linux-kernel - Re: [PATCH 2/6] iommu/amd: Implement DOMAIN_ATTR

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20120127102608.GK19255@amd.com>
Date:	Fri, 27 Jan 2012 11:26:08 +0100
From:	Joerg Roedel <joerg.roedel@....com>
To:	Ohad Ben-Cohen <ohad@...ery.com>
CC:	<iommu@...ts.linux-foundation.org>, <linux-kernel@...r.kernel.org>,
	David Woodhouse <dwmw2@...radead.org>,
	David Brown <davidb@...eaurora.org>,
	Tony Lindgren <tony@...mide.com>,
	Laurent Pinchart <laurent.pinchart@...asonboard.com>,
	Stuart Yoder <b08248@...il.com>,
	Scott Wood <scottwood@...escale.com>,
	Hiroshi Doyu <hdoyu@...dia.com>
Subject: Re: [PATCH 2/6] iommu/amd: Implement DOMAIN_ATTR_GEOMETRY attribute

Hi Ohad,

On Fri, Jan 27, 2012 at 09:08:44AM +0200, Ohad Ben-Cohen wrote:
> On Thu, Jan 26, 2012 at 8:40 PM, Joerg Roedel <joerg.roedel@....com> wrote:
> > Implement the attribute itself and add the code for the
> > AMD IOMMU driver.
> 
> It's somewhat non-intuitive to have the generic attribute code bundled
> with the AMD implementation in the same patch (took me some time to
> find it here :).
> 
> Maybe split this to two patches ?

Yes, that is not ideal. I'll probably split it or move the generic part
to the first patch.

> > diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
> > index ef54718..3d0b0bf 100644
> > --- a/drivers/iommu/iommu.c
> > +++ b/drivers/iommu/iommu.c
> > @@ -345,10 +345,23 @@ EXPORT_SYMBOL_GPL(iommu_device_group);
> >  int iommu_domain_get_attr(struct iommu_domain *domain,
> >                          enum iommu_attr attr, void *data)
> >  {
> > -       if (!domain->ops->domain_get_attr)
> > -               return -EINVAL;
> > +       struct iommu_domain_geometry *geometry;
> > +       int ret = 0;
> > +
> > +       switch (attr) {
> > +       case DOMAIN_ATTR_GEOMETRY:
> > +               geometry  = data;
> > +               *geometry = domain->geometry;
> 
> This is not type-safe and will surely bite someone some day.
> 
> Sure, we can't have type safety in a multi-purpose generic interface,
> but I wonder whether this is the right interface for getting the
> geometry ? I.e. why not have a dedicated API for this ?

I fell also uncomfortable with the missing type-safety of this
interface. But the alternative is to have dedicated functions for
set/get each attribute. Well, it depends on how many attributes we have
in the end, but given that the PAMU guys already have need for a number
of hardware specific attributes it is likely that having individual
functions makes the api too complex in the end.

But probably we can replace the 'void *data' with a 'union
domain_attr'? This will give us some type-safety.


	Joerg

-- 
AMD Operating System Research Center

Advanced Micro Devices GmbH Einsteinring 24 85609 Dornach
General Managers: Alberto Bozzo
Registration: Dornach, Landkr. Muenchen; Registerger. Muenchen, HRB Nr. 43632

--
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/