thunderbolt: Make iommu_dma_protection more accurate

VT-d's dmar_platform_optin() actually represents a combination of properties fairly well standardised by Microsoft as "Pre-boot DMA Protection" and "Kernel DMA Protection"[1]. As such, we can provide interested consumers with an abstracted capability rather than driver-specific interfaces that won't scale. We name it for the former aspect since that's what external callers are most likely to be interested in; the latter is for the IOMMU layer to handle itself.

Also use this as an opportunity to draw a line in the sand and add a new interface so as not to introduce any more callers of iommu_capable() which I also want to get rid of. For now it's a quick'n'dirty wrapper function, but will evolve to subsume the internal interface in future.

[1] https://docs.microsoft.com/en-us/windows-hardware/design/device-experiences/oem-kernel-dma-protection

Suggested-by Christoph Hellwig <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

Quoted 13 lines · Robin Murphy

VT-d's dmar_platform_optin() actually represents a combination of properties fairly well standardised by Microsoft as "Pre-boot DMA Protection" and "Kernel DMA Protection"[1]. As such, we can provide interested consumers with an abstracted capability rather than driver-specific interfaces that won't scale. We name it for the former aspect since that's what external callers are most likely to be interested in; the latter is for the IOMMU layer to handle itself.

Also use this as an opportunity to draw a line in the sand and add a new interface so as not to introduce any more callers of iommu_capable() which I also want to get rid of. For now it's a quick'n'dirty wrapper function, but will evolve to subsume the internal interface in future.

Quoted 4 lines

[1] https://docs.microsoft.com/en-us/windows-hardware/design/device-experiences/oem-kernel-dma-protection

Suggested-by Christoph Hellwig <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

I can't really think of a way in which I suggested this, but it does looks like a good interface:

Reviewed-by Christoph Hellwig <[email protected]>

Between me trying to get rid of iommu_present() and Mario wanting to support the AMD equivalent of DMAR_PLATFORM_OPT_IN, scrutiny has shown
that the iommu_dma_protection attribute is being far too optimistic. Even if an IOMMU might be present for some PCI segment in the system, that doesn't necessarily mean it provides translation for the device(s) we care about. Furthermore, all that DMAR_PLATFORM_OPT_IN really does is tell us that memory was protected before the kernel was loaded, and prevent the user from disabling the intel-iommu driver entirely. While that lets us assume kernel integrity, what matters for actual runtime DMA protection is whether we trust individual devices, based on the "external facing" property that we expect firmware to describe for Thunderbolt ports.

It's proven challenging to determine the appropriate ports accurately given the variety of possible topologies, so while still not getting a perfect answer, by putting enough faith in firmware we can at least get a good bit closer. If we can see that any device near a Thunderbolt NHI has all the requisites for Kernel DMA Protection, chances are that it *is* a relevant port, but moreover that implies that firmware is playing the game overall, so we'll use that to assume that all Thunderbolt ports should be correctly marked and thus will end up fully protected.

CC Mario Limonciello <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

---

v2: Give up trying to look for specific devices, just look for evidence
that firmware cares at all.

 drivers/thunderbolt/domain.c | 12 +++--------
 drivers/thunderbolt/nhi.c    | 41
++++++++++++++++++++++++++++++++++++
 include/linux/thunderbolt.h  |  2 ++
 3 files changed, 46 insertions(+), 9 deletions(-)

drivers/thunderbolt/domain.c +3 -9

index 7018d959f775..2889a214dadc 100644
@@ -7,9 +7,7 @@
  */
 
 #include <linux/device.h>
-#include <linux/dmar.h>
 #include <linux/idr.h>
-#include <linux/iommu.h>
 #include <linux/module.h>
 #include <linux/pm_runtime.h>
 #include <linux/slab.h>

