[Fork] FreshTomato-MIPS

Discussion in 'Tomato Firmware' started by kille72, Apr 26, 2018.

  1. Maxima623

    Maxima623 Network Newbie Member

    Hello! Thanks for your hard work on this. Works great on my Linksys E1500, although I experience problems with QoS. Bandwidth limits set on QoS page are not respected. Speedtests show that I still use my full bandwidth. Other than this, all is well :)
     
  2. hkwakernaak

    hkwakernaak Serious Server Member

    Same results here with an RT-AC66U did a reset to factory -> Clear NVRAM but that doesn't resolve the issue.

     
  3. M_ars

    M_ars Network Guru Member

    You can NOT use any backup. You have to start from scratch. Cleared browser cache?

    Please see also first page of this thread.
     
    pedro311 and kille72 like this.
  4. hkwakernaak

    hkwakernaak Serious Server Member

    Didn't use any backup and cleared NVRAM as written on first page

     
  5. txnative

    txnative Addicted to LI Member

    Can you show how your settings are done in the bandwidth limiter.
     
  6. M_ars

    M_ars Network Guru Member

    Can you show / explain your setup ?
    BR
     
  7. hkwakernaak

    hkwakernaak Serious Server Member

    The RT-AC66U is behind a Fibre router. The fibre router is connected with the WAN port of the RT-AC66U. One 2.4Ghz wifi is defined and one 5Ghz wifi is defined. The rest of the FreshTomato settings of the RT-AC66U is untouched. When the WAN port is not connected the wifi is connecting fine.

     
  8. Maxima623

    Maxima623 Network Newbie Member

    I didn't enable bandwidth limiter. I only use QoS. Even with default settings, it doesn't follow the maximum rates entered on the QoS page. I also use a fresh configuration, meaning I cleared NVRAM thoroughly. QoS worked perfectly with the same setup on last Toastman build.
     
  9. txnative

    txnative Addicted to LI Member

    I see. Are you referring to incoming outgoing rates entered? (Edited) The default settings usually work well pending on how much bandwidth you have . The E1500 model isn't very powerful but if it worked with toastman then it should work with freshtomato. I could replicate to see if there is a problem but I don't have the exact same model, but it is a mips based and will use the same code and my bandwidth maybe a bit more to handle, but I'm checking if Qos is respectful to the default settings . Regards

    Edited:
    I've done what I said I'd do, I've installed the 2018.5 on my belkin f7d4302 and linksys e3200, both would have a problem with what they could do to handle my 200/10,so I placed some reasonable incoming and outgoing numbers that what they could handle 50000 incoming and 9000 outgoing as they performed nicely on the default settings using speedtest.net and dslreports speedtest(bufferbloat)This doesn't mean that there is a problem with the builds or at least of them, but maybe a bug on certain models? I don't know, since I don't have your particular model, I know that it worked fine the models I have. I'd continue to use toastman if you can't resolve on the current fresthomato build. Regards
     
    Last edited: Dec 26, 2018
  10. st_ty_

    st_ty_ Network Newbie Member

    Had very similar problem with my Netgear WNDR4500V2. Was related to the combination of wan-connection, wireless-N and newer dnsmaq-versions. To help you and to check, if we have had identical problem with our routers at all, first disable 5Ghz-wifi and test 2.4Ghz-wifi only with wireless-B/G/N (Do not choose "auto" as it results also in the "constant-reboot-with wireless and wan"-behavior.) Also test, if this "constant-reboot" -behavior happens with all of the freshtomato versions and not with (older) shibby tomato - version (as with my router). Rem. : This has nothing to do with fresh tomato code but with newer dnsmaq-versions (>=2.77) fresh tomato uses. Regards.
     
  11. DracoMilesX

    DracoMilesX Networkin' Nut Member

    Just tried to update from the latest AT to the 2018.5 build.

    Everything looks to be going fine but it seems when I try to change the network settings from 192.168.1.1 to 172.16.0.1 (including dhcp, etc...) and make multiple vlan's the network ports(/router) keeps on rebooting continuously.

    Only other thing that I changed with it at that moment are the wifi's (WPA2 personal/EAS+passwords)

    Got an ASUS RT-N66U trying to flash freshtomato-RT-N66U_RT-AC6x-2018.5-AIO-64K had the same issue the first time i tried to flash FreshTomato too when it was the 2018.3.1 version.

    Anyone else seen this issue?

    For now I reverted back to the latest AT build

    edit: guess it might be similar to the previous mentioned issue by kuthil (but it feels like he was just talking about wifi? and my issue is with lan ports reconnecting too)
     
  12. st_ty_

    st_ty_ Network Newbie Member

    Sorry. Typo in my first post. In first step you have to test if your router works with b/g mixed (5Ghz disabled) and not with "wireles-B/G/N, as I have written in my first post. Regards.
     
  13. Aladino

    Aladino Network Newbie Member

    I’m using latest version Netgear WNDR4500v2
    Internet won't work, I keep trying to connect to my college L2TP server with no luck
    in status, I get IP, DNS etc but no internet
    On the stock gene firmware, L2TP connection used to work perfectly
    Thank you for your effort[​IMG]
     
    Techie007 likes this.
  14. Feliciano

    Feliciano Connected Client Member

    I also think some modules are "hard-coded" to 192.168.1.1, for if I set other IP address, several things break (including some menu options that appear and dissapear deppending on the LAN IP Address set). That don't happen with Toastman's.
     
