Release [Fork] Asuswrt-Merlin 374 LTS release 49E4

[domic]

If it's a MIPS model like the N66U or AC66U it's 2.6.22.19.

Ah.. So much for that wireguard-go idea I had then. Yes, It's a AC66U.

 

[john9527]

RE: watchdog crash

Good data from everyone! Thanks.
Based on that, I think I may have found a memory leak in some code I 'borrowed' from lighttpd to address a possible pointer corruption during nvram reads which went in release 48. I rewrote the code (with some help from StackOverflow ) and it's up to try as 51D2. Fingers crossed.

PS - You'll also get the latest OpenSSL update and updated getdns/stubby. If it muddy's the water, I'll do another build and back them out.

 

[tsanga]

RE: watchdog crash

Good data from everyone! Thanks.
Based on that, I think I may have found a memory leak in some code I 'borrowed' from lighttpd to address a possible pointer corruption during nvram reads which went in release 48. I rewrote the code (with some help from StackOverflow ) and it's up to try as 51D2. Fingers crossed.

PS - You'll also get the latest OpenSSL update and updated getdns/stubby. If it muddy's the water, I'll do another build and back them out.

51D2 did not go well for me. I updated via GUI, then found myself unable to access the router, even after a hard switch reset, static IP, etc. Eventually I put the router into recovery mode and flashed 47EB using the Asus Firmware Restoration Utility to get up and running again.

 

[ColinTaylor]

51D2 did not go well for me. I updated via GUI, then found myself unable to access the router, even after a hard switch reset, static IP, etc. Eventually I put the router into rescue mode and flashed 47EB using the Asus Firmware Restoration Utility to get up and running again.

Same here with my N66U. Did a normal update from 47EB and it was bricked. Even a WPS reset wouldn't revive it. I'm currently waiting on my second attempt at trying to flash 51D2 in recovery mode as the first attempt still resulted in a brick.

 

[Maverickcdn]

Me three. Radios were up and ssid was broadcasting but unit was 100% unresponsive on lan and wlan wouldn't authenticate

 

[ColinTaylor]

My second recovery attempt still resulted in a brick. So I recovered back to 50DA and then attempted to update to 51D2 through the GUI. That also resulted in another brick.

 

[john9527]

This just keeps getting stranger and stranger.....
I've been running the 51D2 on my AC68 all day without a problem (also no problem on the 50E8 release before that).
I then tried to load 51D2 on my N66 and got the same no response as has been reported.

While I've had problems in the past getting something to compile successfully MIPS vs ARM, I've only had one other case I can remember where something would compile and then not work on MIPS vs ARM, and that case involved a specific function library. In this case, it's just basic C code. It appears as if something in the 'old-old' MIPS compiler is generating something strange vs the just 'old' ARM compiler (maybe there's an errata somewhere on pointer manipulations).

Anyway, the patch I found is still my best guess at a possible fix for the reboots. 51D3 is up which just reverts the nvram patch. Verified this loads on my N66.

 

[Maverickcdn]

51D3 flashed on my spare ... Ill flash the other 2 in service probably tomorrow

 

[ColinTaylor]

Anyway, the patch I found is still my best guess at a possible fix for the reboots. 51D3 is up which just reverts the nvram patch. Verified this loads on my N66.

51D3 has been running for 10 hours now and *none* of the processes have increased (or decreased) in size by even a single byte.

 

[tsanga]

Can confirm that 51D3 flashed successfully on my AC66U. Now the clock starts….

 

[Rahul_K]

51D3 has been running for 10 hours now and *none* of the processes have increased (or decreased) in size by even a single byte.

Same here. Literally 0 change in memory consumption for both wanduck and watchdog in the last 13 hours of uptime. 51D3 feels rock solid!

 

[frazor77]

Been on a family break... here's the log from 50DA: https://pastebin.com/Ecg2xcAZ

Will now try moving on to 51D3

 

[RMerlin]

I keep reading that as "soda"...

 

[ColinTaylor]

51D3 has been running for 10 hours now and *none* of the processes have increased (or decreased) in size by even a single byte.

2 days and 16 hours and still no change in memory usage.

 

[frazor77]

I keep reading that as "soda"...

So do you read 51D3 as SIDE?

 

[frazor77]

2 days and 16 hours and still no change in memory usage.

I'm a mere 28h in... but so far so good.

 

[frazor77]

So ~50-60h in my ring pro doorbell and my (Shelly) relay switch that controls my garage door both dropped off the network overnight (around 5.30am) and refused to reconnect (I tried rebooting and resetting both devices several times), last resort was router reboot that brought them back online. The log shows the Ring Spot Cam (another Ring device on my network, but which was working OK) and a few other devices doing a lot of DNS overnight. Log here: https://pastebin.com/iRBeqgR

 

[john9527]

So ~50-60h in my ring pro doorbell and my (Shelly) relay switch that controls my garage door both dropped off the network overnight (around 5.30am) and refused to reconnect (I tried rebooting and resetting both devices several times), last resort was router reboot that brought them back online. The log shows the Ring Spot Cam (another Ring device on my network, but which was working OK) and a few other devices doing a lot of DNS overnight. Log here: https://pastebin.com/iRBeqgR

Nothing for me to investigate there.....router did everything right responding with a valid ACK to all the requests (wireless connect was OK)c. At least some internet access was working since NTP sync was working. I've seen cases where a single device (even one that appears to be OK) can get 'stuck' and cause similar problems (I had an HDHomerun Prime that would get stuck generating an arpc storm a couple times a year).

Since you didn't have any watchdog issues, I'm considering that fixed and am going to do a formal release for that change.

 

[john9527]

Release 51E3 is now available.

Release - [Fork] Asuswrt-Merlin 374 LTS release 51E3

LATEST RELEASE: Update-51E3 28 August 2021 Merlin fork 374.43_51E3j9527 ============================ Update-51E3 Highlights Formal release equivalent to development 51D3j9527 No need to update if already running 51D3 Fixed a memory leak which was causing intermittent reboots on MIPS based...

www.snbforums.com

@RMerlin, please lock this thread.