From fa6ec315c83a93ffc7ea4bdf64e0d0527f81a145 Mon Sep 17 00:00:00 2001
From: Asger Gitz-Johansen <asger.gitz@hotmail.com>
Date: Thu, 29 Aug 2024 20:51:11 +0200
Subject: [PATCH] feat: unique directory for pipelines

Also, start using fork/chdir/exec idiom instead of posix_spawn, because
as we all know: posix_spawn is stupid as a system call:

https://lwn.net/Articles/360556/
---
 include/cli.h      |  1 +
 include/executor.h |  1 +
 src/cli.c          | 15 ++++++++++++---
 src/executor.c     | 36 +++++++++++++++++++++++++++++-------
 src/main.c         |  2 ++
 5 files changed, 45 insertions(+), 10 deletions(-)

diff --git a/include/cli.h b/include/cli.h
index f3f511c..348dda9 100644
--- a/include/cli.h
+++ b/include/cli.h
@@ -29,6 +29,7 @@ typedef struct {
     bool use_colors;
     optional_str log_file;
     optional_str pipeline_log_dir;
+    optional_str pipeline_cwd;
     optional_strlist environment_vars;
 } cli_options;
 
diff --git a/include/executor.h b/include/executor.h
index c32b8b7..9bc3887 100644
--- a/include/executor.h
+++ b/include/executor.h
@@ -4,6 +4,7 @@
 
 void executor(void* pipeline_event);
 void set_logdir(const char* logdir);
+void set_working_directory(const char* cwd);
 void set_shared_environment(const strlist_node* root);
 
 #endif
diff --git a/src/cli.c b/src/cli.c
index 317b936..fe4e010 100644
--- a/src/cli.c
+++ b/src/cli.c
@@ -51,6 +51,10 @@ cli_options new_options() {
     result.pipeline_log_dir.has_value = pipeline_log_dir != NULL;
     result.pipeline_log_dir.value = pipeline_log_dir;
 
+    char* pipeline_cwd = getenv("SCI_PIPELINE_CWD");
+    result.pipeline_cwd.has_value = pipeline_cwd != NULL;
+    result.pipeline_cwd.value = pipeline_cwd;
+
     char* environment_vars = getenv("SCI_PIPELINE_ENV_VARS");
     if(environment_vars == NULL) {
         result.environment_vars.has_value = false;
@@ -78,17 +82,18 @@ void destroy_options(cli_options v) {
 }
 
 //                                                         <max
-const char* optstring = "f:L:w:v:Cl:e:hV";
+const char* optstring = "f:L:P:w:v:Cl:e:hV";
 const char* help_msg = 
     "%s %s\n"
-    "Usage: [-f file] [-L dir] [-w count] [-v level] \n"
-    "       [-C] [-l file] [-e ENV] [-h] [-V]\n"
+    "Usage: [-f file] [-L dir] [-P dir] [-w count]\n\n"
+    "       [-v level] [-C] [-l file] [-e ENV] [-h] [-V]\n"
     "\n"
     SCI_DESCRIPTION "\n"
     "\n"
     "OPTIONS:\n"
     "  -f file     Set sci config file\n"
     "  -L dir      Set pipeline log output directory\n"
+    "  -P dir      Set pipeline working directory prefix\n"
     "  -w count    Set the amount of worker threads\n"
     "  -v level    Set verbosity level [0-4]\n"
     "  -C          Force color output, ignoring $NO_COLOR\n"
@@ -119,6 +124,10 @@ cli_options parse(int argc, char** argv) {
                 options.pipeline_log_dir.value = strdup(optarg);
                 options.pipeline_log_dir.has_value = true;
                 break;
+            case 'P':
+                options.pipeline_cwd.value = strdup(optarg);
+                options.pipeline_cwd.has_value = true;
+                break;
             case 'v':
                 options.verbosity = atoi(optarg);
                 break;
diff --git a/src/executor.c b/src/executor.c
index dc53ca5..719f3cf 100644
--- a/src/executor.c
+++ b/src/executor.c
@@ -15,6 +15,7 @@
 #include <uuid/uuid.h>
 
 const char* log_dir = ".";
+const char* cwd = "/tmp";
 const strlist_node* shared_environment = NULL;
 
 void set_shared_environment(const strlist_node* root) {
@@ -28,6 +29,13 @@ void set_logdir(const char* logdir) {
         mkdir(log_dir, 0700);
 }
 
+void set_working_directory(const char* _cwd) {
+    cwd = _cwd;
+    struct stat st = {0};
+    if(stat(cwd, &st) == -1)
+        mkdir(cwd, 0700);
+}
+
 char* create_pipeline_id() {
     uuid_t uuid;
     uuid_generate(uuid);
@@ -98,10 +106,6 @@ void executor(void* data) {
 
     // spawn the process
     pid_t pid;
-    posix_spawn_file_actions_t actions;
-    posix_spawn_file_actions_init(&actions);
-    posix_spawn_file_actions_adddup2(&actions, fd.value, STDOUT_FILENO);
-    posix_spawn_file_actions_adddup2(&actions, fd.value, STDERR_FILENO);
     const pipeline_event* const e = data;
     char** envp = create_environment(e, pipeline_id);
     int argc;
@@ -112,16 +116,34 @@ void executor(void* data) {
     char arg0[PATH_MAX];
     if(which(argv[0], arg0, PATH_MAX) == -1)
         goto end;
-    if(posix_spawnp(&pid, arg0, &actions, NULL, argv, envp) != 0) {
-        perror("posix_spawn");
+
+    // fork / cwd / exec idiom
+    pid = fork();
+    if(pid < 0) {
+        perror("fork");
         goto end; // I know. The raptors have picked up the scent. I'll just have to mask it with more stinky code.
     }
+
+    if(pid == 0) {
+        // child process
+        dup2(fd.value, STDOUT_FILENO);
+        dup2(fd.value, STDERR_FILENO);
+        char* pipeline_cwd = join3(cwd, "/", pipeline_id);
+        struct stat st = {0};
+        if(stat(pipeline_cwd, &st) == -1)
+            mkdir(pipeline_cwd, 0700);
+        chdir(pipeline_cwd);
+        free(pipeline_cwd);
+        execvpe(arg0, argv, envp);
+        return;
+    }
+
     log_info("{%s} (%s) spawned", pipeline_id, e->name);
 
     // Wait for process to complete
     int status;
     waitpid(pid, &status, 0);
-    log_info("{%s} (%s) exited with status %d", pipeline_id, e->name, status);
+    log_info("{%s} (%s) [pid=%d] exited with status %d", pipeline_id, e->name, pid, status);
     char buf[32];
     sprintf(buf, "exited with status %d", status);
     if(write(fd.value, buf, strnlen(buf, 32)) == -1)
diff --git a/src/main.c b/src/main.c
index e0b41a6..d69f572 100644
--- a/src/main.c
+++ b/src/main.c
@@ -101,6 +101,8 @@ int main(int argc, char** argv) {
 
     if(args.pipeline_log_dir.has_value)
         set_logdir(args.pipeline_log_dir.value);
+    if(args.pipeline_cwd.has_value)
+        set_working_directory(args.pipeline_cwd.value);
 
     struct stat st = {0};
     if(stat("/tmp/sci", &st) == -1)