PERF_OUTPUT_MACHINE_PID     = 1ULL << 37,
        PERF_OUTPUT_VCPU            = 1ULL << 38,
        PERF_OUTPUT_CGROUP          = 1ULL << 39,
+       PERF_OUTPUT_RETIRE_LAT      = 1ULL << 40,
 };
 
 struct perf_script {
        {.str = "machine_pid", .field = PERF_OUTPUT_MACHINE_PID},
        {.str = "vcpu", .field = PERF_OUTPUT_VCPU},
        {.str = "cgroup", .field = PERF_OUTPUT_CGROUP},
+       {.str = "retire_lat", .field = PERF_OUTPUT_RETIRE_LAT},
 };
 
 enum {
                              PERF_OUTPUT_ADDR | PERF_OUTPUT_DATA_SRC |
                              PERF_OUTPUT_WEIGHT | PERF_OUTPUT_PHYS_ADDR |
                              PERF_OUTPUT_DATA_PAGE_SIZE | PERF_OUTPUT_CODE_PAGE_SIZE |
-                             PERF_OUTPUT_INS_LAT,
+                             PERF_OUTPUT_INS_LAT | PERF_OUTPUT_RETIRE_LAT,
 
                .invalid_fields = PERF_OUTPUT_TRACE | PERF_OUTPUT_BPF_OUTPUT,
        },
                return -EINVAL;
        }
 
+       if (PRINT_FIELD(RETIRE_LAT) &&
+           evsel__check_stype(evsel, PERF_SAMPLE_WEIGHT_STRUCT, "WEIGHT_STRUCT", PERF_OUTPUT_RETIRE_LAT))
+               return -EINVAL;
+
        return 0;
 }
 
        if (PRINT_FIELD(INS_LAT))
                fprintf(fp, "%16" PRIu16, sample->ins_lat);
 
+       if (PRINT_FIELD(RETIRE_LAT))
+               fprintf(fp, "%16" PRIu16, sample->retire_lat);
+
        if (PRINT_FIELD(IP)) {
                struct callchain_cursor *cursor = NULL;
 
                     "brstacksym,flags,data_src,weight,bpf-output,brstackinsn,"
                     "brstackinsnlen,brstackoff,callindent,insn,insnlen,synth,"
                     "phys_addr,metric,misc,srccode,ipc,tod,data_page_size,"
-                    "code_page_size,ins_lat,machine_pid,vcpu,cgroup",
+                    "code_page_size,ins_lat,machine_pid,vcpu,cgroup,retire_lat",
                     parse_output_fields),
        OPT_BOOLEAN('a', "all-cpus", &system_wide,
                    "system-wide collection from all CPUs"),