Re: BUG #13985: Segmentation fault on PREPARE TRANSACTION

"Shulgin, Oleksandr"  writes:
> On Wed, Feb 24, 2016 at 10:52 PM, Andres Freund  wrote:
> At the very least ISTM that we have to make pgprocno volatile (or use a
>> memory barrier - but we don't have sufficient support for those in the
>> older branches), and move the PGPROC/PGXACT lookups after the == -1
>> check.

> Use of volatile doesn't change the resulting code dramatically for me.

Marking pgprocno volatile is silly.  What *is* missing is this:

-	ProcArrayStruct *arrayP = procArray;
+	volatile ProcArrayStruct *arrayP = procArray;

which corresponds directly to what the problem is: the storage arrayP
is pointing at may change asynchronously.

			regards, tom lane