Re: [PATCH v2] aspeed: Add boot stub for smp booting

[Peter Maydell]

On Thu, 9 Apr 2020 at 07:31, Joel Stanley <address@hidden> wrote:
>
> This is a boot stub that is similar to the code u-boot runs, allowing
> the kernel to boot the secondary CPU.

> +static void aspeed_write_smpboot(ARMCPU *cpu,
> +                                 const struct arm_boot_info *info)
> +{
> +    static const uint32_t poll_mailbox_ready[] = {
> +        /*
> +         * r2 = per-cpu go sign value
> +         * r1 = AST_SMP_MBOX_FIELD_ENTRY
> +         * r0 = AST_SMP_MBOX_FIELD_GOSIGN
> +         */
> +        0xee100fb0,  /* mrc     p15, 0, r0, c0, c0, 5 */
> +        0xe21000ff,  /* ands    r0, r0, #255          */
> +        0xe59f201c,  /* ldr     r2, [pc, #28]         */
> +        0xe1822000,  /* orr     r2, r2, r0            */
> +
> +        0xe59f1018,  /* ldr     r1, [pc, #24]         */
> +        0xe59f0018,  /* ldr     r0, [pc, #24]         */
> +
> +        0xe320f002,  /* wfe                           */
> +        0xe5904000,  /* ldr     r4, [r0]              */
> +        0xe1520004,  /* cmp     r2, r4                */
> +        0x1afffffb,  /* bne     <wfe>                 */

Note that unlike "wfi", QEMU's "wfe" implementation is merely
a 'yield', so a secondary-CPU boot loop that has wfe in it
will basically be a busy-loop of those vcpu threads.
(This is why the smpboot code in hw/arm/boot.c uses wfi.)

I don't suppose the secondary boot protocol on these boards
is such that a wfi loop will work ? (Depends on what the
primary code in the kernel does to prod the secondary after
writing the magic value.)


> +        0xe591f000,  /* ldr     pc, [r1]              */
> +        AST_SMP_MBOX_GOSIGN,
> +        AST_SMP_MBOX_FIELD_ENTRY,
> +        AST_SMP_MBOX_FIELD_GOSIGN,
> +    };

thanks
-- PMM