  15. Techie007

    Techie007 Networkin' Nut Member

    I upgraded a Linksys E1500 from 2018.4 to 2018.5, and it immediately lost the ability to connect to the Internet via PPPoE. Previously, on 2018.4, it would sit there forever saying "Renewing", but it would gain Internet connectivity within a minute and rotate through public IP addresses every couple hours. Now it never gets an IP address and can't connect to the Internet. It also got some GUI issues, where half of the pages would show using the red theme. Upon clearing NVRAM and starting over, the GUI issues went away, but it still couldn't connect with PPPoE. Reverting back to 2018.4 firmware, clearing NVRAM and setting it back up again did NOT restore functionality. But the ISP assures me that nothing is wrong on their side and the problem is my router. I am now unable to connect to the Internet at that location without using the DHCP option, which routes through their modem instead of giving the router a public IP address.
     
    Aladino likes this.
  16. Aladino

    Aladino Network Newbie Member

    This problem does occur on Tomato and DDWRT, but on the stock firmware it connect perfectly with no problem
    I just write the username password and server and it connect
    I’m sure there is a solution for this problem, we need to wait for him to answer. :)
     
  17. Aladino

    Aladino Network Newbie Member

    I fixed it by disabling Use internal DNS and Prevent DNS-rebind attacks in
    DHCP / DNS Server (LAN) section.

    now I can connect to the internet on lan but I got the wifi problem when I connect to the wifi the whole router reboot.
    now I'm using Shibby 1.28 until there is a fix for that problem. :D
     
  18. txnative

    txnative Addicted to LI Member

    I usually do that, and since I manually set in dns, I select "use receive dns with user-entered dns" and deselect everything else in advanced/DHCP/DNS that doesn't apply to my setup. Those issues are problematic for newer builds? I suppose this occurs when selecting other than "DHCP" on the wan port and using L2TP, PPPOE?
     
  19. Aladino

    Aladino Network Newbie Member

    I guess so since that connecting to L2TP on the stock firmware works perfectly.
    Maybe because DNS protection in the stock genie firmware is outdated comapring to Tomato
     
  20. kernel-panic69

    kernel-panic69 Connected Client Member

    Unless some other package or something in the UI is clobbering or colliding with rp-ppoe, which was updated to the most recent release, there shouldn't be any (new) issues with non-DHCP wan setups. Really wished I had a way to test all wan modes to be sure, but. There may be some magical configuration issue. Are you using 'route(r) modem IP'.... or is it blank? On DHCP wan, not having that info in those fields breaks internet for me, if I remember right.
     
  21. Henkie de Groot

    Henkie de Groot Reformed Router Member

    Thank you!!
    Just upgraded my Asus RT-N66U from Shibby MIPSR2-140_K26_USB_VPN to Freshtomato-K26USB_RT-N5x-MIPSR2-2018.5-VPN-64K. My setup is rather extensive: dual WAN (4G & DSL), USB's, VPN servers & clients, custom routes, QoS, Tons of static stuff - the works. So against all advise (!) I restored my "old" config file and voila - everything runs like a dream :D
    Super cool. Thank you guys - beers are on the way. Happy holidays!
     
  22. Aladino

    Aladino Network Newbie Member

    Anyone who can get OpenVPN to work with the firmware?
    I can't get it working by any chance
    I attached the screenshots of my config &
    attached the log.
     

    Attached Files:

    • 1.png
      1.png
      File size:
      178.4 KB
      Views:
      27
    • 2.png
      2.png
      File size:
      191.4 KB
      Views:
      27
    • 3.png
      3.png
      File size:
      223.1 KB
      Views:
      26
    • 4.png
      4.png
      File size:
      266.4 KB
      Views:
      25
    • syslog (11).txt
      File size:
      70.4 KB
      Views:
      12
  23. Mjenks

    Mjenks New Member Member

    I just installed 2018.5 (freshtomato-K26-NVRAM32K_RT-N5x-MIPSR2-2018.5-Mini.trx) on a Asus RT-N12 B1.

    If I leave the WAN on DHCP, it will never get an IP and stays at renewing. If I set it static, it works just fine.

    Single WAN, inside lan re-ip'd to 192.168.10.x/24. But that shouldn't matter, even a factory reset with all defaults still won't get an external IP..

    EDIT: Using manual external DNS

    Worked fine with 2018.4 on a RT-N16. Tomorrow I will upgrade the 16 and see if it's a .5 issue, or a hardware dislike of the firmware.

    On another note, on 2018.5, I have to use IE. If I used Chrome, I can't click the save button at all.
     
    Last edited: Dec 29, 2018
  24. Mjenks

    Mjenks New Member Member

    Okay 2018.5 (freshtomato-K26USB-NVRAM32K_RT-N5x-MIPSR2-2018.5-Max) works fine on my RT-N16 for WAN DHCP. It does not work on my N12.

    Also, the save button works in Chrome on my N16 with 2018.5.

    I'm going to try Max on my N12 next to see what works. Could be the bundle, or hardware.

    Edit: So, N12 Mini to my N16 Max gets WAN DHCP just fine. It must be something to do with how my ISP is giving out DHCP?
     
    Last edited: Dec 29, 2018
  25. kernel-panic69

    kernel-panic69 Connected Client Member

    Your N16 isn't assigning IPs per MAC address, whereas your ISP probably is. If your N16 works fine with the ISP, then use that as your outer router, or get the MAC from the N12 and get on the phone with your ISP. Been through this fun plenty of times with Cable and DSL providers over the years.
     
  26. st_ty_

    st_ty_ Network Newbie Member

    I have a Netgear WNDR4500v2 and a Asus RT-N66W. By default, I use these routers in router-mode (i.e. with WAN port connected to external DSL-modem/Internet or to another router in cascading mode).
    With both routers I had this "constant reboot by wifi connecting on LAN side"-phenomena.

