zc: read jar file path and service name from a config file (open-tele…

…metry#1255)

* Read jar file path and service name from a config file
* zc: rename config example file

instrumentation/Dockerfile

@@ -6,4 +6,5 @@ RUN apt-get update && \
 WORKDIR /libsplunk
 
 COPY src /libsplunk/src
+COPY testdata/splunk.conf /libsplunk/testdata/splunk.conf
 COPY Makefile /libsplunk/Makefile

instrumentation/Makefile

@@ -13,17 +13,20 @@ obj:
 clean:
 	rm -f tests so/* obj/*
 
-obj/logger.o: obj src/logger.c
+obj/logger.o: obj src/logger.c src/logger.h
 	gcc -c -Wall -Werror -fpic -o obj/logger.o src/logger.c
 
-obj/splunk.o: obj src/splunk.c
+obj/config.o: obj src/config.c src/config.h
+	gcc -c -Wall -Werror -fpic -o obj/config.o src/config.c
+
+obj/splunk.o: obj src/splunk.c src/splunk.h
 	gcc -c -Wall -Werror -fpic -o obj/splunk.o src/splunk.c
 
-so/libsplunk.so: obj so obj/splunk.o obj/logger.o
-	gcc -shared -o so/libsplunk.so obj/splunk.o obj/logger.o
+so/libsplunk.so: obj so obj/splunk.o obj/logger.o obj/config.o
+	gcc -shared -o so/libsplunk.so obj/splunk.o obj/logger.o obj/config.o
 
-tests: src/splunk.c src/test_main.c
-	gcc -o tests src/test_main.c src/logger_test.c src/splunk.c
+tests: src/test_main.h src/test_main.c src/logger.h src/logger_test.c obj/config.o obj/splunk.o
+	gcc -o tests src/test_main.c src/logger_test.c obj/config.o obj/splunk.o
 
 .PHONY: test
 test: tests

instrumentation/src/config.c

@@ -0,0 +1,79 @@
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include "config.h"
+#include "splunk.h"
+
+struct kv {
+    char *k;
+    char *v;
+};
+
+void read_config_file(logger log, struct config *cfg, char *file_name);
+
+void read_lines(struct config *cfg, FILE *fp);
+
+void split_on_eq(char *string, struct kv *pair);
+
+void load_config(logger log, struct config *cfg, char *file_name) {
+    read_config_file(log, cfg, file_name);
+    char log_line[MAX_LOG_LINE_LEN];
+    if (cfg->service_name == NULL) {
+        strcpy(log_line, "service_name not found in config, using default: ");
+        strncat(log_line, default_service_name, MAX_LOG_LINE_LEN - strlen(log_line) - 1);
+        log_debug(log, log_line);
+        cfg->service_name = strdup(default_service_name);
+    }
+    if (cfg->java_agent_jar == NULL) {
+        strcpy(log_line, "java_agent_jar not found in config, using default: ");
+        strncat(log_line, default_jar_path, MAX_LOG_LINE_LEN - strlen(log_line) - 1);
+        log_debug(log, log_line);
+        cfg->java_agent_jar = strdup(default_jar_path);
+    }
+}
+
+void read_config_file(logger log, struct config *cfg, char *file_name) {
+    FILE *fp = fopen(file_name, "r");
+    char log_line[MAX_LOG_LINE_LEN];
+    if (fp == NULL) {
+        strcpy(log_line, "file not found: ");
+        strncat(log_line, file_name, MAX_LOG_LINE_LEN - strlen(log_line) - 1);
+        log_debug(log, log_line);
+        return;
+    }
+
+    strcpy(log_line, "reading config file: ");
+    strncat(log_line, file_name, MAX_LOG_LINE_LEN - strlen(log_line) - 1);
+    log_debug(log, log_line);
+    read_lines(cfg, fp);
+    fclose(fp);
+}
+
+void read_lines(struct config *cfg, FILE *fp) {
+    static const int buflen = 255;
+    char buf[buflen];
+    struct kv pair = {.k = NULL, .v = NULL};
+    while ((fgets(buf, buflen, fp)) != NULL) {
+        buf[strcspn(buf, "\n")] = 0;
+        split_on_eq(buf, &pair);
+        if (streq(pair.k, "java_agent_jar")) {
+            cfg->java_agent_jar = strdup(pair.v);
+        } else if (streq(pair.k, "service_name")) {
+            cfg->service_name = strdup(pair.v);
+        }
+    }
+}
+
+void split_on_eq(char *string, struct kv *pair) {
+    pair->k = strsep(&string, "=");
+    pair->v = string;
+}
+
+void free_config(struct config *cfg) {
+    if (cfg->java_agent_jar != NULL) {
+        free(cfg->java_agent_jar);
+    }
+    if (cfg->service_name != NULL) {
+        free(cfg->service_name);
+    }
+}

