diff options
author | Eric Blake <ebb9@byu.net> | 2009-12-21 07:00:13 -0700 |
---|---|---|
committer | Eric Blake <ebb9@byu.net> | 2009-12-21 07:02:45 -0700 |
commit | c7362ef0beb434d7a3063b5a7ac7b956988dc435 (patch) | |
tree | 5ac5e5ab921710d955583f9927563cbb80f44770 /tests/test-utimens-common.h | |
parent | 9a669cf64253a2b2149d7f7cc5e0664c1bc7dda9 (diff) | |
download | gnulib-c7362ef0beb434d7a3063b5a7ac7b956988dc435.tar.gz |
test-utimens: avoid spurious failure
Fixes a spurious failure on ext3, with one-second resolution,
now that ctime effects are being tested for inequality.
* tests/test-chown.h (nap): Factor...
* tests/nap.h: ...into new file.
* tests/test-lchown.h (nap): Avoid duplication.
* tests/test-utimens-common.h (nap): Use shared implementation,
necessary on file systems with 1-second resolution.
* modules/chown-tests (Files): Include new file.
* modules/fdutimensat-tests (Files): Likewise.
* modules/futimens-tests (Files): Likewise.
* modules/lchown-tests (Files): Likewise.
* modules/openat-tests (Files): Likewise.
* modules/utimens-tests (Files): Likewise.
* modules/utimensat-tests (Files): Likewise.
Signed-off-by: Eric Blake <ebb9@byu.net>
Diffstat (limited to 'tests/test-utimens-common.h')
-rw-r--r-- | tests/test-utimens-common.h | 27 |
1 files changed, 4 insertions, 23 deletions
diff --git a/tests/test-utimens-common.h b/tests/test-utimens-common.h index 707971abe4..30fd886748 100644 --- a/tests/test-utimens-common.h +++ b/tests/test-utimens-common.h @@ -24,10 +24,14 @@ # include <string.h> # include <unistd.h> +/* Gnulib modules. */ # include "stat-time.h" # include "timespec.h" # include "utimecmp.h" +/* Gnulib test header. */ +# include "nap.h" + enum { BILLION = 1000 * 1000 * 1000, @@ -44,29 +48,6 @@ enum { : 0) }; -/* Sleep long enough to cross a timestamp quantization boundary on - most known systems with subsecond timestamp resolution. For - example, ext4 has a quantization of 10 milliseconds, but a - resolution of 1 nanosecond. Likewise, NTFS has a quantization as - slow as 15.25 milliseconds, but a resolution of 100 nanoseconds. - This is necessary on systems where creat or utimens with NULL - rounds down to the quantization boundary, but where gettime and - hence utimensat can inject timestamps between quantization - boundaries. By ensuring we cross a boundary, we are less likely to - confuse utimecmp for two times that would round to the same - quantization boundary but are distinct based on resolution. */ -static void -nap (void) -{ - /* Systems that lack usleep also lack subsecond timestamps, and have - a quantization boundary equal to the resolution. Our usage of - utimecmp allows equality, so no need to waste 980 milliseconds - if the replacement usleep rounds to 1 second. */ -# if HAVE_USLEEP - usleep (20 * 1000); /* 20 milliseconds. */ -# endif -} - # if (defined _WIN32 || defined __WIN32__) && !defined __CYGWIN__ /* Skip ctime tests on native Windows, since it is either a copy of mtime or birth time (depending on the file system), rather than a |