Fix unfortunate NULL in spa_update_dspace

After 1325434b, we can in certain circumstances end up calling
spa_update_dspace with vd->vdev_mg NULL, which ends poorly during
vdev removal.

So let's not do that further space adjustment when we can't.

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes #12380 
Closes #12428
This commit is contained in:
Rich Ercolani 2021-07-26 13:51:30 -04:00 committed by Tony Hutter
parent ccf6d0a59b
commit 05c96b438a
1 changed files with 8 additions and 1 deletions

View File

@ -1849,7 +1849,14 @@ spa_update_dspace(spa_t *spa)
spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
vdev_t *vd = vdev_t *vd =
vdev_lookup_top(spa, spa->spa_vdev_removal->svr_vdev_id); vdev_lookup_top(spa, spa->spa_vdev_removal->svr_vdev_id);
if (vd->vdev_mg->mg_class == spa_normal_class(spa)) { /*
* If the stars align, we can wind up here after
* vdev_remove_complete() has cleared vd->vdev_mg but before
* spa->spa_vdev_removal gets cleared, so we must check before
* we dereference.
*/
if (vd->vdev_mg &&
vd->vdev_mg->mg_class == spa_normal_class(spa)) {
spa->spa_dspace -= spa_deflate(spa) ? spa->spa_dspace -= spa_deflate(spa) ?
vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space; vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space;
} }