diff options
author | Andrei Vagin <avagin@google.com> | 2019-10-18 15:00:07 -0700 |
---|---|---|
committer | gVisor bot <gvisor-bot@google.com> | 2019-10-18 15:01:40 -0700 |
commit | 8ae70f864d7ab9ca6aa2b47d144d1a2671857603 (patch) | |
tree | 1d70c2f2a59b0abe09c04b9f53619cb57ed81789 /test/syscalls/linux/socket_ipv4_udp_unbound.h | |
parent | 4e6f3a0c71e4107efb5c9446508876f942b30ec7 (diff) |
test/perf: optimize the getdents test
* Use mknod instead of open&close to create an empty file.
* Limit a number of files to (1<<16) instead of 100K.
In this case, a test set is (1, 8, 64, 512, 4K, 32K, 64K) instead of (1, 8, 64,
512, 4K, 32K, 98K). I think it is easier to compare results for 32K and 64K
than 32K and 98K. And results for 98K doesn't give us more information than for
54K.
PiperOrigin-RevId: 275552507
Diffstat (limited to 'test/syscalls/linux/socket_ipv4_udp_unbound.h')
0 files changed, 0 insertions, 0 deletions