    My experience so far:
    Three conditions must be fullfilled to trigger this "constant reboot"-effect on WNDR4500v2 /RT-N66W (AC-branch):
    - router mode (i.e. "WAN port used" with pppoe or dhcp selected; other WAN options like PPTP, L2TP, 3G/4G-modem not checked now)
    - DHCP-mode on LAN-side (one of main tasks of dnsmasq)
    - selection of "wireless-N" or "Auto" on 2,4GHz or 5 GHz wifi
    If only one of these conditions is not fullfilled, my routers will work already as expected.
    So preliminary (but not satisfying) workarouds are: "Use wireless B/G instead of wireless-N" or "manual release of IP-address on LAN side".
    All of the Freshtomato versions (2018.1-2018.5) for WNDR4500v2 or RT-N66W are effected by this "constant reboot"-phenomena, and all of them use dnsnmasq versions >=2.77 for DHCP on LAN-side.
    Older Tomato versions (Shibby v138, v140) never show this effect, in any router configuration. All these Tomato builds use dnsnmasq versions <=2.76 for DHCP on LAN-side.
    So I have built some freshtomato versions from the Freshtomato repo but using a "backported" dnsmasq version 2.76 (of the older Tomato-versions): These builds do not show this "constant reboot"-effect anymore!
    Vice versa, by injecting newer versions of dnsmasq (>=2.77) in older Tomato versions (<140/Shibby), these older Tomato versions will show all of a sudden this "constant reboot"-effect as the freshtomato version will always do.
    So by testing an older Tomato version (V140) with all dnsmasq snapshots between versions 2.76 and 2.77 leads to a commit (commit 503c6091...; 09-04-2017; author Florian Bos), where special "pxe-boot" releated code was introduced. With this code implemented into dnsmasq, the effect on WNDR4500v2/RT-N66W begins. Checking the code parts updated by this commit of leads to the following three lines in dhcp.c, causing the "constant reboot"-phenomena in WNDR4500v2 and RT-N66W (AC-branch):

    ...
    155 struct timeval tv;

    182 if (ioctl(fd, SIOCGSTAMP, &tv) == 0)
    183 recvtime = tv.tv_sec;
    ...

    By commenting out these three lines (they are not relevant for routers and have never been changed any more up to the newest dnsmasq commits(2.81test2)), will result in full working builds for all freshtomato version for WNDR4500v2 and RT-N66W (AC-branch). Regards.
     
    Chrushev and Magister like this.
  27. kernel-panic69

    kernel-panic69 Connected Client Member

    I presume that has got to be hardware-specific (BCM4706, 4331), because it 'should' affect ALL Broadcom devices, not just specific ones. I'm also curious to know why those lines 'have no effect'. Anyhow, the WAN port should be using udhcpc via BusyBox, nothing connected to dnsmasq whatsoever -- maybe there is a conflict there? E4200v1 does NOT display this phenomena with those same configuration points -- BCM4718, 4331.
     
    M_ars likes this.
  28. M_ars

    M_ars Network Guru Member

    checked with my old E4200
    Setup: 2018.5 VPN build
    WLAN: set to Auto, channel 1, 20 MHz, WPA2 Personal with AES
    Connection: PPPoE and DHCPv6 PD
    DNSMASQ on for LAN (DHCP and SLAAC/DHCPv6)

    no reboots so far (working internet-connection)
     
    kernel-panic69 likes this.
  29. M_ars

    M_ars Network Guru Member

    intresting, both have Broadcom BCM4706
    @st_ty_ can you re-produce this every time?
     
  30. st_ty_

    st_ty_ Network Newbie Member

    No, I don't think so, not all Broadcom devices, but maybe only Broadcom devices using Freshtomato code of mips-RT-AC-branch. Looking into this thread I can detect a cumulation of routers, also having this "constant reboot"-effect.
    I believe these lines in dnsmasq code 'have no effect' because:
    - WNDR4500 is running now for about 2 weeks w/o any problem.
    - pxe-boot has nothing to do with routers, it is related rather to net-booted PC-machines (as you can also read in the comment of commit 503c6091...; in the dnsmasq repo (http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=commit;h=503c609149fcbde42e345116844fe55db59c233f)
    I know that there exists two different DCHP-related software in Freshtomato: udhcpc-client of busybox for DHCP on WAN side and dnsmasq for DHCP on LAN side. And I can't explain, why DHCP on LAN side in combination with a WAN connection crashes the router, but this relationship definitely exists and is 100% reproducible.
    I only wanted to report a weird effect in Freshtomato (without having an explanation for this effect) and tried to explain, how to circumvent it.
     
  31. st_ty_

    st_ty_ Network Newbie Member

    Do you mean the "constant reboot"-effect or the proposed solution? ;) Both of them, I'm working on this since August/September '18, as for the WNDR4500 no other software with actual versions of dnsmasq (>2.78) exits.
    OK, there are new versions of DD-WRT for WNDR4500, but for me they are too buggy, so we have to make this Freshtomato working on WNDR4500.
    Btw, the RT-N-version of Freshtomato for Asus RT-N66W (with older Broadcom wl- drivers) does not show this "constant reboot"-effect, only the RT-AC-version! Regards.
     
    M_ars likes this.
  32. kernel-panic69

    kernel-panic69 Connected Client Member

    All I run is from RT-AC branch. The VPN mega N5x builds are compiled from RT-AC branch for E4200v1. I've even run test builds I've compiled myself from the main FreshTomato repo source in a local repo with NO modifications from the RT-AC branch. So, it's got to be something with that specific hardware (or something in those particular builds for those devices?) IMO. The only time my FT router reboots is when I flash an upgrade or test build. The mips-master branch is only for pushing updates to packages to the RT-AC branch, as I understand it. All builds are compiled from the RT-AC branch. Maybe @pedro311 can correct me if I am mistaken.

