Re: Use pg_nextpower2_* in a few more places - Mailing list pgsql-hackers

From Zhihong Yu
Subject Re: Use pg_nextpower2_* in a few more places
Date
Msg-id CALNJ-vQfeHXHEXO=pmOOPcFba5DoO-tkQ8sZnxcRwzt_Q9=NyQ@mail.gmail.com
Whole thread Raw
In response to Re: Use pg_nextpower2_* in a few more places  (David Rowley <dgrowleyml@gmail.com>)
Responses Re: Use pg_nextpower2_* in a few more places
List pgsql-hackers


On Sat, Jun 12, 2021 at 6:40 AM David Rowley <dgrowleyml@gmail.com> wrote:
Thanks for having a look.

On Sun, 13 Jun 2021 at 00:50, Zhihong Yu <zyu@yugabyte.com> wrote:
> -       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) ?

I think the new code will produce the same result as the old code in all cases.

All the old code did was finding the next power of 2 that's >= 8 and
larger than tranche_id.  LWLockTrancheNamesAllocated is just a hint at
where the old code should start searching from.  The new code does not
need that hint. All it seems to do is save the old code from having to
start the loop at 8 each time we need more space.

David
Hi,
Maybe add an assertion after the assignment, that newalloc >=  LWLockTrancheNamesAllocated.

Cheers

pgsql-hackers by date:

Previous
From: Ranier Vilela
Date:
Subject: Signed vs Unsigned (take 2) (src/backend/storage/ipc/procarray.c)
Next
From: Tom Lane
Date:
Subject: Re: Race condition in recovery?