instrumentation/src/config.h

@@ -0,0 +1,17 @@
+#ifndef INSTRUMENTATION_CONFIG_H
+#define INSTRUMENTATION_CONFIG_H
+
+#include <stdbool.h>
+
+#include "logger.h"
+
+struct config {
+    char *java_agent_jar;
+    char *service_name;
+};
+
+void load_config(logger log, struct config *cfg, char *file_name);
+
+void free_config(struct config *cfg);
+
+#endif //INSTRUMENTATION_CONFIG_H

instrumentation/src/logger.c

@@ -20,6 +20,10 @@ void log_info(logger l, char *s) {
     syslog(LOG_INFO, prefix, s);
 }
 
+void log_warning(logger l, char *s) {
+    syslog(LOG_WARNING, prefix, s);
+}
+
 void free_logger(logger l) {
     free(l);
 }

instrumentation/src/logger.h

@@ -1,7 +1,7 @@
 #ifndef SPLUNK_INSTRUMENTATION_LOGGER_H
 #define SPLUNK_INSTRUMENTATION_LOGGER_H
 
-#define TEST_LOGS_MAX_LEN 256
+#define MAX_LOG_LINE_LEN 1024
 
 typedef struct logger_impl *logger;
 
@@ -11,7 +11,9 @@ void log_debug(logger l, char *s);
 
 void log_info(logger l, char *s);
 
-int get_logs(logger l, char *buf[256]);
+void log_warning(logger l, char *s);
+
+int get_logs(logger l, char **buf);
 
 void free_logger(logger l);
 

instrumentation/src/logger_test.c

@@ -1,9 +1,10 @@
 #include <stdlib.h>
+#include <string.h>
 #include "logger.h"
 
 struct logger_impl {
     int i;
-    char *logs[TEST_LOGS_MAX_LEN];
+    char *logs[MAX_LOG_LINE_LEN];
 };
 
 logger new_logger() {
@@ -14,15 +15,26 @@ logger new_logger() {
     return ref;
 }
 
+void save_log(logger l, char *s) {
+    l->logs[l->i++] = strdup(s);
+}
+
 void log_info(logger l, char *s) {
-    l->logs[l->i++] = s;
+    save_log(l, s);
 }
 
 void log_debug(logger l, char *s) {
-    l->logs[l->i++] = s;
+    save_log(l, s);
+}
+
+void log_warning(logger l, char *s) {
+    save_log(l, s);
 }
 
 void free_logger(logger l) {
+    for (int i = 0; i < l->i; ++i) {
+        free(l->logs[i]);
+    }
     free(l);
 }
 

instrumentation/src/splunk.c

@@ -1,55 +1,104 @@
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include <stdio.h>
 
+#include "config.h"
 #include "splunk.h"
 
-#define FLAG_LEN 256
+#define ENV_VAR_LEN 512
+#define MAX_SERVICE_NAME_LEN 256
+#define MAX_JAR_PATH_LEN 256
 
-#define FLAG_PREFIX "-javaagent:";
+#define JAVA_TOOL_OPTIONS_PREFIX "-javaagent:";
+#define OTEL_RESOURCE_ATTRIBUTES_PREFIX "service.name="
 
 extern char *program_invocation_short_name;
-char *const AGENT_JAR_PATH = "/usr/lib/splunk-instrumentation/splunk-otel-javaagent.jar";
+
+bool has_read_access(const char *s);
+
+void set_java_tool_options(logger log, struct config *cfg);
+
+void set_service_name(logger log, struct config *cfg);
+
+bool is_disable_env_set();
 
 // The entry point for all executables prior to their execution. If the executable is named "java", we
 // set JAVA_TOOL_OPTIONS to the path of the java agent jar.
 void __attribute__((constructor)) splunk_instrumentation_enter() {
     logger l = new_logger();
-    auto_instrument(l, has_read_access, program_invocation_short_name, AGENT_JAR_PATH);
+    auto_instrument(l, has_read_access, program_invocation_short_name, load_config);
     free_logger(l);
 }
 
