linux-kernel - Re: [RFC PATCH 0/2] return value from shrinkers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 16 May 2013 09:52:05 +0200
From:	Oskar Andero <oskar.andero@...ymobile.com>
To:	Andrew Morton <akpm@...ux-foundation.org>
CC:	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"linux-mm@...ck.org" <linux-mm@...ck.org>,
	Hugh Dickins <hughd@...gle.com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	"Lekanovic, Radovan" <Radovan.Lekanovic@...ymobile.com>,
	David Rientjes <rientjes@...gle.com>,
	Glauber Costa <glommer@...allels.com>
Subject: Re: [RFC PATCH 0/2] return value from shrinkers

On 01:05 Thu 16 May     , Andrew Morton wrote:
> On Mon, 13 May 2013 16:16:33 +0200 Oskar Andero <oskar.andero@...ymobile.com> wrote:
> 
> > In a previous discussion on lkml it was noted that the shrinkers use the
> > magic value "-1" to signal that something went wrong.
> > 
> > This patch-set implements the suggestion of instead using errno.h values
> > to return something more meaningful.
> > 
> > The first patch simply changes the check from -1 to any negative value and
> > updates the comment accordingly.
> > 
> > The second patch updates the shrinkers to return an errno.h value instead
> > of -1. Since this one spans over many different areas I need input on what is
> > a meaningful return value. Right now I used -EBUSY on everything for consitency.
> > 
> > What do you say? Is this a good idea or does it make no sense at all?
> 
> I don't see much point in it, really.  Returning an errno implies that
> the errno will eventually be returned to userspace.  But that isn't the
> case, so such a change is somewhat misleading.

Yes. Glauber Costa pointed that out and I agree - errno.h is probably not
the right way to go.

> If we want the capability to return more than a binary yes/no message
> to callers then yes, we could/should enumerate the shrinker return
> values.  But as that is a different concept from errnos, it should be
> done with a different and shrinker-specific namespace.

Agreed, but even if there right now is only a binary return message, is a
hardcoded -1 considered to be acceptable for an interface? IMHO, it is not
very readable nor intuitive for the users of the interface. Why not, as you
mention, add a define or enum in shrinker.h instead, e.g. SHRINKER_STOP or
something.

-Oskar
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/