X-Git-Url: http://git.tuebingen.mpg.de/?p=paraslash.git;a=blobdiff_plain;f=server.c;h=e08b0661d14ddc4a8ef1cf2386cb75efe0655d19;hp=79e67818c87a1573b312c00091c8199f41c286ed;hb=337d7784b0f8bc30b1fa9af56ace503acd97d40d;hpb=b27f3e8dd53af7209626a9f0e98eb4770602c2e7

diff --git a/server.c b/server.c
index 79e67818..e08b0661 100644
--- a/server.c
+++ b/server.c
@@ -14,7 +14,7 @@
  *
  *
  * 	- The main programs: \ref server.c, \ref audiod.c, \ref client.c,
- *	  \ref audioc.c, \ref fsck.c, \ref afh.c
+ *	  \ref audioc.c, \ref afh.c
  *	- Server: \ref server_command, \ref sender,
  *	- Audio file selector: \ref audio_format_handler, \ref mood, \ref afs_table,
  *	- Client: \ref receiver, \ref receiver_node, \ref filter, \ref filter_node.
@@ -47,7 +47,6 @@
  *	- Time: \ref time.c,
  *	- Spawning processes: \ref exec.c,
  *	- Inter process communication: \ref ipc.c,
- *	- The object storage layer: \ref osl.c,
  *	- Blob tables: \ref blob.c,
  *	- The error subssystem: \ref error.h.
  *	- Access control for paraslash senders: \ref acl.c, \ref acl.h.
@@ -55,16 +54,16 @@
  * Low-level data structures:
  *
  *	- Doubly linked lists: \ref list.h,
- *	- Red-black trees: \ref rbtree.h, \ref rbtree.c,
  *	- Ring buffer: \ref ringbuffer.c, \ref ringbuffer.h,
  *	- Hashing: \ref hash.h, \ref sha1.h, \ref sha1.c,
  *	- Crypto: \ref crypt.c.
- *
+ *	- Forward error correction: \ref fec.c
  */
 
 #include <signal.h>
 #include <dirent.h>
 #include <sys/time.h>
+#include <osl.h>
 
 #include "para.h"
 #include "error.h"
@@ -291,17 +290,33 @@ static void signal_post_select(struct sched *s, struct task *t)
 			if (pid != mmd->afs_pid)
 				continue;
 			PARA_EMERG_LOG("fatal: afs died\n");
-			goto genocide;
+			kill(0, SIGTERM);
+			goto cleanup;
 		}
 		break;
 	/* die on sigint/sigterm. Kill all children too. */
 	case SIGINT:
 	case SIGTERM:
 		PARA_EMERG_LOG("terminating on signal %d\n", st->signum);
-genocide:
 		kill(0, SIGTERM);
+		/*
+		 * We must wait for afs because afs catches SIGINT/SIGTERM.
+		 * Before reacting to the signal, afs might want to use the
+		 * shared memory area and the mmd mutex.  If we destroy this
+		 * mutex too early and afs tries to lock the shared memory
+		 * area, the call to mutex_lock() will fail and terminate the
+		 * afs process. This leads to dirty osl tables.
+		 *
+		 * There's no such problem with the other children of the
+		 * server process (the command handlers) as these reset their
+		 * SIGINT/SIGTERM handlers to the default action, i.e.  these
+		 * processes get killed immediately by the above kill().
+		 */
+		PARA_INFO_LOG("waiting for afs (pid %d) to die\n",
+			(int)mmd->afs_pid);
+		waitpid(mmd->afs_pid, NULL, 0);
+cleanup:
 		free(mmd->afd.afhi.chunk_table);
-		free(mmd->afd.afhi.info_string);
 		close_listed_fds();
 		mutex_destroy(mmd_mutex);
 		shm_detach(mmd);
@@ -342,6 +357,7 @@ static void command_post_select(struct sched *s, struct task *t)
 	int new_fd, ret, i;
 	char *peer_name;
 	pid_t child_pid;
+	uint32_t *chunk_table;
 
 	if (!FD_ISSET(sct->listen_fd, &s->rfds))
 		return;
@@ -354,6 +370,16 @@ static void command_post_select(struct sched *s, struct task *t)
 	mmd->num_connects++;
 	mmd->active_connections++;
 	random();
+	/*
+	 * The chunk table is a pointer located in the mmd struct that point to
+	 * dynamically allocated memory that must be freed by the parent and
+	 * the child. However, as the mmd struct is in a shared memory area,
+	 * there's no guarantee that after the fork these pointers are still
+	 * valid in child context. As this pointer is not used in the child
+	 * anyway, we save it to a local variable and free the memory via that
+	 * copy in the child.
+	 */
+	chunk_table = mmd->afd.afhi.chunk_table;
 	child_pid = fork();
 	if (child_pid < 0) {
 		ret = -ERRNO_TO_PARA_ERROR(errno);
@@ -364,6 +390,8 @@ static void command_post_select(struct sched *s, struct task *t)
 		/* parent keeps accepting connections */
 		return;
 	}
+	/* mmd might already have changed at this point */
+	free(chunk_table);
 	alarm(ALARM_TIMEOUT);
 	close_listed_fds();
 	para_signal_shutdown();