find proccess with largest rss (#300)

* show the cmdline of the process which was sent SIGKILL or SIGTERMIGTERM

* add cmdline to struct procinfo

* add test code

* add 'get-cmdline' to args

* add env EARLYOOM_CMDLINE in notify_ext()

* delete the flag -get-cmdline; fix some log message

* delete get_cmdline from struct poll_loop_args_t

* add env EARLYOOM_CMDLINE to MANPAGE.md

* add get_cmdline to the end of function is_larger

* add build.sh

* static build with version

* To get the latest tag using git command

* add flag '--sort-by-rss': find process with the largest rss (default oom_score)

* reset Makefile and delete build.sh

* delete ignore-root-user from MANPAGE.md

* delete duplicate code

* make --avoid / --prefer work with --sort-by-rss

* add VMRSS_PREFER and VMRSS_AVOID to support prefer_regex and avoid_regex

MANPAGE.md

@@ -159,6 +159,9 @@ that might have occurred due to the processes attaining a high oom_score.
 Use this option with caution as other processes might be sacrificed in place of the ignored
 processes when earlyoom determines to kill processes.
 
+### \-\-sort-by-rss
+find process with the largest rss (default oom_score)
+
 #### \-\-dryrun
 dry run (do not kill any processes)
 

kill.c

@@ -24,6 +24,11 @@
 // Processes matching "--avoid REGEX" get BADNESS_AVOID added to their badness
 #define BADNESS_AVOID -300
 
+// Processes matching "--prefer REGEX" get VMRSS_PREFER added to their VmRSSkiB
+#define VMRSS_PREFER 3145728
+// Processes matching "--avoid REGEX" get VMRSS_AVOID added to their VmRSSkiB
+#define VMRSS_AVOID -3145728
+
 // Buffer size for UID/GID/PID string conversion
 #define UID_BUFSIZ 128
 // At most 1 notification per second when --dryrun is active
@@ -289,42 +294,63 @@ bool is_larger(const poll_loop_args_t* args, const procinfo_t* victim, procinfo_
         cur->badness = res;
     }
 
+    {
+        long long res = get_vm_rss_kib(cur->pid);
+        if (res < 0) {
+            debug("pid %d: error reading rss: %s\n", cur->pid, strerror((int)-res));
+            return false;
+        }
+        cur->VmRSSkiB = res;
+    }
+
     if ((args->prefer_regex || args->avoid_regex || args->ignore_regex)) {
         int res = get_comm(cur->pid, cur->name, sizeof(cur->name));
         if (res < 0) {
             debug("pid %d: error reading process name: %s\n", cur->pid, strerror(-res));
             return false;
         }
         if (args->prefer_regex && regexec(args->prefer_regex, cur->name, (size_t)0, NULL, 0) == 0) {
-            cur->badness += BADNESS_PREFER;
+            if (args->sort_by_rss) {
+                cur->VmRSSkiB += VMRSS_PREFER;
+            } else {
+                cur->badness += BADNESS_PREFER;
+            }
         }
         if (args->avoid_regex && regexec(args->avoid_regex, cur->name, (size_t)0, NULL, 0) == 0) {
-            cur->badness += BADNESS_AVOID;
+            if (args->sort_by_rss) {
+                cur->VmRSSkiB += VMRSS_AVOID;
+            } else {
+                cur->badness += BADNESS_AVOID;
+            }
         }
         if (args->ignore_regex && regexec(args->ignore_regex, cur->name, (size_t)0, NULL, 0) == 0) {
             return false;
         }
     }
 
-    if (cur->badness < victim->badness) {
+    if (cur->VmRSSkiB == 0) {
+        // Kernel threads have zero rss
         return false;
     }
 
-    {
-        long long res = get_vm_rss_kib(cur->pid);
-        if (res < 0) {
-            debug("pid %d: error reading rss: %s\n", cur->pid, strerror((int)-res));
+    if (args->sort_by_rss) {
+         /* find process with the largest rss */
+        if (cur->VmRSSkiB < victim->VmRSSkiB) {
             return false;
         }
-        cur->VmRSSkiB = res;
-    }
 
-    if (cur->VmRSSkiB == 0) {
-        // Kernel threads have zero rss
-        return false;
-    }
-    if (cur->badness == victim->badness && cur->VmRSSkiB <= victim->VmRSSkiB) {
-        return false;
+        if (cur->VmRSSkiB == victim->VmRSSkiB && cur->badness <= victim->badness) {
+            return false;
+        }
+    } else {
+        /* find process with the largest oom_score */
+        if (cur->badness < victim->badness) {
+            return false;
+        }
+
+        if (cur->badness == victim->badness && cur->VmRSSkiB <= victim->VmRSSkiB) {
+            return false;
+        }
     }
 
