llkd: bootstat: propagate detailed livelock canonical boot reason

Report kernel_panic,sysrq,livelock,<state> reboot reason via last
dmesg (pstore console).  Add ro.llk.killtest property, which will
allow reliable ABA platforms to drop kill test and go directly
to kernel panic.  This should also allow some manual unit testing
of the canonical boot reason report.

New canonical boot reasons from llkd are:
- kernel_panic,sysrq,livelock,alarm llkd itself locked up (Hail Mary)
- kernel_panic,sysrq,livelock,driver uninterrruptible D state
- kernel_panic,sysrq,livelock,zombie uninterrruptible Z state

Manual test assumptions:
- llkd is built by the platform and landed on system partition
- unit test is built and landed in /data/nativetest (could
  land in /data/nativetest64, adjust test correspondingly)
- llkd not enabled, ro.llk.enable and ro.llk.killtest
  are not set by platform allowing test to adjust all the
  configuration properties and start llkd.
- or, llkd is enabled, ro.llk.enable is true, and killtest is
  disabled, ro.llk.killtest is false, setup by the platform.
  This breaks the go/apct generic operations of the unit test
  for llk.zombie and llk.driver as kernel panic results
  requiring manual intervention otherwise.  If test moves to
  go/apct, then we will be forced to bypass these tests under
  this condition (but allow them to run if ro.llk.killtest
  is "off" so specific testing above/below can be run).

for i in driver zombie; do
        adb shell su root setprop ro.llk.killtest off
        adb shell /data/nativetest/llkd_unit_test/llkd_unit_test --gtest_filter=llkd.${i}
        adb wait-for-device
        adb shell su root setprop ro.llk.killtest off
        sleep 60
        adb shell getprop sys.boot.reason
        adb shell /data/nativetest/llkd_unit_test/llkd_unit_test --gtest_filter=llkd.${i}
done

Test: llkd_unit_test (see test assumptions)
Bug: 33808187
Bug: 72838192
Change-Id: I2b24875376ddfdbc282ba3da5c5b3567de85dbc0

bootstat/bootstat.cpp

@@ -303,6 +303,9 @@ const std::map<std::string, int32_t> kBootReasonMap = {
     {"kernel_panic,init", 158},
     {"kernel_panic,oom", 159},
     {"kernel_panic,stack", 160},
+    {"kernel_panic,sysrq,livelock,alarm", 161},   // llkd
+    {"kernel_panic,sysrq,livelock,driver", 162},  // llkd
+    {"kernel_panic,sysrq,livelock,zombie", 163},  // llkd
 };
 
 // Converts a string value representing the reason the system booted to an

llkd/README.md

@@ -53,7 +53,9 @@ on purpose, and llkd effectively sweeps up processes that create these
 conditions.  If the test can, it will reconfigure llkd to expedite the test
 duration by adjusting the ro.llk.* Android properties.  Tests run the D state
 with some scheduling progress to ensure that ABA checking prevents false
-triggers.
+triggers. If 100% reliable ABA on platform, then ro.llk.killtest can be
+set to false; however this will result in some of the unit tests to panic
+kernel instead of deal with more graceful kill operation.
 
 Android Properties
 ------------------
@@ -108,13 +110,6 @@ default <empty>, comma separated list of uid numbers or names.
 Architectural Concerns
 ----------------------
 
-- Figure out how to communicate the kernel panic better to bootstat canonical
-  boot reason determination.  This may require an alteration to bootstat, or
-  some logging from llkd.  Would like to see boot reason to be
-  watchdog,livelock as a minimum requirement.  Or more specifically would want
-  watchdog,livelock,device or watchdog,livelock,zombie be reported.
-  Currently reports panic,sysrq (user requested panic) or panic depending on
-  system support of pstore.
 - Create kernel module and associated gTest to actually test panic.
 - Create gTest to test out blacklist (ro.llk.blacklist.<properties> generally
   not be inputs).  Could require more test-only interfaces to libllkd.

