On Mon, 24 Apr 2017, Clark Williams wrote:
> On a numa system we allocate a stack from the closest NUMA node
> for each measurement thread. We need to touch those stack pages
> to prevent them faulting when first run. Not a big deal on a
> system with just a few cores, but overhead adds up with 32+ cores.
>
> Signed-off-by: Clark Williams <williams@xxxxxxxxxx>
> ---
> src/cyclictest/cyclictest.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/src/cyclictest/cyclictest.c b/src/cyclictest/cyclictest.c
> index 5e23fc58f602..49a81f168cdb 100644
> --- a/src/cyclictest/cyclictest.c
> +++ b/src/cyclictest/cyclictest.c
> @@ -2376,6 +2376,9 @@ int main(int argc, char **argv)
> /* allocate memory for a stack on appropriate node */
> stack = rt_numa_numa_alloc_onnode(stksize, node, i);
>
> + /* touch the stack pages to pre-fault them in */
> + memset(stack, 0, stksize);
> +
> /* set the thread's stack */
> if (pthread_attr_setstack(&attr, stack, stksize))
> fatal("failed to set stack addr for thread %d to 0x%x\n",
> --
> 2.9.3
>
> --
Signed-off-by: John Kacur <jkacur@xxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
