1:52 p.m.
To ensure proper processing by virGetUserID() and virGetGroupID()
of a uid/gid add a "+" prior to the uid/gid to denote it's really
a uid/gid for the label.
Signed-off-by: John Ferlan <jferlan(a)redhat.com>
---
src/security/security_dac.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/security/security_dac.c b/src/security/security_dac.c
index 019c789..cb7d322 100644
--- a/src/security/security_dac.c
+++ b/src/security/security_dac.c
@@ -60,7 +60,7 @@ virSecurityDACSetUserAndGroup(virSecurityManagerPtr mgr,
priv->user = user;
priv->group = group;
- if (virAsprintf(&priv->baselabel, "%u:%u",
+ if (virAsprintf(&priv->baselabel, "+%u:+%u",
(unsigned int) user,
(unsigned int) group) < 0)
return -1;
@@ -1064,7 +1064,7 @@ virSecurityDACGenLabel(virSecurityManagerPtr mgr,
}
break;
case VIR_DOMAIN_SECLABEL_DYNAMIC:
- if (virAsprintf(&seclabel->label, "%u:%u",
+ if (virAsprintf(&seclabel->label, "+%u:+%u",
(unsigned int) priv->user,
(unsigned int) priv->group) < 0)
return rc;
--
1.8.3.1
3:13 p.m.
On 10/29/2013 12:52 PM, John Ferlan wrote:
To ensure proper processing by virGetUserID() and virGetGroupID()
of a uid/gid add a "+" prior to the uid/gid to denote it's really
a uid/gid for the label.
Signed-off-by: John Ferlan <jferlan(a)redhat.com>
---
src/security/security_dac.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
ACK. Although unlikely, it is possible to have a system with a username
that is purely digits, and where those digits don't match the underlying
uid, so it is indeed nice that when we know we have a uid that we force
the parser to skip a name lookup (which harmlessly fails on 99.99% of
the systems, but which could potentially get wrong credentials on the
rare system with odd usernames). Worth including in 1.1.4.
diff --git a/src/security/security_dac.c b/src/security/security_dac.c
index 019c789..cb7d322 100644
--- a/src/security/security_dac.c
+++ b/src/security/security_dac.c
@@ -60,7 +60,7 @@ virSecurityDACSetUserAndGroup(virSecurityManagerPtr mgr,
priv->user = user;
priv->group = group;
- if (virAsprintf(&priv->baselabel, "%u:%u",
+ if (virAsprintf(&priv->baselabel, "+%u:+%u",
(unsigned int) user,
(unsigned int) group) < 0)
return -1;
@@ -1064,7 +1064,7 @@ virSecurityDACGenLabel(virSecurityManagerPtr mgr,
}
break;
case VIR_DOMAIN_SECLABEL_DYNAMIC:
- if (virAsprintf(&seclabel->label, "%u:%u",
+ if (virAsprintf(&seclabel->label, "+%u:+%u",
(unsigned int) priv->user,
(unsigned int) priv->group) < 0)
return rc;
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
4:18 p.m.
On 10/29/2013 04:13 PM, Eric Blake wrote:
On 10/29/2013 12:52 PM, John Ferlan wrote:
> To ensure proper processing by virGetUserID() and virGetGroupID()
> of a uid/gid add a "+" prior to the uid/gid to denote it's really
> a uid/gid for the label.
>
> Signed-off-by: John Ferlan <jferlan(a)redhat.com>
> ---
> src/security/security_dac.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
ACK. Although unlikely, it is possible to have a system with a username
that is purely digits, and where those digits don't match the underlying
uid, so it is indeed nice that when we know we have a uid that we force
the parser to skip a name lookup (which harmlessly fails on 99.99% of
the systems, but which could potentially get wrong credentials on the
rare system with odd usernames). Worth including in 1.1.4.
Thanks - this is now pushed.
John
It's also notable that without the patch, messages would be sent to
/var/log/messages such as:
Oct 19 10:13:21 myhost libvirtd[4055]: User record for user '1000'
was not found: No such file or directory
Oct 19 10:13:21 myhost libvirtd[4055]: Group record for user '1000'
was not found: No such file or directory
4042
days inactive
4042
days old
2 comments
2 participants
participants (2)
-
Eric Blake -
John Ferlan