Sporadic Reset caused by efusA9 onboard Watchdog

    Hello,


    I am observing occasional resets of the EFUSA9 Solo CPU Board. Occasional means: after 3 hours, after 1 day and after nearly 2 weeks.


    I have been observing the serial debug output. There is no error or crash printed. The output from the rebooting board reveals the reason for the restart: WDOG or Watchdog


    How can i find out what causes the "WDOG" to reset the board? My 5 V supply (up to 3 A) measures 4.983 V, the supply of the TFT display measures 3.232 V.


    I am using the NSPI, NI2C, Ethernet and a LVDS attached display NEC NL10276BC13-01C which is powered by the EFUSA9.



    Serial Output:


    UFN: Version 1.1, ActiveKey = Drivers\Active\74
    INFO:OALLogSetZones: dpCurSettings.ulZoneMask: 0xb


    Microsoft Windows CE Bootloader Common Library Version 1.2 Built Oct 9 2015 17:44:45
    Microsoft Windows CE Bootloader for efusA9 Built Oct 9 2015
    Portions copyright (c) 2012 F&S Elektronik Systeme GmbH
    Boot Loader, Version 1.20
    NBoot, Version VN25
    HW rev. 1.10


    HW-Watchdog: ON
    System ready!
    Preparing for download...
    Press >S< to step into monitor...
    AUTO BOOT enabled
    +ReadKernelRegionFromNandFlash
    Image Signature in Flash Memory found (dwSig=0x43454345)
    TOC pointer=0x80491A48


    ROMHDR (cTOC = 0x00271a48) ---------------------
    DLL First : 0x4001efd7
    DLL Last : 0x4013f000
    Physical First : 0x80220000
    Physical Last : 0x804fa518
    Num Modules : 32
    RAM Start : 0x80620000
    RAM Free : 0x8066b000
    RAM End : 0x90620000
    Num Copy Entries : 2
    Copy Entries Offset : 0x802bdfe0
    Prof Symbol Length : 0x00000000
    Prof Symbol Offset : 0x00000000
    Num Files : 8
    Kernel Flags : 0x00000001
    FileSys RAM Percent : 0x20202020
    Driver Glob Start : 0x00000000
    Driver Glob Length : 0x00000000
    CPU : 0x01c4
    MiscFlags : 0x0002
    Extensions : 0x80221f00
    Tracking Mem Start : 0x00000000
    Tracking Mem Length : 0x00000000
    Kernel (2921kB) read from flash disk started finished in 1000 milliseconds
    Kernel read from NAND
    INFO: OEMLaunch: Jumping to Physical Address 0x10220000h (Virtual Address 0x10220000h)...


    Jumping to Kernel @ 0x10220000
    INFO:OALLogSetZones: dpCurSettings.ulZoneMask: 0xb
    Windows CE Kernel for ARM (Thumb Enabled)
    Enabling L2 cache
    OEMInit: silicon rev = 0x12
    SMP support disabled


