#include <sys/types.h>
#include <fcntl.h>
#include <errno.h>
-
-/*
- * libsysfs
- */
-#include <sysfs/libsysfs.h>
-#include <sysfs/dlist.h>
+#include <sys/time.h>
+#include <sys/resource.h>
/*
* libcheckers
#include <structs_vec.h>
#include <dmparser.h>
#include <devmapper.h>
+#include <sysfs.h>
#include <dict.h>
#include <discovery.h>
#include <debug.h>
#include <switchgroup.h>
#include <print.h>
#include <configure.h>
+#include <prio.h>
#include "main.h"
#include "pidfile.h"
#include "uxclnt.h"
#include "cli.h"
#include "cli_handlers.h"
+#include "lock.h"
+#include "waiter.h"
#define FILE_NAME_SIZE 256
#define CMDSIZE 160
#define LOG_MSG(a,b) \
- if (strlen(b)) condlog(a, "%s: %s", pp->dev_t, b);
-
-#ifdef LCKDBG
-#define lock(a) \
- fprintf(stderr, "%s:%s(%i) lock %p\n", __FILE__, __FUNCTION__, __LINE__, a); \
- pthread_mutex_lock(a)
-#define unlock(a) \
- fprintf(stderr, "%s:%s(%i) unlock %p\n", __FILE__, __FUNCTION__, __LINE__, a); \
- pthread_mutex_unlock(a)
-#define lock_cleanup_pop(a) \
- fprintf(stderr, "%s:%s(%i) unlock %p\n", __FILE__, __FUNCTION__, __LINE__, a); \
- pthread_cleanup_pop(1);
-#else
-#define lock(a) pthread_mutex_lock(a)
-#define unlock(a) pthread_mutex_unlock(a)
-#define lock_cleanup_pop(a) pthread_cleanup_pop(1);
-#endif
+ if (strlen(b)) condlog(a, "%s: %s", pp->dev, b);
pthread_cond_t exit_cond = PTHREAD_COND_INITIALIZER;
pthread_mutex_t exit_mutex = PTHREAD_MUTEX_INITIALIZER;
-/*
- * structs
- */
-struct vectors * gvecs; /* global copy of vecs for use in sig handlers */
-
-static struct event_thread *
-alloc_waiter (void)
-{
-
- struct event_thread * wp;
-
- wp = (struct event_thread *)MALLOC(sizeof(struct event_thread));
-
- return wp;
-}
-
-static void
-free_waiter (void * data)
-{
- struct event_thread * wp = (struct event_thread *)data;
-
- /*
- * indicate in mpp that the wp is already freed storage
- */
- lock(wp->vecs->lock);
-
- if (wp->mpp)
- /*
- * be careful, mpp may already be freed -- null if so
- */
- wp->mpp->waiter = NULL;
- else
- condlog(3, "free_waiter, mpp freed before wp=%p,", wp);
-
- unlock(wp->vecs->lock);
-
- if (wp->dmt)
- dm_task_destroy(wp->dmt);
-
- FREE(wp);
-}
-
-static void
-stop_waiter_thread (struct multipath * mpp, struct vectors * vecs)
-{
- struct event_thread * wp = (struct event_thread *)mpp->waiter;
-
- if (!wp) {
- condlog(3, "%s: no waiter thread", mpp->alias);
- return;
- }
- condlog(2, "%s: stop event checker thread", wp->mapname);
- pthread_kill((pthread_t)wp->thread, SIGUSR1);
-}
-
-static void
-cleanup_lock (void * data)
-{
- unlock((pthread_mutex_t *)data);
-}
+int logsink;
/*
- * mpp->no_path_retry:
- * -2 (QUEUE) : queue_if_no_path enabled, never turned off
- * -1 (FAIL) : fail_if_no_path
- * 0 (UNDEF) : nothing
- * >0 : queue_if_no_path enabled, turned off after polling n times
+ * global copy of vecs for use in sig handlers
*/
-static void
-update_queue_mode_del_path(struct multipath *mpp)
-{
- if (--mpp->nr_active == 0 && mpp->no_path_retry > 0) {
- /*
- * Enter retry mode.
- * meaning of +1: retry_tick may be decremented in
- * checkerloop before starting retry.
- */
- mpp->stat_queueing_timeouts++;
- mpp->retry_tick = mpp->no_path_retry * conf->checkint + 1;
- condlog(1, "%s: Entering recovery mode: max_retries=%d",
- mpp->alias, mpp->no_path_retry);
- }
- condlog(2, "%s: remaining active paths: %d", mpp->alias, mpp->nr_active);
-}
-
-static void
-update_queue_mode_add_path(struct multipath *mpp)
-{
- if (mpp->nr_active++ == 0 && mpp->no_path_retry > 0) {
- /* come back to normal mode from retry mode */
- mpp->retry_tick = 0;
- dm_queue_if_no_path(mpp->alias, 1);
- condlog(2, "%s: queue_if_no_path enabled", mpp->alias);
- condlog(1, "%s: Recovered to normal mode", mpp->alias);
- }
- condlog(2, "%s: remaining active paths: %d", mpp->alias, mpp->nr_active);
-}
+struct vectors * gvecs;
static int
need_switch_pathgroup (struct multipath * mpp, int refresh)
{
struct pathgroup * pgp;
struct path * pp;
- int i, j;
+ unsigned int i, j;
if (!mpp || mpp->pgfailback == -FAILBACK_MANUAL)
return 0;
{
struct multipath * ompp;
vector ompv = vecs->mpvec;
- int i, j;
+ unsigned int i;
+ int j;
vector_foreach_slot (ompv, ompp, i) {
if (!find_mp_by_wwid(nmpv, ompp->wwid)) {
* remove all current maps not allowed by the
* current configuration
*/
- if (dm_flush_map(ompp->alias, DEFAULT_TARGET)) {
+ if (dm_flush_map(ompp->alias)) {
condlog(0, "%s: unable to flush devmap",
ompp->alias);
/*
}
else {
dm_lib_release();
- condlog(3, "%s devmap removed", ompp->alias);
+ condlog(2, "%s devmap removed", ompp->alias);
}
}
}
return 0;
}
-static int
-update_multipath (struct vectors *vecs, char *mapname)
+static void
+sync_map_state(struct multipath *mpp)
{
- struct multipath *mpp;
- struct pathgroup *pgp;
+ struct pathgroup *pgp;
struct path *pp;
- int i, j;
- int r = 1;
-
- mpp = find_mp_by_alias(vecs->mpvec, mapname);
-
- if (!mpp)
- goto out;
+ unsigned int i, j;
- free_pgvec(mpp->pg, KEEP_PATHS);
- mpp->pg = NULL;
-
- if (setup_multipath(vecs, mpp))
- goto out; /* mpp freed in setup_multipath */
+ if (!mpp->pg)
+ return;
- /*
- * compare checkers states with DM states
