A number of issues were found by running static analysers on the code of trace-cmd with openscanhub[1]. Mostly ressource leaks. [1] https://fedoraproject.org/wiki/OpenScanHub Jerome Marchand (15): libtracefs: call va_end() before exiting tracefs_hist_set_sort_key() libtracefs: prevent memory leak in append_filer() libtracefs: prevent a memory leak in update_fields() libtracefs: prevent a memory leak in tracefs_synth_add_end_field() libtracefs: prevent memory leak in tracefs_event_systems() libtracefs: don't leak socket file descriptor in open_vsock() libtracefs: prevent a memory leak in add_func_str() libtracefs: prevent a memory leak in tracefs_system_events() libtracefs: prevent a memory leak in open_cpu_files() libtracefs: prevent memory leak in tracefs_instance_create() libtracefs: my_yyinput() should return 0 when no data can be read libtracefs: prevent memory leak in tracefs_dynevent_get_all() libtracefs: close dir in the error path in tracefs_system_events() libtracefs: close dir in the error path in tracefs_event_systems() libtracefs: initialize val in build_filter() src/tracefs-dynevents.c | 1 + src/tracefs-events.c | 21 ++++++++++----------- src/tracefs-filter.c | 6 ++++-- src/tracefs-hist.c | 10 ++++++---- src/tracefs-instance.c | 1 + src/tracefs-sqlhist.c | 6 +++--- src/tracefs-tools.c | 4 +++- src/tracefs-vsock.c | 4 +++- 8 files changed, 31 insertions(+), 22 deletions(-) -- 2.44.0