MDEV-17844 recs_off_validate() fails in page_zip_write_trx_id_and_roll_ptr()

In commit 0e5a4ac2532c64a545796c787354dc41d61d0e62 (MDEV-15562)
we introduced was a bogus debug check failure that does not affect
the correctness of the release build.

With a fixed-length PRIMARY KEY, we do not have to recompute
the rec_get_offsets() after restarting the mini-transaction,
because the offsets of DB_TRX_ID,DB_ROLL_PTR are not going
to change.

row_undo_mod_clust(): Invoke rec_offs_make_valid() to keep the
debug check in page_zip_write_trx_id_and_roll_ptr() happy.

The scenario to reproduce this bug should be rather unlikely:
In the time frame when row_undo_mod_clust() has committed its
first mini-transaction and has not yet started the next one,
another mini-transaction must do something that causes the page
to be reorganized, split or merged.

1 files changed, 15 insertions, 0 deletions

diff --git a/storage/innobase/row/row0umod.cc b/storage/innobase/row/row0umod.cc
index e028e0ccb56..d2707130439 100644
--- a/storage/innobase/row/row0umod.cc
+++ b/storage/innobase/row/row0umod.cc
@@ -457,6 +457,21 @@ row_undo_mod_clust(
 		2 columns so that we can access DB_TRX_ID, DB_ROLL_PTR. */
 		offset_t offsets_[REC_OFFS_HEADER_SIZE + MAX_REF_PARTS + 2];
 		if (trx_id_offset) {
+#ifdef UNIV_DEBUG
+			ut_ad(rec_offs_validate(NULL, index, offsets));
+			if (buf_block_get_page_zip(
+				    btr_pcur_get_block(&node->pcur))) {
+				/* Below, page_zip_write_trx_id_and_roll_ptr()
+				needs offsets to access DB_TRX_ID,DB_ROLL_PTR.
+				We already computed offsets for possibly
+				another record in the clustered index.
+				Because the PRIMARY KEY is fixed-length,
+				the offsets for the PRIMARY KEY and
+				DB_TRX_ID,DB_ROLL_PTR are still valid.
+				Silence the rec_offs_validate() assertion. */
+				rec_offs_make_valid(rec, index, true, offsets);
+			}
+#endif
 		} else if (rec_is_metadata(rec, *index)) {
 			ut_ad(!buf_block_get_page_zip(btr_pcur_get_block(
 							      &node->pcur)));