    It sounds as if you are running an RT-AC66 build or something on your hardware to me?
     
  33. st_ty_

    st_ty_ Network Newbie Member

    E4200v1 uses the RT-N branch. IMO you cannot compare the behaviour of this machine with an router using the RT-AC-branch.
    Yes, ok, it sounds like running an RT-AC66 build. But for the Asus RT-N66W there are two different builds available (can also be seen in the Makefiles of the source code of Freshtomato): one of the RTN-branch (works flawless) and one of the RT-AC-branch (having this "constant reboot"-effect when trying to couple a WiFi-device with wireless-N-mode to the WAN-connected router).
     
  34. kernel-panic69

    kernel-panic69 Connected Client Member

    ./release/src; /release/src-rt; /release/src-rt-6.x; From the mips-RT-AC branch of the SOURCE CODE repo, not the binary firmware file download tree. The only other branch in pedro's repo is mips-master. If you look through the commits and even the online source tree, you will see 'build mips-N and AC coexistence' or some such. 'src' is the main code directory, src-rt is the 5.x driver version directory, rt-6.x is the 6.x drivers. You are using N devices, why the heck are you trying to use an AC device build, other than the fact that they are there? I'm leaning towards more of a driver issue, coupled with those lines from dnsmasq code, but it is definitely a possible bug or glitch that may fix others' problems.
     
  35. st_ty_

    st_ty_ Network Newbie Member

    Thank you very much for your reply. Seems that you are nearly the only one interested in this issue.
    To build Freshtomato for WNDR4500 you have to start making the build in/release/src-rt-6.x (!!) , even WNDR4500 is not an AC-wireless supporting machine. There is no NAND memory-type support in Broadcom 5.x-SDK. As WNDR4500 has NAND memory, you must use the Broadcom 6.x - drivers.
    For RT-N66W - not having NAND memory type - you can start makeing a build in either /src-rt (leading to an build with older 5.x Broadcom driver) or in /release/src-rt-6.x (leading to build with newer 6.x drivers). Done this about 50 times up to now. Regards.
     
  36. txnative

    txnative Addicted to LI Member

    what was your conclusion after building the wndr4500 now?
     
  37. st_ty_

    st_ty_ Network Newbie Member

    Like I said before: With commenting out these three lines in source code you an get a working build, without commenting out these lines you will result in a never ending rebooting machine (in router mode, withy trying to access by wireless-N; maybe in AP-mode this correction is not needed, I never checked this, as AP-mode is not what I need). Regards.
     
  38. M_ars

    M_ars Network Guru Member

    Nice finding and lots of work to find that problem/bug. :)
    I think commenting out the three lines you mentioned should work. We dont use the delay option and therefore we dont need the exact receive time of the upd packet. (recvtime = now; //is also there)
    I think that could be a mips specific patch (workaround for the unknown reboot problem)
    @pedro311 what do you think?

    Edit:
    maybe instead of commenting out something, use #ifdef BCMARM ?
    --> that way, stay as close as possible to arm

    Code:
    #ifdef HAVE_LINUX_NETWORK
       struct arpreq arp_req;
    /* only for Tomato MIPS (RT-AC) - workaround for unknown reboots */
    #ifdef (BCMARM)
       struct timeval tv;
    #endif //BCMARM
    #endif
    
    ...
    #if defined (HAVE_LINUX_NETWORK)
    /* only for Tomato MIPS (RT-AC) - workaround for unknown reboots
    see https://www.linksysinfo.org/index.php?threads/fork-freshtomato-mips.74145/page-8#post-301674
    */
    #ifdef (BCMARM)
    if (ioctl(fd, SIOCGSTAMP, &tv) == 0)
    recvtime = tv.tv_sec;
    #endif //BCMARM
    ...
    
     
    Last edited: Dec 30, 2018
    kernel-panic69 and kille72 like this.
  39. danielhaden

    danielhaden Network Guru Member

    WNR3500v2 (no 'l') is successful with freshtomato-K26_RT-MIPSR2-2018.5-MiniIPv6 that was small enough to fit in the 35??KB limit. Excellent!!!
     
    Last edited: Dec 30, 2018
    kernel-panic69 likes this.
  40. pedro311

    pedro311 Networkin' Nut Member

    Great catch! I will add such patch ASAP. Thanks.
     
  41. kernel-panic69

    kernel-panic69 Connected Client Member

    The wndr4500V2 and RT-N66 builds should be in the 5.x driver section of the makefile or configured to use the 5.x driver, perhaps? Both are configured to use the 6.x driver, but why? I can tell you running the 6.x driver on an N device causes all kinds of havoc save and except for the roboswitch.

    Anyhow, the patch has been applied:

    https://bitbucket.org/pedro311/freshtomato-mips/commits/bcccc155ba96d0fbdc9f007bc313e34afae5ffae

    On a side note, I haven't run a test build on the last merge yet @pedro311, got several things going on here atm.
     
    M_ars likes this.
  42. Henkie de Groot

    Henkie de Groot Reformed Router Member

    @Aladino For me the Router works as an OpenVPN client just fine. I had to force a DDNS update once though, but all works (for me) as it should.
     
  43. ripley7

    ripley7 New Member Member

    pedro311,

    Thank you for your work on this project!

    I can confirm #73 NyNe's issue. I used freshtomato-RT-AC66U_RT-AC6x-2018.5-AIO-64K.trx on an Asus RT-AC66U.
    With a fresh, unmodified install and cleared nvram, DHCP works fine for wired clients. Configuring wireless clients with a static ip address also works fine. However, when wireless clients configured with dhcp attempt to connect, the router reboots. I procured a dnsmasq binary (version 2.7.2, not 2.8) from an older version of shibby from another RT-AC66U. I stopped the dnsmasq service and ran the dnsmasq 2.7.2 binary from the command line and wireless clients were able to obtain an ip address and all worked fine, and the router did not reboot. It appears that dnsmasq is at least part of the issue. Thanks again!

