Re: Fw: Re: heap_force_common in contrib/pg_surgery/heap_surgery.c has an off by one stack buffer overflow

On Thu, Jun 04, 2026 at 02:42:23PM +0530, Ashutosh Sharma wrote:
> These are admittedly small risks, but they are real ones. Keeping the
> array 1-based eliminates that entire class of potential confusion and
> makes the code easier to maintain going forward. I'd lean toward
> Surya's approach for that reason.

That depends on the code path involved:
- pruneheap.c has "processed" and "htsv", that use a +1 index to avoid
the substract, where we also worry about performance.
- heapam_handler.c has in_index, that uses a -1 index.

At the end, the first pattern is an outlier, we don't need to worry
about performance in pg_surgery, and we're talking about three lines
of code in pg_surgery to change (two for include_this_tid, one for the
assertion). With all that in mind, I'd just do a -1 conversion and
call it a day. :)
--
Michael