On 20/02/25 8:33 pm, Brendan Jackman wrote:
This calculation divides a fixed parameter by an environment-dependent
parameter i.e. the number of CPUs.
The simple way to avoid machine-specific failures here is to just put a
cap on the max value of the latter.
I haven't read the test, but if nr_cpus is being computed, then this
value must be important to the test somehow? Would it potentially be
wrong to let the test run for nr_cpus != actual number of cpus?
Also, if the patch is correct then will it be better to also print a
diagnostic telling the user that the number of cpus is going to be
capped for the test to run?
Suggested-by: Mateusz Guzik <mjguzik@xxxxxxxxx>
Signed-off-by: Brendan Jackman <jackmanb@xxxxxxxxxx>
---
tools/testing/selftests/mm/uffd-stress.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/tools/testing/selftests/mm/uffd-stress.c b/tools/testing/selftests/mm/uffd-stress.c
index 1facfb79e09aa4113e344d7d90dec06a37264058..f306accbef255c79bc3eeba8b9e42161a88fc10e 100644
--- a/tools/testing/selftests/mm/uffd-stress.c
+++ b/tools/testing/selftests/mm/uffd-stress.c
@@ -453,6 +453,10 @@ int main(int argc, char **argv)
}
nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
+ if (nr_cpus > 32) {
+ /* Don't let calculation below go to zero. */
+ nr_cpus = 32;
+ }
nr_pages_per_cpu = bytes / page_size / nr_cpus;
if (!nr_pages_per_cpu) {
