All of lore.kernel.org
 help / color / mirror / Atom feed
From: Nicolin Chen <nicolinc@nvidia.com>
To: Jason Gunthorpe <jgg@nvidia.com>
Cc: <dan.j.williams@intel.com>, "Tian, Kevin" <kevin.tian@intel.com>,
	"Jonathan Cameron" <jonathan.cameron@huawei.com>,
	"will@kernel.org" <will@kernel.org>,
	"robin.murphy@arm.com" <robin.murphy@arm.com>,
	"bhelgaas@google.com" <bhelgaas@google.com>,
	"joro@8bytes.org" <joro@8bytes.org>,
	"praan@google.com" <praan@google.com>,
	"baolu.lu@linux.intel.com" <baolu.lu@linux.intel.com>,
	"miko.lenczewski@arm.com" <miko.lenczewski@arm.com>,
	"linux-arm-kernel@lists.infradead.org"
	<linux-arm-kernel@lists.infradead.org>,
	"iommu@lists.linux.dev" <iommu@lists.linux.dev>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-pci@vger.kernel.org" <linux-pci@vger.kernel.org>,
	"linux-cxl@vger.kernel.org" <linux-cxl@vger.kernel.org>
Subject: Re: [PATCH RFCv1 1/3] PCI: Allow ATS to be always on for CXL.cache capable devices
Date: Thu, 19 Feb 2026 20:52:56 -0800	[thread overview]
Message-ID: <aZfoqETxb31jp/Pk@nvidia.com> (raw)
In-Reply-To: <20260219174139.GI723117@nvidia.com>

On Thu, Feb 19, 2026 at 01:41:39PM -0400, Jason Gunthorpe wrote:
> On Thu, Feb 19, 2026 at 08:53:19AM -0800, Nicolin Chen wrote:
> > The thing is that those driver_managed_dma callbacks don't call
> > iommu_device_use_default_domain(). So, the iommu core loses the
> > trigger to switch domain from BLOCKED/empty-DMA to DMA/IDENTITY.
> 
> But they don't use DMA API at all so it doesn't matter to them.
> 
> Your issue is that BLOCKED breaks MSI on ARM.

Thanks for the hint!

It actually failed in iommu_dma_prepare_msi() due to having an
IOMMU_COOKIE_NONE in the blocking_domain.

My implementation sets group->domain to group->blocking_domain,
and keeps group->default_domain=NULL to retain the EPROBE_DEFER
validation in iommu_device_use_default_domain().

Then in iommu_dma_prepare_msi(), the group->domain now becomes
valid so it failed due to its unsupported iommu cookie, which
I entirely missed.

I could simply fix this by adding:

@@ -3892,7 +3892,8 @@ int iommu_dma_prepare_msi(struct msi_desc *desc, phys_addr_t msi_addr)

        mutex_lock(&group->mutex);
        /* An IDENTITY domain must pass through */
-       if (group->domain && group->domain->type != IOMMU_DOMAIN_IDENTITY) {
+       if (group->default_domain && group->domain &&
+           group->domain->type != IOMMU_DOMAIN_IDENTITY) {
                switch (group->domain->cookie_type) {
                case IOMMU_COOKIE_DMA_MSI:
                case IOMMU_COOKIE_DMA_IOVA:

> That is fixed by using
> an empty-DMA API domain as default.

Hmm, even if we set arm_smmu_blocked_domain.type to an empty DMA
(IOMMU_DOMAIN_UNMANAGED?), it still doesn't allocate a cookie?

> What is missing is to bring back the IDENTITY performance optimization
> in a secure way.

I might have got it wrong (from the last part below).
https://lore.kernel.org/linux-iommu/20260127150440.GF1134360@nvidia.com/.

You mean to disable ATS on IDENTITY domains? 