@@ -257,13 +255,9 @@ static ssize_t iommu_dma_protection_show(struct
 device *dev,
 					 struct device_attribute *attr,
 					 char *buf)
 {
-	/*
-	 * Kernel DMA protection is a feature where Thunderbolt security is
-	 * handled natively using IOMMU. It is enabled when IOMMU is
-	 * enabled and ACPI DMAR table has DMAR_PLATFORM_OPT_IN set.
-	 */
-	return sprintf(buf, "%d\n",
-		       iommu_present(&pci_bus_type) &&
 dmar_platform_optin());
+	struct tb *tb = container_of(dev, struct tb, dev);
+
+	return sysfs_emit(buf, "%d\n", tb->nhi->iommu_dma_protection);
 }

static DEVICE_ATTR_RO(iommu_dma_protection);

drivers/thunderbolt/nhi.c +40 -0

index c73da0532be4..9e396e283792 100644
@@ -14,6 +14,7 @@
 #include <linux/errno.h>
 #include <linux/pci.h>
 #include <linux/interrupt.h>
+#include <linux/iommu.h>
 #include <linux/module.h>
 #include <linux/delay.h>
 #include <linux/property.h>

@@ -1102,6 +1103,45 @@ static void nhi_check_quirks(struct tb_nhi *nhi)
 		nhi->quirks |= QUIRK_AUTO_CLEAR_INT;
 }
 
+static int nhi_check_iommu_pdev(struct pci_dev *pdev, void *data)
+{
+	if (!pdev->untrusted ||
+	    !dev_iommu_capable(&pdev->dev,
 IOMMU_CAP_PRE_BOOT_PROTECTION))
+		return 0;
+	*(bool *)data = true;
+	return 1; /* Stop walking */
+}
+
+static void nhi_check_iommu(struct tb_nhi *nhi)
+{
+	struct pci_bus *bus = nhi->pdev->bus;
+	bool port_ok = false;
+
+	/*
+	 * Ideally what we'd do here is grab every PCI device that
+	 * represents a tunnelling adapter for this NHI and check their
+	 * status directly, but unfortunately USB4 seems to make it
+	 * obnoxiously difficult to reliably make any correlation.
+	 *
+	 * So for now we'll have to bodge it... Hoping that the system
+	 * is at least sane enough that an adapter is in the same PCI
+	 * segment as its NHI, if we can find *something* on that segment
+	 * which meets the requirements for Kernel DMA Protection, we'll
+	 * take that to imply that firmware is aware and has (hopefully)
+	 * done the right thing in general. We need to know that the PCI
+	 * layer has seen the ExternalFacingPort property and propagated
+	 * it to the "untrusted" flag that the IOMMU layer will then
+	 * enforce, but also that the IOMMU driver itself can be trusted
+	 * not to have been subverted by a pre-boot DMA attack.
+	 */
+	while (bus->parent)
+		bus = bus->parent;
+
+	pci_walk_bus(bus, nhi_check_iommu_pdev, &port_ok);
+
+	nhi->iommu_dma_protection = port_ok;
+}
+
 static int nhi_init_msi(struct tb_nhi *nhi)
 {

drivers/thunderbolt/nhi.c +1 -0

 	struct pci_dev *pdev = nhi->pdev;

@@ -1219,6 +1259,7 @@ static int nhi_probe(struct pci_dev *pdev, const
 struct pci_device_id *id)
 		return -ENOMEM;
 
 	nhi_check_quirks(nhi);
+	nhi_check_iommu(nhi);
 
 	res = nhi_init_msi(nhi);

drivers/thunderbolt/nhi.c +0 -0

 	if (res) {

include/linux/thunderbolt.h +1 -0

index 124e13cb1469..7a8ad984e651 100644
@@ -465,6 +465,7 @@ static inline struct tb_xdomain
 *tb_service_parent(struct tb_service *svc)
  * @msix_ida: Used to allocate MSI-X vectors for rings
  * @going_away: The host controller device is about to disappear so when
  *		this flag is set, avoid touching the hardware anymore.
+ * @iommu_dma_protection: An IOMMU will isolate external-facing ports.
  * @interrupt_work: Work scheduled to handle ring interrupt when no
  *		    MSI-X is used.