     // Skip processes with oom_score_adj = -1000, like the
@@ -371,7 +397,7 @@ void debug_print_procinfo(const procinfo_t* cur)
 }
 
 /*
- * Find the process with the largest oom_score.
+ * Find the process with the largest oom_score or rss(when flag --sort-by-rss is set).
  */
 procinfo_t find_largest_process(const poll_loop_args_t* args)
 {

kill.h

@@ -22,6 +22,8 @@ typedef struct {
     bool kill_process_group;
     /* do not kill processes owned by root */
     bool ignore_root_user;
+    /* find process with the largest rss */
+    bool sort_by_rss;
     /* prefer/avoid killing these processes. NULL = no-op. */
     regex_t* prefer_regex;
     regex_t* avoid_regex;

main.c

@@ -42,6 +42,7 @@ enum {
     LONG_OPT_IGNORE,
     LONG_OPT_IGNORE_ROOT,
     LONG_OPT_USE_SYSLOG,
+    LONG_OPT_SORT_BY_RSS,
 };
 
 static int set_oom_score_adj(int);
@@ -89,6 +90,7 @@ int main(int argc, char* argv[])
         .swap_kill_percent = 5,
         .report_interval_ms = 1000,
         .ignore_root_user = false,
+        .sort_by_rss = false,
         /* omitted fields are set to zero */
     };
     int set_my_priority = 0;
@@ -131,6 +133,7 @@ int main(int argc, char* argv[])
         { "ignore", required_argument, NULL, LONG_OPT_IGNORE },
         { "dryrun", no_argument, NULL, LONG_OPT_DRYRUN },
         { "ignore-root-user", no_argument, NULL, LONG_OPT_IGNORE_ROOT },
+        { "sort-by-rss", no_argument, NULL, LONG_OPT_SORT_BY_RSS },
         { "syslog", no_argument, NULL, LONG_OPT_USE_SYSLOG },
         { "help", no_argument, NULL, 'h' },
         { 0, 0, NULL, 0 } /* end-of-array marker */
@@ -224,6 +227,10 @@ int main(int argc, char* argv[])
             args.ignore_root_user = true;
             fprintf(stderr, "Processes owned by root will not be killed\n");
             break;
+        case LONG_OPT_SORT_BY_RSS:
+            args.sort_by_rss = true;
+            fprintf(stderr, "Find process with the largest rss\n");
+            break;
         case LONG_OPT_PREFER:
             prefer_cmds = optarg;
             break;
@@ -264,6 +271,7 @@ int main(int argc, char* argv[])
                 "  -p                        set niceness of earlyoom to -20 and oom_score_adj to\n"
                 "                            -100\n"
                 "  --ignore-root-user        do not kill processes owned by root\n"
+                "  --sort-by-rss             find process with the largest rss (default oom_score)\n"
                 "  --prefer REGEX            prefer to kill processes matching REGEX\n"
                 "  --avoid REGEX             avoid killing processes matching REGEX\n"
                 "  --ignore REGEX            ignore processes matching REGEX\n"

testsuite_cli_test.go

@@ -102,10 +102,11 @@ func TestCli(t *testing.T) {
 		// We use {"-r=0"} instead of {"-r", "0"} so runEarlyoom() can detect that there will be no output
 		{args: []string{"-r=0"}, code: -1, stderrContains: startupMsg, stdoutEmpty: true},
 		{args: []string{"-r", "0.1"}, code: -1, stderrContains: startupMsg, stdoutContains: memReport},
-		// Test --avoid, --prefer and -ignore-root-user
+		// Test --avoid, --prefer, --ignore-root-user and --sort-by-rss
 		{args: []string{"--avoid", "MyProcess1"}, code: -1, stderrContains: "Will avoid killing", stdoutContains: memReport},
 		{args: []string{"--prefer", "MyProcess2"}, code: -1, stderrContains: "Preferring to kill", stdoutContains: memReport},
 		{args: []string{"--ignore-root-user"}, code: -1, stderrContains: "Processes owned by root will not be killed", stdoutContains: memReport},
+		{args: []string{"--sort-by-rss"}, code: -1, stderrContains: "Find process with the largest rss", stdoutContains: memReport},
 		{args: []string{"-i"}, code: -1, stderrContains: "Option -i is ignored"},
 		// Extra arguments should error out
 		{args: []string{"xyz"}, code: 13, stderrContains: "extra argument not understood", stdoutEmpty: true},