Asuswrt-Merlin 3.0.0.4.374.33 Beta 3 available

[RogerSC]

Functionally, beta3b is great. Signal strength on 5GHz. is okay (rt-n66u). Big loss on 2.4GHz., so I'm going back to beta1, since I have a .CFG file for that, and it was working really well for me.

Could use the experimental build equivalent to beta3b with the old SDK, but beta1 was so solid, might as well go back to that.

Thanks to RMerlin for carrying your features forward. Hopefully Asus will put out something better to build on soon.

 

[JPAC99]

Just to be 100%... what pops up when you put cursor over wifi icon? The Control Channel would show set for 1 on the Settings but the wifi icon would show 6, so would inSSIDer show me as being on channel 6.

Hi Pulp, Try changing bandwidth to 20 mhz and then change to channel 1.

 

[jsmiddleton4]

I appreciate folks making suggestions, tweak this, tweak that, but how is needing to do those kinds of tweaks progress?

A firmware should not need a bunch of tweaks in order for end users to maximize the ability of their hardware.

Setup information, etc.? Well sure.

But the level of tweaks apparently required in order for people to use their hardware to the fullest is telling me Asus is not paying attention to details and to what we end users have been saying about stuff like issues with the 5ghz band.

I don't get it but there it is.

 

[pulp]

Hi Pulp, Try changing bandwidth to 20 mhz and then change to channel 1.

Yep, will do but wifey watching Stargate via PS3 Media Server and kids have a LAN gaming party going on. I'm locked out of my network!

 

[JGrana]

See my reply on this in the other thread. I can't reproduce the issue here, which makes me think you might have kept the database in its default location, which is in RAM. That's probably what is causing your router to run out of memory as it fills up the database, you should relocate the database to a directory on your HDD.

Hmm, maybe the number of photos? I double checked and my minidlna.conf says to put the log and database files on my USB drive (formatted EXT3). When it runs, the log file grows as well as files.db and the art_cache directory. Like I said, I have lots of photos, 267GB at last count.

image_utils.c gets the malloc fail around 1/3 of the way through the directory.

Thanks for the work!

 

[SkyNet]

When i read it in the changelog:

- NEW: RT-N66U is now based on SDK6.x code (like Asus's own releases).
A separate SDK5 based on the old 5.100 driver from FW 270 is still
available in the Experimental folder on the download site.

I immediately realized that the issue with Ethernet ping on my AC66U succesfully solved and tests confirmed it. I don't know how it's relate to my AC66U but issue started in 3.0.0.4.372.30 with SDK changes for n66u... and now with 3.0.0.4.374.33 Beta 3 it solved finally with change SDK for...

 

[DrTeeth]

Concerning the power output issues, I'm not entirely sure if this this information is accurate <snip>

What are those commands meant to show?

 

[bluepoint]

Concerning the power output issues, I'm not entirely sure if this this information is accurate (with all that stuff being closed source and such, and various wl commands being obsolete or specific to different Broadcom devices), but I found the results interesting:

SDK6.30 (RT-N66U):
admin@stargate2:/tmp/home/root# wl -i eth1 txpwrlimit
84 mW (19.25 dBm)

Here's my result from your Beta3b.

admin@RT-N66U:/tmp/home/root# wl -i eth1 txpwrlimit
1000 mW (30.0 dBm)

 

[bluepoint]

A firmware should not need a bunch of tweaks in order for end users to maximize the ability of their hardware.

I don't think you should expect Asus to support your darft-N wireless card forever? It's about the time they started dropping non standard compliant hardwares and it will not get better since their focus now is AC.

 

[jsmiddleton4]

My broadcom 4322 is not some off brand non-standard card.

So you're saying Asus as it releases firwmares is going to lose a section after section of end users and force people who want to use Asus routers to update their laptops, wireless cards or go with some kind of USB wireless adapter?

This is your point of view?

 

[RMerlin]

What are those commands meant to show?

At first it seemed like the power limit value is different based on the SDK version and the router. Trying to find an explanation as to why increasing the power output doesn't seem to do much. Reducing the value definitely works, my signal strength immediately dropped by many dBs when I set it to 20 mW (lowest possible value).

Oddly enough however, that value seems to vary. A reboot once again brought back the maximum power output to 1000 mW. I fail to understand how a limit can change so drastically over time.

I've been also doing signal strength compares between SDK5 and SDK6. So far I get maybe a 4-5 dB attenuation increase on SDK6.

Even weirder: I just moved my laptop by about 2cm, and signal strength instantly dropped from -56 dBm to -62 dBm, and it's staying there. It seems that the router and/or NIC dynamically readjusts based on some variable criteria.

I'm afraid this is an area where there is simply nothing more I can do. The signal strength of both my AC66 and N66U standing on the same desk gives me roughly the same signal strength at the other end of the apartment, with maybe a 2-3 dB difference between both, so the issue isn't due to a misconfiguration of the RT-N66U radios caused by the SDK switch. This is a case where only someone with 1) the driver source code 2) an engineering-level understanding on how the wireless radios operate (something that maybe only Broadcom's own engineers might have?) could do anything about it.

 

[RMerlin]

Hmm, maybe the number of photos? I double checked and my minidlna.conf says to put the log and database files on my USB drive (formatted EXT3). When it runs, the log file grows as well as files.db and the art_cache directory. Like I said, I have lots of photos, 267GB at last count.

image_utils.c gets the malloc fail around 1/3 of the way through the directory.

Thanks for the work!

