run_gc_task has been mistakenly left uninitialized in previous
patch for LU-7340. This has been silently ignored by gcc even
if -Wall option is used during build, possibly because no
optimization level/option requested where -Wuninitialized
option/check may only pe performed.
The side effect is that generated assembly code completelly
avoids run_gc_task usage from source, and thus a kthread
for ChangeLogs garbage-collection is created upon each
record creation and this without any of the garbage-collection
conditions are triggered.
Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: Ieb9ce062ba6ebf0c365c1e6f8a57f89dd39e0a9d
Reviewed-on: https://review.whamcloud.com/31347
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Stephan Thiell <sthiell@stanford.edu>
struct llog_ctxt *ctxt;
struct thandle *llog_th;
int rc;
- bool run_gc_task;
+ bool run_gc_task = false;
rec->cr_hdr.lrh_len = llog_data_len(sizeof(*rec) +
changelog_rec_varsize(&rec->cr));