From: Jason Gunthorpe <jgg@nvidia.com>
To: "Tian, Kevin" <kevin.tian@intel.com>
Cc: "iommu@lists.linux.dev" <iommu@lists.linux.dev>,
"linux-kselftest@vger.kernel.org"
<linux-kselftest@vger.kernel.org>,
Lu Baolu <baolu.lu@linux.intel.com>,
"kvm@vger.kernel.org" <kvm@vger.kernel.org>,
Nicolin Chen <nicolinc@nvidia.com>,
"Liu, Yi L" <yi.l.liu@intel.com>
Subject: Re: [PATCH v3 03/17] iommufd: Replace the hwpt->devices list with iommufd_group
Date: Tue, 28 Mar 2023 08:38:35 -0300 [thread overview]
Message-ID: <ZCLRu2NiVWNcdMk9@nvidia.com> (raw)
In-Reply-To: <BN9PR11MB52768F348A2814083BC5F5328C889@BN9PR11MB5276.namprd11.prod.outlook.com>
On Tue, Mar 28, 2023 at 02:32:22AM +0000, Tian, Kevin wrote:
> > From: Jason Gunthorpe <jgg@nvidia.com>
> > Sent: Friday, March 24, 2023 11:03 PM
> >
> > On Fri, Mar 24, 2023 at 01:37:51AM +0000, Tian, Kevin wrote:
> >
> > > If vfio races attach/detach then lots of things are messed.
> > >
> > > e.g. iommufd_device_detach() directly calls list_del(&idev->group_item)
> > > w/o checking whether the device has been attached.
> >
> > Yeah, you obviously can't race attach/detach or detach/replace
> >
> > > And with that race UAF could occur if we narrow down the lock scope
> > > to iommufd_hw_pagetable_attach():
> > >
> > > cpu0 cpu1
> > > vfio_iommufd_attach()
> > > iommufd_device_attach()
> > > iommufd_device_auto_get_domain()
> > > mutex_lock(&ioas->mutex);
> > > iommufd_hw_pagetable_alloc()
> > > hwpt = iommufd_object_alloc() //hwpt.users=1
> > > hwpt->domain = iommu_domain_alloc(idev->dev->bus);
> > > iommufd_hw_pagetable_attach() //hwpt.users=2
> > > vfio_iommufd_detach()
> > > iommufd_device_detach()
> > > mutex_lock(&idev->igroup->lock);
> > > hwpt = iommufd_hw_pagetable_detach()
> > > mutex_unlock(&idev->igroup->lock);
> > > iommufd_hw_pagetable_put(hwpt)
> > > iommufd_object_destroy_user(hwpt)
> > //hwpt.users=0
> > > iommufd_hw_pagetable_destroy(hwpt)
> > > iommu_domain_free(hwpt->domain);
> > > iopt_table_add_domain(&hwpt->ioas->iopt, hwpt->domain); //UAF
> >
> > You didn't balance the refcounts properly, the cpu1 put will get to
> > hwpt.users=1
> >
>
> iommufd_object_destroy_user() decrements the count twice if the value
> is two:
>
> refcount_dec(&obj->users);
> if (!refcount_dec_if_one(&obj->users)) {
Ohh, it isn't allowed to call iommufd_object_destroy_user() until
finalize has passed..
Jason
next prev parent reply other threads:[~2023-03-28 11:38 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-03-21 19:14 [PATCH v3 00/17] Add iommufd physical device operations for replace and alloc hwpt Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 01/17] iommufd: Move isolated msi enforcement to iommufd_device_bind() Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 02/17] iommufd: Add iommufd_group Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 03/17] iommufd: Replace the hwpt->devices list with iommufd_group Jason Gunthorpe
2023-03-23 7:21 ` Tian, Kevin
2023-03-23 14:23 ` Jason Gunthorpe
2023-03-24 1:37 ` Tian, Kevin
2023-03-24 15:02 ` Jason Gunthorpe
2023-03-28 2:32 ` Tian, Kevin
2023-03-28 11:38 ` Jason Gunthorpe [this message]
2023-03-29 3:03 ` Tian, Kevin
2023-04-11 14:31 ` Jason Gunthorpe
2023-04-12 8:27 ` Tian, Kevin
2023-04-12 11:17 ` Jason Gunthorpe
2023-04-13 2:52 ` Tian, Kevin
2023-04-14 13:31 ` Jason Gunthorpe
2023-04-20 6:15 ` Tian, Kevin
2023-04-20 15:34 ` Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 04/17] iommu: Export iommu_get_resv_regions() Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 05/17] iommufd: Keep track of each device's reserved regions instead of groups Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 06/17] iommufd: Use the iommufd_group to avoid duplicate MSI setup Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 07/17] iommufd: Make sw_msi_start a group global Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 08/17] iommufd: Move putting a hwpt to a helper function Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 09/17] iommufd: Add enforced_cache_coherency to iommufd_hw_pagetable_alloc() Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 10/17] iommufd: Reorganize iommufd_device_attach into iommufd_device_change_pt Jason Gunthorpe
2023-03-23 7:25 ` Tian, Kevin
2023-03-23 14:26 ` Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 11/17] iommu: Introduce a new iommu_group_replace_domain() API Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 12/17] iommufd: Add iommufd_device_replace() Jason Gunthorpe
2023-03-23 7:31 ` Tian, Kevin
2023-03-23 14:30 ` Jason Gunthorpe
2023-03-24 1:42 ` Tian, Kevin
2023-03-24 15:03 ` Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 13/17] iommufd: Make destroy_rwsem use a lock class per object type Jason Gunthorpe
2023-03-23 7:54 ` Tian, Kevin
2023-03-21 19:14 ` [PATCH v3 14/17] iommufd/selftest: Test iommufd_device_replace() Jason Gunthorpe
2023-03-23 7:57 ` Tian, Kevin
2023-03-23 14:32 ` Jason Gunthorpe
2023-03-21 19:14 ` [PATCH v3 15/17] iommufd: Add IOMMU_HWPT_ALLOC Jason Gunthorpe
2023-03-23 8:00 ` Tian, Kevin
2023-03-21 19:14 ` [PATCH v3 16/17] iommufd/selftest: Return the real idev id from selftest mock_domain Jason Gunthorpe
2023-03-23 8:02 ` Tian, Kevin
2023-03-21 19:14 ` [PATCH v3 17/17] iommufd/selftest: Add a selftest for IOMMU_HWPT_ALLOC Jason Gunthorpe
2023-03-23 8:03 ` Tian, Kevin
2023-03-23 8:04 ` [PATCH v3 00/17] Add iommufd physical device operations for replace and alloc hwpt Tian, Kevin
2023-03-23 14:35 ` Jason Gunthorpe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ZCLRu2NiVWNcdMk9@nvidia.com \
--to=jgg@nvidia.com \
--cc=baolu.lu@linux.intel.com \
--cc=iommu@lists.linux.dev \
--cc=kevin.tian@intel.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=nicolinc@nvidia.com \
--cc=yi.l.liu@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).