Asuswrt-Merlin 376.46 is out (RT-AC87 only)

[RMerlin]

Merlin...excuse me if this is a dumb/wrong observation, but it looks like you loaded the binaries into a 'prebuild' directory.....shouldn't it be 'prebuilt'?

When I made that change (and a few others like going back to the paragon drivers) I was able to get a build to complete under my fork (maybe call it a Frankenstein build - I have no idea if it will actually work)

Yes, that folder on Git was wrong, however I was getting various link errors anyway with other components such as emf/igs which aren't compiled for non-SMP kernels. That's why I never actually fixed it on the repo, and just reverted the original commit.

Same errors happened with the RT-AC56S, another uniprocessor ARM profile.

 

[RMerlin]

Edit: In order to replicate this issue, PPTP has to be selected from the drop-down menu.

Hitting "Apply" doesn't activate the VPN server when the selected mode is OpenVPN.

That could be a critical detail as to which VPN server you are actually talking about. I'll see if I can reproduce it with the PPTP page.

 

[RMerlin]

Yes, looks like for some reason Asus was automatically enabling the PPTPD server when applying settings, and the code got carried over to mine. I now removed that line.

 

[z1022]

Merlin...excuse me if this is a dumb/wrong observation, but it looks like you loaded the binaries into a 'prebuild' directory.....shouldn't it be 'prebuilt'?

When I made that change (and a few others like going back to the paragon drivers) I was able to get a build to complete under my fork (maybe call it a Frankenstein build - I have no idea if it will actually work)

Hi john,

Can you provide the RT-N18U build for testing?

Thanks

 

[john9527]

Hi john,

Can you provide the RT-N18U build for testing?

Thanks

I'm always a little bit nervous when something works for me and not Merlin

EDIT: I double checked my build parameters and now duplicate Merlin's results....looks like the N18U is a no go.

 

[RMerlin]

I'm always a little bit nervous when something works for me and not Merlin

I'll probably give it another try. One possibility is that when the driver fails to link, the compiler might be spitting non-sense about other modules also failing to link. I'll see if maybe it might be a domino effect caused by the wl_arm/default/ driver failing to link against a non-SMP kernel.

Wouldn't be the first oddity with this build environment...

 

[RMerlin]

I'm always a little bit nervous when something works for me and not Merlin

But, if someone wants to try (at their own risk), send me a private message and I will send a link. I don't want to post a link for general testing until we at least know the router survives the load.

Nope, it still fails linking emf and igs.

ERROR: "_raw_spin_lock_bh" [drivers/net/igs/igs.ko] undefined!
ERROR: "_raw_spin_unlock_bh" [drivers/net/igs/igs.ko] undefined!
ERROR: "_raw_spin_lock_bh" [drivers/net/emf/emf.ko] undefined!
ERROR: "_raw_spin_unlock_bh" [drivers/net/emf/emf.ko] undefined!

If your build manages to complete then either you are compiling your kernel with SMP still enabled (which is a bad idea performance-wise since the RT-N18U is a single core device), or your build will contain versions of emf.ko and igs.ko that will fail to load on an RT-N18U.

 

[john9527]

If your build manages to complete then either you are compiling your kernel with SMP still enabled (which is a bad idea performance-wise since the RT-N18U is a single core device), or your build will contain versions of emf.ko and igs.ko that will fail to load on an RT-N18U.

Double checked and part of SMP was still enabled....good call. Now get the same results. Sorry for the fire drill.

 

[RMerlin]

Double checked and part of SMP was still enabled....good call. Now get the same results. Sorry for the fire drill.

No problem.

Note that it might be interesting to see how things would work out if SMP was left enabled. I'd expect the Linux kernel to be intelligent enough to deal with a single core, not sure it would be optimal performance-wise tho.

 

[john9527]

No problem.

Note that it might be interesting to see how things would work out if SMP was left enabled. I'd expect the Linux kernel to be intelligent enough to deal with a single core, not sure it would be optimal performance-wise tho.

Well, I didn't delete the fw image yet if anyone wants to give it a go...

 

[AndreyPopov]

RT-AC68U have key BCMSMP=y

RT-N18U ARMCPUSMP=up

and RT-N18U also have RTAC56U=y

 

[AndreyPopov]

https://github.com/RMerl/asuswrt-merlin/commit/e404eac647cf4ccf22ec42f7477fead55339789c

not resolved problem with adding username/password on PPTP Server mode.

on 376.2061 all work. on 376.2678 not work.

