OK - before anyone points it out, the comment in the code below just
before the call to "main" is wrong: interrupts are disabled (as you'd
expect, before anything has been setup) before jumping to "main".
Just goes to prove the point about never trusting comments, only the
code....
By the way, memory and other hardware initialisation is all done
in 'C' in this case: the less assembler the better, I think.
Brendan
--- In lpc2000@yahoogroups.com, "brendanmurphy37"
<brendan.murphy@i...> wrote:
>
> Hi,
>
> Below is a simple startup sequence that fills the stack areas and
> initialises overall state, before jumping to 'C'.
>
> Please don't bother pointing out there are more efficient ways of
> doing this: I know there are. It gets the job done....
>
> Note that this is designed to work with a GNU link file which
> reserves the space for the stacks as follows:
>
> /* next section is .bss, which is uninitialised data */
> .bss :
> {
> __start_bss = . ;
>
> /* include all uninitialised writable data */
>
> *(.bss)
>
> /* include ramaining uninitialised data */
>
> *(COMMON)
>
> __end_bss = . ;
>
> /* required allignment for stacks */
>
> . = ALIGN(0x10) ;
>
> /* allocate room for supervisor stack */
>
> __tos_svc = . ;
>
> . += 0x800 ;
>
> __stack_svc = . ;
>
> /* allocate room for interrupt stack */
>
> __tos_irq = . ;
>
> . += 0x200 ;
>
> __stack_irq = . ;
>
> /* allocate room for fast interrupt stack */
>
> __tos_fiq = . ;
>
> . += 0x100 ;
>
> __stack_fiq = . ;
>
> } > ram
>
> Hope this of some use to people.
>
> Regards
> Brendan
>
> /* define values to use to zap stacks */
>
> SVC_STACK_VALUE = 0xaaaaaaaa
> IRQ_STACK_VALUE = 0xbbbbbbbb
> FIQ_STACK_VALUE = 0xcccccccc
>
> /*
> * Exception vector table, common to all ARM-based systems
> *
> * See ARM Architecture Reference Manual, Programmer's Model
section
> for
> * details. Table entries just jump to handlers (using full 32-bit
> addressing)
> */
>
> _int_vectors:
>
> ldr pc, do_reset_addr
> ldr pc, do_reset_addr
> ldr pc, do_reset_addr
> ldr pc, do_reset_addr
> ldr pc, do_reset_addr
> nop
> ldr pc, do_reset_addr
> ldr pc, do_reset_addr
>
> /* table of interrupt handler adresses */
>
> do_reset_addr: .long do_reset
>
> /*
> * System reset handler
> */
>
> do_reset:
>
> /*
> * Set stack pointers for all modes used (supervisor, IRQ and FIQ)
> */
>
> msr cpsr_c, #0xd3 /* ensure we're in supervisor mode */
> ldr sp, =__stack_svc /* set supervisor mode stack pointer */
>
> msr cpsr_c, #0xd2 /* enter IRQ mode, with interrupts disabled
> */
> ldr sp, =__stack_irq /* set IRQ mode stack pointer */
>
> msr cpsr_c, #0xd1 /* enter FIQ mode, with interrupts disabled
> */
> ldr sp, =__stack_fiq /* set FIQ mode stack pointer */
>
> /* now zap all stacks with distinctive pattern, to help detect
> overflow */
>
> ldr r0, =__tos_svc
> ldr r1, =__stack_svc
> ldr r2, =SVC_STACK_VALUE
> bl zap_mem
>
> ldr r0, =__tos_irq
> ldr r1, =__stack_irq
> ldr r2, =IRQ_STACK_VALUE
> bl zap_mem
>
> ldr r0, =__tos_fiq
> ldr r1, =__stack_fiq
> ldr r2, =FIQ_STACK_VALUE
> bl zap_mem
>
> /*
> * Enable interrupts, enter supervisor mode and branch to start
> * of 'C' code
> */
>
> msr cpsr_c, #0xc3 /* I=1 F=1 T=0 MODE=supervisor */
> b main
>
> /*
> * Function to zap a block of memory with specified value
> * Note that block MUST be word alligned
> *
> * r0: start of block
> * r1: end of block
> * r2: value to load
> * lr: return address
> */
>
> zap_mem:
>
> cmp r0, r1
> strlo r2, [r0], #4
> blo zap_mem
> mov pc, lr
>
>
>
> --- In lpc2000@yahoogroups.com, "xjag74" <detlef.weidner@w...>
wrote:
> > All right, thank you Charles and Brendan (parallel replies),
> > if that's the usual way then I'll code test pattern writing
before
> > RAM copying in the startup.s file.
> >
> > Regards
> > xjag
> >
> > --- In lpc2000@yahoogroups.com, "xjag74" <detlef.weidner@w...>
> wrote:
> > > Now I realized that the interrupts use another stack (IRQ
stack) Show quoted textHide quoted text
> > than
> > > a normal routine (User stack).
> > >
> > > Is it possible to read the User stack pointer within the ISR?
> > >
> > > What I try to do is to capture the SP every timer1 ISR entry
> > (200us)
> > > and to store the maximum value to figure out if the stack
> collides
> > > with the data area.
> > >
> > > The stack check option of the GNU compiler is not working.
> > >
> > > Regards
> > > xjag
> > >
> > >
> > >
> > >