    Thank you: The since- released freshtomato-mips 2019.1.015-beta appears to have fixed this issue.
     
    Last edited: Feb 7, 2019
  44. M_ars

    M_ars Network Guru Member

    I think RT-N66U user should be able to choose between 5.x and 6.x driver. Shibby added it starting with build 114
    BR
    M_ars
     
    pedro311 and kernel-panic69 like this.
  45. kernel-panic69

    kernel-panic69 Connected Client Member

    Going back to @st_ty_ 's comment on DD-WRT. The buggy problems revolve around the drivers and "I do it my way", otherwise I think even the E4200v1 wouldn't have issue with the 6.x drivers. Only way to tell for sure is to conjure up a test build for FT and test, perhaps.
     
  46. st_ty_

    st_ty_ Network Newbie Member

    Faster for testing: You can build this "test FT" by yourself.
    Check out mips-RT-AC branch from repo, do git clean -dxfq, change to /release/srt-rt-6.x folder and start with e.g. make n60z (AIO build) for Linksys E4200.
    The building process is working. At the end you will receive a firmware file called freshtomato-E4200USB-NVRAM60K_RT-AC6x-MIPSR2-2018.5-AIO.bin in /release/src-rt-6.x/image folder (building time 50min.)
    If you got a working firmware with this procedure? I don't know, I don't have any Linksys E4200.
    You still have the risk for testing and maybe risk for bricking one of your Linksys E4200 devices.
    BR
    st_ty_
     
  47. kernel-panic69

    kernel-panic69 Connected Client Member

    1) I already have mips-master and mips-RT-AC local repos - I guess you missed that part?
    2) The n60z AIO build will brick an E4200v1 because the bin file size exceeds the flash size (20+MB bin, 16MB flash). That build option needs to be removed or changed to reduce firmware size @pedro311, @kille72
    3) DD-WRT K3X builds for the E4200v1 have the 6.x drivers, hence my earlier comment regarding things getting screwy and havoc.
    4) Since FT seems a little less manipulated as far as the Broadcom source, I may work on porting the 6.x drivers over to build an n60o Mega-VPN after I make progress on other things to help the team...
     
  48. MarMed

    MarMed New Member Member

    @pedro311 I may be mistaken, but on E3200USB-NVRAM32K_RT-N5x-MIPSR2-2018.5-Mega-VPN, for the E3200 actually has 60k nvram, but this build only has 32K
     
  49. kernel-panic69

    kernel-panic69 Connected Client Member

    You are correct. The 32K configuration is for 5GHz band wi-fi stability or some such.
     
  50. pedro311

    pedro311 Networkin' Nut Member

    Exactly.
     
    txnative likes this.
  51. danielhaden

    danielhaden Network Guru Member

    I'm having a little trouble with QOS increasing lag in the 2018.5 version. It seems to stumble a bit. Any tips?
     
  52. txnative

    txnative Addicted to LI Member

    Post in Using Qos - tutorial and discussions, you may need to post a screen shot of your setting and don't have people guess at what of the the type of tips scenero you have going on.
     
  53. danielhaden

    danielhaden Network Guru Member

    Thanks.
    However, a particular setting isn't what I needed. The problem is failing downloads, so I had to disable QOS, because it is slightly broken. Fortunately, bandwidth limiter does just as well at preventing bufferbloat... without causing downloads to fail and without adding lags.
    I did turn on the multicast features. Perhaps that's what broke QOS?
     
  54. kernel-panic69

    kernel-panic69 Connected Client Member

  55. txnative

    txnative Addicted to LI Member

    I use tomato qos but I customize it to my network and device needs. Using the default settings will work but a user must understand how tomato qos works with the defaults as well some to all, a user isn't going to need all those classifications, so you'll need to do device specific as well as knowing the using dst or src port settings etc just to name a few things, you don't have to be a genius or network admin but willing to learn, bandwidth limiter does indeed help for anyone to use as long you know the proper way of setting it up, it nice that you have it tailored to your needs. Regards as it's evening here.
     
  56. danielhaden

    danielhaden Network Guru Member

    Last edited: Jan 3, 2019
  57. pedro311

    pedro311 Networkin' Nut Member

    kille72 and kernel-panic69 like this.
  58. dadaniel

    dadaniel Network Guru Member

    There is a DHCP server problem in recent versions:

    It will still send its own IP address for DNS-Server to clients - when "Use internal DNS" is disabled and dnsmasq isn't running!! @pedro311, @kille72
     
  59. kernel-panic69

    kernel-panic69 Connected Client Member

    Tomato doesn't use udhcpd that I know of, the only dhcp server is dnsmasq. You have to either use the 'no-resolv' directive in the custom config with 'server=x.x.x.x' lines below it, or check 'use received dns'. Otherwise, that is default behavior AFAIK.
     
  60. MongooseProXC

    MongooseProXC Connected Client Member

    kille72 and EasyIsEZ like this.
  61. EasyIsEZ

    EasyIsEZ Serious Server Member

    Same situation here when I tested with the 2018.4 build on my RT-AC66U router. Assuming the configuration in NVRAM would migrate seamlessly, I updated, the router came back but wireless was terrible. :( It switched back and forth between on-line/off-line with the power led on the router going on and off at the same time. Hard resetting using the button on the back didn't work so eventually, I connected a cable, reapplied the firmware while ticking the box to clear the NVRAM after the upgrade. The box came up with the default Tomato24/50 networks but even after manually configuring it, it wasn't smooth and wireless connectivity dropped frequently, especially after connecting it back to the WAN. Whenever the WAN was connected, the behavior appeared to worsen with more disconnects.

