chore: bump libs to the latest tag 0.13.2

Signed-off-by: Andrea Terzolo <andreaterzolo3@gmail.com>

CHANGELOG.md

@@ -22,7 +22,7 @@ Released on 2023-09-25
 
 
 * new(falco-driver-loader): --source-only now prints the values as env vars [[#2353](https://github.com/falcosecurity/falco/pull/2353)] - [@steakunderscore](https://github.com/steakunderscore)
-* new(docker): allow passing options to falco-driver-loader from the driver loader cointainer [[#2781](https://github.com/falcosecurity/falco/pull/2781)] - [@LucaGuerra](https://github.com/LucaGuerra)
+* new(docker): allow passing options to falco-driver-loader from the driver loader container [[#2781](https://github.com/falcosecurity/falco/pull/2781)] - [@LucaGuerra](https://github.com/LucaGuerra)
 * new(docker): add experimental falco-distroless image based on Wolfi [[#2768](https://github.com/falcosecurity/falco/pull/2768)] - [@LucaGuerra](https://github.com/LucaGuerra)
 * new: the legacy falco image is available as driver-loader-legacy [[#2718](https://github.com/falcosecurity/falco/pull/2718)] - [@LucaGuerra](https://github.com/LucaGuerra)
 * new: added option to enable/disable echoing of server answer to stdout (disabled by default) when using HTTP output [[#2602](https://github.com/falcosecurity/falco/pull/2602)] - [@FedeDP](https://github.com/FedeDP)
@@ -1086,7 +1086,7 @@ Released on 2021-01-18
 ### Minor Changes
 
 * build: bump b64 to v2.0.0.1 [[#1441](https://github.com/falcosecurity/falco/pull/1441)] - [@fntlnz](https://github.com/fntlnz)
-* rules(macro container_started): re-use `spawned_process` macro inside `container_started` macro [[#1449](https://github.com/falcosecurity/falco/pull/1449)] - [@leodido](https://github.com/leodido)
+* rules(macro container_started): reuse `spawned_process` macro inside `container_started` macro [[#1449](https://github.com/falcosecurity/falco/pull/1449)] - [@leodido](https://github.com/leodido)
 * docs: reach out documentation [[#1472](https://github.com/falcosecurity/falco/pull/1472)] - [@fntlnz](https://github.com/fntlnz)
 * docs: Broken outputs.proto link [[#1493](https://github.com/falcosecurity/falco/pull/1493)] - [@deepskyblue86](https://github.com/deepskyblue86)
 * docs(README.md): correct broken links [[#1506](https://github.com/falcosecurity/falco/pull/1506)] - [@leogr](https://github.com/leogr)

cmake/modules/falcosecurity-libs.cmake

@@ -35,8 +35,8 @@ else()
   # In case you want to test against another falcosecurity/libs version (or branch, or commit) just pass the variable -
   # ie., `cmake -DFALCOSECURITY_LIBS_VERSION=dev ..` 
   if(NOT FALCOSECURITY_LIBS_VERSION)
-    set(FALCOSECURITY_LIBS_VERSION "0.13.1")
-    set(FALCOSECURITY_LIBS_CHECKSUM "SHA256=2be42a27be3ffe6bd7e53eaa5d8358cab05a0dca821819c6e9059e51b9786219")
+    set(FALCOSECURITY_LIBS_VERSION "0.13.2")
+    set(FALCOSECURITY_LIBS_CHECKSUM "SHA256=442810cc09a63406053ec3506ca16b59b5f5514ab08d478632f65beac6f167b5")
   endif()
 
   # cd /path/to/build && cmake /path/to/source

falco.yaml

@@ -331,24 +331,17 @@ rule_matching: first
 # If it does, it is most likely happening due to the entire event flow being too slow,
 # indicating that the server is under heavy load.
 # 
-# Lowering the number of items can prevent memory from steadily increasing until the OOM
-# killer stops the Falco process. We provide recovery actions to self-limit or self-kill
-# in order to handle this situation earlier, similar to how we expose the kernel buffer size
-# as a parameter. However, it will not address the root cause of the event pipe not keeping up.
-# 
 # `capacity`: the maximum number of items allowed in the queue is determined by this value. 
 # Setting the value to 0 (which is the default) is equivalent to keeping the queue unbounded. 
-# In other words, when this configuration is set to 0, the number of allowed items is effectively 
-# set to the largest possible long value, disabling this setting.
+# In other words, when this configuration is set to 0, the number of allowed items is 
+# effectively set to the largest possible long value, disabling this setting.
 # 
-# `recovery`: strategy to follow when the queue becomes filled up. It applies only when the
-# queue is bounded and there is still available system memory. In the case of an unbounded
-# queue, if the available memory on the system is consumed, the Falco process would be
-# OOM killed. The value `exit` is the default, `continue` does nothing special and `empty` 
-# empties the queue and then continues.
+# In the case of an unbounded queue, if the available memory on the system is consumed, 
+# the Falco process would be OOM killed. When using this option and setting the capacity, 
+# the current event would be dropped, and the event loop would continue. This behavior mirrors 
+# kernel-side event drops when the buffer between kernel space and user space is full.
 outputs_queue:
   capacity: 0
-  recovery: exit
 
