On 03/25/2013 08:25 AM, Paolo Bonzini wrote:
The need_prctl variable is not really needed. If it is false,
capng_apply will be called twice with the same set, causing
a little extra work but no problem. This keeps the code a bit
simpler.
It is also clearer to invoke capng_apply(CAPNG_SELECT_BOUNDS)
separately, to make sure it is done while we have CAP_SETPCAP.
Signed-off-by: Paolo Bonzini <pbonzini(a)redhat.com>
---
src/util/virutil.c | 18 +++++++++++-------
1 file changed, 11 insertions(+), 7 deletions(-)
/* Change to the temp capabilities */
- if ((capng_ret = capng_apply(CAPNG_SELECT_BOTH)) < 0) {
+ if ((capng_ret = capng_apply(CAPNG_SELECT_CAPS)) < 0) {
Beforehand, we limited both caps and bounding set, with an overlarge
set, now you are limiting just caps...
virReportError(VIR_ERR_INTERNAL_ERROR,
_("cannot apply process capabilities %d"), capng_ret);
goto cleanup;
@@ -3063,12 +3061,18 @@ virSetUIDGIDWithCaps(uid_t uid, gid_t gid, unsigned long long
capBits,
goto cleanup;
/* Tell it we are done keeping capabilities */
- if (need_prctl && prctl(PR_SET_KEEPCAPS, 0, 0, 0, 0)) {
+ if (prctl(PR_SET_KEEPCAPS, 0, 0, 0, 0)) {
virReportSystemError(errno, "%s",
_("prctl failed to reset KEEPCAPS"));
goto cleanup;
}
+ /* Set bounding set while we have CAP_SETPCAP. Unfortunately we cannot
+ * do this if we failed to get the capability above, so ignore the
+ * return value.
+ */
+ capng_apply(CAPNG_SELECT_BOUNDS);
...and then separately limiting bounds, but still while having an
overlarge set.
+
/* Drop the caps that allow setuid/gid (unless they were requested) */
if (need_setgid)
capng_update(CAPNG_DROP, CAPNG_EFFECTIVE|CAPNG_PERMITTED, CAP_SETGID);
@@ -3078,7 +3082,7 @@ virSetUIDGIDWithCaps(uid_t uid, gid_t gid, unsigned long long
capBits,
if (need_setpcap)
capng_update(CAPNG_DROP, CAPNG_EFFECTIVE|CAPNG_PERMITTED, CAP_SETPCAP);
Here, the set is now pruned to size...
- if (need_prctl && ((capng_ret = capng_apply(CAPNG_SELECT_BOTH)) < 0)) {
+ if (((capng_ret = capng_apply(CAPNG_SELECT_CAPS)) < 0)) {
...but you are now only limiting caps, not the bounding set. Is that
correct?
Does this need to be considered for 1.0.4, or can we delay it to
post-release?
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org