* @hop_count: Number of rings (end point hops) supported by NHI.

include/linux/thunderbolt.h +1 -1

@@ -479,6 +480,7 @@ struct tb_nhi {
 	struct tb_ring **rx_rings;
 	struct ida msix_ida;
 	bool going_away;
+	bool iommu_dma_protection;
 	struct work_struct interrupt_work;
 	u32 hop_count;
 	unsigned long quirks;
--

2.28.0.dirty

[Public]

Quoted 14 lines

Between me trying to get rid of iommu_present() and Mario wanting to support the AMD equivalent of DMAR_PLATFORM_OPT_IN, scrutiny has shown
that the iommu_dma_protection attribute is being far too optimistic. Even if an IOMMU might be present for some PCI segment in the system, that doesn't necessarily mean it provides translation for the device(s) we care about. Furthermore, all that DMAR_PLATFORM_OPT_IN really does is tell us that memory was protected before the kernel was loaded, and prevent the user from disabling the intel-iommu driver entirely. While that lets us assume kernel integrity, what matters for actual runtime DMA protection is whether we trust individual devices, based on the "external facing" property that we expect firmware to describe for Thunderbolt ports.

Quoted 9 lines

It's proven challenging to determine the appropriate ports accurately given the variety of possible topologies, so while still not getting a perfect answer, by putting enough faith in firmware we can at least get a good bit closer. If we can see that any device near a Thunderbolt NHI has all the requisites for Kernel DMA Protection, chances are that it *is* a relevant port, but moreover that implies that firmware is playing the game overall, so we'll use that to assume that all Thunderbolt ports should be correctly marked and thus will end up fully protected.

This approach looks generally good to me. I do worry a little bit about older systems that didn't set ExternalFacingPort in the FW but were previously setting iommu_dma_protection, but I think that those could be treated on a quirk basis to set PCI IDs for those root ports as external facing if/when they come up.

I'll send up a follow up patch that adds the AMD ACPI table check. If it looks good can roll it into your series for v3, or if this series goes as is for v2 it can come on its own.

Quoted 4 lines

CC Mario Limonciello <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

---

Quoted 2 lines

v2: Give up trying to look for specific devices, just look for evidence
that firmware cares at all.

I still do think you could know exactly which devices to use if you're in SW CM mode, but I guess the consensus is to not bifurcate the way this can be checked.

Hi Mario,

Quoted 26 lines · Mario

[Public]

Quoted 23 lines

Between me trying to get rid of iommu_present() and Mario wanting to support the AMD equivalent of DMAR_PLATFORM_OPT_IN, scrutiny has shown
that the iommu_dma_protection attribute is being far too optimistic. Even if an IOMMU might be present for some PCI segment in the system, that doesn't necessarily mean it provides translation for the device(s) we care about. Furthermore, all that DMAR_PLATFORM_OPT_IN really does is tell us that memory was protected before the kernel was loaded, and prevent the user from disabling the intel-iommu driver entirely. While that lets us assume kernel integrity, what matters for actual runtime DMA protection is whether we trust individual devices, based on the "external facing" property that we expect firmware to describe for Thunderbolt ports.

It's proven challenging to determine the appropriate ports accurately given the variety of possible topologies, so while still not getting a perfect answer, by putting enough faith in firmware we can at least get a good bit closer. If we can see that any device near a Thunderbolt NHI has all the requisites for Kernel DMA Protection, chances are that it *is* a relevant port, but moreover that implies that firmware is playing the game overall, so we'll use that to assume that all Thunderbolt ports should be correctly marked and thus will end up fully protected.

Quoted 5 lines

This approach looks generally good to me. I do worry a little bit about older systems that didn't set ExternalFacingPort in the FW but were previously setting iommu_dma_protection, but I think that those could be treated on a quirk basis to set PCI IDs for those root ports as external facing if/when they come up.

