[okl4-developer] hardware debugging support added in traps.spp leads to OKL4 crash

Geoffrey Lee glee at ok-labs.com
Thu Oct 16 02:47:46 EST 2008


On Wed, Oct 15, 2008 at 03:53:23AM -0700, Gabi Voiculescu wrote:
> Hello.
> 
> I am trying to add hardware debugging support to OKL4-2.1.0. on my ARM1176 development chip processor (the debugging hardware is present and working).
> 
> I achieve this by filling with a task specific, non-zero value the PROCID bitfield of the ContextID  register (p15, 0, xxx, c13, c0, 1) in traps.spp.
> 
> Reading the ARM documentation and interacting with ARM technical support I was assured that the PROCID bitfield has no impact on the memory management unit, being just useful for hardware debugging.
> 
> When using my modified traps.spp, OKL4 freezes.
> 

It appears you are overwriting the your pagetable base rather than
setting a context id?

	-gl

> 
> Questions:
> 
> Do you have any previous experience with this particular functionality?
> 
> Can you give me a hint on why my two instructions could lead to an OKL4 crash?
> 
> Can you help me in fixing this?
> 
> Thank you,
> Gabi Voiculescu
> Newgate Design
> 
> 
> 
> Below I give the diff output.
> ------------------------------------------------------------------------------
> $ diff traps.spp traps.spp.gabi
> 649c649,655
> < 
> ---
> >         
> >         ldr     tmp1,   [to_tcb, #OFS_TCB_MYSELF_GLOBAL]
> >         
> >         
> >         add     tmp2,   tmp2,   tmp1,   lsl #8*/  /* CHECKME: tmp2 = tmp2 + (tmp1<<8) -gabi */
> >         
> >          
> 651c657
> <         mcr     p15, 0, tmp6, c13, c0, 1
> ---
> >         mcr     p15, 0, tmp6, c13, c0, 1 
> 1486c1492,1497
> < 
> ---
> >         
> >         ldr     tmp1,   [to_tcb, #OFS_TCB_MYSELF_GLOBAL] 
> >         
> >         
> >         add     tmp2,   tmp2,   tmp1,   lsl #8*/    /* CHECKME: tmp2 = tmp2 + (tmp1<<8) -gabi */
> 
> 
> 
> 
> 
> 
> 
> 
>       
> _______________________________________________
> Developer mailing list
> Developer at okl4.org
> https://lists.okl4.org/mailman/listinfo/developer


-- 




More information about the Developer mailing list