llkd/include/llkd.h

@@ -37,6 +37,8 @@ unsigned llkCheckMilliseconds(void);
 #define KHT_ENABLE_PROPERTY            "ro." KHT_ENABLE_WRITEABLE_PROPERTY
 #define LLK_MLOCKALL_PROPERTY          "ro.llk.mlockall"
 #define LLK_MLOCKALL_DEFAULT           true
+#define LLK_KILLTEST_PROPERTY          "ro.llk.killtest"
+#define LLK_KILLTEST_DEFAULT           true
 #define LLK_TIMEOUT_MS_PROPERTY        "ro.llk.timeout_ms"
 #define KHT_TIMEOUT_PROPERTY           "ro.khungtask.timeout"
 #define LLK_D_TIMEOUT_MS_PROPERTY      "ro.llk.D.timeout_ms"

llkd/libllkd.cpp

@@ -70,6 +70,7 @@ milliseconds llkCycle;                               // ms to next thread check
 bool llkEnable = LLK_ENABLE_DEFAULT;                 // llk daemon enabled
 bool llkRunning = false;                             // thread is running
 bool llkMlockall = LLK_MLOCKALL_DEFAULT;             // run mlocked
+bool llkTestWithKill = LLK_KILLTEST_DEFAULT;         // issue test kills
 milliseconds llkTimeoutMs = LLK_TIMEOUT_MS_DEFAULT;  // default timeout
 enum { llkStateD, llkStateZ, llkNumStates };         // state indexes
 milliseconds llkStateTimeoutMs[llkNumStates];        // timeout override for each detection state
@@ -292,7 +293,7 @@ struct proc {
           exeMissingValid(false),
           cmdlineValid(false),
           updated(true),
-          killed(false) {
+          killed(!llkTestWithKill) {
         memset(comm, '\0', sizeof(comm));
         setComm(_comm);
     }
@@ -475,8 +476,8 @@ bool llkWriteStringToFileConfirm(const std::string& string, const std::string& f
     return android::base::Trim(content) == string;
 }
 
-void llkPanicKernel(bool dump, pid_t tid) __noreturn;
-void llkPanicKernel(bool dump, pid_t tid) {
+void llkPanicKernel(bool dump, pid_t tid, const char* state) __noreturn;
+void llkPanicKernel(bool dump, pid_t tid, const char* state) {
     auto sysrqTriggerFd = llkFileToWriteFd("/proc/sysrq-trigger");
     if (sysrqTriggerFd < 0) {
         // DYB
@@ -496,6 +497,8 @@ void llkPanicKernel(bool dump, pid_t tid) {
         }
         ::usleep(200000);  // let everything settle
     }
+    llkWriteStringToFile(std::string("SysRq : Trigger a crash : 'livelock,") + state + "'\n",
+                         "/dev/kmsg");
     android::base::WriteStringToFd("c", sysrqTriggerFd);
     // NOTREACHED
     // DYB
@@ -507,7 +510,7 @@ void llkPanicKernel(bool dump, pid_t tid) {
 }
 
 void llkAlarmHandler(int) {
-    llkPanicKernel(false, ::getpid());
+    llkPanicKernel(false, ::getpid(), "alarm");
 }
 
 milliseconds GetUintProperty(const std::string& key, milliseconds def) {
@@ -686,7 +689,7 @@ void llkCheckSchedUpdate(proc* procp, const std::string& piddir) {
             (val != procp->nrSwitches)) {
             procp->nrSwitches = val;
             procp->count = 0ms;
-            procp->killed = false;
+            procp->killed = !llkTestWithKill;
         }
         return;
     }
@@ -700,7 +703,7 @@ void llkCheckSchedUpdate(proc* procp, const std::string& piddir) {
         if (schedUpdate != procp->schedUpdate) {
             procp->schedUpdate = schedUpdate;
             procp->count = 0ms;
-            procp->killed = false;
+            procp->killed = !llkTestWithKill;
         }
     }
 