-void auto_instrument(logger log, has_access_t has_access, const char *program_name, const char *jar_path) {
-    if (strcmp(program_name, "java") != 0) {
+void auto_instrument(logger log, has_access_func_t has_access, const char *program_name, load_config_func_t load_config_func) {
+    if (!streq(program_name, "java")) {
         return;
     }
     if (is_disable_env_set()) {
         log_debug(log, "disable_env_set, quitting");
         return;
     }
-    if (!has_access(jar_path)) {
+
+    struct config cfg = {.java_agent_jar = NULL, .service_name = NULL};
+    load_config_func(log, &cfg, conf_file);
+
+    if (!has_access(cfg.java_agent_jar)) {
         log_info(log, "agent jar not found or no read access, quitting");
         return;
     }
-    char opts[FLAG_LEN] = FLAG_PREFIX;
-    if (strlen(opts) + strlen(jar_path) > FLAG_LEN) {
-        log_info(log, "jar path too long");
+
+    set_java_tool_options(log, &cfg);
+    set_service_name(log, &cfg);
+
+    free_config(&cfg);
+}
+
+void set_service_name(logger log, struct config *cfg) {
+    char otel_resource_attributes[ENV_VAR_LEN] = OTEL_RESOURCE_ATTRIBUTES_PREFIX;
+    char log_line[MAX_LOG_LINE_LEN];
+    size_t service_name_len = strlen(((*cfg).service_name));
+    if (service_name_len > MAX_SERVICE_NAME_LEN) {
+        sprintf(log_line, "service_name too long: got %zu chars, max %d chars", service_name_len, MAX_SERVICE_NAME_LEN);
+        log_warning(log, log_line);
         return;
     }
-    strncat(opts, jar_path, FLAG_LEN);
-    log_debug(log, "setting JAVA_TOOL_OPTIONS");
-    setenv("JAVA_TOOL_OPTIONS", opts, 0);
+    strcat(otel_resource_attributes, (*cfg).service_name);
+    sprintf(log_line, "setting OTEL_RESOURCE_ATTRIBUTES='%s'", otel_resource_attributes);
+    log_debug(log, log_line);
+    setenv("OTEL_RESOURCE_ATTRIBUTES", otel_resource_attributes, 0);
 }
 
-bool streq(char *a, char *b) {
-    return strcmp(a, b) == 0;
+void set_java_tool_options(logger log, struct config *cfg) {
+    char java_tool_options[ENV_VAR_LEN] = JAVA_TOOL_OPTIONS_PREFIX;
+    char log_line[MAX_LOG_LINE_LEN];
+    size_t jar_path_len = strlen(cfg->java_agent_jar);
+    if (jar_path_len > MAX_JAR_PATH_LEN) {
+        sprintf(log_line, "jar_path too long: got %zu chars, max %d chars", jar_path_len, MAX_JAR_PATH_LEN);
+        log_warning(log, log_line);
+        return;
+    }
+    strcat(java_tool_options, (*cfg).java_agent_jar);
+    sprintf(log_line, "setting JAVA_TOOL_OPTIONS='%s'", java_tool_options);
+    log_debug(log, log_line);
+    setenv("JAVA_TOOL_OPTIONS", java_tool_options, 0);
 }
 
 bool is_disable_env_set() {
-    char *env = getenv(DISABLE_ENV_VAR_NAME);
+    char *env = getenv(disable_env_var_name);
     return env && !streq("false", env) && !streq("FALSE", env) && !streq("0", env);
 }
 
 bool has_read_access(const char *s) {
     return access(s, R_OK) == 0;
 }
+
+bool streq(const char *expected, const char *actual) {
+    if (expected == NULL && actual == NULL) {
+        return true;
+    }
+    if (expected == NULL || actual == NULL) {
+        return false;
+    }
+    return strcmp(expected, actual) == 0;
+}

instrumentation/src/splunk.h

@@ -3,15 +3,22 @@
 
 #include <stdbool.h>
 #include "logger.h"
+#include "config.h"
 
-static char *const DISABLE_ENV_VAR_NAME = "DISABLE_SPLUNK_AUTOINSTRUMENTATION";
+static char *const conf_file = "/usr/lib/splunk-instrumentation/splunk.conf";
 
-typedef bool (*has_access_t)(const char *);
+static char *const default_jar_path = "/usr/lib/splunk-instrumentation/splunk-otel-javaagent.jar";
 
-void auto_instrument(logger log, has_access_t has_access, const char *program_name, const char *jar_path);
+static char *const default_service_name = "default.service";
 
-bool has_read_access(const char *s);
+static char *const disable_env_var_name = "DISABLE_SPLUNK_AUTOINSTRUMENTATION";
 
-bool is_disable_env_set();
+typedef bool (*has_access_func_t)(const char *);
+
+typedef void (*load_config_func_t)(logger log, struct config *, char *);
+
+void auto_instrument(logger log, has_access_func_t has_access, const char *program_name, load_config_func_t load_config_func);
+
+bool streq(const char *expected, const char *actual);
 
 #endif //SPLUNK_INSTRUMENTATION_SPLUNK_H