    efusA9 V1.60 - Firmware Init
    Copyright (c) 2013 F&S Elektronik Systeme GmbH
    Build: Oct 16 2015/20:01:55
    [OAL] MACB: Disabled
    [OAL] RestartReason: WDOG
    FSPART: FS partition driver loaded
    BINFS: RegisterVolume - Mounted volume '\BINFS'
    PM-NETDCU: STARTED
    BE2: Version 1.4, ActiveKey = Drivers\Active\01
    BE2: Version 1.4, ActiveKey = Drivers\Active\04
    IPU: Version 1.3, ActiveKey = Drivers\Active\10
    HCD: Version 1.1, ActiveKey = Drivers\Active\17
    PP: Version 1.1, ActiveKey = Drivers\Active\19
    VDI: Version 1.1, ActiveKey = Drivers\Active\20
    ENET: Version 1.3, ActiveKey = Comm\ETHNETA
    BE2: Version 1.4, ActiveKey = Drivers\Active\34
    NI2C: Version 0.1, ActiveKey = Drivers\Active\35
    NI2C: Version 0.1, ActiveKey = Drivers\Active\36
    NI2C: Version 0.1, ActiveKey = Drivers\Active\37
    Serial: Version 1.4, ActiveKey =
    Serial: Version 1.4, ActiveKey =
    Serial: Port disabled. Serial debug is on !
    Serial: Version 1.4, ActiveKey =
    SHC: Version 1.2, ActiveKey = Drivers\Active\42
    SHC: Version 1.2, ActiveKey = Drivers\Active\43
    SHC: Version 1.2, ActiveKey = Drivers\Active\44
    CID: Version 2.5, ActiveKey = Drivers\Active\45
    CID: Version 2.5, ActiveKey = Drivers\Active\46
    NSPI: Version 3.7, ActiveKey = Drivers\Active\47
    NSPI: Version 3.7, ActiveKey = Drivers\Active\48
    NSPI: Version 3.7, ActiveKey = Drivers\Active\49
    PWM: Version 1.3, ActiveKey = Drivers\Active\50
    EXTRTC: Version 1.2, ActiveKey = Drivers\Active\51
    SHC: [USDHC3] SD card inserted
    PPU: Version 1.1, ActiveKey = Drivers\Active\53
    BCS: Version 1.4, ActiveKey = Drivers\Active\54
    DIO: Version 1.4, ActiveKey = Drivers\Active\55
    FRW: Version 1.3, ActiveKey = Drivers\Active\58
    GALCORE 4.6.9(9754) (Aug 13 2014 10:51:18)
    Major GPU: SysIntr=34 MemBases=0x130000 MMU Version=0
    2D GPU: SysIntr=35 MemBases=0x134000 MMU Version=0
    Video memory: BaseAddress=0x0 PhysBase=0x106cc000 size=0x7000000 physSize=0x0
    LCD: Version 1.0, ActiveKey = Drivers\Display\LCD
    LCD: Read registry settings from Drivers\Display\LCD
    LCD: Read registry settings from Drivers\Display\LCD
    LCD: Display-Mode 100, Name NEC NL10276BC13
    TchProxy: touch driver cann't be loaded. Check touch driver registry settings.
    NDCUCFG V 61 started. Platform: efusA9
    NDCUCFG Open COM1: at 115200 Baud
    CreateFile() failed -> ERROR COM1:
    CheckAutoStart: Version 1.7, LaunchNum = 100
    SoftRTC disabled


    Hello,


    i am just do some test because of your other thread concering the memory leak. Until now i had no sporadic WDOG reboot seen at all.


    Did you deploy any thread with higher priorty then WDOG (<101)? Default priority of I2C and SPI are >=101 and should cause no reboot.


    PS: My colleagues confirmed that this is a known problem under EC2013/V1.60 they just work on a solution.

    F&S Elektronik Systeme GmbH
    As this is an international forum, please try to post in English.
    Da dies ein internationales Forum ist, bitten wir darum, Beiträge möglichst in Englisch zu verfassen.

    Edited once, last by fs-support_ZU ().

    I am using 3 threads with priorities 255 (with quantum 2), 252 (with quantum 5) and 250 (with quantum 0).


    The resets occured after several days of continous operation, up to more than a week. The one I monitored with the serial debug port resetted after about 30 hours.

    Hello,


    sorry i edit the thread above "My colleagues confirmed that this is a known problem under EC2013/V1.60 they just work on a solution".
    We hope to found the solution in short terms.

    F&S Elektronik Systeme GmbH
    As this is an international forum, please try to post in English.
    Da dies ein internationales Forum ist, bitten wir darum, Beiträge möglichst in Englisch zu verfassen.

    Hello,


    thank you for the Information.


    Is there a way to disable the watchdog? I do not need it, we have our own watchdog on the carrier board.

    I provide the following Information as it might help your colleagues finding a solution:


    After one of my boards was reset by WDOG, the board behaved strange: Yesterday in the evening the USB Keyboard was working but, today in the morning it didn't work anymore. I left the Board powered up over the night. I reset the board with the reset button (connected to RESET IN). My custom application started up as usual. The Serial Debug Port shows the following messages upon every access of the I2C B Bus:

    Start condition or Repeated Start can't be launch.NI2C3: The bus idle.
    Start condition or Repeated Start can't be launch.NI2C3: The bus idle.
    ...


    The only devices connected are one PCA9555 and one TMP175.


    My oscilloscope shows continoues LOW for I2C_B_DAT and continous HIGH for I2C_B_CLK.


    This error state is preserved even after hardware resets!
    After power down and power up the I2C is working again.

    I just had a look at the EC2013 Roadmap which notes the fix for the WDOG issue in the 1.70 release planned for February 2016.


    We do not have the time to wait until then.


    Is there any workaround (HW or SW) which I can apply on my own?

    Hello,


    you cannot disable HW watchdog from kernel it is alread activataed in the EBootloader. Maybe to disable WDOG is also not a solution. We have to locate the source of the problem.


    Think you don't have to wait until February my colleagues are *just* working on this problem. We move release date of V1.70 (or 1.80) forward if the problem is solved!

    F&S Elektronik Systeme GmbH
    As this is an international forum, please try to post in English.
    Da dies ein internationales Forum ist, bitten wir darum, Beiträge möglichst in Englisch zu verfassen.