There are no such systems out there AFAICT.

Quoted 8 lines

I'll send up a follow up patch that adds the AMD ACPI table check. If it looks good can roll it into your series for v3, or if this series goes as is for v2 it can come on its own.

Quoted 4 lines

CC Mario Limonciello <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

---

Quoted 6 lines

Quoted 2 lines

v2: Give up trying to look for specific devices, just look for evidence
that firmware cares at all.

I still do think you could know exactly which devices to use if you're in SW CM mode, but I guess the consensus is to not bifurcate the way this can be checked.

Indeed.

The patch looks good to me now. I will give it a try on a couple of systems later today or tomorrow and let you guys know how it went. I don't expect any problems but let's see.

Thanks a lot Robin for working on this :)

-----Original Message-----
From: Robin Murphy <[email protected]>
Sent: Monday, March 21, 2022 06:12
To: [email protected]; Limonciello, Mario <[email protected]>
Cc: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected] foundation.org; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH v2 2/2] thunderbolt: Make iommu_dma_protection more accurate

Quoted 10 lines · On 2022-03-21 10:58, [email protected] wrote:

Hi Mario,

Quoted 7 lines · Mario

[Public]

Quoted 5 lines

Between me trying to get rid of iommu_present() and Mario wanting to support the AMD equivalent of DMAR_PLATFORM_OPT_IN, scrutiny has shown
that the iommu_dma_protection attribute is being far too optimistic. Even if an IOMMU might be present for some PCI segment in the

system,

Quoted 2 lines

Quoted 2 lines

Quoted 2 lines

that doesn't necessarily mean it provides translation for the device(s) we care about. Furthermore, all that DMAR_PLATFORM_OPT_IN really

does

Quoted 17 lines

Quoted 17 lines

Quoted 16 lines

is tell us that memory was protected before the kernel was loaded, and prevent the user from disabling the intel-iommu driver entirely. While that lets us assume kernel integrity, what matters for actual runtime DMA protection is whether we trust individual devices, based on the "external facing" property that we expect firmware to describe for Thunderbolt ports.

It's proven challenging to determine the appropriate ports accurately given the variety of possible topologies, so while still not getting a perfect answer, by putting enough faith in firmware we can at least get a good bit closer. If we can see that any device near a Thunderbolt NHI has all the requisites for Kernel DMA Protection, chances are that it *is* a relevant port, but moreover that implies that firmware is playing the game overall, so we'll use that to assume that all Thunderbolt ports should be correctly marked and thus will end up fully protected.

Quoted 1 line

Quoted 1 line

This approach looks generally good to me. I do worry a little bit about

older

Quoted 1 line

Quoted 1 line

systems that didn't set ExternalFacingPort in the FW but were previously

setting

Quoted 1 line

Quoted 1 line

iommu_dma_protection, but I think that those could be treated on a

quirk

Quoted 1 line

Quoted 1 line

basis to set PCI IDs for those root ports as external facing if/when they

come

Quoted 3 lines

Quoted 1 line

up.

There are no such systems out there AFAICT.

And even if there are, as above they've never actually been fully protected and still won't be, so it's arguably a good thing for them to stop thinking so.

Quoted 22 lines

Quoted 9 lines

I'll send up a follow up patch that adds the AMD ACPI table check. If it looks good can roll it into your series for v3, or if this series goes as is for v2 it can come on its own.

Quoted 4 lines

CC Mario Limonciello <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

---

Quoted 6 lines

Quoted 2 lines

v2: Give up trying to look for specific devices, just look for evidence
that firmware cares at all.

I still do think you could know exactly which devices to use if you're in SW CM mode, but I guess the consensus is to not bifurcate the way this can be checked.

Indeed.

The patch looks good to me now. I will give it a try on a couple of systems later today or tomorrow and let you guys know how it went. I don't expect any problems but let's see.

