram?

I thought only the .128 has this? Do they both have the 64 nvram?

I find the .128 to be much faster than the .144......


CC

 

[RMerlin]

Merlin,

I didn't think that .144 had the 64 nvram?

I thought only the .128 has this? Do they both have the 64 nvram?

I find the .128 to be much faster than the .144......


CC

The code for 64K support is in the sources, but Asus chose (for some reason) not to enable it when they released 3.0.0.3.144.

I completely recompile the firmware from the sources, so I'm able to enable/disable different options than their official release.

There shouldn't be much differences in performances between the existing releases, since as far as I know the wifi driver hasn't changed between any of these releases.

 

[jyavenard]

Would you be so kind as telling me what you had to do to enable 64kB nvram on this firmware?

I'll what I can do to get it going on tomato firmware.

Thanks
JY

 

[jyavenard]

The code for 64K support is in the sources, but Asus chose (for some reason) not to enable it when they released 3.0.0.3.144.

I completely recompile the firmware from the sources, so I'm able to enable/disable different options than their official release.

There shouldn't be much differences in performances between the existing releases, since as far as I know the wifi driver hasn't changed between any of these releases.

Something you could add in your IMPORTANT.txt in order to tell people how to archive/restore their nvram values
rather than using nvram show.

do:
nvram export --set

This will generate an output that can be passed on directly to a shell script. Very easy to restore with that.
followed by
nvram commit

 

[RMerlin]

Would you be so kind as telling me what you had to do to enable 64kB nvram on this firmware?

I'll what I can do to get it going on tomato firmware.

Thanks
JY

I set "NVRAM_64K=y" in target.mak. So, nothing that can help other custom firmwares I'm afraid Asus implemented support both at the kernel level and in the related tools. They didn't touch the CFE. So you would have to backport all the kernel-level changes to Tomato to be able to implement it.

 

[RMerlin]

Something you could add in your IMPORTANT.txt in order to tell people how to archive/restore their nvram values
rather than using nvram show.

do:
nvram export --set

This will generate an output that can be passed on directly to a shell script. Very easy to restore with that.
followed by
nvram commit

That option doesn't exist with Asus's version of the nvram tool.

As for the "nvram save filename" option it supports, that will generate a binary file similar to what is used by the webui, so it's unsuitable for nvram size changes.

 

[thexile]

Are you using a US version? I suspect this might be specific to the European limitations...

Power levels of the transmitter are NOT legally imposed in Europe. The only limit is on the effective radiated signal strength (0.1W). So if you used a lower gain antenna it would be legal with a higher power transmission level.

Also your firmware seems to limit to 80mW whereas the original firmware allows 100mW for Europe.

It is perfectly legal to own radio equipment capable of illegal transmissions - it is only illegal to use it in a forbidden way. i.e. modifying potential output levels in itself puts you at no legal risk.

Also is it possible to switch the router between US and EU settings? Say if I relocated i might want to switch....An option to set the location would be appreciated.

Yes, I second this request!

 

[RMerlin]

Try this, but I'm not sure if it will work as the bootloader might be overwriting it, in which case there isn't much I can do:

nvram set regulation_domain=US
nvram set wl0_country_code=US
nvram set wl1_country_code=Q2
nvram commit
reboot

 

[richto]

As presumably the EU power settings are not fixed, i migrated from V9 of this straight to version 151 original firmware which seems to be fast and stable - and my 5GHz is working properly again @ 100mW

 

[xrecar]

Why can't I select /mnt/ anymore to see BOTH of my hard drives via media server? I bought a router with two USB ports for a reason. Is there any way to fix this?

 

[RMerlin]

Why can't I select /mnt/ anymore to see BOTH of my hard drives via media server? I bought a router with two USB ports for a reason. Is there any way to fix this?

I haven't tested it, but see what happens if you connect through telnet and issue the following commands:

nvram set dms_dir=/mnt
nvram commit

Then stop and start the Media server.

 

[xrecar]

interesting solution. Trying it right now. It appears both drives are labeled the same thing instead of "my book'" and "elements" I get "elements" and "elements(1)". I will wait for it to stop scanning and see if it straightens out. Thanks again for the temp workaround.

 

[richto]

As well as the 5GHz power settings working properly (actually allows 101mW for EU!), whereas they dont on this custom version, be aware that original build 151 is also MUCH faster than this build...

 

[RMerlin]

As well as the 5GHz power settings working properly (actually allows 101mW for EU!), whereas they dont on this custom version, be aware that original build 151 is also MUCH faster than this build...

Except mine doesn't crash half of the time

Anyway, just waiting on Asus to release newer GPL code to rebase my work on it. Then I'll have the best of both worlds.

 

[richto]

Zero crashes for me so far, and no major issues reported on the 151 thread...

 

[RogerSC]

Zero crashes for me so far, and no major issues reported on the 151 thread...

Probably depends on the specific things that you do with your router...I use my router as simply as I can, and have also had no crashes, none with .108, and none with .151 yet. But from what I've been reading about the problems in the firmware source code, I'm not sure how much ASUS does code reviews, new feature testing, and regression testing when they make changes.

 

[RMerlin]

Zero crashes for me so far, and no major issues reported on the 151 thread...

The crashes were caused by very specific scenarios. I tracked down and fixed two router crashes that occurred if you had SMB file sharing enabled (which, in turn, enabled GRO support in the networking stack), and I either tried to connect my laptop over wifi, or tried to access a device/PC on the LAN over VPN while the VPN was within the same IP network. Those bugs (especially the first one) meant that any firmware newer than 108 was impossible to use. I haven't tried build 151 yet (was hoping to see Asus do an official release + release the source code this week), so I don't know if they finally fixed that one or not.