From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mimir.eigenstate.org ([206.124.132.107]) by ewsd; Sun Mar  8 17:42:42 EDT 2020
Received: from stockyard.org (cpe-74-66-65-8.nyc.res.rr.com [74.66.65.8])
	by mimir.eigenstate.org (OpenSMTPD) with ESMTPSA id a6471b05 (TLSv1.2:ECDHE-RSA-AES256-SHA:256:NO);
	Sun, 8 Mar 2020 14:42:34 -0700 (PDT)
Message-ID: <78179A023495F178ECAB0D2302D514F8@eigenstate.org>
To: plan9fullfrontal@qs.co.nz, 9front@9front.org
Subject: Re: [9front] APE malloc on amd64 not 64bit alligned.
Date: Sun, 8 Mar 2020 14:42:32 -0700
From: ori@eigenstate.org
In-Reply-To: <0fa7882b-d950-5217-a15b-1e2ba09bd73a@qs.co.nz>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
List-ID: <9front.9front.org>
List-Help: <http://lists.9front.org>
X-Glyph: ➈
X-Bullshit: self-healing proven element-scale map/reduce interface 

> I don't know if this is just an FYI or if it should be classified as a bug.
> 
> 99% of APE mallocs return XXXX4 or XXXXC , with 1% being XXXXXX0.
> 
> The kernel returns 96% XXXXX0 and 4% XXXXX8 (in one test of 1M random 
> sized mallocs).
> 
> Should APE malloc on amd64 return 64bit alligned pointers?
> 
> 
> Are many people using amd64 9front ?

we're *technically* within spec, since amd64
doesn't care about alignment unless you're
using the vector registers, so we're within
our rights to define the max alignment of a
type as 4.

it'd probably be a good idea to change it, since
some ported software may make assumptions.

yes, I'm on amd64.