linux-kernel - Re: [PATCH v4 10/19] iommu: Add set_platform_dma

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: <Y7bNidP/px0NmvR+@nvidia.com>
Date:   Thu, 5 Jan 2023 09:15:53 -0400
From:   Jason Gunthorpe <jgg@...dia.com>
To:     Baolu Lu <baolu.lu@...ux.intel.com>
Cc:     Joerg Roedel <joro@...tes.org>,
        Christoph Hellwig <hch@...radead.org>,
        Kevin Tian <kevin.tian@...el.com>,
        Will Deacon <will@...nel.org>,
        Robin Murphy <robin.murphy@....com>,
        Jean-Philippe Brucker <jean-philippe@...aro.org>,
        Suravee Suthikulpanit <suravee.suthikulpanit@....com>,
        Hector Martin <marcan@...can.st>,
        Sven Peter <sven@...npeter.dev>,
        Rob Clark <robdclark@...il.com>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>,
        Andy Gross <agross@...nel.org>,
        Bjorn Andersson <andersson@...nel.org>,
        Yong Wu <yong.wu@...iatek.com>,
        Matthias Brugger <matthias.bgg@...il.com>,
        Heiko Stuebner <heiko@...ech.de>,
        Matthew Rosato <mjrosato@...ux.ibm.com>,
        Orson Zhai <orsonzhai@...il.com>,
        Baolin Wang <baolin.wang@...ux.alibaba.com>,
        Chunyan Zhang <zhang.lyra@...il.com>,
        Chen-Yu Tsai <wens@...e.org>,
        Thierry Reding <thierry.reding@...il.com>,
        iommu@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4 10/19] iommu: Add set_platform_dma_ops iommu ops

On Thu, Jan 05, 2023 at 01:58:42PM +0800, Baolu Lu wrote:
> Hi Jason,
> 
> On 2023/1/4 21:17, Jason Gunthorpe wrote:
> > On Wed, Jan 04, 2023 at 08:57:16PM +0800, Lu Baolu wrote:
> > 
> > > diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
> > > index de91dd88705b..4e35a9f94873 100644
> > > --- a/drivers/iommu/iommu.c
> > > +++ b/drivers/iommu/iommu.c
> > > @@ -2163,6 +2163,17 @@ static int iommu_group_do_detach_device(struct device *dev, void *data)
> > >   	return 0;
> > >   }
> > > +static int iommu_group_do_set_platform_dma(struct device *dev, void *data)
> > > +{
> > > +	const struct iommu_ops *ops = dev_iommu_ops(dev);
> > > +
> > > +	if (!ops->set_platform_dma_ops)
> > > +		return -EINVAL;
> > > +
> > > +	ops->set_platform_dma_ops(dev);
> > > +	return 0;
> > > +}
> > > +
> > >   static int __iommu_group_set_domain(struct iommu_group *group,
> > >   				    struct iommu_domain *new_domain)
> > >   {
> > > @@ -2177,10 +2188,14 @@ static int __iommu_group_set_domain(struct iommu_group *group,
> > >   	 * platform specific behavior.
> > >   	 */
> > >   	if (!new_domain) {
> > > -		if (WARN_ON(!group->domain->ops->detach_dev))
> > > -			return -EINVAL;
> > This should still have the WARN_ON..
> > 
> > if (WARN_ON(!group->domain->ops->detach_dev && !dev_iommu_ops(dev)->set_platform_dma_ops)
> 
> This has been implicitly included in the code.
> 
> iommu_group_do_set_platform_dma() returns -EINVAL if the iommu driver
> doesn't support set_platform_dma_ops (otherwise always return success).
> Then, the domain->ops->detach_dev is required and a WARN_ON was there.
> 
>         if (!new_domain) {
>                 ret = __iommu_group_for_each_dev(group, NULL,
>                                 iommu_group_do_set_platform_dma);
>                 if (ret) {
>                         if (WARN_ON(!group->domain->ops->detach_dev))
>                                 return -EINVAL;
>                         __iommu_group_for_each_dev(group, group->domain,
>                                 iommu_group_do_detach_device);
>                 }
>                 group->domain = NULL;
>                 return 0;
>         }
> 
> Perhaps I should add a comment to explain this?

But you delete this later when you remove this.

I think testing the op directly is much clearer, get rid of the whole
ret and EINVAL thinig:

if (dev_iommu_ops(dev)->set_platform_dma_ops)
   __iommu_group_for_each_dev(group, NULL,
                                 iommu_group_do_set_platform_dma); //	 Can't fail!
else if (group->domain->ops->detach_dev)
     __iommu_group_for_each_dev(group, group->domain,
                                 iommu_group_do_detach_device);
else
   WARN(true)

Jason