[PATCH 1/3] bsps/riscv: Make SMP start more robust

[Alan Cudmore]

Hi Sebastian,
I applied these three patches after my patches and ran them on my K210
board and simulator. I have a set of 12 tests including benchmarks, SMP01,
SMP08, ticker, etc. Everything ran OK.
Is there anything in particular I can try to test them like setting the
maximum CPUs to 1? (K210 is a dual core)
Alan


On Thu, Mar 16, 2023 at 4:59 AM Sebastian Huber <
sebastian.huber at embedded-brains.de> wrote:

> In SMP configurations, check that we run on a configured processor.  If
> not,
> then there is not much what can be done since we do not have a stack
> available
> for this processor.  Just loop forever in this case.  Do this in assemlby
> to
> ensure that no stack memory is used.
> ---
>  bsps/riscv/riscv/start/bspsmp.c |  5 +----
>  bsps/riscv/shared/start/start.S | 16 ++++++++++++++--
>  2 files changed, 15 insertions(+), 6 deletions(-)
>
> diff --git a/bsps/riscv/riscv/start/bspsmp.c
> b/bsps/riscv/riscv/start/bspsmp.c
> index 91f4f7b96a..ce5792f5b8 100644
> --- a/bsps/riscv/riscv/start/bspsmp.c
> +++ b/bsps/riscv/riscv/start/bspsmp.c
> @@ -36,10 +36,7 @@ void bsp_start_on_secondary_processor(Per_CPU_Control
> *cpu_self)
>
>    cpu_index_self = _Per_CPU_Get_index(cpu_self);
>
> -  if (
> -    cpu_index_self < rtems_configuration_get_maximum_processors()
> -      && _SMP_Should_start_processor(cpu_index_self)
> -  ) {
> +  if (_SMP_Should_start_processor(cpu_index_self)) {
>      set_csr(mie, MIP_MSIP | MIP_MEIP);
>      _SMP_Start_multitasking_on_secondary_processor(cpu_self);
>    } else {
> diff --git a/bsps/riscv/shared/start/start.S
> b/bsps/riscv/shared/start/start.S
> index 34e1839ca1..42e4348cd0 100644
> --- a/bsps/riscv/shared/start/start.S
> +++ b/bsps/riscv/shared/start/start.S
> @@ -66,8 +66,17 @@ SYM(_start):
>         LADDR   sp, _ISR_Stack_area_begin
>         LADDR   t2, _ISR_Stack_size
>         csrr    s0, mhartid
> -       li      t3, RISCV_BOOT_HARTID
> -       sub     s0, s0, t3
> +       li      t3, RISCV_BOOT_HARTID
> +       sub     s0, s0, t3
> +
> +       /*
> +        * Check that this is a configured processor.  If not, then there
> is
> +        * not much what can be done since we do not have a stack
> available for
> +        * this processor.  Just loop forever in this case.
> +        */
> +       LREG    t3, _SMP_Processor_configured_maximum
> +       bgeu    s0, t3, .Lwfi
> +
>         LADDR   t0, _Per_CPU_Information
>         slli    t1, s0, PER_CPU_CONTROL_SIZE_LOG2
>         add     s1, t0, t1
> @@ -100,6 +109,9 @@ SYM(_start):
>         tail    boot_card
>
>  #ifdef RTEMS_SMP
> +.Lwfi:
> +       wfi
> +       j       .Lwfi
>
>  .Lstart_on_secondary_processor:
>
> --
> 2.35.3
>
> _______________________________________________
> devel mailing list
> devel at rtems.org
> http://lists.rtems.org/mailman/listinfo/devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20230316/832f57cc/attachment-0001.htm>