- */
- vector_foreach_slot (mpp->pg, pgp, i) {
- vector_foreach_slot (pgp->paths, pp, j) {
- if (pp->dmstate != PSTATE_FAILED)
+ vector_foreach_slot (mpp->pg, pgp, i){
+ vector_foreach_slot (pgp->paths, pp, j){
+ if (pp->state == PATH_UNCHECKED ||
+ pp->state == PATH_WILD)
continue;
-
- if (pp->state != PATH_DOWN) {
- condlog(2, "%s: mark as failed", pp->dev_t);
- mpp->stat_path_failures++;
- pp->state = PATH_DOWN;
- update_queue_mode_del_path(mpp);
-
- /*
- * if opportune,
- * schedule the next check earlier
- */
- if (pp->tick > conf->checkint)
- pp->tick = conf->checkint;
- }
+ if ((pp->dmstate == PSTATE_FAILED ||
+ pp->dmstate == PSTATE_UNDEF) &&
+ (pp->state == PATH_UP || pp->state == PATH_GHOST))
+ dm_reinstate_path(mpp->alias, pp->dev_t);
+ else if ((pp->dmstate == PSTATE_ACTIVE ||
+ pp->dmstate == PSTATE_UNDEF) &&
+ (pp->state == PATH_DOWN ||
+ pp->state == PATH_SHAKY))
+ dm_fail_path(mpp->alias, pp->dev_t);
}
}
- r = 0;
-out:
- if (r)
- condlog(0, "failed to update multipath");
-
- return r;
-}
-
-static sigset_t unblock_signals(void)
-{
- sigset_t set, old;
-
- sigemptyset(&set);
- sigaddset(&set, SIGHUP);
- sigaddset(&set, SIGUSR1);
- pthread_sigmask(SIG_UNBLOCK, &set, &old);
- return old;
}
-/*
- * returns the reschedule delay
- * negative means *stop*
- */
-static int
-waiteventloop (struct event_thread * waiter)
-{
- sigset_t set;
- int event_nr;
- int r;
-
- if (!waiter->event_nr)
- waiter->event_nr = dm_geteventnr(waiter->mapname);
-
- if (!(waiter->dmt = dm_task_create(DM_DEVICE_WAITEVENT))) {
- condlog(0, "%s: devmap event #%i dm_task_create error",
- waiter->mapname, waiter->event_nr);
- return 1;
- }
-
- if (!dm_task_set_name(waiter->dmt, waiter->mapname)) {
- condlog(0, "%s: devmap event #%i dm_task_set_name error",
- waiter->mapname, waiter->event_nr);
- dm_task_destroy(waiter->dmt);
- return 1;
- }
-
- if (waiter->event_nr && !dm_task_set_event_nr(waiter->dmt,
- waiter->event_nr)) {
- condlog(0, "%s: devmap event #%i dm_task_set_event_nr error",
- waiter->mapname, waiter->event_nr);
- dm_task_destroy(waiter->dmt);
- return 1;
- }
-
- dm_task_no_open_count(waiter->dmt);
-
- /* accept wait interruption */
- set = unblock_signals();
-
- /* interruption spits messages */
- dm_shut_log();
-
- /* wait */
- r = dm_task_run(waiter->dmt);
-
- /* wait is over : event or interrupt */
- pthread_sigmask(SIG_SETMASK, &set, NULL);
- //dm_restore_log();
-
- if (!r) /* wait interrupted by signal */
- return -1;
-
- dm_task_destroy(waiter->dmt);
- waiter->dmt = NULL;
- waiter->event_nr++;
-
- /*
- * upon event ...
- */
- while (1) {
- condlog(3, "%s: devmap event #%i",
- waiter->mapname, waiter->event_nr);
-
- /*
- * event might be :
- *
- * 1) a table reload, which means our mpp structure is
- * obsolete : refresh it through update_multipath()
- * 2) a path failed by DM : mark as such through
- * update_multipath()
- * 3) map has gone away : stop the thread.
- * 4) a path reinstate : nothing to do
- * 5) a switch group : nothing to do
- */
- pthread_cleanup_push(cleanup_lock, waiter->vecs->lock);
- lock(waiter->vecs->lock);
- r = update_multipath(waiter->vecs, waiter->mapname);
- lock_cleanup_pop(waiter->vecs->lock);
-
- if (r)
- return -1; /* stop the thread */
-
- event_nr = dm_geteventnr(waiter->mapname);
-
- if (waiter->event_nr == event_nr)
- return 1; /* upon problem reschedule 1s later */
-
- waiter->event_nr = event_nr;
- }
- return -1; /* never reach there */
-}
-
-static void *
-waitevent (void * et)
-{
- int r;
- struct event_thread *waiter;
-
- mlockall(MCL_CURRENT | MCL_FUTURE);
-
- waiter = (struct event_thread *)et;
- pthread_cleanup_push(free_waiter, et);
-
- while (1) {
- r = waiteventloop(waiter);
-
- if (r < 0)
- break;
-
- sleep(r);
- }
-
- pthread_cleanup_pop(1);
- return NULL;
-}
-
-static int
-start_waiter_thread (struct multipath * mpp, struct vectors * vecs)
+static void
+sync_maps_state(vector mpvec)
{
- pthread_attr_t attr;
- struct event_thread * wp;
-
- if (!mpp)
- return 0;
-
- if (pthread_attr_init(&attr))
- goto out;
-
- pthread_attr_setstacksize(&attr, 32 * 1024);
- pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
-
- wp = alloc_waiter();
-
- if (!wp)
- goto out;
-
- mpp->waiter = (void *)wp;
- strncpy(wp->mapname, mpp->alias, WWID_SIZE);
- wp->vecs = vecs;
- wp->mpp = mpp;
-
- if (pthread_create(&wp->thread, &attr, waitevent, wp)) {
- condlog(0, "%s: cannot create event checker", wp->mapname);
- goto out1;
- }
- condlog(2, "%s: event checker started", wp->mapname);
+ unsigned int i;
+ struct multipath *mpp;
- return 0;
-out1:
- free_waiter(wp);
- mpp->waiter = NULL;
-out:
- condlog(0, "failed to start waiter thread");
- return 1;
+ vector_foreach_slot (mpvec, mpp, i)
+ sync_map_state(mpp);
}
static int
* clear references to this map before flushing so we can ignore
* the spurious uevent we may generate with the dm_flush_map call below
*/
- if (dm_flush_map(mpp->alias, DEFAULT_TARGET)) {
+ if (dm_flush_map(mpp->alias)) {
/*
* May not really be an error -- if the map was already flushed
* from the device mapper by dmsetup(8) for instance.