Best test case I could re-create here was with a few gigs (I had to copy the same folder multiple times - I'm really not a picture collector ). I was monitoring memory usage throughout the actual scanning process, and didn't see a tendency for it to keep dropping - it would constantly vary between 50 MB and 80 MB free at most. If you want to do more tests, try monitoring it while scanning:

watch free -m

Keep an eye on the free memory field, see if it has a tendency toward dropping, or if it's just oscillating between a very low value and a higher one - could be that you have some very large photos that require a lot of RAM to scan. Those I worked with weren't anywhere over 1920x1200 at most.


Just to be 100% sure the service isn't storing any data in RAM, check the content of /var/cache/minidlna after you index part of your collection.

 

[bluepoint]

My broadcom 4322 is not some off brand non-standard card.

So you're saying Asus as it releases firwmares is going to lose a section after section of end users and force people who want to use Asus routers to update their laptops, wireless cards or go with some kind of USB wireless adapter?

This is your point of view?

Your card is only good to SDK5.100 as there is support for your non standard wireless, up until SDK5.110 they have extended your cards support. Unfortunately, it seems SDK6 drops support for draft-N so now you're having problems. Sorry to say, for you to use SDK6 based firmware you have to buy a compliant wireless client or use SDK5 based firmware.
If you look around most newer routers not only Asus drop supports for Draft-N.

 

[RMerlin]

I appreciate folks making suggestions, tweak this, tweak that, but how is needing to do those kinds of tweaks progress?

A firmware should not need a bunch of tweaks in order for end users to maximize the ability of their hardware.

Well, this thread is about a beta firmware, so experimentations are to be expected. The switch to a new SDK by Asus also means that there might be issues in need of tracking down.

As I always say, interoperability will always be a major issue with wireless technology, especially when dealing with products from either different manufacturers, or different generations.

This isn't specific to Asus products. A few months ago, I went to a customer's home to try to set up their new wireless network. Turns out it was simply impossible to connect the customer's laptop to their brand new Linksys E6400 (or some similar model, I forgot the exact model number). The issue was widely reported as being a compatibility issue between the Atheros card that laptop used and a wide range of recent routers, including that specific Linksys model.

The initial solution was to install a cheap Asus RT-N12 as repeater in her office (since I needed the repeater to cover their backyard anyway), which worked pretty well. Eventually we opted to using a recent USB adapter on her laptop, which allowed here then to connect without any issues to their main Linksys router, and not just at half-speed on the repeater.

Perfect, trouble-free wireless is pretty much an utopia, based on my personal experience. Tinkering will almost certainly be involved at some point. If you need trouble-free, Ethernet is the only game in town at this point.

 

[mikey]

Oddly enough however, that value seems to vary. A reboot once again brought back the maximum power output to 1000 mW. I fail to understand how a limit can change so drastically over time.

Does this mean that it is possible that some sequence might restore the 1000 mW limit for those who are having trouble? The value that I'm currently using is 100 mW and when I reboot, I still get drastically lower signal level. Do you think setting it to the default 80 mW and rebooting might fix it? Maybe setting it higher than 80 is causing the problem and when it is set to higher than 80, it actually ends up using a value much lower than 80?

Mike

 

[Deleted member 22229]

Yeah same here this version cripples my 2.4 Ghz signal unusable at distance im flashing back to the older Merlin build 3.0.0.4.372.32bata1 I have tried them all and this build is rock solid on both radio bands and no issues with any of my devices computers, laptops, iphones, galaxy tablets, roku boxes all 100%. I will no longer bother with any build that has the SDK6 drivers.

 

[JGrana]

Best test case I could re-create here was with a few gigs (I had to copy the same folder multiple times - I'm really not a picture collector ). I was monitoring memory usage throughout the actual scanning process, and didn't see a tendency for it to keep dropping - it would constantly vary between 50 MB and 80 MB free at most. If you want to do more tests, try monitoring it while scanning:

watch free -m

Keep an eye on the free memory field, see if it has a tendency toward dropping, or if it's just oscillating between a very low value and a higher one - could be that you have some very large photos that require a lot of RAM to scan. Those I worked with weren't anywhere over 1920x1200 at most.


Just to be 100% sure the service isn't storing any data in RAM, check the content of /var/cache/minidlna after you index part of your collection.

Ok, did as you instructed. I ran 2 Putty sessions, one with watch free -m and the other doing a tail -f minidlna.log.

Interesting. Swap didn't change much, free memory wobbled between ~80000 and 61000, but buffers seemed to slowly creep down (from ~4000 then getting down to around 200) around the 200 buffers mark I saw the first malloc fail.

In looking at /var/cache - there is no minidlna file. The closest would be a file "songs.gdb".

Hope this helps!

 

[mikey]

Does this mean that it is possible that some sequence might restore the 1000 mW limit for those who are having trouble? The value that I'm currently using is 100 mW and when I reboot, I still get drastically lower signal level. Do you think setting it to the default 80 mW and rebooting might fix it? Maybe setting it higher than 80 is causing the problem and when it is set to higher than 80, it actually ends up using a value much lower than 80?

Mike

Never mind. I tried all that and signal strength is still just as bad. Interestingly, my N66U reports 1000 mW when I run that command: yet the signal strength is still way down.

I installed the beta3-SDK5 in the experimental folder and signal strength is back up. Is it OK to use that version?

Mike

 

[Deleted member 22229]

As long as that version has the ipv6 firewall it should be good. I use Merlin 372.32 bata1 works great. I just hope in the future there are no security issues that forces me to use the SDK6 firmwares because here they just don't work good.

 

[RMerlin]

As long as that version has the ipv6 firewall it should be good. I use Merlin 372.32 bata1 works great. I just hope in the future there are no security issues that forces me to use the SDK6 firmwares because here they just don't work good.

The SDK5 build is the exact same FW code as the regular build, except for the wireless driver, switch driver, HW acceleration code (CTF), and the other closed-source components from Broadcom. That means you get no support for the XBox optimisation switch, and no HW PPPoE acceleration.