From 90149552b1ee50c2915305f869cbf09da271e186 Mon Sep 17 00:00:00 2001 From: Alexander Motin Date: Wed, 20 Sep 2023 14:17:11 -0400 Subject: [PATCH] ZIL: Fix potential race on flush deferring. zil_lwb_set_zio_dependency() can not set write ZIO dependency on previous LWB's write ZIO if one is already in done handler and set state to LWB_STATE_WRITE_DONE. So theoretically done handler of next LWB's write ZIO may run before done handler of previous LWB write ZIO completes. In such case we can not defer flushes, since the flush issue process is not locked. This may fix some reported assertions of lwb_vdev_tree not being empty inside zil_free_lwb(). Reviewed-by: Brian Behlendorf Signed-off-by: Alexander Motin Sponsored by: iXsystems, Inc. Closes #15278 --- module/zfs/zil.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/module/zfs/zil.c b/module/zfs/zil.c index b30676b42d..9e9c9c2254 100644 --- a/module/zfs/zil.c +++ b/module/zfs/zil.c @@ -1550,7 +1550,16 @@ zil_lwb_write_done(zio_t *zio) lwb->lwb_state = LWB_STATE_WRITE_DONE; lwb->lwb_child_zio = NULL; lwb->lwb_write_zio = NULL; + + /* + * If nlwb is not yet issued, zil_lwb_set_zio_dependency() is not + * called for it yet, and when it will be, it won't be able to make + * its write ZIO a parent this ZIO. In such case we can not defer + * our flushes or below may be a race between the done callbacks. + */ nlwb = list_next(&zilog->zl_lwb_list, lwb); + if (nlwb && nlwb->lwb_state != LWB_STATE_ISSUED) + nlwb = NULL; mutex_exit(&zilog->zl_lock); if (avl_numnodes(t) == 0)