1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Memory Leak

Discussion in 'Tomato Firmware' started by Riddlah, Aug 14, 2012.

  1. Dark_Shadow

    Dark_Shadow Addicted to LI Member

    Let me know if you need other testers.
     
  2. shibby20

    shibby20 Network Guru Member

    @jya i asked @guardian to check my RT + fix (pptp.c and pppl2tp.c) on RT-N16
    http://update.groov.pl/ipv6/tomato-K26USB-1.28.RT-MIPSR2-100.8-AIO.trx

    answer:
    Conclusion:
    RT + my fix on RT-N16 works correct. No leak, L2TP works
    RT-N + my fix on RT-N66U works correct. No leak, L2TP untested.

    Can you tell us in which file(s) did you find a problem with leak?
    I think i should push my changes to GIT. Maybe we have two leaks? What do you think?

    @guardian if you can please check this image:
    http://update.groov.pl/ipv6/tomato-K26USB-1.28.RT-N5x-MIPSR2-100.7-Mega-VPN-64K.trx

    on RT-N66U +64k and let US know is L2TP works. Thank You.
     
  3. jakey

    jakey Networkin' Nut Member

    Code:
    ASUS RT-N16 Firmware Version 3.0.0.3.162
    Fix bugs:
    8. Sonos audio system and network map compatibility issue. Special thanks to Merlin for his contribution.
    
    @Shibby, slightly off topic but can please include the above fix for the RT-N66 as I'm having problems with my Sonos kit.

    Thanks
     
  4. jyavenard

    jyavenard Network Guru Member

    Shibby, there's no way that you're IPv6 patch (which only touch l2tp and pptp) fix the 6in4 leak.

    I tested your patch myself and it leaks just the same as without.

    Obviously this is with the patch you provided... if you made other changes, can't say for sure..
     
  5. Dark_Shadow

    Dark_Shadow Addicted to LI Member

    Running shibby's Tomato Firmware 1.28.0000 MIPSR2-100.7 K26 USB AIO-64K

    Wan uptime 15hrs+ with 6to4 tunnel at HE on RT-N66U, No Leak
     
  6. Cyberian75

    Cyberian75 Network Guru Member

    Same here with Shadow.
     
  7. jyavenard

    jyavenard Network Guru Member

    Ok.

    RT-N66 with 64K patch:
    http://www.avenard.org/wrt54-tomato...501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT-64K.trx

    RT-N16 and other 32k nvram routers:
    http://www.avenard.org/wrt54-tomato...28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.trx

    The bug was in linux_osl.[ch] kernel module.

    Calling for testers...
    Cheers
    Jean-Yves

    Shibby: I made a mistake when I pushed the change to git, I had merged your change to radvd from tomato-shibby-RT to tomato-shibby-RT-N and had forgotten about it, and it ended up in your branch.. I reverted that modification as soon as I noticed. Sorry
     
    kthaddock likes this.
  8. pharma

    pharma Network Guru Member

    @jya,

    Would like to test, but not sure whether can use the "RT-N16 and others" firmware above on E4200 with 60K. I have flashed .trx's before without issue. If not can wait till issue is identified and fixed for all supported routers.

    Pharma
     
  9. jyavenard

    jyavenard Network Guru Member

    let me build an image for the E4200 ... The RT-N16 and others are for 32k nvram router
     
  10. jyavenard

    jyavenard Network Guru Member

    it's uploading now:
    there:
    http://www.avenard.org/wrt54-tomato/tomatoUSB/
    you'll find:
    tomato-E1550USB-NVRAM60K-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.bin
    tomato-E2000USB-NVRAM60K-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.bin
    tomato-E2500USB-NVRAM60K-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.bin
    tomato-E3000USB-NVRAM60K-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.bin
    tomato-E3200USB-NVRAM60K-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.bin
    tomato-E4200USB-NVRAM60K-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.bin
    tomato-K26USB-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT-64K.trx
    tomato-K26USB-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.trx
    tomato-K26USB-NVRAM60K-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT.trx

    version 1.28.0501.11 is the one you want
     
  11. Cyberian75

    Cyberian75 Network Guru Member

  12. maurer

    maurer LI Guru Member

    RT-N16 with build 100 AIO - uptime 17 days;
    DHCPv6 with prefix delegation (native dynamic IPv6 from ISP)
    working as AP behind openwrt 1043nd
     
  13. jyavenard

    jyavenard Network Guru Member

    Native IPv6 didn't expose any memory leak issues, so not surprising that you haven't had problems for so long.
     