 
 ##########################

proposals/20221129-artifacts-distribution.md

@@ -69,7 +69,7 @@ The allowed publishing channels are:
 Both channels are equivalent and may publish the same artifacts. However, for historical reasons and to avoid confusion, the **`docker.io` registry should only be used for container images** and not for other kinds of artifacts (e.g., plugins, rules, etc.).
 
 
-Mirrors are allowed and encouraged if they facilitate artifacts consumption by our users. This proposal reccomends to enable mirrors on the major public OCI registry, such as [Amazon ECR](https://gallery.ecr.aws/) (which is already implentend in our infra at the time of writing).
+Mirrors are allowed and encouraged if they facilitate artifacts consumption by our users. This proposal recommends to enable mirrors on the major public OCI registry, such as [Amazon ECR](https://gallery.ecr.aws/) (which is already implentend in our infra at the time of writing).
 
 
 Official **channels and mirrors must be listed at [falco.org](https://falco.org/)**. 

userspace/engine/falco_common.cpp

@@ -33,12 +33,6 @@ static std::vector<std::string> rule_matching_names = {
 	"all"
 };
 
-static std::vector<std::string> outputs_queue_recovery_names = {
-		"continue",
-		"exit",
-		"empty",
-};
-
 bool falco_common::parse_priority(std::string v, priority_type& out)
 {
 	for (size_t i = 0; i < priority_names.size(); i++)
@@ -66,19 +60,6 @@ falco_common::priority_type falco_common::parse_priority(std::string v)
 	return out;
 }
 
-bool falco_common::parse_queue_recovery(std::string v, outputs_queue_recovery_type& out)
-{
-	for (size_t i = 0; i < outputs_queue_recovery_names.size(); i++)
-	{
-		if (!strcasecmp(v.c_str(), outputs_queue_recovery_names[i].c_str()))
-		{
-			out = (outputs_queue_recovery_type) i;
-			return true;
-		}
-	}
-	return false;
-}
-
 bool falco_common::format_priority(priority_type v, std::string& out, bool shortfmt)
 {
 	if ((size_t) v < priority_names.size())

userspace/engine/falco_common.h

@@ -60,12 +60,6 @@ struct falco_exception : std::exception
 namespace falco_common
 {
 
-	enum outputs_queue_recovery_type {
-		RECOVERY_CONTINUE = 0,  /* outputs_queue_capacity recovery strategy of continuing on. */
-		RECOVERY_EXIT = 1,  /* outputs_queue_capacity recovery strategy of exiting, self OOM kill. */
-		RECOVERY_EMPTY = 2,  /* outputs_queue_capacity recovery strategy of emptying queue then continuing. */
-	};
-
 	const std::string syscall_source = sinsp_syscall_event_source_name;
 
 	// Same as numbers/indices into the above vector
@@ -83,7 +77,6 @@ namespace falco_common
 	
 	bool parse_priority(std::string v, priority_type& out);
 	priority_type parse_priority(std::string v);
-	bool parse_queue_recovery(std::string v, outputs_queue_recovery_type& out);
 	bool format_priority(priority_type v, std::string& out, bool shortfmt=false);
 	std::string format_priority(priority_type v, bool shortfmt=false);
 

userspace/falco/app/actions/init_outputs.cpp

@@ -65,7 +65,6 @@ falco::app::run_result falco::app::actions::init_outputs(falco::app::state& s)
 		s.config->m_output_timeout,
 		s.config->m_buffered_outputs,
 		s.config->m_outputs_queue_capacity,
-		s.config->m_outputs_queue_recovery,
 		s.config->m_time_format_iso_8601,
 		hostname));
 