    Since multiple people with RT-AC66Us experience the same issues, this seems to require further investigations. If desired, I can give the firmware another go and collect some data to troubleshoot it. Alternatively, perhaps somebody already knows what is causing the trouble and can shine some light on this shady subject. :)
     
  62. EasyIsEZ

    EasyIsEZ Serious Server Member

    Sounds like you nailed it. Thanks for researching this as this is killing my Asus RT-AC66u as well.

     
  63. EasyIsEZ

    EasyIsEZ Serious Server Member

    Yes, but by now you also have a fix for the issue. I'm looking forward to deploying the first 2019 build of FreshTomato!

     
  64. ddimitrov

    ddimitrov Network Newbie Member

    Hello guys,

    I installed FreshTomato Firmware 2018.5 MIPSR2 K26 USB Mega-VPN on my Linksys E3200.
    When I enable both AdBlock and Tor modules, CPU Load jumps to sky (reaching over 4.00 in a couple of minutes, going upwards, while there is only 1 CPU core), which causes the router to hang or Tor to autoclose (e.g. the Tor process to stop and disappear). The configurations of AdBlock and Tor modules do not matter. The problem happens with arbitrary configurations - with default configurations, with an empty list of AdBlock Blacklist URLs, etc.

    If I enable AdBlock only, or if I enable Tor only, everything runs fine (CPU Load stays around 0.00 on idle).

    I even tried with a fresh new installation of FreshTomato, after 30/30/30 cleanup, but the problem happened again.

    I have successfully run such combination (e.g. both AdBlock and Tor enabled) for several years on the same router with Shibby Tomato builds (builds 138 and 140).

    Has anybody experienced such problem with FreshTomato?

    Thanks.
     
  65. pedro311

    pedro311 Networkin' Nut Member

    I hope that was caused by dnsmasq (on some routers/configurations even reboots), and is fixed here: https://bitbucket.org/pedro311/freshtomato-mips/commits/bcccc155ba96d0fbdc9f007bc313e34afae5ffae
    We'll see...
     
    M_ars and kille72 like this.
  66. st_ty_

    st_ty_ Network Newbie Member

    Can confirm same behaviour with NETGEAR R6300v1 (bought this week at ebay and just tested). Workaround suggested for RT-N66W and WNDR4500v2 (commit bcccc15) does its job with this router as well !! :)
    Btw, stdout of building process of FT nearly always shows some strange comment lines ("x out of y hunk FAILED")when applying patches for the features in /src/router directory, e.g. for ppd:
    "…
    find patches/pppd -maxdepth 1 -type f -name '*.patch' | sort -t '\0' -n | while read FILE; do ( if ! patch -p0 -Rf --dry-run --silent < $FILE 2>/dev/null; then patch -p0 < $FILE; fi ) done
    1 out of 1 hunk FAILED
    patching file pppd/pppd/Makefile.linux
    …"
    Regular observations?
    BR
    st_ty_
     
  67. pedro311

    pedro311 Networkin' Nut Member

    It's expected bahavior ;)
     
  68. NineEyes

    NineEyes Reformed Router Member

    Here's my experience with the Asus RT-AC66U "boot loop"... I have 3 of these routers on my network - two which I believe are rev A1 and one which I believe is rev A2. All were running Shibby 138 AIO for years. My first two updates to freshtomato-RT-AC66U_RT-AC6x-2018.5-AIO-64K (erasing NVRAM before and during update through the Web UI) went fine. The remaining rev A1 however could not stay up for more than a minute with freshtomato-RT-AC66U_RT-AC6x-2018.5-AIO-64K. I tried installing multiple times through the Web UI and via the Asus Firmware Updater and the errant behavior didn't change. The reboots occurred even with no cables connected (but power). (I can't say with 100% certainty my laptop wasn't auto-connecting wirelessly to "Tomato24" :eek:). This lone rev A1 is back running Shibby 138 AIO for now.

    The moral of the story is you may need to test more than 2 or 3 RT-AC66Us before declaring victory with any fix or workaround. If there are any betas that need testing, I'll do my part.

    Thank you devs for pushing forward this important project! I'll be watching for your updates.
     
    ddimitrov likes this.
  69. golf247

    golf247 Network Newbie Member

    Not sure whether this will help but came across your note and wanted to give you some ideas based on some recent N12D1 work I did in case you were still stuck.

    Recovery for N10 and N12: set manual IP at 192.168.1.15, Hold reset, plug in power. Power light starts blinking slowly after 12sec.

    WPS button 30/30/30 is NVRAM clear.

    TFTP method: (FYI in win10, have to turn TFTP client on) Start CMD in folder that holds new firmware file. get ready to CMD "tftp -i 192.168.1.1 put newN10p.trx", go into recovery mode,hit enter. Wait about 30 sec and it will complete. Wait another 5 mins to allow firmware to install (no indications on leds) and then power off/on. Let router do it's boot up for a few mins. Then power off/on and do WPS NVRAM clear.
     
  70. Chrushev

    Chrushev Network Newbie Member

    Tried switching from Shibby's to 2018.5 build. Netgear 6300v1 loads and boots fine at first. but as soon as you try to change the non default SSID (Tomato24) to something else on 2.4Ghz WiFi it goes into an infinite reboot loop. Only way to recover is reset through physical button.

    Shibby's latest build worked flawlessly with this router.

    EDIT: Tried 2018.4, same issue. So FreshTomato is not usable on NetGear R6300v1 even though there are builds for it.