Quoted 1 line

Thanks a lot Robin for working on this :)

Heh, let's just hope the other half-dozen or so subsystems I need to touch for this IOMMU cleanup aren't all quite as involved as this turned out to be :)

Cheers,
Robin.

Between me trying to get rid of iommu_present() and Mario wanting to support the AMD equivalent of DMAR_PLATFORM_OPT_IN, scrutiny has shown that the iommu_dma_protection attribute is being far too optimistic. Even if an IOMMU might be present for some PCI segment in the system, that doesn't necessarily mean it provides translation for the device(s) we care about. Furthermore, all that DMAR_PLATFORM_OPT_IN really does is tell us that memory was protected before the kernel was loaded, and prevent the user from disabling the intel-iommu driver entirely. While that lets us assume kernel integrity, what matters for actual runtime DMA protection is whether we trust individual devices, based on the "external facing" property that we expect firmware to describe for Thunderbolt ports.

It's proven challenging to determine the appropriate ports accurately given the variety of possible topologies, so while still not getting a perfect answer, by putting enough faith in firmware we can at least get a good bit closer. If we can see that any device near a Thunderbolt NHI has all the requisites for Kernel DMA Protection, chances are that it *is* a relevant port, but moreover that implies that firmware is playing the game overall, so we'll use that to assume that all Thunderbolt ports should be correctly marked and thus will end up fully protected.

CC Mario Limonciello <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

Between me trying to get rid of iommu_present() and Mario wanting to support the AMD equivalent of DMAR_PLATFORM_OPT_IN, scrutiny has shown that the iommu_dma_protection attribute is being far too optimistic. Even if an IOMMU might be present for some PCI segment in the system, that doesn't necessarily mean it provides translation for the device(s) we care about. Furthermore, all that DMAR_PLATFORM_OPT_IN really does is tell us that memory was protected before the kernel was loaded, and prevent the user from disabling the intel-iommu driver entirely. While that lets us assume kernel integrity, what matters for actual runtime DMA protection is whether we trust individual devices, based on the "external facing" property that we expect firmware to describe for Thunderbolt ports.

It's proven challenging to determine the appropriate ports accurately given the variety of possible topologies, so while still not getting a perfect answer, by putting enough faith in firmware we can at least get a good bit closer. If we can see that any device near a Thunderbolt NHI has all the requisites for Kernel DMA Protection, chances are that it *is* a relevant port, but moreover that implies that firmware is playing the game overall, so we'll use that to assume that all Thunderbolt ports should be correctly marked and thus will end up fully protected.

CC Mario Limonciello <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

---

v2: Give up trying to look for specific devices, just look for evidence
that firmware cares at all.

 drivers/thunderbolt/domain.c | 12 +++--------
 drivers/thunderbolt/nhi.c    | 41 ++++++++++++++++++++++++++++++++++++
 include/linux/thunderbolt.h  |  2 ++
 3 files changed, 46 insertions(+), 9 deletions(-)

drivers/thunderbolt/domain.c +3 -9

index 7018d959f775..2889a214dadc 100644
@@ -7,9 +7,7 @@
  */
 
 #include <linux/device.h>
-#include <linux/dmar.h>
 #include <linux/idr.h>
-#include <linux/iommu.h>
 #include <linux/module.h>
 #include <linux/pm_runtime.h>
 #include <linux/slab.h>

@@ -257,13 +255,9 @@ static ssize_t iommu_dma_protection_show(struct device *dev,
 					 struct device_attribute *attr,
 					 char *buf)
 {
-	/*
-	 * Kernel DMA protection is a feature where Thunderbolt security is
-	 * handled natively using IOMMU. It is enabled when IOMMU is
-	 * enabled and ACPI DMAR table has DMAR_PLATFORM_OPT_IN set.
-	 */
-	return sprintf(buf, "%d\n",
-		       iommu_present(&pci_bus_type) && dmar_platform_optin());
+	struct tb *tb = container_of(dev, struct tb, dev);
+
+	return sysfs_emit(buf, "%d\n", tb->nhi->iommu_dma_protection);
 }
 static DEVICE_ATTR_RO(iommu_dma_protection);
 