userspace/falco/atomic_signal_handler.h

@@ -87,9 +87,9 @@ namespace falco
         /**
          * @brief If a signal is triggered, performs an handler action.
          * The action function will be invoked exactly once among all the
-         * simultaneus calls. The action will not be performed if the
+         * simultaneous calls. The action will not be performed if the
          * signal is not triggered, or if the triggered has already been
-         * handled. When an action is being performed, all the simultaneus
+         * handled. When an action is being performed, all the simultaneous
          * callers will wait and be blocked up until its execution is finished.
          * If the handler action throws an exception, it will be considered
          * performed. After the first handler has been performed, every

userspace/falco/configuration.cpp

@@ -42,7 +42,6 @@ falco_configuration::falco_configuration():
 	m_watch_config_files(true),
 	m_buffered_outputs(false),
 	m_outputs_queue_capacity(DEFAULT_OUTPUTS_QUEUE_CAPACITY_UNBOUNDED_MAX_LONG_VALUE),
-	m_outputs_queue_recovery(falco_common::RECOVERY_EXIT),
 	m_time_format_iso_8601(false),
 	m_output_timeout(2000),
 	m_grpc_enabled(false),
@@ -290,11 +289,6 @@ void falco_configuration::load_yaml(const std::string& config_name, const yaml_h
 	{
 		m_outputs_queue_capacity = DEFAULT_OUTPUTS_QUEUE_CAPACITY_UNBOUNDED_MAX_LONG_VALUE;
 	}
-	std::string recovery = config.get_scalar<std::string>("outputs_queue.recovery", "exit");
-	if (!falco_common::parse_queue_recovery(recovery, m_outputs_queue_recovery))
-	{
-		throw std::logic_error("Unknown recovery \"" + recovery + "\"--must be one of exit, continue, empty");
-	}
 
 	m_time_format_iso_8601 = config.get_scalar<bool>("time_format_iso_8601", false);
 

userspace/falco/configuration.h

@@ -74,7 +74,6 @@ public:
 	bool m_watch_config_files;
 	bool m_buffered_outputs;
 	size_t m_outputs_queue_capacity;
-	falco_common::outputs_queue_recovery_type m_outputs_queue_recovery;
 	bool m_time_format_iso_8601;
 	uint32_t m_output_timeout;
 

userspace/falco/falco_outputs.cpp

@@ -48,7 +48,6 @@ falco_outputs::falco_outputs(
 	uint32_t timeout,
 	bool buffered,
 	size_t outputs_queue_capacity,
-	falco_common::outputs_queue_recovery_type outputs_queue_recovery,
 	bool time_format_iso_8601,
 	const std::string& hostname)
 {
@@ -67,7 +66,6 @@ falco_outputs::falco_outputs(
 		add_output(output);
 	}
 	m_outputs_queue_num_drops = {0};
-	m_outputs_queue_recovery = outputs_queue_recovery;
 #ifndef __EMSCRIPTEN__
 	m_queue.set_capacity(outputs_queue_capacity);
 	m_worker_thread = std::thread(&falco_outputs::worker, this);
@@ -287,29 +285,11 @@ inline void falco_outputs::push(const ctrl_msg& cmsg)
 #ifndef __EMSCRIPTEN__
 	if (!m_queue.try_push(cmsg))
 	{
-		switch (m_outputs_queue_recovery)
+		if(m_outputs_queue_num_drops.load() == 0)
 		{
-		case falco_common::RECOVERY_EXIT:
-			throw falco_exception("Fatal error: Output queue out of memory. Exiting ...");
-		case falco_common::RECOVERY_EMPTY:
-			/* Print a log just the first time */
-			if(m_outputs_queue_num_drops.load() == 0)
-			{
-				falco_logger::log(LOG_ERR, "Output queue out of memory. Drop event plus events in queue due to emptying the queue; continue on ...");
-			}
-			m_outputs_queue_num_drops += m_queue.size() + 1;
-			m_queue.clear();
-			break;
-		case falco_common::RECOVERY_CONTINUE:
-			if(m_outputs_queue_num_drops.load() == 0)
-			{
-				falco_logger::log(LOG_ERR, "Output queue out of memory. Drop event and continue on ...");
-			}
-			m_outputs_queue_num_drops++;
-			break;
-		default:
-			throw falco_exception("Fatal error: strategy unknown. Exiting ...");
+			falco_logger::log(LOG_ERR, "Outputs queue out of memory. Drop event and continue on ...");
 		}
+		m_outputs_queue_num_drops++;
 	}
 #else
 	for (auto o : m_outputs)

userspace/falco/falco_outputs.h

@@ -36,7 +36,7 @@ limitations under the License.
 