}
else {
dm_lib_release();
- condlog(3, "%s: devmap removed", mpp->alias);
+ condlog(2, "%s: devmap removed", mpp->alias);
}
orphan_paths(vecs->pathvec, mpp);
- remove_map(mpp, vecs, stop_waiter_thread, 1);
+ remove_map_and_stop_waiter(mpp, vecs, 1);
return 0;
}
static int
-uev_add_map (char * devname, struct vectors * vecs)
+uev_add_map (struct sysfs_device * dev, struct vectors * vecs)
{
- condlog(2, "%s: add map (uevent)", devname);
- return ev_add_map(devname, vecs);
+ condlog(2, "%s: add map (uevent)", dev->kernel);
+ return ev_add_map(dev, vecs);
}
int
-ev_add_map (char * devname, struct vectors * vecs)
+ev_add_map (struct sysfs_device * dev, struct vectors * vecs)
{
- int major, minor;
- char dev_t[BLK_DEV_SIZE];
char * alias;
+ char *dev_t;
+ int major, minor;
char * refwwid;
struct multipath * mpp;
int map_present;
int r = 1;
- if (sscanf(devname, "dm-%d", &minor) == 1 &&
- !sysfs_get_dev(sysfs_path, devname, dev_t, BLK_DEV_SIZE) &&
- sscanf(dev_t, "%d:%d", &major, &minor) == 2)
- alias = dm_mapname(major, minor);
- else
- alias = STRDUP(devname);
-
+ dev_t = sysfs_attr_get_value(dev->devpath, "dev");
+
+ if (!dev_t || sscanf(dev_t, "%d:%d", &major, &minor) != 2)
+ return 1;
+
+ alias = dm_mapname(major, minor);
+
if (!alias)
return 1;
-
+
map_present = dm_map_present(alias);
- if (map_present && dm_type(alias, DEFAULT_TARGET) <= 0) {
+ if (map_present && dm_type(alias, TGT_MPATH) <= 0) {
condlog(4, "%s: not a multipath map", alias);
- FREE(alias);
return 0;
}
mpp = find_mp_by_alias(vecs->mpvec, alias);
if (mpp) {
- /*
+ /*
* Not really an error -- we generate our own uevent
* if we create a multipath mapped device as a result
* of uev_add_path
*/
condlog(0, "%s: devmap already registered",
- devname);
- FREE(alias);
+ dev->kernel);
return 0;
}
/*
* now we can register the map
*/
- if (map_present && (mpp = add_map_without_path(vecs, minor, alias,
- start_waiter_thread))) {
- condlog(3, "%s: devmap %s added", alias, devname);
+ if (map_present && (mpp = add_map_without_path(vecs, minor, alias))) {
+ sync_map_state(mpp);
+ condlog(2, "%s: devmap %s added", alias, dev->kernel);
return 0;
}
- refwwid = get_refwwid(devname, DEV_DEVMAP, vecs->pathvec);
+ refwwid = get_refwwid(dev->kernel, DEV_DEVMAP, vecs->pathvec);
if (refwwid) {
- r = coalesce_paths(vecs, NULL, refwwid);
+ r = coalesce_paths(vecs, NULL, refwwid, 0);
dm_lib_release();
}
-
+
if (!r)
- condlog(3, "%s: devmap %s added", alias, devname);
+ condlog(2, "%s: devmap %s added", alias, dev->kernel);
else
- condlog(0, "%s: uev_add_map %s failed", alias, devname);
+ condlog(0, "%s: uev_add_map %s failed", alias, dev->kernel);
FREE(refwwid);
- FREE(alias);
return r;
}
static int
-uev_remove_map (char * devname, struct vectors * vecs)
+uev_remove_map (struct sysfs_device * dev, struct vectors * vecs)
{
- condlog(2, "%s: remove map (uevent)", devname);
- return ev_remove_map(devname, vecs);
+ condlog(2, "%s: remove map (uevent)", dev->kernel);
+ return ev_remove_map(dev->kernel, vecs);
}
int
mpp = find_mp_by_str(vecs->mpvec, devname);
if (!mpp) {
- condlog(3, "%s: devmap not registered, can't remove",
+ condlog(2, "%s: devmap not registered, can't remove",
devname);
return 0;
}
}
static int
-uev_umount_map (char * devname, struct vectors * vecs)
+uev_umount_map (struct sysfs_device * dev, struct vectors * vecs)
{
struct multipath * mpp;
- condlog(2, "%s: umount map (uevent)", devname);
+ condlog(2, "%s: umount map (uevent)", dev->kernel);
- mpp = find_mp_by_str(vecs->mpvec, devname);
+ mpp = find_mp_by_str(vecs->mpvec, dev->kernel);
if (!mpp)
return 0;
return 0;
}
-
+
static int
-uev_add_path (char * devname, struct vectors * vecs)
+uev_add_path (struct sysfs_device * dev, struct vectors * vecs)
{