@@ -709,7 +712,7 @@ void llkCheckSchedUpdate(proc* procp, const std::string& piddir) {
         if (static_cast<uint64_t>(val) != procp->nrSwitches) {
             procp->nrSwitches = val;
             procp->count = 0ms;
-            procp->killed = false;
+            procp->killed = !llkTestWithKill;
         }
     }
 }
@@ -719,6 +722,7 @@ void llkLogConfig(void) {
               << LLK_ENABLE_PROPERTY "=" << llkFormat(llkEnable) << "\n"
               << KHT_ENABLE_PROPERTY "=" << llkFormat(khtEnable) << "\n"
               << LLK_MLOCKALL_PROPERTY "=" << llkFormat(llkMlockall) << "\n"
+              << LLK_KILLTEST_PROPERTY "=" << llkFormat(llkTestWithKill) << "\n"
               << KHT_TIMEOUT_PROPERTY "=" << llkFormat(khtTimeout) << "\n"
               << LLK_TIMEOUT_MS_PROPERTY "=" << llkFormat(llkTimeoutMs) << "\n"
               << LLK_D_TIMEOUT_MS_PROPERTY "=" << llkFormat(llkStateTimeoutMs[llkStateD]) << "\n"
@@ -869,7 +873,7 @@ milliseconds llkCheck(bool checkRunning) {
                 procp->time = utime + stime;
                 if (procp->state != state) {
                     procp->count = 0ms;
-                    procp->killed = false;
+                    procp->killed = !llkTestWithKill;
                     procp->state = state;
                 } else {
                     procp->count += llkCycle;
@@ -973,7 +977,7 @@ milliseconds llkCheck(bool checkRunning) {
             // We are here because we have confirmed kernel live-lock
             LOG(ERROR) << state << ' ' << llkFormat(procp->count) << ' ' << ppid << "->" << pid
                        << "->" << tid << ' ' << procp->getComm() << " [panic]";
-            llkPanicKernel(true, tid);
+            llkPanicKernel(true, tid, (state == 'Z') ? "zombie" : "driver");
         }
         LOG(VERBOSE) << "+closedir()";
     }
@@ -1045,6 +1049,7 @@ bool llkInit(const char* threadname) {
     }
     khtEnable = android::base::GetBoolProperty(KHT_ENABLE_PROPERTY, khtEnable);
     llkMlockall = android::base::GetBoolProperty(LLK_MLOCKALL_PROPERTY, llkMlockall);
+    llkTestWithKill = android::base::GetBoolProperty(LLK_KILLTEST_PROPERTY, llkTestWithKill);
     // if LLK_TIMOUT_MS_PROPERTY was not set, we will use a set
     // KHT_TIMEOUT_PROPERTY as co-operative guidance for the default value.
     khtTimeout = GetUintProperty(KHT_TIMEOUT_PROPERTY, khtTimeout);

llkd/llkd.rc

@@ -44,5 +44,6 @@ service llkd /system/bin/llkd
     user llkd
     group llkd readproc
     capabilities KILL IPC_LOCK
+    file /dev/kmsg w
     file /proc/sysrq-trigger w
     writepid /dev/cpuset/system-background/tasks

llkd/tests/llkd_test.cpp

@@ -154,6 +154,27 @@ inline void waitForPid(pid_t child_pid) {
     ASSERT_EQ(WTERMSIG(wstatus), SIGKILL);
 }
 
+bool checkKill(const char* reason) {
+    if (android::base::GetBoolProperty(LLK_KILLTEST_PROPERTY, LLK_KILLTEST_DEFAULT)) {
+        return false;
+    }
+    auto bootreason = android::base::GetProperty("sys.boot.reason", "nothing");
+    if (bootreason == reason) {
+        GTEST_LOG_INFO << "Expected test result confirmed " << reason << "\n";
+        return true;
+    }
+    GTEST_LOG_WARNING << "Expected test result is " << reason << "\n";
+
+    // apct adjustment if needed (set LLK_KILLTEST_PROPERTY to "off" to allow test)
+    //
+    // if (android::base::GetProperty(LLK_KILLTEST_PROPERTY, "") == "false") {
+    //     GTEST_LOG_WARNING << "Bypassing test\n";
+    //     return true;
+    // }
+
+    return false;
+}
+
 }  // namespace
 
 // The tests that use this helper are to simulate processes stuck in 'D'
@@ -221,6 +242,10 @@ TEST(llkd, driver_ABA_glacial) {
 // is that llkd will perform kill mitigation and not progress to kernel_panic.
 
 TEST(llkd, zombie) {
+    if (checkKill("kernel_panic,sysrq,livelock,zombie")) {
+        return;
+    }
+
     const auto period = llkdSleepPeriod('Z');
 
     /* Create a Persistent Zombie Process */
@@ -241,6 +266,10 @@ TEST(llkd, zombie) {
 }
 
 TEST(llkd, driver) {
+    if (checkKill("kernel_panic,sysrq,livelock,driver")) {
+        return;
+    }
+
     const auto period = llkdSleepPeriod('D');
 
     /* Create a Persistent Device Process */