    My steps are this:

    1) Load 2018.5
    2) Do complete NVRAM reset via Config Menu dropdown
    3) Do a reboot (otherwise only 2.4 Ghz) shows up, after reboot via GUI both show up
    4) Change the SSID on 2.4Ghz to anything else and click Save.
    5) After about 30 seconds the router will go into infinite reboot cycle.
     
    Last edited: Jan 7, 2019
  71. pedro311

    pedro311 Networkin' Nut Member

  72. Chrushev

    Chrushev Network Newbie Member

  73. pedro311

    pedro311 Networkin' Nut Member

    I'll release 2019.1 beta this week, so be patient.
     
    M_ars, kernel-panic69 and Chrushev like this.
  74. Chrushev

    Chrushev Network Newbie Member

    Just compiled off of latest commit. Can confirm issue is resolved.
     
    M_ars and kernel-panic69 like this.
  75. danielhaden

    danielhaden Network Guru Member

    WRT320N v1, working perfectly (as 5ghz AP) with freshtomato-K26_RT-MIPSR2-2018.5.083-beta-MiniIPv6.bin
    At first it limped, but I nvram set clkfreq=400,200,100 and turned off extra monitor services (why default on?).
    Excellent 5ghz! Performance Far Exceeds Expectations.
     
    Last edited: Jan 15, 2019
  76. danielhaden

    danielhaden Network Guru Member

    freshtomato-E4200USB-NVRAM60K_RT-N5x-MIPSR2-2018.5.083-beta-VPN.bin has Broken QOS, because gui forbids typing port range with more than 5 characters. I wanted to type 1-65535, got 1-655 clang clang clang clang clang.
     
  77. pedro311

    pedro311 Networkin' Nut Member

    So use 2018.5 ;)
     
    rs232 likes this.
  78. danielhaden

    danielhaden Network Guru Member

    Thanks. I put that version on. QOS dst port shows 0 on graphs.

    Edit: Found out this happened due to conflicting classifications/rules. The graph itself works. The default rules, not as helpful as expected, because far too many.
     
    Last edited: Jan 15, 2019
  79. Chrushev

    Chrushev Network Newbie Member

    Hadnt had much time last night to look at the build I made, looking at it now.

    2 questions:

    1) Whats with weird port forwarding entries under Port Forwarding options?
    2) Why is Wifi on and unsecured by default? Shibby's and ddwrt builds always generate a random password. If desire is to keep unsecured wifi then disable the radios by default.
     
    Last edited: Jan 8, 2019
  80. kernel-panic69

    kernel-panic69 Connected Client Member

    Which build of DD-WRT generates a wifi password by default? Must be a <Kong> build, because that is not default behavior for any BS builds that I know of, unless it's a 'Netgear thing'.....
     
  81. pedro311

    pedro311 Networkin' Nut Member

    What do you mean? Where?
     
  82. danielhaden

    danielhaden Network Guru Member

    The /qos-graphs.asp page.

    P.S.
    For 2019 version it would be nice to delete the huge number of mostly outdated classifications and put just 10 classifications (one classification per each QOS class). Then the graphs page makes sense so that everyone could really see what the QOS is doing.
     
    Last edited: Jan 8, 2019
    MongooseProXC and Elfew like this.
  83. pedro311

    pedro311 Networkin' Nut Member

    pharma, M_ars, kille72 and 2 others like this.
  84. NineEyes

    NineEyes Reformed Router Member

    I put Kong's r37495 on a Netgear R7800 yesterday and the WiFi came up secured and the passwords (2.4 and 5GHz) were equal to the password printed on the bottom of the modem ("excotictomato901" for me). I'm not sure if this password is common or if it was read from ROM in the Atheros chip.

    Using default radio and OoS settings and using Android "WiFi Speed Test" at 5GHz, it seems my Asus RT-AC66U APs running freshtomato-RT-AC66U_RT-AC6x-2018.5-AIO-64K are 30-50% faster than the R7800 running DD-WRT. Go Tomato!
     
    danielhaden likes this.
  85. Radojevic

    Radojevic Network Newbie Member

    I never ran across a Tomato Shibby, nor DDWRT version that generated a wifi password by default.
    The last Tomato Shibby verion I used was:
    tomato-K26USB-1.28.RT-N5x-MIPSR2-140-AIO-64K
     
    kille72 likes this.
  86. Chrushev

    Chrushev Network Newbie Member

    Can you explain why this forwarding is in the defaults?

    Code:
    { "portforward",        "0<3<1.1.1.0/24<1000:2000<<192.168.1.2<ex: 1000 to 2000, restricted>0<2<<1000,2000<<192.168.1.2<ex: 1000 and 2000>0<1<<1000<2000<192.168.1.2<ex: different internal port>0<3<<1000:2000,3000<<192.168.1.2<ex: 1000 to 2000, and 3000>" },
    { "trigforward",        "0<1<3000:4000<5000:6000<ex: open 5000-6000 if 3000-4000>"    },
     
  87. pedro311

    pedro311 Networkin' Nut Member

    Look closer, it's inactive, just an example.
     
    kille72 likes this.
  88. Elfew

    Elfew Network Guru Member

    I agree, is it possible?
     
  89. ddimitrov

    ddimitrov Network Newbie Member

    Anything about my problem with AdBlock + Tor in Linksys E3200?
    Nobody?
     
  90. pedro311

    pedro311 Networkin' Nut Member

    Yes, it will be changed, to more "modern" one rules.
     