- condlog(2, "%s: add path (uevent)", devname);
- return (ev_add_path(devname, vecs) != 1)? 0 : 1;
+ condlog(2, "%s: add path (uevent)", dev->kernel);
+ return (ev_add_path(dev->kernel, vecs) != 1)? 0 : 1;
}
pp = find_path_by_dev(vecs->pathvec, devname);
if (pp) {
- condlog(0, "%s: spurious uevent, path already in pathvec, %p",
- devname, pp->mpp);
- /*
- * allow reconfig of orphaned path here
- */
- if (pp->mpp) return 1;
+ condlog(0, "%s: spurious uevent, path already in pathvec",
+ devname);
+ if (pp->mpp)
+ return 0;
}
else {
/*
condlog(0, "%s: failed to get path uid", devname);
return 1; /* leave path added to pathvec */
}
- if (blacklist_path(conf, pp)){
+ if (filter_path(conf, pp) > 0){
int i = find_slot(vecs->pathvec, (void *)pp);
if (i != -1)
vector_del_slot(vecs->pathvec, i);
free_path(pp);
return 2;
- }
+ }
mpp = pp->mpp = find_mp_by_wwid(vecs->mpvec, pp->wwid);
rescan:
if (mpp) {
condlog(0, "%s: failed in domap for addition of new "
"path %s", mpp->alias, devname);
/*
- * deal with asynchronous uevents :((
- */
+ * deal with asynchronous uevents :((
+ */
if (mpp->action == ACT_RELOAD) {
condlog(0, "%s: uev_add_path sleep", mpp->alias);
sleep(1);
if (setup_multipath(vecs, mpp))
goto out;
+ sync_map_state(mpp);
+
if (mpp->action == ACT_CREATE &&
start_waiter_thread(mpp, vecs))
goto out;
- condlog(3, "%s path added to devmap %s", devname, mpp->alias);
+ condlog(2, "%s path added to devmap %s", devname, mpp->alias);
return 0;
out:
- remove_map(mpp, vecs, NULL, 1);
+ remove_map(mpp, vecs, 1);
return 1;
}
static int
-uev_remove_path (char * devname, struct vectors * vecs)
+uev_remove_path (struct sysfs_device * dev, struct vectors * vecs)
{
- condlog(2, "%s: remove path (uevent)", devname);
- return ev_remove_path(devname, vecs);
+ int retval;
+
+ condlog(2, "%s: remove path (uevent)", dev->kernel);
+ retval = ev_remove_path(dev->kernel, vecs);
+ if (!retval)
+ sysfs_device_put(dev);
+
+ return retval;
}
int
{
struct multipath * mpp;
struct path * pp;
- int i;
- int rm_path = 1;
+ int i, retval = 0;
pp = find_path_by_dev(vecs->pathvec, devname);
}
/*
- * avoid referring to the map of an orphanned path
+ * avoid referring to the map of an orphaned path
*/
if ((mpp = pp->mpp)) {
+ /*
+ * transform the mp->pg vector of vectors of paths
+ * into a mp->params string to feed the device-mapper
+ */
+ if (update_mpp_paths(mpp, vecs->pathvec)) {
+ condlog(0, "%s: failed to update paths",
+ mpp->alias);
+ goto out;
+ }
+ if ((i = find_slot(mpp->paths, (void *)pp)) != -1)
+ vector_del_slot(mpp->paths, i);
/*
* remove the map IFF removing the last path
*/
- if (pathcount(mpp, PATH_WILD) > 1) {
- vector rpvec = vector_alloc();
+ if (VECTOR_SIZE(mpp->paths) == 0) {
+ char alias[WWID_SIZE];
/*
- * transform the mp->pg vector of vectors of paths
- * into a mp->params string to feed the device-mapper
- */
- update_mpp_paths(mpp, vecs->pathvec);
- if ((i = find_slot(mpp->paths, (void *)pp)) != -1)
- vector_del_slot(mpp->paths, i);
-
- if (VECTOR_SIZE(mpp->paths) == 0) {
- char alias[WWID_SIZE];
-
- /*
- * flush_map will fail if the device is open
- */
- strncpy(alias, mpp->alias, WWID_SIZE);
- if (flush_map(mpp, vecs))
- rm_path = 0;
- else
- condlog(3, "%s: removed map after removing"
- " multiple paths", alias);
- }
- else {
- if (setup_map(mpp)) {
- condlog(0, "%s: failed to setup map for"
- " removal of path %s", mpp->alias, devname);
- free_pathvec(rpvec, KEEP_PATHS);
- goto out;
- }
- /*
- * reload the map
- */
- mpp->action = ACT_RELOAD;
- if (domap(mpp) <= 0) {
- condlog(0, "%s: failed in domap for "
- "removal of path %s",
- mpp->alias, devname);
- /*
- * Delete path from pathvec so that
- * update_mpp_paths wont find it later
- * when/if another path is removed.