drivers/thunderbolt/nhi.c +1 -0

index c73da0532be4..9e396e283792 100644
@@ -14,6 +14,7 @@
 #include <linux/errno.h>
 #include <linux/pci.h>
 #include <linux/interrupt.h>
+#include <linux/iommu.h>
 #include <linux/module.h>
 #include <linux/delay.h>
 #include <linux/property.h>

drivers/thunderbolt/nhi.c +23 -0

+		return 0;
+	*(bool *)data = true;
+	return 1; /* Stop walking */
+}
+
+static void nhi_check_iommu(struct tb_nhi *nhi)
+{
+	struct pci_bus *bus = nhi->pdev->bus;
+	bool port_ok = false;
+
+	/*
+	 * Ideally what we'd do here is grab every PCI device that
+	 * represents a tunnelling adapter for this NHI and check their
+	 * status directly, but unfortunately USB4 seems to make it
+	 * obnoxiously difficult to reliably make any correlation.
+	 *
+	 * So for now we'll have to bodge it... Hoping that the system
+	 * is at least sane enough that an adapter is in the same PCI
+	 * segment as its NHI, if we can find *something* on that segment
+	 * which meets the requirements for Kernel DMA Protection, we'll
+	 * take that to imply that firmware is aware and has (hopefully)
+	 * done the right thing in general. We need to know that the PCI
+	 * layer has seen the ExternalFacingPort property and propagated

Hi Robin,

I tried this now on two Intel systems. One with integrated Thunderbolt and one with discrete. There was a small issue, see below but once fixed it worked as expected :)

Quoted 78 lines · Robin Murphy

Between me trying to get rid of iommu_present() and Mario wanting to support the AMD equivalent of DMAR_PLATFORM_OPT_IN, scrutiny has shown that the iommu_dma_protection attribute is being far too optimistic. Even if an IOMMU might be present for some PCI segment in the system, that doesn't necessarily mean it provides translation for the device(s) we care about. Furthermore, all that DMAR_PLATFORM_OPT_IN really does is tell us that memory was protected before the kernel was loaded, and prevent the user from disabling the intel-iommu driver entirely. While that lets us assume kernel integrity, what matters for actual runtime DMA protection is whether we trust individual devices, based on the "external facing" property that we expect firmware to describe for Thunderbolt ports.

It's proven challenging to determine the appropriate ports accurately given the variety of possible topologies, so while still not getting a perfect answer, by putting enough faith in firmware we can at least get a good bit closer. If we can see that any device near a Thunderbolt NHI has all the requisites for Kernel DMA Protection, chances are that it *is* a relevant port, but moreover that implies that firmware is playing the game overall, so we'll use that to assume that all Thunderbolt ports should be correctly marked and thus will end up fully protected.

CC Mario Limonciello <[email protected]>

Signed-off-by Robin Murphy <[email protected]>

---

v2: Give up trying to look for specific devices, just look for evidence
that firmware cares at all.

  drivers/thunderbolt/domain.c | 12 +++--------
  drivers/thunderbolt/nhi.c    | 41 ++++++++++++++++++++++++++++++++++++
  include/linux/thunderbolt.h  |  2 ++
  3 files changed, 46 insertions(+), 9 deletions(-)

drivers/thunderbolt/domain.c +3 -9

index 7018d959f775..2889a214dadc 100644
@@ -7,9 +7,7 @@
   */
 
  #include <linux/device.h>
-#include <linux/dmar.h>
  #include <linux/idr.h>
-#include <linux/iommu.h>
  #include <linux/module.h>
  #include <linux/pm_runtime.h>
  #include <linux/slab.h>

