From: | Ranier Vilela <ranier(dot)vf(at)gmail(dot)com> |
---|---|
To: | Daniel Gustafsson <daniel(at)yesql(dot)se> |
Cc: | Pg Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Avoid use deprecated Windows Memory API |
Date: | 2022-09-15 11:26:34 |
Message-ID: | CAEudQAq18WuwSSvwqEru=NFXiYW8jwopYRmhJOOup39+-QQsFg@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Em qui., 15 de set. de 2022 às 05:35, Daniel Gustafsson <daniel(at)yesql(dot)se>
escreveu:
> > On 15 Sep 2022, at 01:19, Ranier Vilela <ranier(dot)vf(at)gmail(dot)com> wrote:
>
> > LocalAlloc is deprecated.
> > So use HeapAlloc instead, once LocalAlloc is an overhead wrapper to
> HeapAlloc.
> >
> > Attached a patch.
>
> Don't forget that patches which aim to reduce overhead are best when
> accompanied with benchmarks which show the effect of the reduction.
>
I'm trusting the API producer.
>
> - pacl = (PACL) LocalAlloc(LPTR, dwNewAclSize);
> + pacl = (PACL) HeapAlloc(hDefaultProcessHeap, 0, dwNewAclSize);
>
> These calls are not equal, the LocalAlloc calls zeroes out the allocated
> memory
> but the HeapAlloc does not unless the HEAP_ZERO_MEMORY flag is passed. I
> haven't read the code enough to know if that matters, but it seems
> relevant to
> at least discuss.
>
Yeah, I missed that.
But works fine and passes all tests.
If really ok, yet another improvement by avoiding useless padding.
CF entry created.
https://commitfest.postgresql.org/40/3893/
regards,
Ranier Vilela
From | Date | Subject | |
---|---|---|---|
Next Message | Japin Li | 2022-09-15 11:36:40 | Re: SUBTRANS: Minimizing calls to SubTransSetParent() |
Previous Message | Bharath Rupireddy | 2022-09-15 10:25:54 | Re: Assertion failure in WaitForWALToBecomeAvailable state machine |