All of lore.kernel.org
 help / color / mirror / Atom feed
From: Joerg Roedel <joro@8bytes.org>
To: iommu@lists.linux-foundation.org
Cc: Qian Cai <cai@lca.pw>,
	Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>,
	linux-kernel@vger.kernel.org, Joerg Roedel <joro@8bytes.org>
Subject: [PATCH 0/5] iommu/amd: Fix race conditions around increase_address_space()
Date: Mon,  4 May 2020 14:54:08 +0200	[thread overview]
Message-ID: <20200504125413.16798-1-joro@8bytes.org> (raw)

Hi,

Qian reported more race conditions around the dma-mapping
code path in the AMD IOMMU driver. The race condtions are
between increas_address_space() and fetch_pte(), there were
two of them:

	1) the 'pt_root' and 'mode' fields of 'struct
	   protection_domain' relate to each other so that
	   they must be updated together atomically.

	2) The increase_address_space() function publishes
	   the updated page-table before it has been written
	   to the DTE. This might cause PTEs to be mapped
	   and addresses handed to the device which are not
	   yet reachable through the DTE entry, causing IO
	   page-faults.

This patch-set fixes these issues, as tested by Qian Cai.

Thanks a lot again for reporting these issued and testing
the fixes!

Regards,

	Joerg


Joerg Roedel (5):
  iommu/amd: Fix race in increase_address_space()/fetch_pte()
  iommu/amd: Do not loop forever when trying to increase address space
  iommu/amd: Call domain_flush_complete() in update_domain()
  iommu/amd: Update Device Table in increase_address_space()
  iommu/amd: Do not flush Device Table in iommu_map_page()

 drivers/iommu/amd_iommu.c       | 198 +++++++++++++++++++++++++-------
 drivers/iommu/amd_iommu_types.h |   9 +-
 2 files changed, 161 insertions(+), 46 deletions(-)

-- 
2.17.1


WARNING: multiple messages have this Message-ID (diff)
From: Joerg Roedel <joro@8bytes.org>
To: iommu@lists.linux-foundation.org
Cc: Qian Cai <cai@lca.pw>, linux-kernel@vger.kernel.org
Subject: [PATCH 0/5] iommu/amd: Fix race conditions around increase_address_space()
Date: Mon,  4 May 2020 14:54:08 +0200	[thread overview]
Message-ID: <20200504125413.16798-1-joro@8bytes.org> (raw)

Hi,

Qian reported more race conditions around the dma-mapping
code path in the AMD IOMMU driver. The race condtions are
between increas_address_space() and fetch_pte(), there were
two of them:

	1) the 'pt_root' and 'mode' fields of 'struct
	   protection_domain' relate to each other so that
	   they must be updated together atomically.

	2) The increase_address_space() function publishes
	   the updated page-table before it has been written
	   to the DTE. This might cause PTEs to be mapped
	   and addresses handed to the device which are not
	   yet reachable through the DTE entry, causing IO
	   page-faults.

This patch-set fixes these issues, as tested by Qian Cai.

Thanks a lot again for reporting these issued and testing
the fixes!

Regards,

	Joerg


Joerg Roedel (5):
  iommu/amd: Fix race in increase_address_space()/fetch_pte()
  iommu/amd: Do not loop forever when trying to increase address space
  iommu/amd: Call domain_flush_complete() in update_domain()
  iommu/amd: Update Device Table in increase_address_space()
  iommu/amd: Do not flush Device Table in iommu_map_page()

 drivers/iommu/amd_iommu.c       | 198 +++++++++++++++++++++++++-------
 drivers/iommu/amd_iommu_types.h |   9 +-
 2 files changed, 161 insertions(+), 46 deletions(-)

-- 
2.17.1

_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu

             reply	other threads:[~2020-05-04 12:55 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-05-04 12:54 Joerg Roedel [this message]
2020-05-04 12:54 ` [PATCH 0/5] iommu/amd: Fix race conditions around increase_address_space() Joerg Roedel
2020-05-04 12:54 ` [PATCH 1/5] iommu/amd: Fix race in increase_address_space()/fetch_pte() Joerg Roedel
2020-05-04 12:54   ` Joerg Roedel
2020-05-04 12:54 ` [PATCH 2/5] iommu/amd: Do not loop forever when trying to increase address space Joerg Roedel
2020-05-04 12:54   ` Joerg Roedel
2020-05-04 12:54 ` [PATCH 3/5] iommu/amd: Call domain_flush_complete() in update_domain() Joerg Roedel
2020-05-04 12:54   ` Joerg Roedel
2020-05-04 12:54 ` [PATCH 4/5] iommu/amd: Update Device Table in increase_address_space() Joerg Roedel
2020-05-04 12:54   ` Joerg Roedel
2020-05-04 12:54 ` [PATCH 5/5] iommu/amd: Do not flush Device Table in iommu_map_page() Joerg Roedel
2020-05-04 12:54   ` Joerg Roedel
2020-05-05 12:40 ` [PATCH 0/5] iommu/amd: Fix race conditions around increase_address_space() Joerg Roedel
2020-05-05 12:40   ` Joerg Roedel

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=20200504125413.16798-1-joro@8bytes.org \
    --to=joro@8bytes.org \
    --cc=cai@lca.pw \
    --cc=iommu@lists.linux-foundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=suravee.suthikulpanit@amd.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.