From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Noah Misch <noah(at)leadboat(dot)com> |
Cc: | pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: memory-related bugs |
Date: | 2011-03-12 17:44:29 |
Message-ID: | 26254.1299951869@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Noah Misch <noah(at)leadboat(dot)com> writes:
> A suitably-instrumented run of "make installcheck-world" under valgrind turned
> up a handful of memory-related bugs:
Hmm, interesting work, but I don't think I believe in the necessity for
this kluge:
> + else if (attributeName != &(att->attname))
> + namestrcpy(&(att->attname), attributeName);
The rules against overlapping memcpy/strcpy's source and destination are
meant to cover the case of partial overlap; I find it hard to imagine an
implementation that will mess up when the source and destination are
identical. If we did think it was important to avoid this situation I
would rather find another way, like modifying the caller. Likewise
the other changes to avoid no-op memcpy's do not appear to me to be
bugs, though possibly they might save enough cycles to be worth doing
anyway.
> ! stats->attrtype = (Form_pg_type) palloc(sizeof(FormData_pg_type));
> ! memcpy(stats->attrtype, GETSTRUCT(typtuple), sizeof(FormData_pg_type));
> ...
> ! stats->attrtype = (Form_pg_type) palloc(TYPE_FIXED_PART_SIZE);
> ! memcpy(stats->attrtype, GETSTRUCT(typtuple), TYPE_FIXED_PART_SIZE);
I wonder whether we should instead fix this by copying the correct tuple
length.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Martijn van Oosterhout | 2011-03-12 18:37:01 | Re: Collations versus user-defined functions |
Previous Message | Tom Lane | 2011-03-12 17:17:11 | Collations versus user-defined functions |