From: Jason Gunthorpe <jgg@nvidia.com>
To: iommu@lists.linux.dev, Kevin Tian <kevin.tian@intel.com>,
linux-kselftest@vger.kernel.org
Cc: kvm@vger.kernel.org, Nicolin Chen <nicolinc@nvidia.com>,
Yi Liu <yi.l.liu@intel.com>
Subject: [PATCH 04/14] iommufd: Use the iommufd_group to avoid duplicate reserved groups and msi setup
Date: Fri, 24 Feb 2023 20:27:49 -0400 [thread overview]
Message-ID: <4-v1-7612f88c19f5+2f21-iommufd_alloc_jgg@nvidia.com> (raw)
In-Reply-To: <0-v1-7612f88c19f5+2f21-iommufd_alloc_jgg@nvidia.com>
These items only needs to be done once per group, not once per device. The
once per device was a way to make the device list work. Since we are
abandoning this we can optimize things a bit.
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
drivers/iommu/iommufd/device.c | 38 ++++++++++++++-----------
drivers/iommu/iommufd/io_pagetable.c | 5 ++--
drivers/iommu/iommufd/iommufd_private.h | 1 -
3 files changed, 23 insertions(+), 21 deletions(-)
diff --git a/drivers/iommu/iommufd/device.c b/drivers/iommu/iommufd/device.c
index 264bfa2212481f..75e8d79678736f 100644
--- a/drivers/iommu/iommufd/device.c
+++ b/drivers/iommu/iommufd/device.c
@@ -298,9 +298,20 @@ int iommufd_hw_pagetable_attach(struct iommufd_hw_pagetable *hwpt,
}
}
- rc = iopt_table_enforce_group_resv_regions(&hwpt->ioas->iopt, idev->dev,
- idev->igroup->group,
- &sw_msi_start);
+ /*
+ * The first device in the group to be attached will do all the work
+ * to setup the hwpt and ioas. Every other device re-uses it through
+ * the shared group attachment. Users are allowed/expected to attach
+ * every device in the group to the same hwpt, that just turns into
+ * a NOP.
+ */
+ if (idev->igroup->devices) {
+ idev->igroup->devices++;
+ return 0;
+ }
+
+ rc = iopt_table_enforce_group_resv_regions(
+ &hwpt->ioas->iopt, idev->igroup->group, &sw_msi_start);
if (rc)
return rc;
@@ -308,22 +319,15 @@ int iommufd_hw_pagetable_attach(struct iommufd_hw_pagetable *hwpt,
if (rc)
goto err_unresv;
- /*
- * Only attach to the group once for the first device that is in the
- * group. All the other devices will follow this attachment.
- * The user can attach every device individually as well.
- */
- if (!idev->igroup->devices) {
- rc = iommu_attach_group(hwpt->domain, idev->igroup->group);
- if (rc)
- goto err_unresv;
- idev->igroup->hwpt = hwpt;
- refcount_inc(&hwpt->obj.users);
- }
+ rc = iommu_attach_group(hwpt->domain, idev->igroup->group);
+ if (rc)
+ goto err_unresv;
+ idev->igroup->hwpt = hwpt;
+ refcount_inc(&hwpt->obj.users);
idev->igroup->devices++;
return 0;
err_unresv:
- iopt_remove_reserved_iova(&hwpt->ioas->iopt, idev->dev);
+ iopt_remove_reserved_iova(&hwpt->ioas->iopt, idev->igroup->group);
return rc;
}
@@ -339,7 +343,7 @@ iommufd_hw_pagetable_detach(struct iommufd_device *idev)
iommu_detach_group(hwpt->domain, idev->igroup->group);
idev->igroup->hwpt = NULL;
}
- iopt_remove_reserved_iova(&hwpt->ioas->iopt, idev->dev);
+ iopt_remove_reserved_iova(&hwpt->ioas->iopt, idev->igroup->group);
return hwpt;
}
diff --git a/drivers/iommu/iommufd/io_pagetable.c b/drivers/iommu/iommufd/io_pagetable.c
index e0ae72b9e67f86..096491bbb5acf5 100644
--- a/drivers/iommu/iommufd/io_pagetable.c
+++ b/drivers/iommu/iommufd/io_pagetable.c
@@ -1163,7 +1163,6 @@ void iopt_remove_access(struct io_pagetable *iopt,
/* Narrow the valid_iova_itree to include reserved ranges from a group. */
int iopt_table_enforce_group_resv_regions(struct io_pagetable *iopt,
- struct device *device,
struct iommu_group *group,
phys_addr_t *sw_msi_start)
{
@@ -1191,7 +1190,7 @@ int iopt_table_enforce_group_resv_regions(struct io_pagetable *iopt,
}
rc = iopt_reserve_iova(iopt, resv->start,
- resv->length - 1 + resv->start, device);
+ resv->length - 1 + resv->start, group);
if (rc)
goto out_reserved;
}
@@ -1206,7 +1205,7 @@ int iopt_table_enforce_group_resv_regions(struct io_pagetable *iopt,
goto out_free_resv;
out_reserved:
- __iopt_remove_reserved_iova(iopt, device);
+ __iopt_remove_reserved_iova(iopt, group);
out_free_resv:
list_for_each_entry_safe(resv, tmp, &group_resv_regions, list)
kfree(resv);
