diff options
| author | Sebastian Huber <sebastian.huber@embedded-brains.de> | 2014-09-22 13:42:26 +0200 |
|---|---|---|
| committer | Sebastian Huber <sebastian.huber@embedded-brains.de> | 2014-09-23 10:03:10 +0200 |
| commit | be43b79fcabb7551677e2d27c75e2a500e2ba622 (patch) | |
| tree | adacb76593743ee5bf58d0ea5df2db97880b5d5a /testsuite/rwlock01 | |
| parent | Include missing header file (diff) | |
| download | rtems-libbsd-be43b79fcabb7551677e2d27c75e2a500e2ba622.tar.bz2 | |
Replace RTEMS objects with custom implementation
Performance analysis revealed that the standard RTEMS objects are a
major bottleneck. The object get mechanism and attribute checks at
runtime have a significant overhead. Use a custom implementation for
synchronization primitives. This drops also the size of the
synchronization primitives considerably.
Diffstat (limited to 'testsuite/rwlock01')
| -rw-r--r-- | testsuite/rwlock01/test_main.c | 6 |
1 files changed, 0 insertions, 6 deletions
diff --git a/testsuite/rwlock01/test_main.c b/testsuite/rwlock01/test_main.c index fc4b2694..3e121c38 100644 --- a/testsuite/rwlock01/test_main.c +++ b/testsuite/rwlock01/test_main.c @@ -188,9 +188,6 @@ test_rw_non_recursive(test_context *ctx) rw_rlock(rw); /* FIXME: We use a mutex implementation */ assert(rw_wowned(rw)); - /* FIXME: We always allow recursion */ - rw_rlock(rw); - rw_runlock(rw); rw_runlock(rw); rw_rlock(rw); @@ -228,9 +225,6 @@ test_rw_non_recursive(test_context *ctx) rw_wlock(rw); assert(rw_wowned(rw)); - /* FIXME: We always allow recursion */ - rw_wlock(rw); - rw_wunlock(rw); rw_wunlock(rw); rw_wlock(rw); |