| commit | 6f722a4e30c3093440a9c92d37d8c83e1786851d | [log] [tgz] |
|---|---|---|
| author | Jan Stancek <jstancek@redhat.com> | Thu Jun 04 13:46:24 2015 +0200 |
| committer | Jan Stancek <jstancek@redhat.com> | Fri Jun 12 10:36:38 2015 +0200 |
| tree | 31906687c7731c09aeaa5e3b604a58f50ffedb41 | |
| parent | a4b989d0e19424f020e67803c54168734073d8c8 [diff] |
mem/oom: allocate memory using multiple threads This is to make testcase faster on big systems, for example, on Lenovo X3950 X6, 256GB RAM, 240 CPUs, 8 NUMA nodes oom01 testcase would previously take ~4 hours to complete. With this patch it takes ~16 minutes. This change spawns CPU-1 (or at least 1) thread(s), which allocate / fault-in memory in same way as before. Signed-off-by: Jan Stancek <jstancek@redhat.com> Acked-by: Cyril Hrubis <chrubis@suse.cz>