On 08/19/2011 08:39 PM, Eric W. Biederman wrote:
> But Kay's suggestion to use netlink_has_listeners() seems like good
> idea. IOW if there is no listener, it should skip quietly and not
> fail the whole call...
> In the case of ESRCH I completely agree.
>
> We are currently ignoring errors in the semantically more interesting
> case when netlink_broadcast does not deliver the packet to one of the
> listening netlink sockets.
>
> How does this patch look?
Well, your version still returns -ESRCH, so it is not helping much :-)
If you meant to handle it like I fixed in patch below (ignore ESRCH)
then I am ok with that approach, patch below fixes the problem
we have.
Thanks,
Milan
diff --git a/lib/kobject_uevent.c b/lib/kobject_uevent.c
index 70af0a7..04b2ed2 100644
--- a/lib/kobject_uevent.c
+++ b/lib/kobject_uevent.c
@@ -139,6 +139,7 @@ int kobject_uevent_env(struct kobject *kobj, enum kobject_action action,
u64 seq;
int i = 0;
int retval = 0;
+ bool delivery_failure;
#ifdef CONFIG_NET
struct uevent_sock *ue_sk;
#endif
@@ -251,6 +252,7 @@ int kobject_uevent_env(struct kobject *kobj, enum kobject_action action,
if (retval)
goto exit;
+ delivery_failure = false;
#if defined(CONFIG_NET)
/* send netlink message */
mutex_lock(&uevent_sock_mutex);
@@ -281,14 +283,17 @@ int kobject_uevent_env(struct kobject *kobj, enum kobject_action action,
0, 1, GFP_KERNEL,
kobj_bcast_filter,
kobj);
- /* ENOBUFS should be handled in userspace */
- if (retval == -ENOBUFS)
+ if (retval == -ESRCH)
retval = 0;
+ else if (retval)
+ delivery_failure = true;
} else
- retval = -ENOMEM;
+ delivery_failure = true;
}
mutex_unlock(&uevent_sock_mutex);
#endif
+ if (delivery_failure)
+ retval = -ENOBUFS;
/* call uevent_helper, usually only enabled during early boot */
if (uevent_helper[0] && !kobj_usermode_filter(kobj)) {
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/dm-devel