diff --git a/drivers/iommu/iommufd/iommufd_private.h b/drivers/iommu/iommufd/iommufd_private.h
index 5f3ad16da819e7..dbecdff013d082 100644
--- a/drivers/iommu/iommufd/iommufd_private.h
+++ b/drivers/iommu/iommufd/iommufd_private.h
@@ -77,7 +77,6 @@ int iopt_table_add_domain(struct io_pagetable *iopt,
void iopt_table_remove_domain(struct io_pagetable *iopt,
struct iommu_domain *domain);
int iopt_table_enforce_group_resv_regions(struct io_pagetable *iopt,
- struct device *device,
struct iommu_group *group,
phys_addr_t *sw_msi_start);
int iopt_set_allow_iova(struct io_pagetable *iopt,
--
2.39.1
next prev parent reply other threads:[~2023-02-25 0:28 UTC|newest]
Thread overview: 64+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-25 0:27 [PATCH 00/14] Add iommufd physical device operations for replace and alloc hwpt Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 01/14] iommufd: Move isolated msi enforcement to iommufd_device_bind() Jason Gunthorpe
2023-03-02 7:45 ` Tian, Kevin
2023-02-25 0:27 ` [PATCH 02/14] iommufd: Add iommufd_group Jason Gunthorpe
2023-03-02 7:55 ` Tian, Kevin
2023-03-02 12:51 ` Jason Gunthorpe
2023-03-03 2:13 ` Tian, Kevin
2023-03-06 19:16 ` Jason Gunthorpe
2023-03-07 2:32 ` Tian, Kevin
2023-02-25 0:27 ` [PATCH 03/14] iommufd: Replace the hwpt->devices list with iommufd_group Jason Gunthorpe
2023-03-02 8:01 ` Tian, Kevin
2023-03-06 20:22 ` Jason Gunthorpe
2023-03-07 2:38 ` Tian, Kevin
2023-03-07 13:53 ` Jason Gunthorpe
2023-03-08 7:29 ` Tian, Kevin
2023-03-08 19:00 ` Jason Gunthorpe
2023-02-25 0:27 ` Jason Gunthorpe [this message]
2023-03-02 8:06 ` [PATCH 04/14] iommufd: Use the iommufd_group to avoid duplicate reserved groups and msi setup Tian, Kevin
2023-03-02 12:55 ` Jason Gunthorpe
2023-03-03 2:16 ` Tian, Kevin
2023-02-25 0:27 ` [PATCH 05/14] iommufd: Make sw_msi_start a group global Jason Gunthorpe
2023-03-02 8:09 ` Tian, Kevin
2023-03-06 20:27 ` Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 06/14] iommufd: Move putting a hwpt to a helper function Jason Gunthorpe
2023-03-02 8:12 ` Tian, Kevin
2023-03-06 20:29 ` Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 07/14] iommufd: Add enforced_cache_coherency to iommufd_hw_pagetable_alloc() Jason Gunthorpe
2023-03-02 8:14 ` Tian, Kevin
2023-02-25 0:27 ` [PATCH 08/14] iommu: Introduce a new iommu_group_replace_domain() API Jason Gunthorpe
2023-03-02 8:16 ` Tian, Kevin
2023-02-25 0:27 ` [PATCH 09/14] iommufd: Add iommufd_device_replace() Jason Gunthorpe
2023-02-26 3:01 ` Baolu Lu
2023-02-27 13:58 ` Jason Gunthorpe
2023-02-28 1:50 ` Baolu Lu
2023-02-28 13:51 ` Jason Gunthorpe
2023-03-01 1:55 ` Baolu Lu
2023-02-26 3:13 ` Baolu Lu
2023-02-27 14:00 ` Jason Gunthorpe
2023-02-28 2:10 ` Baolu Lu
2023-02-28 13:52 ` Jason Gunthorpe
2023-03-01 2:23 ` Baolu Lu
2023-03-02 8:20 ` Tian, Kevin
2023-03-06 20:44 ` Jason Gunthorpe
2023-03-07 2:42 ` Tian, Kevin
2023-03-07 13:54 ` Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 10/14] iommufd: Make destroy_rwsem use a lock class per object type Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 11/14] iommufd/selftest: Test iommufd_device_replace() Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 12/14] iommufd: Add IOMMU_HWPT_ALLOC Jason Gunthorpe
2023-03-06 1:42 ` Nicolin Chen
2023-03-06 20:31 ` Jason Gunthorpe
2023-03-17 3:02 ` Tian, Kevin
2023-03-17 4:02 ` Nicolin Chen
2023-03-17 10:20 ` Tian, Kevin
2023-03-21 17:16 ` Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 13/14] iommufd/selftest: Return the real idev id from selftest mock_domain Jason Gunthorpe
2023-02-25 0:27 ` [PATCH 14/14] iommufd/selftest: Add a selftest for IOMMU_HWPT_ALLOC Jason Gunthorpe
2023-02-26 19:29 ` Nicolin Chen
2023-02-27 15:02 ` Jason Gunthorpe
2023-02-28 0:17 ` Nicolin Chen
2023-03-07 8:42 ` [PATCH 00/14] Add iommufd physical device operations for replace and alloc hwpt Tian, Kevin
2023-03-07 12:46 ` Jason Gunthorpe
2023-03-08 2:08 ` Baolu Lu
2023-03-08 7:38 ` Tian, Kevin
2023-03-08 18:59 ` 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=4-v1-7612f88c19f5+2f21-iommufd_alloc_jgg@nvidia.com \
--to=jgg@nvidia.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).