Can I assume relation would not be invalid during from ExecutorRun to ExecutorEnd

During my recent work,  I need some new stuff attached to Relation.  Rather than adding

some new data structures,  I added it to Relation directly.  Like relation->balabala.  Then

I initialize it during ExecutorRun,  like  table_tuple_insert.. and destroy it at ExecutorEnd.

The above solution works based on 2 assumptions at least: 

1.  During the ExecutorRun & ExecutorEnd,  the relcache will never by invalidated, if not

the old relation->balabala will be lost.  I assume this is correct since I didn't see any places

where we handle such changes in Executor code. 

2.  We need to consider the ExuecotRun raised error,  we need to destroy the balabala resource

as well.  so I added it to the RelationClose function.  

So the overall design works like this:

xxx_table_tuple_insert(Relation rel, ...)

{

   if (rel->balabala == NULL)

        rel->balabala = allocate_bala_resource(rel);  // Allocate the memory xCtx which is under TopTransactionContext.

   do_task_with(rel->balabala); 

}

at the end of the executor,  I run

release_bala_resource(Relation rel)

{

   if (rel->balabala == NULL)

         return;

   do_the_real_task();

   MemoryContextDelete(rel->bala->memctx); 

   rel->balabala = NULL

}

For the failed cases:

RelationClose(..)

{

Will my suluation work?

--