@@ -257,13 +255,9 @@ static ssize_t iommu_dma_protection_show(struct device *dev,
  					 struct device_attribute *attr,
  					 char *buf)
  {
-	/*
-	 * Kernel DMA protection is a feature where Thunderbolt security is
-	 * handled natively using IOMMU. It is enabled when IOMMU is
-	 * enabled and ACPI DMAR table has DMAR_PLATFORM_OPT_IN set.
-	 */
-	return sprintf(buf, "%d\n",
-		       iommu_present(&pci_bus_type) && dmar_platform_optin());
+	struct tb *tb = container_of(dev, struct tb, dev);
+
+	return sysfs_emit(buf, "%d\n", tb->nhi->iommu_dma_protection);
  }
  static DEVICE_ATTR_RO(iommu_dma_protection);
 

drivers/thunderbolt/nhi.c +1 -0

index c73da0532be4..9e396e283792 100644
@@ -14,6 +14,7 @@
  #include <linux/errno.h>
  #include <linux/pci.h>
  #include <linux/interrupt.h>
+#include <linux/iommu.h>
  #include <linux/module.h>
  #include <linux/delay.h>
  #include <linux/property.h>

drivers/thunderbolt/nhi.c static void nhi_check_quirks(struct tb_nhi *nhi) · line 1109

Quoted 7 lines

drivers/thunderbolt/nhi.c +4 -0

@@ -1102,6 +1103,45 @@ static void nhi_check_quirks(struct tb_nhi *nhi)
  		nhi->quirks |= QUIRK_AUTO_CLEAR_INT;
  }
 
+static int nhi_check_iommu_pdev(struct pci_dev *pdev, void *data)
+{
+	if (!pdev->untrusted ||
+	    !dev_iommu_capable(&pdev->dev, IOMMU_CAP_PRE_BOOT_PROTECTION))

This one needs to take the pdev->external_facing into account too because most of the time there are no existing tunnels when the driver is loaded so we only see the PCIe root/downstream port. I think this is enough actually:

	if (!pdev->external_facing ||
	    !dev_iommu_capable(&pdev->dev, IOMMU_CAP_PRE_BOOT_PROTECTION))

Quoted 23 lines

drivers/thunderbolt/nhi.c +23 -0

+		return 0;
+	*(bool *)data = true;
+	return 1; /* Stop walking */
+}
+
+static void nhi_check_iommu(struct tb_nhi *nhi)
+{
+	struct pci_bus *bus = nhi->pdev->bus;
+	bool port_ok = false;
+
+	/*
+	 * Ideally what we'd do here is grab every PCI device that
+	 * represents a tunnelling adapter for this NHI and check their
+	 * status directly, but unfortunately USB4 seems to make it
+	 * obnoxiously difficult to reliably make any correlation.
+	 *
+	 * So for now we'll have to bodge it... Hoping that the system
+	 * is at least sane enough that an adapter is in the same PCI
+	 * segment as its NHI, if we can find *something* on that segment
+	 * which meets the requirements for Kernel DMA Protection, we'll
+	 * take that to imply that firmware is aware and has (hopefully)
+	 * done the right thing in general. We need to know that the PCI
+	 * layer has seen the ExternalFacingPort property and propagated

drivers/thunderbolt/nhi.c static void nhi_check_quirks(struct tb_nhi *nhi) · line 1142

Quoted 10 lines

drivers/thunderbolt/nhi.c +10 -0

+	 * it to the "untrusted" flag that the IOMMU layer will then
+	 * enforce, but also that the IOMMU driver itself can be trusted
+	 * not to have been subverted by a pre-boot DMA attack.
+	 */
+	while (bus->parent)
+		bus = bus->parent;
+
+	pci_walk_bus(bus, nhi_check_iommu_pdev, &port_ok);
+
+	nhi->iommu_dma_protection = port_ok;

I would put here a log debug, something like this:

dev_dbg(&nhi->pdev->dev, "IOMMU DMA protection is %sabled\n",

	port_ok ? "en" : "dis");