Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 17 Oct 2022 13:35:55 -0700
From:   Kees Cook <keescook@chromium.org>
To:     Ard Biesheuvel <ardb@kernel.org>
Cc:     "Guilherme G. Piccoli" <gpiccoli@igalia.com>,
        Anton Vorontsov <anton@enomsg.org>,
        Colin Cross <ccross@android.com>,
        Tony Luck <tony.luck@intel.com>, linux-kernel@vger.kernel.org,
        linux-hardening@vger.kernel.org
Subject: Re: [PATCH] pstore: migrate to crypto acomp interface (take 2)
Message-ID: <202210171333.309A3D9@keescook>
References: <20221006234138.1835739-1-keescook@chromium.org>
 <191ec24d-35d4-e4e5-85f7-d7301984e647@igalia.com>
 <202210171100.5BAC4A5CC8@keescook>
 <CAMj1kXHzrRTVcxb5+hgUPV3tjekPcDWzVf6cG_Mc9JJmYBz2Mw@mail.gmail.com>
 <202210171227.35ED875219@keescook>
 <CAMj1kXEJQ8gh-iXZNL8bNcmV=JCmKHNp5BnhYthhSOyR5h79_g@mail.gmail.com>
 <202210171237.DF5D4A3FD7@keescook>
 <CAMj1kXGmsJNg7En-55aRF+ApicPD_Opkh8Jw+oTorSOSO+cfuw@mail.gmail.com>
 <202210171307.32A5D9C07@keescook>
 <CAMj1kXHced1khwsrHqMUmECh_7irYOckFd+Sx3z9KSmsL7tPxw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAMj1kXHced1khwsrHqMUmECh_7irYOckFd+Sx3z9KSmsL7tPxw@mail.gmail.com>
Precedence: bulk

On Mon, Oct 17, 2022 at 10:13:52PM +0200, Ard Biesheuvel wrote:
> On Mon, 17 Oct 2022 at 22:11, Kees Cook <keescook@chromium.org> wrote:
> >
> > On Mon, Oct 17, 2022 at 09:45:08PM +0200, Ard Biesheuvel wrote:
> > > On Mon, 17 Oct 2022 at 21:40, Kees Cook <keescook@chromium.org> wrote:
> > > > Okay, so strictly speaking, eliminating the per-CPU allocation is an
> > > > improvement. Keeping scomp and doing in-place compression will let
> > > > pstore use "any" compressions method.
> > >
> > > I'm not following the point you are making here.
> >
> > Sorry, I mean to say that if I leave scomp in pstore, nothing is "worse"
> > (i.e. the per-cpu allocation is present in both scomp and acomp). i.e.
> > no regression either way, but if we switch to a distinct library call,
> > it's an improvement on the memory utilization front.
> >
> > > > Is there a crypto API that does _not_ preallocate the per-CPU stuff?
> > > > Because, as you say, it's a huge amount of memory on the bigger
> > > > systems...
> > >
> > > The library interface for each of the respective algorithms.
> >
> > Where is the crypto API for just using the library interfaces, so I
> > don't have to be tied to a specific algo?
> >
> 
> That doesn't exist, that is the point.

Shouldn't something like that exist, though?

> But how does the algo matter when you are dealing with mere kilobytes
> of ASCII text?

Sure, though, this is how we got here -- every couple of years, someone
added another library interface to another compression aglo. I tore all
that out so we could avoid having to choose a single one, but was left
with the zbufsize mess (that, yes, doesn't matter). So now pstore can
just not care what compression is chosen.

-- 
Kees Cook