- */
- if ((i = find_slot(vecs->pathvec, (void *)pp)) != -1)
- vector_del_slot(vecs->pathvec, i);
- free_path(pp);
- return 1;
- }
- /*
- * update our state from kernel
- */
- if (setup_multipath(vecs, mpp)) {
- free_pathvec(rpvec, KEEP_PATHS);
- goto out;
- }
-
- condlog(3, "%s path removed from devmap %s",
- devname, mpp->alias);
+ * flush_map will fail if the device is open
+ */
+ strncpy(alias, mpp->alias, WWID_SIZE);
+ if (!flush_map(mpp, vecs)) {
+ condlog(2, "%s: removed map after"
+ " removing all paths",
+ alias);
+ free_path(pp);
+ return 0;
}
- free_pathvec(rpvec, KEEP_PATHS);
+ /*
+ * Not an error, continue
+ */
}
- else {
- char alias[WWID_SIZE];
+ if (setup_map(mpp)) {
+ condlog(0, "%s: failed to setup map for"
+ " removal of path %s", mpp->alias,
+ devname);
+ goto out;
+ }
+ /*
+ * reload the map
+ */
+ mpp->action = ACT_RELOAD;
+ if (domap(mpp) <= 0) {
+ condlog(0, "%s: failed in domap for "
+ "removal of path %s",
+ mpp->alias, devname);
+ retval = 1;
+ } else {
/*
- * flush_map will fail if the device is open
+ * update our state from kernel
*/
- strncpy(alias, mpp->alias, WWID_SIZE);
- if (flush_map(mpp, vecs))
- rm_path = 0;
- else
- condlog(3, "%s: removed map", alias);
+ if (setup_multipath(vecs, mpp)) {
+ goto out;
+ }
+ sync_map_state(mpp);
+
+ condlog(2, "%s: path removed from map %s",
+ devname, mpp->alias);
}
}
- if (rm_path) {
- if ((i = find_slot(vecs->pathvec, (void *)pp)) != -1)
- vector_del_slot(vecs->pathvec, i);
- free_path(pp);
- }
+ if ((i = find_slot(vecs->pathvec, (void *)pp)) != -1)
+ vector_del_slot(vecs->pathvec, i);
- return 0;
+ free_path(pp);
+
+ return retval;
out:
- remove_map(mpp, vecs, stop_waiter_thread, 1);
+ remove_map_and_stop_waiter(mpp, vecs, 1);
return 1;
}
static int
map_discovery (struct vectors * vecs)
{
- int i;
struct multipath * mpp;
+ unsigned int i;
- if (dm_get_maps(vecs->mpvec, "multipath"))
+ if (dm_get_maps(vecs->mpvec))
return 1;
vector_foreach_slot (vecs->mpvec, mpp, i)
{
struct vectors * vecs;
int r;
-
+
*reply = NULL;
*len = 0;
vecs = (struct vectors *)trigger_data;
return 0;
}
-int
+int
uev_trigger (struct uevent * uev, void * trigger_data)
{
int r = 0;
- char devname[32];
+ struct sysfs_device *sysdev;
struct vectors * vecs;
vecs = (struct vectors *)trigger_data;
if (uev_discard(uev->devpath))
return 0;
- basename(uev->devpath, devname);
+ sysdev = sysfs_device_get(uev->devpath);
+ if(!sysdev)
+ return 0;
+
lock(vecs->lock);
/*
- * device map add/remove event
+ * device map event
+ * Add events are ignored here as the tables
+ * are not fully initialised then.
*/
- if (!strncmp(devname, "dm-", 3)) {
- if (!strncmp(uev->action, "add", 3)) {
- r = uev_add_map(devname, vecs);
+ if (!strncmp(sysdev->kernel, "dm-", 3)) {
+ if (!strncmp(uev->action, "change", 6)) {
+ r = uev_add_map(sysdev, vecs);
goto out;
}
if (!strncmp(uev->action, "remove", 6)) {
- r = uev_remove_map(devname, vecs);
+ r = uev_remove_map(sysdev, vecs);
goto out;
}
if (!strncmp(uev->action, "umount", 6)) {
- r = uev_umount_map(devname, vecs);
+ r = uev_umount_map(sysdev, vecs);
goto out;
}
goto out;
}
-
+
/*
* path add/remove event
*/
- if (blacklist(conf->blist_devnode, devname))
+ if (filter_devnode(conf->blist_devnode, conf->elist_devnode,
+ sysdev->kernel) > 0)
goto out;
if (!strncmp(uev->action, "add", 3)) {
- r = uev_add_path(devname, vecs);
+ r = uev_add_path(sysdev, vecs);
goto out;
}
if (!strncmp(uev->action, "remove", 6)) {
- r = uev_remove_path(devname, vecs);
+ r = uev_remove_path(sysdev, vecs);
goto out;
}
{
if (uevent_listen(&uev_trigger, ap))
fprintf(stderr, "error starting uevent listener");
-
+
return NULL;
}
static void *
uxlsnrloop (void * ap)
{
- if (load_keys())
- return NULL;
-
- if (alloc_handlers())
+ if (cli_init())
return NULL;
- add_handler(LIST+PATHS, cli_list_paths);
- add_handler(LIST+MAPS, cli_list_maps);
- add_handler(LIST+MAPS+STATUS, cli_list_maps_status);
- add_handler(LIST+MAPS+STATS, cli_list_maps_stats);
- add_handler(LIST+MAPS+TOPOLOGY, cli_list_maps_topology);
- add_handler(LIST+TOPOLOGY, cli_list_maps_topology);
- add_handler(LIST+MAP+TOPOLOGY, cli_list_map_topology);
- add_handler(LIST+CONFIG, cli_list_config);
- add_handler(ADD+PATH, cli_add_path);
- add_handler(DEL+PATH, cli_del_path);
- add_handler(ADD+MAP, cli_add_map);
- add_handler(DEL+MAP, cli_del_map);
- add_handler(SWITCH+MAP+GROUP, cli_switch_group);
- add_handler(RECONFIGURE, cli_reconfigure);
- add_handler(SUSPEND+MAP, cli_suspend);
- add_handler(RESUME+MAP, cli_resume);
- add_handler(REINSTATE+PATH, cli_reinstate);
- add_handler(FAIL+PATH, cli_fail);
+ set_handler_callback(LIST+PATHS, cli_list_paths);
+ set_handler_callback(LIST+PATHS+FMT, cli_list_paths_fmt);
+ set_handler_callback(LIST+MAPS, cli_list_maps);
+ set_handler_callback(LIST+STATUS, cli_list_status);
+ set_handler_callback(LIST+MAPS+STATUS, cli_list_maps_status);
+ set_handler_callback(LIST+MAPS+STATS, cli_list_maps_stats);
+ set_handler_callback(LIST+MAPS+TOPOLOGY, cli_list_maps_topology);
+ set_handler_callback(LIST+TOPOLOGY, cli_list_maps_topology);
+ set_handler_callback(LIST+MAP+TOPOLOGY, cli_list_map_topology);
+ set_handler_callback(LIST+CONFIG, cli_list_config);
+ set_handler_callback(LIST+BLACKLIST, cli_list_blacklist);
+ set_handler_callback(LIST+DEVICES, cli_list_devices);
+ set_handler_callback(LIST+WILDCARDS, cli_list_wildcards);
+ set_handler_callback(ADD+PATH, cli_add_path);
+ set_handler_callback(DEL+PATH, cli_del_path);
+ set_handler_callback(ADD+MAP, cli_add_map);
+ set_handler_callback(DEL+MAP, cli_del_map);
+ set_handler_callback(SWITCH+MAP+GROUP, cli_switch_group);
+ set_handler_callback(RECONFIGURE, cli_reconfigure);
+ set_handler_callback(SUSPEND+MAP, cli_suspend);
+ set_handler_callback(RESUME+MAP, cli_resume);
+ set_handler_callback(REINSTATE+PATH, cli_reinstate);
+ set_handler_callback(FAIL+PATH, cli_fail);
uxsock_listen(&uxsock_trigger, ap);
}
static void
-fail_path (struct path * pp)
+fail_path (struct path * pp, int del_active)
{
if (!pp->mpp)
return;
pp->dev_t, pp->mpp->alias);
dm_fail_path(pp->mpp->alias, pp->dev_t);
- update_queue_mode_del_path(pp->mpp);
+ if (del_active)
+ update_queue_mode_del_path(pp->mpp);
}
/*
* caller must have locked the path list before calling that function
*/
static void
-reinstate_path (struct path * pp)
+reinstate_path (struct path * pp, int add_active)
{
if (!pp->mpp)
return;
condlog(0, "%s: reinstate failed", pp->dev_t);
else {
condlog(2, "%s: reinstated", pp->dev_t);
- update_queue_mode_add_path(pp->mpp);
+ if (add_active)
+ update_queue_mode_add_path(pp->mpp);
}
}
*
* we can safely return here, because upon map reload, all
* PG will be enabled.
