From 8e20e0ff39f0492cdc477f15d9d8c1b2f092d562 Mon Sep 17 00:00:00 2001 From: Alexander Motin Date: Fri, 11 Aug 2023 12:04:44 -0400 Subject: [PATCH] ZIL: Replay blocks without next block pointer. If we get next block allocation error during log write, we trigger transaction commit. But the block we have just completed is still written and transactions it covers will be acknowledged normally. If after that we ignore the block during replay just because it is the last in the chain, we may not replay some transactions that we have acknowledged as synced, that is not right. Reviewed-by: Brian Behlendorf Signed-off-by: Alexander Motin Sponsored by: iXsystems, Inc. Closes #15132 --- module/zfs/zil.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/module/zfs/zil.c b/module/zfs/zil.c index 6f04a7d4a7df..567787a19b66 100644 --- a/module/zfs/zil.c +++ b/module/zfs/zil.c @@ -290,7 +290,7 @@ zil_read_log_block(zilog_t *zilog, boolean_t decrypt, const blkptr_t *bp, char *lr = (char *)(zilc + 1); if (memcmp(&cksum, &zilc->zc_next_blk.blk_cksum, - sizeof (cksum)) || BP_IS_HOLE(&zilc->zc_next_blk) || + sizeof (cksum)) || zilc->zc_nused < sizeof (*zilc) || zilc->zc_nused > size) { error = SET_ERROR(ECKSUM); @@ -304,7 +304,7 @@ zil_read_log_block(zilog_t *zilog, boolean_t decrypt, const blkptr_t *bp, zil_chain_t *zilc = (zil_chain_t *)(lr + size) - 1; if (memcmp(&cksum, &zilc->zc_next_blk.blk_cksum, - sizeof (cksum)) || BP_IS_HOLE(&zilc->zc_next_blk) || + sizeof (cksum)) || (zilc->zc_nused > (size - sizeof (*zilc)))) { error = SET_ERROR(ECKSUM); } else {