]> granicus.if.org Git - postgresql/commit
Fix PlanRowMark/ExecRowMark structures to handle inheritance correctly.
authorTom Lane <tgl@sss.pgh.pa.us>
Thu, 13 Jan 2011 01:47:02 +0000 (20:47 -0500)
committerTom Lane <tgl@sss.pgh.pa.us>
Thu, 13 Jan 2011 01:47:02 +0000 (20:47 -0500)
commitd487afbb813b7ca8803e20974b9e45530a1f4ef1
tree81c79a1546e772ad11114d705fcc3cbc3b05822a
parent7a32ff97321408afa0ddfcae1a4a060062956d24
Fix PlanRowMark/ExecRowMark structures to handle inheritance correctly.

In an inherited UPDATE/DELETE, each target table has its own subplan,
because it might have a column set different from other targets.  This
means that the resjunk columns we add to support EvalPlanQual might be
at different physical column numbers in each subplan.  The EvalPlanQual
rewrite I did for 9.0 failed to account for this, resulting in possible
misbehavior or even crashes during concurrent updates to the same row,
as seen in a recent report from Gordon Shannon.  Revise the data structure
so that we track resjunk column numbers separately for each subplan.

I also chose to move responsibility for identifying the physical column
numbers back to executor startup, instead of assuming that numbers derived
during preprocess_targetlist would stay valid throughout subsequent
massaging of the plan.  That's a bit slower, so we might want to consider
undoing it someday; but it would complicate the patch considerably and
didn't seem justifiable in a bug fix that has to be back-patched to 9.0.
12 files changed:
src/backend/executor/execJunk.c
src/backend/executor/execMain.c
src/backend/executor/nodeLockRows.c
src/backend/executor/nodeModifyTable.c
src/backend/nodes/copyfuncs.c
src/backend/nodes/outfuncs.c
src/backend/optimizer/plan/planner.c
src/backend/optimizer/prep/preptlist.c
src/backend/optimizer/prep/prepunion.c
src/include/executor/executor.h
src/include/nodes/execnodes.h
src/include/nodes/plannodes.h