autofs/autofs-5.1.6-dont-prune-offset-map-entries.patch

50 lines
1.5 KiB
Diff

autofs-5.1.6 - dont prune offset map entries
From: Ian Kent <raven@themaw.net>
Indirect maps create offset map entries for multi-mount map entries on
deamnd and remove them when they expire.
Since they are created based on an owning map entry they don't correspond
to an actual map entry so they, and their owning map entry, should never
be pruned.
Signed-off-by: Ian Kent <raven@themaw.net>
---
CHANGELOG | 1 +
daemon/lookup.c | 9 +++++++++
2 files changed, 10 insertions(+)
diff --git a/CHANGELOG b/CHANGELOG
index 3608345..34b160e 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -109,6 +109,7 @@ xx/xx/2018 autofs-5.1.5
- refactor sss getautomntbyname().
- improve sss getautomntbyname() error handling.
- use a valid timeout in lookup_prune_one_cache().
+- dont prune offset map entries.
19/12/2017 autofs-5.1.4
- fix spec file url.
diff --git a/daemon/lookup.c b/daemon/lookup.c
index 8bf1335..2de622e 100644
--- a/daemon/lookup.c
+++ b/daemon/lookup.c
@@ -1355,6 +1355,15 @@ void lookup_prune_one_cache(struct autofs_point *ap, struct mapent_cache *mc, ti
}
if (ap->type == LKP_INDIRECT) {
+ /* Don't prune offset map entries since they are
+ * created on demand and managed by expire and don't
+ * prune the multi-map owner map entry.
+ */
+ if (*me->key == '/' || me->multi == me) {
+ me = cache_enumerate(mc, me);
+ continue;
+ }
+
/* If the map hasn't been read (nobrowse
* indirect mounts) then keep cached entries
* for POSITIVE_TIMEOUT.