 	All methods in this class are thread-safe. The output framework supports
 	a multi-producer model where messages are stored in a queue and consumed
-	by each configured output asynchrounously.
+	by each configured output asynchronously.
 */
 class falco_outputs
 {
@@ -50,7 +50,6 @@ public:
 		uint32_t timeout,
 		bool buffered,
 		size_t outputs_queue_capacity,
-		falco_common::outputs_queue_recovery_type outputs_queue_recovery,
 		bool time_format_iso_8601,
 		const std::string& hostname);
 
@@ -87,8 +86,8 @@ public:
 	void reopen_outputs();
 
 	/*!
-		\brief Return the number of currently dropped events as a result of failed push attempts
-		into the outputs queue when using `continue` or `empty` recovery strategies.
+		\brief Return the number of events currently dropped due to failed push 
+		attempts into the outputs queue
 	*/
 	uint64_t get_outputs_queue_num_drops();
 
@@ -121,7 +120,6 @@ private:
 	falco_outputs_cbq m_queue;
 #endif
 
-	falco_common::outputs_queue_recovery_type m_outputs_queue_recovery;
 	std::atomic<uint64_t> m_outputs_queue_num_drops;
 	std::thread m_worker_thread;
 	inline void push(const ctrl_msg& cmsg);

userspace/falco/stats_writer.cpp

@@ -34,6 +34,12 @@ limitations under the License.
 // check that this value changed since their last observation.
 static std::atomic<stats_writer::ticker_t> s_timer((stats_writer::ticker_t) 0);
 static timer_t s_timerid;
+// note: Workaround for older GLIBC versions (< 2.35), where calling timer_delete() 
+// with an invalid timer ID not returned by timer_create() causes a segfault because of 
+// a bug in GLIBC (https://sourceware.org/bugzilla/show_bug.cgi?id=28257).
+// Just performing a nullptr check is not enough as even after creating the timer, s_timerid
+// remains a nullptr somehow.
+bool s_timerid_exists = false;
 
 static void timer_handler(int signum)
 {
@@ -60,18 +66,31 @@ bool stats_writer::init_ticker(uint32_t interval_msec, std::string &err)
 	sev.sigev_value.sival_ptr = &s_timerid;
 #ifndef __EMSCRIPTEN__
 	// delete any previously set timer
-	timer_delete(s_timerid);
-	if (timer_create(CLOCK_MONOTONIC, &sev, &s_timerid) == -1) {
+	if (s_timerid_exists)
+	{
+		if (timer_delete(s_timerid) == -1) 
+		{
+			err = std::string("Could not delete previous timer: ") + strerror(errno);
+			return false;
+		}
+		s_timerid_exists = false;
+	}
+
+	if (timer_create(CLOCK_MONOTONIC, &sev, &s_timerid) == -1) 
+	{
 		err = std::string("Could not create periodic timer: ") + strerror(errno);
 		return false;
 	}
+	s_timerid_exists = true;
+
 #endif
 	timer.it_value.tv_sec = interval_msec / 1000;
 	timer.it_value.tv_nsec = (interval_msec % 1000) * 1000 * 1000;
 	timer.it_interval = timer.it_value;
 
 #ifndef __EMSCRIPTEN__
-	if (timer_settime(s_timerid, 0, &timer, NULL) == -1) {
+	if (timer_settime(s_timerid, 0, &timer, NULL) == -1) 
+	{
 		err = std::string("Could not set up periodic timer: ") + strerror(errno);
 		return false;
 	}
@@ -113,8 +132,7 @@ stats_writer::stats_writer(
 	if (m_initialized)
 	{
 #ifndef __EMSCRIPTEN__
-		// capacity and controls should not be relevant for stats outputs, adopt capacity
-		// for completeness, but do not implement config recovery strategies.
+		// Adopt capacity for completeness, even if it's likely not relevant
 		m_queue.set_capacity(config->m_outputs_queue_capacity);
 		m_worker = std::thread(&stats_writer::worker, this);
 #endif
@@ -134,7 +152,11 @@ stats_writer::~stats_writer()
 		}
 		// delete timerID and reset timer
 #ifndef __EMSCRIPTEN__
-		timer_delete(s_timerid);
+		if (s_timerid_exists)
+		{
+			timer_delete(s_timerid);
+			s_timerid_exists = false;
+		}
 #endif
 	}
 }