 

[AndreyPopov]

Advanced_VPN_Content differencies between 376.2678 and 376.2061
1.
if(document.getElementById('pptpd_clientlist_table').rows[x].cells[1].title == pptpd_connected_clients[y].username){
if(document.getElementById('pptpd_clientlist_table').rows[x].cells[1].innerHTML == pptpd_connected_clients[y].username){


if($("openvpnd_clientlist_table").rows[x].cells[1].title == openvpnd_connected_clients[y].username){

if($("openvpnd_clientlist_table").rows[x].cells[1].innerHTML == openvpnd_connected_clients[y].username){

--------------
innerHTML replaced by title

2.
}else if(!Block_chars(valid_username, [" ", "@", "*", "+", "|", ":", "?", "<", ">", ",", ".", "/", ";", "[", "]", "\\", "=", "\"", "&" ])){

}else if(!Block_chars(valid_username, [" ", "@", "*", "+", "|", ":", "?", "<", ">", ",", ".", "/", ";", "[", "]", "\\", "=", "\"" ])){

}else if(!Block_chars(valid_password, ["<", ">", "&"])){

}else if(!Block_chars(valid_password, ["<", ">"])){

---------------

added "&" as block symbol


3.
if(rule_num >= upper){

if(rule_num > upper){


if(rule_num >= upper+1){

if(rule_num > upper+1){

------------------------------
">=" replaced by ">"

4.
if(username_obj.value == $(table_id).rows.cells[1].title){

if(username_obj.value == $(table_id).rows.cells[1].innerHTML){

-----------

innerHTML replaced by title

5.

code +='<td width="35%" title="'+pptpd_clientlist_col[0]+'">'+ pptpd_clientlist_col[0] +'</td>';

code +='<td width="35%">'+ pptpd_clientlist_col[0] +'</td>';


code +='<td width="35%" title="'+vpn_server_clientlist_col[0]+'">'+ vpn_server_clientlist_col[0] +'</td>';

code +='<td width="35%">'+ pptpd_clientlist_col[1] +'</td>';

----------------------
added title="'+pptpd_clientlist_col[0] and title="'+vpn_server_clientlist_col[0]
and replace pptpd_clientlist_col[1] by vpn_server_clientlist_col[0]


-----------
-----------------
what of these not able add username/password?

 

[z1022]

Well, I didn't delete the fw image yet if anyone wants to give it a go...

John,

I have a interest to testing. Can you give me a like ?? However, before I flash, I need to test how to use firmware Restoration.

Thanks

 

[john9527]

John,

I have a interest to testing. Can you give me a like ?? However, before I flash, I need to test how to use firmware Restoration.

Thanks

Sent you a PM with the link...you've followed the discussion, so know this isn't a perfectly accurate build (good idea on testing firmware restoration first).
Be interesting to see your results.

 

[z1022]

Sent you a PM with the link...you've followed the discussion, so know this isn't a perfectly accurate build (good idea on testing firmware restoration first).
Be interesting to see your results.

I know this in't a perfectly accurate build. If I have tested, I will report results.

Thanks

 

[AndreyPopov]

I cannot undestant WHY and HOW there:
if(pptpd_clientlist_col[1].length >32){
overlib_str1 += pptpd_clientlist_col[1];
pptpd_clientlist_col[1]=pptpd_clientlist_col[1].substring(0, 30)+"...";
code +='<td width="35%" title="'+overlib_str1+'">'+ pptpd_clientlist_col[1] +'</td>';
}else
code +='<td width="35%" title="'+vpn_server_clientlist_col[0]+'">'+ vpn_server_clientlist_col[0] +'</td>';

pptpd_clientlist_col[1] replaced by vpn_server_clientlist_col[0]

is this ASUS bug?

 

[z1022]

Sent you a PM with the link...you've followed the discussion, so know this isn't a perfectly accurate build (good idea on testing firmware restoration first).
Be interesting to see your results.

Testing 376.47 build.

system status show 1 core CPU, it should be OK.
However, the wireless is disable and no wireless light off.

Build 376.43 also cannot enable wireless.

Please refer attached pic.

 

[RMerlin]

Testing 376.47 build.

system status show 1 core CPU, it should be OK.
However, the wireless is disable and no wireless light off.

Build 376.43 also cannot enable wireless.

Please refer attached pic.


View attachment 2680
View attachment 2681

Try running this over SSH:

modprobe wl
dmesg | tail -n 20

I'm curious to see what happens there.

 

[john9527]

Maybe you should break this out into a separate thread to avoid confusing people?

I'm sure you noted that the Wireless driver version was blank.....