Thanks
Nicolin


  reply	other threads:[~2026-02-20  4:53 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-17  4:56 [PATCH RFCv1 0/3] Allow ATS to be always on for certain ATS-capable devices Nicolin Chen
2026-01-17  4:56 ` [PATCH RFCv1 1/3] PCI: Allow ATS to be always on for CXL.cache capable devices Nicolin Chen
2026-01-19 17:58   ` Jason Gunthorpe
2026-01-21  8:01   ` Tian, Kevin
2026-01-21 10:03     ` Jonathan Cameron
2026-01-21 13:03       ` Jason Gunthorpe
2026-01-22  1:17         ` Baolu Lu
2026-01-22 13:15           ` Jason Gunthorpe
2026-01-22  5:44         ` dan.j.williams
2026-01-22 13:14           ` Jason Gunthorpe
2026-01-22 16:29             ` Nicolin Chen
2026-01-22 16:58               ` Jason Gunthorpe
2026-01-22 19:46             ` dan.j.williams
2026-01-27  8:10               ` Tian, Kevin
2026-01-27 15:04                 ` Jason Gunthorpe
2026-01-28  0:49                   ` dan.j.williams
2026-01-28 13:05                     ` Jason Gunthorpe
2026-02-03  5:13                       ` Nicolin Chen
2026-02-03 14:33                         ` Jason Gunthorpe
2026-02-03 17:45                           ` Nicolin Chen
2026-02-03 17:55                             ` Jason Gunthorpe
2026-02-03 18:50                               ` Nicolin Chen
2026-02-04 13:21                                 ` Jason Gunthorpe
2026-02-03 18:59                               ` Robin Murphy
2026-02-03 19:24                                 ` Nicolin Chen
2026-02-03 23:16                                 ` Jason Gunthorpe
2026-02-04 12:18                                   ` Robin Murphy
2026-02-04 13:20                                     ` Jason Gunthorpe
2026-02-18 22:56                               ` Nicolin Chen
2026-02-19 14:37                                 ` Jason Gunthorpe
2026-02-19 16:53                                   ` Nicolin Chen
2026-02-19 17:41                                     ` Jason Gunthorpe
2026-02-20  4:52                                       ` Nicolin Chen [this message]
2026-02-20 12:50                                         ` Jason Gunthorpe
2026-02-20 13:22                                           ` Robin Murphy
2026-02-20 13:51                                             ` Jason Gunthorpe
2026-02-20 14:45                                               ` Robin Murphy
2026-02-26 15:10                                                 ` Jason Gunthorpe
2026-02-20 18:49                                           ` Nicolin Chen
2026-02-24 14:38                                             ` Jason Gunthorpe
2026-01-28  0:57                   ` Tian, Kevin
2026-01-28 13:11                     ` Jason Gunthorpe
2026-01-29  3:28                       ` Tian, Kevin
2026-01-22 10:24         ` Alejandro Lucero Palau
2026-01-17  4:56 ` [PATCH RFCv1 2/3] PCI: Allow ATS to be always on for non-CXL NVIDIA GPUs Nicolin Chen
2026-01-19 18:00   ` Jason Gunthorpe
2026-01-19 18:09     ` Nicolin Chen
2026-01-17  4:56 ` [PATCH RFCv1 3/3] iommu/arm-smmu-v3: Allow ATS to be always on Nicolin Chen
2026-01-19 20:06   ` Jason Gunthorpe
2026-01-26 12:39   ` Will Deacon
2026-01-26 17:20     ` Jason Gunthorpe
2026-01-26 18:40       ` Nicolin Chen
2026-01-26 19:16         ` Jason Gunthorpe
2026-01-26 18:49       ` Robin Murphy
2026-01-26 19:09         ` Jason Gunthorpe
2026-01-27 13:10           ` Will Deacon
2026-01-27 13:26             ` Robin Murphy
2026-01-27 13:50               ` Will Deacon
2026-01-27 14:49                 ` Jason Gunthorpe
2026-01-26 18:21     ` Nicolin Chen

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=aZfoqETxb31jp/Pk@nvidia.com \
    --to=nicolinc@nvidia.com \
    --cc=baolu.lu@linux.intel.com \
    --cc=bhelgaas@google.com \
    --cc=dan.j.williams@intel.com \
    --cc=iommu@lists.linux.dev \
    --cc=jgg@nvidia.com \
    --cc=jonathan.cameron@huawei.com \
    --cc=joro@8bytes.org \
    --cc=kevin.tian@intel.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-cxl@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=miko.lenczewski@arm.com \
    --cc=praan@google.com \
    --cc=robin.murphy@arm.com \
    --cc=will@kernel.org \
    /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.