MVME6100 problems? - VxWorks

This is a discussion on MVME6100 problems? - VxWorks ; Anyone have experience with the MVME6100 SBC, or hear of any problems it may have? I've been told it's Tundra TSi148 VMEbus interface chip is to blame for the problem I'm having (trying to move RAM_LOW/HIGH_ADRS past 0x2000000 so I ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: MVME6100 problems?

  1. MVME6100 problems?

    Anyone have experience with the MVME6100 SBC, or hear of any problems
    it may have?

    I've been told it's Tundra TSi148 VMEbus interface chip is to blame for
    the
    problem I'm having (trying to move RAM_LOW/HIGH_ADRS past 0x2000000
    so I can increase the shared memory size).


  2. Re: MVME6100 problems?

    As you know, RAM_HIGH_ADRS is used to host the copy of ROM in RAM in
    the "high" part of RAM. After uncompressing the ROM content into
    RAM_HIGH_ADRS, a jump instruction is made.

    Recall that PowerPC instructions are 32 bit wide. 26 bits are used for
    addressing. the Branch instruction can only go upto 32 mega bytes. So
    the jump will fail after 0x2000000.

    In order to jump further, you may have to use gcc option for longcall.
    Or use assembly code.


    MikeU wrote:
    > Anyone have experience with the MVME6100 SBC, or hear of any problems
    > it may have?
    >
    > I've been told it's Tundra TSi148 VMEbus interface chip is to blame for
    > the
    > problem I'm having (trying to move RAM_LOW/HIGH_ADRS past 0x2000000
    > so I can increase the shared memory size).



  3. Re: MVME6100 problems?

    Thanks for the response. I tried the gcc -mlongcall option without
    success. I also tried the WindRiver compiler option
    -Xcode-absolute-far.
    Both resulted in bootrom that seems to hang. I thought is should
    have worked. Surprised me when it didn't.

    hjb wrote:
    > As you know, RAM_HIGH_ADRS is used to host the copy of ROM in RAM in
    > the "high" part of RAM. After uncompressing the ROM content into
    > RAM_HIGH_ADRS, a jump instruction is made.
    >
    > Recall that PowerPC instructions are 32 bit wide. 26 bits are used for
    > addressing. the Branch instruction can only go upto 32 mega bytes. So
    > the jump will fail after 0x2000000.
    >
    > In order to jump further, you may have to use gcc option for longcall.
    > Or use assembly code.
    >
    >
    > MikeU wrote:
    > > Anyone have experience with the MVME6100 SBC, or hear of any problems
    > > it may have?
    > >
    > > I've been told it's Tundra TSi148 VMEbus interface chip is to blame for
    > > the
    > > problem I'm having (trying to move RAM_LOW/HIGH_ADRS past 0x2000000
    > > so I can increase the shared memory size).



  4. Re: MVME6100 problems?

    Use the shell to debug the branch at aseembly level, to make sure long
    branch works. Create assembly code :

    addis r6, r0, func@hi
    addi r0, r6, func@l(r6)
    mtlr r0
    blrl

    Load and single step through.
    If there are no bugs in the VxWorks dbgLib for PPC it should show you
    how things go.


    MikeU wrote:
    > Thanks for the response. I tried the gcc -mlongcall option without
    > success. I also tried the WindRiver compiler option
    > -Xcode-absolute-far.
    > Both resulted in bootrom that seems to hang. I thought is should
    > have worked. Surprised me when it didn't.
    >
    > hjb wrote:
    > > As you know, RAM_HIGH_ADRS is used to host the copy of ROM in RAM in
    > > the "high" part of RAM. After uncompressing the ROM content into
    > > RAM_HIGH_ADRS, a jump instruction is made.
    > >
    > > Recall that PowerPC instructions are 32 bit wide. 26 bits are used for
    > > addressing. the Branch instruction can only go upto 32 mega bytes. So
    > > the jump will fail after 0x2000000.
    > >
    > > In order to jump further, you may have to use gcc option for longcall.
    > > Or use assembly code.
    > >
    > >
    > > MikeU wrote:
    > > > Anyone have experience with the MVME6100 SBC, or hear of any problems
    > > > it may have?
    > > >
    > > > I've been told it's Tundra TSi148 VMEbus interface chip is to blame for
    > > > the
    > > > problem I'm having (trying to move RAM_LOW/HIGH_ADRS past 0x2000000
    > > > so I can increase the shared memory size).



  5. Re: MVME6100 problems?

    MikeU wrote:
    > Anyone have experience with the MVME6100 SBC, or hear of any problems
    > it may have?
    >
    > I've been told it's Tundra TSi148 VMEbus interface chip is to blame for
    > the
    > problem I'm having (trying to move RAM_LOW/HIGH_ADRS past 0x2000000
    > so I can increase the shared memory size).


    If you move the kernel above 0x2000000 (32Meg), then you need to either
    set global excExtendedVectors = TRUE; or, for project builds, enable
    INCLUDE_EXC_EXTENDED_VECTORS component.
    See the PPC arch supplement for Tornado or search for
    INCLUDE_EXC_EXTENDED_VECTORS in Workbench help.

    GV


+ Reply to Thread