  14. guardian

    guardian Serious Server Member

    jya:

    have tested tomato-K26USB-1.28.0501.11-MIPSR2-jya-RT-N-VLAN-VPN-NOCAT-64K.trx on the Asus RT-N66 router.

    L2TP connection and 6in4 tunnel work ok, no memory leak. But I've experienced router hangs after nearly each "save" button clicking in Basic menu. After clicking "save" router still responds to ping, but no ssh/http access is possible, and no wan connection. After power reset it boots and works ok, having settings saved. I had to power reset it several times to save all changes in submenus of "Basic". It's possible that there's some kind of bug in nvram saving/commiting/services restarting code.

    Same behaviour is with Shibby's latest testing build.
     
  15. jyavenard

    jyavenard Network Guru Member

    weird.. I'm not experiencing that at all on my RT-N66...
    it's probably unrelated to the memory leak fix.

    going back to earlier build, are you still seeing that?
     
  16. guardian

    guardian Serious Server Member

    I have a suspect that hangs may be somehow related to the fact that I use jumbo frames on my network. That gives me additional speed for my Synology NAS access. Can the switch be somehow reset to "non-jumbo" MTU after clicking "save" on the router's control panel?

    Edit: no, it isn't related to jumbo frames. It looks like it starts to "hang" on saving after filling (and saving) the form at Basic/IPv6. Looks like with IPv6 enabled the restart of the services messes something up so services like http and ssh become unaccessible.

    Add: Toastman's 500.2 is ok, no hangs.
     
  17. Toastman

    Toastman Super Moderator Staff Member Member

    The fix from jyavenard is now in Toastman RT-N 1.28.0500.5 issue

    Thanks JY, this was a lot of work. Appreciated.
     
  18. Dark_Shadow

    Dark_Shadow Addicted to LI Member

    Flashed shibby's Tomato Firmware 1.28.0000 MIPSR2-101 K26 USB Mega-VPN-64K this morning, 6 hrs uptime all good
     
  19. USNetboy

    USNetboy Networkin' Nut Member

    Cherry-picked @jya's fix and the radvd upgrade into the Toastman-RT-N build that was leaking for me on RT-N66U (500.4). No other changes were pulled into this build. It's been running solid for the past 4 hours. Not experiencing any problems with 6in4 traffic or the UI. @jya, seems like you nailed it. Amazing job!
     
  20. pharma

    pharma Network Guru Member

    Does this include the newer Broadcom driver? Just asked because for some reason the wireless signal strength is not as strong when compared to .0500.2 with wireless setting exactly the same.
     
  21. kthaddock

    kthaddock Network Guru Member

    Yes, you can check with
     
  22. shibby20

    shibby20 Network Guru Member

    v101 on RT-N66U and 12h uptime without leak.
     
  23. Cyberian75

    Cyberian75 Network Guru Member

    Same here.
     
  24. pharma

    pharma Network Guru Member

    Thank you! The version in .0500.2 is

    5.100 RC138.9
    wl0: Aug 9 2011 09:29:42 version 5.100.138.9
     
  25. Gitsum

    Gitsum LI Guru Member

    Where is version 101?
     
  26. Cyberian75

    Cyberian75 Network Guru Member

  27. Gitsum

    Gitsum LI Guru Member

  28. kthaddock

    kthaddock Network Guru Member

    Are you sure about that ? RT- suppose to have 5.100.138.9 and RT-N 5.100.138.20 same as shibbys builds.
    I'm using 1.28.0501.11 - jya-VLAN-RT-N-VPN-NOCAT and that use 5.100.138.2

    Maby Toastman can tell what driver is in his build this time?
     
  29. jyavenard

    jyavenard Network Guru Member

    Nah, RT us using much older drivers. .5.100.138.9 was used in RT-N just a few revisions ago.
     
  30. shibby20

    shibby20 Network Guru Member

    Because this is only for testers: ) I'm compiling at the moment all builds and planing to release new version tomorrow
     
  31. kthaddock

    kthaddock Network Guru Member

  32. pharma

    pharma Network Guru Member

    Yes, I just checked and the .0500.2 version has the 5.100.138.9 Broadcom version. I'll try to play with settings with newer driver in .0500.5, but so far wireless signal strength is much stronger with older driver.
     
