Branch HEAD

b=15576
i=johann, adilger

Description: Resolve device initialization race
Details    : Prevent proc handler from accessing devices added to the
             obd_devs array but yet be intialized.

diff --git a/lustre/ChangeLog b/lustre/ChangeLog
index 3c698ef..e1ab172 100644 (file)
--- a/lustre/ChangeLog
+++ b/lustre/ChangeLog
@@ -12,6 +12,12 @@ tbd  Sun Microsystems, Inc.
        * RHEL 4 and RHEL 5/SLES 10 clients behaves differently on 'cd' to a
         removed cwd "./" (refer to Bugzilla 14399).
 
+Severity   : normal
+Bugzilla   : 15576
+Description: Resolve device initialization race
+Details    : Prevent proc handler from accessing devices added to the
+             obd_devs array but yet be intialized.
+
 Severity   : enhancement
 Bugzilla   : 15308
 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.

diff --git a/lustre/obdclass/genops.c b/lustre/obdclass/genops.c
index fcb0e1b..8fd27b0 100644 (file)
--- a/lustre/obdclass/genops.c
+++ b/lustre/obdclass/genops.c
@@ -254,6 +254,17 @@ int class_unregister_type(const char *name)
         RETURN(0);
 } /* class_unregister_type */
 
+/**
+ * Create a new obd device.
+ *
+ * Find an empty slot in ::obd_devs[], create a new obd device in it.
+ *
+ * \param typename [in] obd device type string.
+ * \param name     [in] obd device name.
+ *
+ * \retval NULL if create fails, otherwise return the obd device
+ *         pointer created.
+ */
 struct obd_device *class_newdev(const char *type_name, const char *name)
 {
         struct obd_device *result = NULL;
@@ -262,7 +273,7 @@ struct obd_device *class_newdev(const char *type_name, const char *name)
         int i;
         int new_obd_minor = 0;
 
-        if (strlen(name) > MAX_OBD_NAME) {
+        if (strlen(name) >= MAX_OBD_NAME) {
                 CERROR("name/uuid must be < %u bytes long\n", MAX_OBD_NAME);
                 RETURN(ERR_PTR(-EINVAL));
         }
@@ -305,7 +316,8 @@ struct obd_device *class_newdev(const char *type_name, const char *name)
                         result->obd_minor = i;
                         new_obd_minor = i;
                         result->obd_type = type;
-                        memcpy(result->obd_name, name, strlen(name));
+                        strncpy(result->obd_name, name,
+                                sizeof(result->obd_name));
                         obd_devs[i] = result;
                 }
         }
@@ -409,15 +421,22 @@ struct obd_device *class_uuid2obd(struct obd_uuid *uuid)
         return class_num2obd(dev);
 }
 
+/**
+ * Get obd device from ::obd_devs[]
+ *
+ * \param num [in] array index
+ * 
+ * \retval NULL if ::obd_devs[\a num] does not contains an obd device
+ *         otherwise return the obd device there.
+ */
 struct obd_device *class_num2obd(int num)
 {
         struct obd_device *obd = NULL;
 
         if (num < class_devno_max()) {
                 obd = obd_devs[num];
-                if (obd == NULL) {
+                if (obd == NULL)
                         return NULL;
-                }
 
                 LASSERTF(obd->obd_magic == OBD_DEVICE_MAGIC,
                          "%p obd_magic %08x != %08x\n",

diff --git a/lustre/obdclass/linux/linux-module.c b/lustre/obdclass/linux/linux-module.c
index 714a714..137449c 100644 (file)
--- a/lustre/obdclass/linux/linux-module.c
+++ b/lustre/obdclass/linux/linux-module.c
@@ -262,6 +262,21 @@ int obd_proc_read_pinger(char *page, char **start, off_t off, int count,
                        );
 }
 
+/**
+ * Check all obd devices health
+ *
+ * \param page
+ * \param start
+ * \param off
+ * \param count
+ * \param eof
+ * \param data
+ *                  proc read function parameters, please refer to kernel
+ *                  code fs/proc/generic.c proc_file_read()
+ * \param data [in] unused
+ *
+ * \retval number of characters printed
+ */
 static int obd_proc_read_health(char *page, char **start, off_t off,
                                 int count, int *eof, void *data)
 {
@@ -276,7 +291,7 @@ static int obd_proc_read_health(char *page, char **start, off_t off,
                 struct obd_device *obd;
 
                 obd = class_num2obd(i);
-                if (obd == NULL)
+                if (obd == NULL || !obd->obd_attached || !obd->obd_set_up)
                         continue;
 
                 LASSERT(obd->obd_magic == OBD_DEVICE_MAGIC);

diff --git a/lustre/obdclass/obd_mount.c b/lustre/obdclass/obd_mount.c
index 8c8da6f..3ba46c0 100644 (file)
--- a/lustre/obdclass/obd_mount.c
+++ b/lustre/obdclass/obd_mount.c
@@ -556,7 +556,12 @@ static int server_stop_mgs(struct super_block *sb)
 
 DECLARE_MUTEX(mgc_start_lock);
 
-/* Set up a mgcobd to process startup logs */
+/** Set up a mgc obd to process startup logs
+ *
+ * \param sb [in] super block of the mgc obd
+ *
+ * \retval 0 success, otherwise error code
+ */
 static int lustre_start_mgc(struct super_block *sb)
 {
         struct lustre_handle mgc_conn = {0, };
@@ -613,7 +618,7 @@ static int lustre_start_mgc(struct super_block *sb)
         mutex_down(&mgc_start_lock);
 
         obd = class_name2obd(mgcname);
-        if (obd) {
+        if (obd && !obd->obd_stopping) {
                 /* Re-using an existing MGC */
                 atomic_inc(&obd->u.cli.cl_mgc_refcount);