Subject: Re: [BUGS] Bug #514: Backend crashes periodically 
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Warren Volz <wrv@po.cwru.edu>
cc: Barry Lind <barry@xythos.com>, pgsql-bugs@postgresql.org
Date: Mon, 12 Nov 2001 12:23:30 -0500
Comments: In-reply-to Warren Volz <wrv@po.cwru.edu>	message dated "Sun, 11 Nov 2001 23:20:20 -0500"

I believe I have found and fixed the crash-in-EvalPlanQual problem you
two reported.  The patch for current sources is attached --- it should
work in 7.1.* too, although line numbers might be off a bit.

			regards, tom lane


*** src/backend/executor/nodeIndexscan.c.orig	Mon Oct 29 15:30:53 2001
--- src/backend/executor/nodeIndexscan.c	Mon Nov 12 12:18:06 2001
***************
*** 339,348 ****
  	{
  		/*
  		 * If we are being passed an outer tuple, save it for runtime key
! 		 * calc
  		 */
  		if (exprCtxt != NULL)
  			econtext->ecxt_outertuple = exprCtxt->ecxt_outertuple;
  
  		/*
  		 * Reset the runtime-key context so we don't leak memory as each
--- 339,355 ----
  	{
  		/*
  		 * If we are being passed an outer tuple, save it for runtime key
! 		 * calc.  We also need to link it into the "regular" per-tuple
! 		 * econtext, so it can be used during indexqualorig evaluations.
  		 */
  		if (exprCtxt != NULL)
+ 		{
+ 			ExprContext *stdecontext;
+ 
  			econtext->ecxt_outertuple = exprCtxt->ecxt_outertuple;
+ 			stdecontext = node->scan.scanstate->cstate.cs_ExprContext;
+ 			stdecontext->ecxt_outertuple = exprCtxt->ecxt_outertuple;
+ 		}
  
  		/*
  		 * Reset the runtime-key context so we don't leak memory as each

---------------------------(end of broadcast)---------------------------
TIP 2: you can get off all lists at once with the unregister command
    (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)