  33. kthaddock

    kthaddock Network Guru Member

    Okey I don't have experience. Have you done ?
    I get:
     
  34. pharma

    pharma Network Guru Member

    When I execute "nvram export --set | grep xpwr=" I get:

    Old Drivers (.0500.2)
    nvram set wl1_txpwr="59"
    nvram set wl0_txpwr="59"
    nvram set wl_txpwr="42"
    nvram set pa0maxpwr="400"

    New Drivers (.0500.5)
    nvram set wl1_txpwr="59"
    nvram set wl0_txpwr="59"
    nvram set wl_txpwr="42"
    nvram set pa0maxpwr="400"

    It may be how my router reacts to newer broadcom, not sure. But an example is for the wireless printer signal strength I get RSSI -13dBm with Quality at 86, and with older drivers the RSSI was -4 and Quality 96. I took both readings today before I upgraded firmware to Toastman's .0500.5. Wireless/Advanced settings are identical. If no one else (especially with E4200) notices any difference in signal strength than it might be my router.
     
  35. kthaddock

    kthaddock Network Guru Member

    Okey, with that value (86) I think routers rx will be saturated. Qualtity over 50 is not usefull.
    In your case I should lower power so qualitiy is 40-50.
     
  36. pharma

    pharma Network Guru Member

    But my issue is not router being saturated, but signal strength with different broadcom drivers. With 1 router, same (identical) wireless settings, different firmware (.0500.2 vs .0500.5) ... the signal strength should be close or almost the same to printer (fixed distance from router to printer). But I find alot of difference in signal strength --- .0500.2 (5.100.138.9) seems to give stronger wireless signal for some reason with my E4200v1 router. I'm not sure, but thought I saw a post that mentioned Teddy Bear did some manually changes to driver because it was so hard to get it to work at that time with dual band routers. Not sure if this would have made a difference and I naturally want to use newer drivers, but for now will keep using .0500.2.

    Edit: After flashing back to .0500.2 about 10 minutes ago, I'm getting RSSI -2, Quality 98 for the wireless signal strength from router to printer. I have Tx set to 59 (for many years now) with no problem. Anyway, maybe the next time the broadcom driver is changed I'll try again.

    nvram set wl1_txpwr="59"
    nvram set wl0_txpwr="59"
    nvram set wl_txpwr="42"
    nvram set pa0maxpwr="400"
     
  37. kthaddock

    kthaddock Network Guru Member

    You can test where saturated occurs. lower your signal to a minimum.
    Then rise signal and measure where signal not rise anymore, then you have your saturrated point
     
  38. gingernut

    gingernut LI Guru Member

    The newer drivers give lower output power compared to the version that Teddy Bear updated last year.

    FWIW I built from RMerlins source a bin for my RT-N66U to test, using the drivers for the RT-AC66U which are even more recent, and the power output was terrible.
     
  39. kthaddock

    kthaddock Network Guru Member

    That can depend on different tx power table, only right thing is to measure power at the antenna contact.
    Make sure that is the same power output when testing.

    Have you built any RT-N16 whith RT-N66U driver ? That would be intresting to test !
     
  40. kthaddock

    kthaddock Network Guru Member

     
  41. gingernut

    gingernut LI Guru Member

    I'll build an image for the the RT-N16 from RMerlin's source using the newer 6.30.39.29 Broadcom driver to test.

    Asus just released an update for the RT-AC66U with a more recent driver, v6.30.39.31, but only the image, no source yet available.

    There seemed to be alot of user problem reports using the previous version.
     
  42. kthaddock

    kthaddock Network Guru Member

    Nice is it possible to get one to test :D Pm me pls
     
  43. shibby20

    shibby20 Network Guru Member

    IMO Leave new driver for new routers.
     
  44. jyavenard

    jyavenard Network Guru Member

    The new driver is only used by Asus when you build for the rt-66ac. It's not used for the n16 and n66
     
  45. kthaddock

    kthaddock Network Guru Member

    yes I know that, it's just for fun to test. I don't expect any miracle.
     
  46. gingernut

    gingernut LI Guru Member

    If you compile from the src-rt-6x directory you can build an image for the n16/n66 using the newer drivers.
     
  47. pharma

    pharma Network Guru Member

    Is the solution to increase the Tx power when using the newer driver? Or doesn't matter and will be the same result?

    Thanks,
    Pharma
     
  48. gingernut

    gingernut LI Guru Member

    Same result I'm afraid, uping output power makes no difference, as if it dosen't work.
     

Share This Page