On 05/03/2013 08:53 AM, Michal Privoznik wrote:
--- daemon/libvirtd-config.c | 50 +++++---------- daemon/libvirtd.c | 29 +++++---- daemon/remote.c | 161 ++++++++++++++++++----------------------------- 3 files changed, 92 insertions(+), 148 deletions(-)
Good sign that our interface is nice, because using it shrinks the code base.
@@ -287,10 +287,10 @@ daemonUnixSocketPaths(struct daemonConfig *config, goto no_memory; } else { if (privileged) { - if (!(*sockfile = strdup(LOCALSTATEDIR "/run/libvirt/libvirt-sock"))) - goto no_memory; - if (!(*rosockfile = strdup(LOCALSTATEDIR "/run/libvirt/libvirt-sock-ro"))) - goto no_memory; + if (VIR_STRDUP(*sockfile, LOCALSTATEDIR "/run/libvirt/libvirt-sock") < 0) + goto error; + if (VIR_STRDUP(*rosockfile, LOCALSTATEDIR "/run/libvirt/libvirt-sock-ro") < 0) + goto error;
Could merge these two 'if' into one, especially since you were doing it elsewhere in this patch, but not essential.
@@ -1172,7 +1173,7 @@ int main(int argc, char **argv) {
case 'p': VIR_FREE(pid_file); - if (!(pid_file = strdup(optarg))) { + if (VIR_STRDUP_QUIET(pid_file, optarg) < 0) { VIR_ERROR(_("Can't allocate memory")); exit(EXIT_FAILURE);
One of the few cases where _QUIET is right, since main() is early enough that normal error reporting may not be initialized yet :)
@@ -473,18 +447,18 @@ static int remoteRelayDomainEventDiskChange(virConnectPtr conn ATTRIBUTE_UNUSED, memset(&data, 0, sizeof(data)); if (oldSrcPath && ((VIR_ALLOC(oldSrcPath_p) < 0) || - !(*oldSrcPath_p = strdup(oldSrcPath)))) + VIR_STRDUP(*oldSrcPath_p, oldSrcPath) < 0)) goto mem_error;
Yep, I said I could live with the double OOM for now, and that we'll clean it up when we make our next pass over VIR_ALLOC.
@@ -4779,14 +4739,14 @@ static void make_nonnull_domain(remote_nonnull_domain *dom_dst, virDomainPtr dom_src) { dom_dst->id = dom_src->id; - dom_dst->name = strdup(dom_src->name); + ignore_value(VIR_STRDUP_QUIET(dom_dst->name, dom_src->name)); memcpy(dom_dst->uuid, dom_src->uuid, VIR_UUID_BUFLEN);
Aargh, I promised to work on this, and still haven't gotten around to it. But it's an independent cleanup to make the make_nonnull_ functions not return half-baked objects, so doesn't hold up your patch. ACK. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org