- */
+ */
if (!pp->mpp->pg || !pp->pgindex)
return;
pgp = VECTOR_SLOT(pp->mpp->pg, pp->pgindex - 1);
-
+
if (pgp->status == PGSTATE_DISABLED) {
condlog(2, "%s: enable group #%i", pp->mpp->alias, pp->pgindex);
dm_enablegroup(pp->mpp->alias, pp->pgindex);
mpvec_garbage_collector (struct vectors * vecs)
{
struct multipath * mpp;
- int i;
+ unsigned int i;
vector_foreach_slot (vecs->mpvec, mpp, i) {
if (mpp && mpp->alias && !dm_map_present(mpp->alias)) {
condlog(2, "%s: remove dead map", mpp->alias);
- remove_map(mpp, vecs, stop_waiter_thread, 1);
+ remove_map_and_stop_waiter(mpp, vecs, 1);
i--;
}
}
defered_failback_tick (vector mpvec)
{
struct multipath * mpp;
- int i;
+ unsigned int i;
vector_foreach_slot (mpvec, mpp, i) {
/*
retry_count_tick(vector mpvec)
{
struct multipath *mpp;
- int i;
+ unsigned int i;
vector_foreach_slot (mpvec, mpp, i) {
if (mpp->retry_tick) {
}
}
-static void *
-checkerloop (void *ap)
+void
+check_path (struct vectors * vecs, struct path * pp)
{
- struct vectors *vecs;
- struct path *pp;
- int i, count = 0;
int newstate;
- mlockall(MCL_CURRENT | MCL_FUTURE);
- vecs = (struct vectors *)ap;
- condlog(2, "path checkers start up");
+ if (!pp->mpp)
+ return;
+
+ if (pp->tick && --pp->tick)
+ return; /* don't check this path yet */
/*
- * init the path check interval
+ * provision a next check soonest,
+ * in case we exit abnormaly from here
*/
- vector_foreach_slot (vecs->pathvec, pp, i) {
- pp->checkint = conf->checkint;
+ pp->tick = conf->checkint;
+
+ if (!checker_selected(&pp->checker)) {
+ pathinfo(pp, conf->hwtable, DI_SYSFS);
+ select_checker(pp);
+ }
+ if (!checker_selected(&pp->checker)) {
+ condlog(0, "%s: checker is not set", pp->dev);
+ return;
}
+ /*
+ * Set checker in async mode.
+ * Honored only by checker implementing the said mode.
+ */
+ checker_set_async(&pp->checker);
- while (1) {
- pthread_cleanup_push(cleanup_lock, vecs->lock);
- lock(vecs->lock);
- condlog(4, "tick");
+ newstate = checker_check(&pp->checker);
- vector_foreach_slot (vecs->pathvec, pp, i) {
- if (!pp->mpp)
- continue;
+ if (newstate < 0) {
+ condlog(2, "%s: unusable path", pp->dev);
+ pathinfo(pp, conf->hwtable, 0);
+ return;
+ }
+ /*
+ * Async IO in flight. Keep the previous path state
+ * and reschedule as soon as possible
+ */
+ if (newstate == PATH_PENDING) {
+ pp->tick = 1;
+ return;
+ }
+ if (newstate != pp->state) {
+ int oldstate = pp->state;
+ pp->state = newstate;
+ LOG_MSG(1, checker_message(&pp->checker));
- if (pp->tick && --pp->tick)
- continue; /* don't check this path yet */
+ /*
+ * upon state change, reset the checkint
+ * to the shortest delay
+ */
+ pp->checkint = conf->checkint;
+ if (newstate == PATH_DOWN || newstate == PATH_SHAKY ||
+ update_multipath_strings(pp->mpp, vecs->pathvec)) {
/*
- * provision a next check soonest,
- * in case we exit abnormaly from here
+ * proactively fail path in the DM
*/
- pp->tick = conf->checkint;
-
- if (!checker_selected(&pp->checker)) {
- pathinfo(pp, conf->hwtable, DI_SYSFS);
- select_checker(pp);
- }
+ if (oldstate == PATH_UP ||
+ oldstate == PATH_GHOST)
+ fail_path(pp, 1);
+ else
+ fail_path(pp, 0);
- if (!checker_selected(&pp->checker)) {
- condlog(0, "%s: checker is not set", pp->dev);
- continue;
- }
- newstate = checker_check(&pp->checker);
-
- if (newstate < 0) {
- condlog(2, "%s: unusable path", pp->dev);
- pathinfo(pp, conf->hwtable, 0);
- continue;
- }
+ /*
+ * cancel scheduled failback
+ */
+ pp->mpp->failback_tick = 0;
- if (newstate != pp->state) {
- pp->state = newstate;
- LOG_MSG(1, checker_message(&pp->checker));
+ pp->mpp->stat_path_failures++;
+ return;
+ }
- /*
- * upon state change, reset the checkint
- * to the shortest delay
- */
- pp->checkint = conf->checkint;
-
- if (newstate == PATH_DOWN ||
- newstate == PATH_SHAKY ||
- update_multipath_strings(pp->mpp,
- vecs->pathvec)) {
- /*
- * proactively fail path in the DM
- */
- fail_path(pp);
-
- /*
- * cancel scheduled failback
- */
- pp->mpp->failback_tick = 0;
-
- pp->mpp->stat_path_failures++;
- continue;
- }
+ /*
+ * reinstate this path
+ */
+ if (oldstate != PATH_UP &&
