On Sat, Jun 12, 2021 at 5:32 AM David Rowley <dgrowle...@gmail.com> wrote:
> Back in f0705bb62, we added pg_nextpower2_32 and pg_nextpower2_64 to
> efficiently obtain the next power of 2 of a given number using an
> intrinsic function to find the left-most 1 bit.
>
> In d025cf88b and 02a2e8b44, I added some usages of these new functions
> but I didn't quite get all of them done. The attached replaces all
> of the remaining ones that I'm happy enough to go near.
>
> The ones that I left behind are ones in the form of:
>
> while (reqsize >= buflen)
> {
> buflen *= 2;
> buf = repalloc(buf, buflen);
> }
>
> The reason I left those behind is that I was too scared that I might
> introduce an opportunity to wrap buflen back around to zero again. At
> the moment the repalloc() would prevent that as we'd go above
> MaxAllocSize before we wrapped buflen back to zero again. All the
> other places I touched does not change the risk of that happening.
>
> It would be nice to get rid of doing that repalloc() in a loop, but it
> would need a bit more study to ensure we couldn't wrap or we'd need to
> add some error checking code that raised an ERROR if it did wrap. I
> don't want to touch those as part of this effort.
>
> I've also fixed up a few places that were just doubling the size of a
> buffer but used a "while" loop to do this when a simple "if" would
> have done. Using an "if" is ever so slightly more optimal since the
> condition will be checked once rather than twice when the buffer needs
> to increase in size.
>
> I'd like to fix these for PG15.
>
> David
>
Hi,
- newalloc = Max(LWLockTrancheNamesAllocated, 8);
- while (newalloc <= tranche_id)
- newalloc *= 2;
+ newalloc = pg_nextpower2_32(Max(8, tranche_id + 1));
Should LWLockTrancheNamesAllocated be included in the Max() expression (in
case it gets to a high value) ?
Cheers
