Revert part of "Log I/Os longer than zio_delay_max (30s default)"

This reverts commit 9dcb97198338ba2d8764dd5604b278118612f74
which was originally introduced to debug occasional slow I/Os.
These I/Os would complete eventually but were observed to take
several 100 seconds.

The root cause of this issue was the CFQ scheduler which can,
under certain conditions, excessively delay an I/O from being
issued to the device.  This issue was mitigated somewhat by
commit 84daaddedbfc9cf4bd1490d8a6f4b2967051e308 which ensures
the I/O elevator gets changed even for DM style devices.

This change isn't in any way harmful but it does conflict with
a required change to properly account from I/O wait time.
Because Linux does not export the io_schedule_timeout() function
we must instead rely  on io_schedule() via cv_wait_io().

The additional debugging information which was added to the
delay event has been intentionally left in place.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>

diff --git a/module/zfs/zio.c b/module/zfs/zio.c
index bcfc081..ece3329 100644 (file)
--- a/module/zfs/zio.c
+++ b/module/zfs/zio.c
@@ -1305,34 +1305,18 @@ __zio_execute(zio_t *zio)
 int
 zio_wait(zio_t *zio)
 {
-       uint64_t timeout;
        int error;
 
        ASSERT(zio->io_stage == ZIO_STAGE_OPEN);
        ASSERT(zio->io_executor == NULL);
 
        zio->io_waiter = curthread;
-       timeout = ddi_get_lbolt() + (zio_delay_max / MILLISEC * hz);
 
        __zio_execute(zio);
 
        mutex_enter(&zio->io_lock);
-       while (zio->io_executor != NULL) {
-               /*
-                * Wake up periodically to prevent the kernel from complaining
-                * about a blocked task.  However, check zio_delay_max to see
-                * if the I/O has exceeded the timeout and post an ereport.
-                */
-               cv_timedwait_interruptible(&zio->io_cv, &zio->io_lock,
-                   ddi_get_lbolt() + hz);
-
-               if (timeout && (ddi_get_lbolt() > timeout)) {
-                       zio->io_delay = zio_delay_max;
-                       zfs_ereport_post(FM_EREPORT_ZFS_DELAY,
-                           zio->io_spa, zio->io_vd, zio, 0, 0);
-                       timeout = 0;
-               }
-       }
+       while (zio->io_executor != NULL)
+               cv_wait(&zio->io_cv, &zio->io_lock);
        mutex_exit(&zio->io_lock);
 
        error = zio->io_error;
@@ -2905,11 +2889,15 @@ zio_done(zio_t *zio)
        vdev_stat_update(zio, zio->io_size);
 
        /*
-        * When an I/O completes but was slow post an ereport.
+        * If this I/O is attached to a particular vdev is slow, exeeding
+        * 30 seconds to complete, post an error described the I/O delay.
+        * We ignore these errors if the device is currently unavailable.
         */
-       if (zio->io_delay >= zio_delay_max)
-               zfs_ereport_post(FM_EREPORT_ZFS_DELAY, zio->io_spa,
-                   zio->io_vd, zio, 0, 0);
+       if (zio->io_delay >= zio_delay_max) {
+               if (zio->io_vd != NULL && !vdev_is_dead(zio->io_vd))
+                       zfs_ereport_post(FM_EREPORT_ZFS_DELAY, zio->io_spa,
+                                         zio->io_vd, zio, 0, 0);
+       }
 
        if (zio->io_error) {
                /*