On Tue, Jun 20, 2017 at 17:03:30 +0200, Erik Skultety wrote:
During investigation of [1] I saw nothing in the logs that would help me get to the root cause. Then I found out that we don't log anything when lstat fails. Sure, doesn't happen often, but if it happens we should reflect that in the logs to prevent spurious behaviour.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1463285
Signed-off-by: Erik Skultety <eskultet@redhat.com> --- src/util/virfile.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/util/virfile.c b/src/util/virfile.c index d444b32f8..6bbcc3d15 100644 --- a/src/util/virfile.c +++ b/src/util/virfile.c @@ -1560,8 +1560,10 @@ virFileResolveLinkHelper(const char *linkpath, * directories, if linkpath is absolute and the basename is * already a non-symlink. */ if (IS_ABSOLUTE_FILE_NAME(linkpath) && !intermediatePaths) { - if (lstat(linkpath, &st) < 0) + if (lstat(linkpath, &st) < 0) { + virReportSystemError(errno, "%s", linkpath); return -1; + }
NACK, this function is designed not to report errors. It's even documented so. Some other callers even report their own errors. At most a VIR_DEBUG is appropriate here.