+ oldstate != PATH_GHOST)
+ reinstate_path(pp, 1);
+ else
+ reinstate_path(pp, 0);
- /*
- * reinstate this path
- */
- reinstate_path(pp);
+ /*
+ * schedule [defered] failback
+ */
+ if (pp->mpp->pgfailback > 0)
+ pp->mpp->failback_tick =
+ pp->mpp->pgfailback + 1;
+ else if (pp->mpp->pgfailback == -FAILBACK_IMMEDIATE &&
+ need_switch_pathgroup(pp->mpp, 1))
+ switch_pathgroup(pp->mpp);
- /*
- * schedule [defered] failback
- */
- if (pp->mpp->pgfailback > 0)
- pp->mpp->failback_tick =
- pp->mpp->pgfailback + 1;
- else if (pp->mpp->pgfailback == -FAILBACK_IMMEDIATE &&
- need_switch_pathgroup(pp->mpp, 1))
- switch_pathgroup(pp->mpp);
+ /*
+ * if at least one path is up in a group, and
+ * the group is disabled, re-enable it
+ */
+ if (newstate == PATH_UP)
+ enable_group(pp);
+ }
+ else if (newstate == PATH_UP || newstate == PATH_GHOST) {
+ LOG_MSG(4, checker_message(&pp->checker));
+ /*
+ * double the next check delay.
+ * max at conf->max_checkint
+ */
+ if (pp->checkint < (conf->max_checkint / 2))
+ pp->checkint = 2 * pp->checkint;
+ else
+ pp->checkint = conf->max_checkint;
- /*
- * if at least one path is up in a group, and
- * the group is disabled, re-enable it
- */
- if (newstate == PATH_UP)
- enable_group(pp);
- }
- else if (newstate == PATH_UP || newstate == PATH_GHOST) {
- LOG_MSG(4, checker_message(&pp->checker));
- /*
- * double the next check delay.
- * max at conf->max_checkint
- */
- if (pp->checkint < (conf->max_checkint / 2))
- pp->checkint = 2 * pp->checkint;
- else
- pp->checkint = conf->max_checkint;
+ pp->tick = pp->checkint;
+ condlog(4, "%s: delay next check %is",
+ pp->dev_t, pp->tick);
+ }
+ else if (newstate == PATH_DOWN)
+ LOG_MSG(2, checker_message(&pp->checker));
- pp->tick = pp->checkint;
- condlog(4, "%s: delay next check %is",
- pp->dev_t, pp->tick);
+ pp->state = newstate;
- }
- pp->state = newstate;
+ /*
+ * path prio refreshing
+ */
+ condlog(4, "path prio refresh");
+ pathinfo(pp, conf->hwtable, DI_PRIO);
- /*
- * path prio refreshing
- */
- condlog(4, "path prio refresh");
- pathinfo(pp, conf->hwtable, DI_PRIO);
-
- if (need_switch_pathgroup(pp->mpp, 0)) {
- if (pp->mpp->pgfailback > 0 &&
- pp->mpp->failback_tick <= 0)
- pp->mpp->failback_tick =
- pp->mpp->pgfailback + 1;
- else if (pp->mpp->pgfailback ==
- -FAILBACK_IMMEDIATE)
- switch_pathgroup(pp->mpp);
+ /*
+ * pathgroup failback policy
+ */
+ if (need_switch_pathgroup(pp->mpp, 0)) {
+ if (pp->mpp->pgfailback > 0 &&
+ pp->mpp->failback_tick <= 0)
+ pp->mpp->failback_tick =
+ pp->mpp->pgfailback + 1;
+ else if (pp->mpp->pgfailback ==
+ -FAILBACK_IMMEDIATE)
+ switch_pathgroup(pp->mpp);
+ }
+}
+
+static void *
+checkerloop (void *ap)
+{
+ struct vectors *vecs;
+ struct path *pp;
+ int count = 0;
+ unsigned int i;
+
+ mlockall(MCL_CURRENT | MCL_FUTURE);
+ vecs = (struct vectors *)ap;
+ condlog(2, "path checkers start up");
+
+ /*
+ * init the path check interval
+ */
+ vector_foreach_slot (vecs->pathvec, pp, i) {
+ pp->checkint = conf->checkint;
+ }
+
+ while (1) {
+ pthread_cleanup_push(cleanup_lock, vecs->lock);
+ lock(vecs->lock);
+ condlog(4, "tick");
+
+ if (vecs->pathvec) {
+ vector_foreach_slot (vecs->pathvec, pp, i) {
+ check_path(vecs, pp);
}
}
- defered_failback_tick(vecs->mpvec);
- retry_count_tick(vecs->mpvec);
-
+ if (vecs->mpvec) {
+ defered_failback_tick(vecs->mpvec);
+ retry_count_tick(vecs->mpvec);
+ }
if (count)
count--;
else {
mpvec_garbage_collector(vecs);
count = MAPGCINT;
}
-
+
lock_cleanup_pop(vecs->lock);
sleep(1);
}
vector mpvec;
int i;
- if (!(vecs->pathvec = vector_alloc()))
+ if (!vecs->pathvec && !(vecs->pathvec = vector_alloc()))
return 1;
-
- if (!(vecs->mpvec = vector_alloc()))
+
+ if (!vecs->mpvec && !(vecs->mpvec = vector_alloc()))
return 1;
-
+
if (!(mpvec = vector_alloc()))
return 1;
path_discovery(vecs->pathvec, conf, DI_ALL);
vector_foreach_slot (vecs->pathvec, pp, i){
- if (blacklist_path(conf, pp)){
+ if (filter_path(conf, pp) > 0){
vector_del_slot(vecs->pathvec, i);
free_path(pp);
i--;
- }
+ }
else
pp->checkint = conf->checkint;
}
/*
* create new set of maps & push changed ones into dm
*/
- if (coalesce_paths(vecs, mpvec, NULL))
+ if (coalesce_paths(vecs, mpvec, NULL, 0))
return 1;
/*
dm_lib_release();
- if (conf->verbosity > 2)
- vector_foreach_slot(mpvec, mpp, i)
- print_map(mpp);
+ sync_maps_state(mpvec);
/*
* purge dm of old maps
*/
- remove_maps(vecs, NULL);
+ remove_maps(vecs);
/*
* save new set of maps formed by considering current path state
*/
+ vector_free(vecs->mpvec);
vecs->mpvec = mpvec;
/*
* free old map and path vectors ... they use old conf state
*/
if (VECTOR_SIZE(vecs->mpvec))
- remove_maps(vecs, stop_waiter_thread);
+ remove_maps_and_stop_waiters(vecs);
if (VECTOR_SIZE(vecs->pathvec))
free_pathvec(vecs->pathvec, FREE_PATHS);
+ vecs->pathvec = NULL;
conf = NULL;
if (load_config(DEFAULT_CONFIGFILE))
if (!vecs)
return NULL;
- vecs->lock =
+ vecs->lock =
(pthread_mutex_t *)MALLOC(sizeof(pthread_mutex_t));
if (!vecs->lock)
goto out;
- vecs->pathvec = vector_alloc();
-
- if (!vecs->pathvec)
- goto out1;
-
- vecs->mpvec = vector_alloc();
-
- if (!vecs->mpvec)
- goto out2;
-
pthread_mutex_init(vecs->lock, NULL);
return vecs;
-out2:
- vector_free(vecs->pathvec);
-out1:
- FREE(vecs->lock);
out:
FREE(vecs);
condlog(0, "failed to init paths");
signal_set(SIGUSR1, sigusr1);
signal_set(SIGINT, sigend);
signal_set(SIGTERM, sigend);
- signal_set(SIGKILL, sigend);
+ signal(SIGPIPE, SIG_IGN);
}
static void
setscheduler (void)
{
- int res;
+ int res;
static struct sched_param sched_param = {
- sched_priority: 99
+ .sched_priority = 99
};
- res = sched_setscheduler (0, SCHED_RR, &sched_param);
+ res = sched_setscheduler (0, SCHED_RR, &sched_param);
- if (res == -1)
- condlog(LOG_WARNING, "Could not set SCHED_RR at priority 99");
+ if (res == -1)
+ condlog(LOG_WARNING, "Could not set SCHED_RR at priority 99");
return;
}
fprintf(fp, "%i", val);
fclose(fp);
}
-
+
static int
child (void * param)
{
if (load_config(DEFAULT_CONFIGFILE))
exit(1);
+ if (init_checkers()) {
+ condlog(0, "failed to initialize checkers");
+ exit(1);
+ }
+ if (init_prio()) {
+ condlog(0, "failed to initialize prioritizers");
+ exit(1);
+ }
+
setlogmask(LOG_UPTO(conf->verbosity + 3));
/*
conf->max_checkint = MAX_CHECKINT(conf->checkint);
}
+ if (conf->max_fds) {
+ struct rlimit fd_limit;
+ if (conf->max_fds > 0) {
+ fd_limit.rlim_cur = conf->max_fds;
+ fd_limit.rlim_max = conf->max_fds;
+ }
+ else {
+ fd_limit.rlim_cur = RLIM_INFINITY;
+ fd_limit.rlim_max = RLIM_INFINITY;
+ }
+ if (setrlimit(RLIMIT_NOFILE, &fd_limit) < 0)
+ condlog(0, "can't set open fds limit to %d : %s\n",
+ conf->max_fds, strerror(errno));
+ }
+
if (pidfile_create(DEFAULT_PIDFILE, getpid())) {
if (logsink)
log_thread_stop();
if (!vecs)
exit(1);
- if (sysfs_get_mnt_path(sysfs_path, FILE_NAME_SIZE)) {
+ if (sysfs_init(conf->sysfs_dir, FILE_NAME_SIZE)) {
condlog(0, "can not find sysfs mount point");
exit(1);
}
pthread_attr_init(&attr);
pthread_attr_setstacksize(&attr, 64 * 1024);
pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
-
+
pthread_create(&check_thr, &attr, checkerloop, vecs);
pthread_create(&uevent_thr, &attr, ueventloop, vecs);
pthread_create(&uxlsnr_thr, &attr, uxlsnrloop, vecs);
* exit path
*/
lock(vecs->lock);
- remove_maps(vecs, stop_waiter_thread);
+ remove_maps_and_stop_waiters(vecs);
free_pathvec(vecs->pathvec, FREE_PATHS);
pthread_cancel(check_thr);
pthread_cancel(uevent_thr);
pthread_cancel(uxlsnr_thr);
+ sysfs_cleanup();
+
free_keys(keys);
keys = NULL;
free_handlers(handlers);
vecs->lock = NULL;
FREE(vecs);
vecs = NULL;
- free_config(conf);
- conf = NULL;
condlog(2, "--------shut down-------");
-
+
if (logsink)
log_thread_stop();
dm_lib_release();
dm_lib_exit();
+ /*
+ * Freeing config must be done after condlog() and dm_lib_exit(),
+ * because logging functions like dlog() and dm_write_log()
+ * reference the config.
+ */
+ free_config(conf);
+ conf = NULL;
+
#ifdef _DEBUG_
dbg_free_final(NULL);
#endif
extern int optind;
int arg;
int err;
-
+
logsink = 1;
+ dm_init();
if (getuid() != 0) {
fprintf(stderr, "need to be root\n");
err = 0;
else
err = daemonize();
-
+
if (err < 0)
/* error */
exit(1);