From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Bruce Momjian <maillist(at)candle(dot)pha(dot)pa(dot)us> |
Cc: | pgsql-hackers(at)postgreSQL(dot)org |
Subject: | Re: [HACKERS] an older problem? hash table out of memory |
Date: | 1999-05-04 14:04:10 |
Message-ID: | 29625.925826650@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Bruce Momjian <maillist(at)candle(dot)pha(dot)pa(dot)us> writes:
>> The "hashtable out of memory" problem is reproducible, however.
>> I'm on it.
> Historically, no one knows much about the hash routines.
Well, I've been learning some unpleasant truths :-(. Hope to have
some fixes to commit in the next few days.
The immediate cause of one coredump I saw was that someone who was
overenthusiastically replacing sprintf's with snprintf's had written
snprintf(tempname, strlen(tempname), ...);
where tempname points to just-allocated, quite uninitialized
memory. Exercise for the student: how many different ways can
this go wrong? Unsettling question: how many other places did
that someone make the same mistake??
I don't have time for this right now, but it'd be a real good idea
to grep the source for strlen near snprintf to see if this same
problem appears anywhere else...
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Thomas Lockhart | 1999-05-04 14:19:45 | Re: [HACKERS] numeric data type on 6.5 |
Previous Message | Thomas Lockhart | 1999-05-04 13:54:33 | Re: [HACKERS] pg_dump bug (was Re: [SQL] Slow Inserts Again) |