  91. danielhaden

    danielhaden Network Guru Member

    Proposal for new default QOS: Auto-QOS--less specifying makes it fit more uses.
    Instead of 50 rules, here's 5

    Classification/rules
    tcp/udp, port 53, 0-10kb, class 1, dns
    tcp/udp, src port 1-65535, 0-1kb, class 1, fastlittle
    tcp/udp, dst port 80,443,8080, 0-512kb, class 4, httpsmall
    tcp/udp, dst port 80,443,8080, 512kb+, class 8, httplarge
    tcp/udp, dst port 1-65535, class 10, bulk

    classes
    1,16%,100%
    2,12%,100%
    3, 8%,100%
    4, 6%,100%
    5, 4%,100%
    6, 3%,100%
    7, 2%,100%
    8, 1%, 96%
    9, 1%, 96% default
    10,1%, 96%

    Today, people expect QOS to go faster, so this does! Confirmed at Netflix's fast.com. This auto-QOS on Tomato is faster than Gargoyle's modern QOS, and tomato doesn't scramble internet radio streams. Seems like an almost epic boost. Also, the graphs started working. Yay! Solid win! Seems good for defaults in Freshtomato 2019.
    P.S.
    Skype, Sip, and VOIP phones are ports 3478-3481,1718-1720,5060-5070 however, I didn't include that in the example, because most people have cell phones. Examples and old references like that probably should go in a 'Notes: Click To Expand, examples' section at the bottom of the page (like several other pages in the GUI).
     
    Last edited: Jan 9, 2019
    kernel-panic69 likes this.
  92. kernel-panic69

    kernel-panic69 Connected Client Member

    Atheros is not Broadcom, Broadcom is not Atheros. FT is Broadcom-only. BUT, I do believe Netgear Broadcom devices may have the default password on the sticker, although I would have to verify that.
     
  93. danielhaden

    danielhaden Network Guru Member

    Still testing, but this time with wee little old E2000@400MHz and 5.8ghz-only use, as main router, and the proposed new QOS defaults. Wow! That thing was punching way above its class.
    Gargoyle 1.11 on Archer c7v2 defeated during simulcast (internet radios cracked up).
    DD-WRT on newer hardware, defeated during simulcast (internet radios cracked up).
    Freshtomato running the proposed QOS default changes, didn't fault any connections. The proposed new QOS defaults (my prior post) seem to be working nicely.
    For now, the takeaway should be that 5 rules performs better than 50; and, a cleanup effort is necessary.
     
    Last edited: Jan 15, 2019
    kernel-panic69 likes this.
  94. kernel-panic69

    kernel-panic69 Connected Client Member

    There have been a lot of commits between 2018.5 and the upcoming beta. May wish to try and wait for the upcoming beta release to see if it's fixed. Personally, I use uBlock Origin on my clients and I haven't compiled any test builds since the last release as I have been busy building patches to update busybox to the latest version.

    @danielhaden Do I know you by another handle on another forum, perhaps?
     
    danielhaden likes this.
  95. danielhaden

    danielhaden Network Guru Member

    Oh yeah, since 2003, although the tux-cat avatar is a newer upgrade. Meanwhile, back to updating the Tomato QOS defaults so that it can compete.
    Just so that I'm not annoying without enjoying it, here again are the proposed new QOS defaults for 2019 Freshtomato:
    Classification/rules (five is faster than fifty)
    tcp/udp, port 53, 0-10kb, class 1, dns
    tcp/udp, src port 1-65535, 0-1kb, class 1, fastlittle
    tcp/udp, dst port 80,443,8080, 0-512kb, class 4, httpsmall
    tcp/udp, dst port 80,443,8080, 512kb+, class 8, httplarge
    tcp/udp, dst port 1-65535, class 10, bulk

    classes
    1,16%,100%
    2,12%,100%
    3, 8%,100%
    4, 6%,100%
    5, 4%,100%
    6, 3%,100%
    7, 2%,100%
    8, 1%, 96%
    9, 1%, 96% default
    10,1%, 96%

    Might as well enjoy it.
     
    Last edited: Jan 9, 2019
    kernel-panic69 likes this.
  96. ghoffman

    ghoffman LI Guru Member

    these look good.

    how would these compare to a cake/sqm engine alone ?
     
  97. MongooseProXC

    MongooseProXC Connected Client Member

    This should save a lot of NVRam too! One of the first things I do is delete these rules to open up some space for that 32k limit.
     
  98. the_tourist

    the_tourist Network Newbie Member

    Hi,

    Although we have 2 cell phones, we also have two VOIP home lines, which are our main telephone lines and by far the most used here. We also use Skype a lot for our video communications.

    In fact, my expectations regarding QOS are, first and foremost, that it gives priority to VOIP and Skype; for these communications do not suffer any disruption (or as little as possible) caused by our other Internet uses.

    I would therefore find it essential that the new QOS rules include these priorities (VOIP and Skype) or at least offer a configuration option that is very easy to set up (for non-experts).
     
  99. txnative

    txnative Addicted to LI Member

    I feel that it is a waste of pedros time to do such a task for Qos. Users can take examples as shown here and throughout tomato forum that users have taken the time to show examples also, some tutorials. New users can be directed when needed but as in this topic alone users have shown that deleting the default rules and creating ones in place will achieve optimum connections for their network, but what works for them may not work for that new user and will find themselves in the same place as to being confused stating that Qos doesn't work. If you want it easy buy a router supported by openwrt were cake and piece of cake is turn it on place your inbound, outbound numbers in and that's it, but in some cases things aren't as easy as one would might like to think when it comes to setting and maintaining a home network.
     
    kille72 likes this.
  100. kernel-panic69

    kernel-panic69 Connected Client Member

    The problem is that Broadcom dual- and multi-band routers and 5Ghz band functionality on OpenWRT is a crap shoot, if you can even get a workaround functional at all. And, I am pretty sure that most likely, Broadcom-MIPS is probably going to have little to zero possibility there (I never got 5Ghz working on E4200v1 out of the box). OpenWRT doesn't have any proprietary Broadcom code whatsoever, it is completely open source, which is both good and bad.
     
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice