Skip to contentSkip to conversation

iommu/vt-d: Remove caching mode check before device TLB flush

PATCH In review Lu Baolu 4 participants 11 messages Apr 15, 2024 – Apr 24, 2024

Subsystems (2)
lore
Share & tools
lore

Lu Baolu

author maintainer
#0 lore raw

PATCH v3 1/1 iommu/vt-d: Remove caching mode check before device TLB flush

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

---
 drivers/iommu/intel/iommu.c | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

Change log:
v3:
- It turned out that the removals don't change the driver's behavior, hence change it from a fix patch to a cleanup one.
- No functionality changes.
v2: https://lore.kernel.org/lkml/[email protected]/
- Squash two patches into a single one.
- No functionality changes.
v1: https://lore.kernel.org/linux-iommu/[email protected]/

drivers/iommu/intel/iommu.c +2 -7
index a7ecd90303dc..f0a67e9d9faf 100644
@@ -1501,11 +1501,7 @@ static void iommu_flush_iotlb_psi(struct intel_iommu *iommu,
 	else
 		__iommu_flush_iotlb_psi(iommu, did, pfn, pages, ih);
 
-	/*
-	 * In caching mode, changes of pages from non-present to present require
-	 * flush. However, device IOTLB doesn't need to be flushed in this case.
-	 */
-	if (!cap_caching_mode(iommu->cap) || !map)
+	if (!map)
 		iommu_flush_dev_iotlb(domain, addr, mask);
 }
 
@@ -1579,8 +1575,7 @@ static void intel_flush_iotlb_all(struct iommu_domain *domain)
 			iommu->flush.flush_iotlb(iommu, did, 0, 0,
 						 DMA_TLB_DSI_FLUSH);
 
-		if (!cap_caching_mode(iommu->cap))
-			iommu_flush_dev_iotlb(dmar_domain, 0, MAX_AGAW_PFN_WIDTH);
+		iommu_flush_dev_iotlb(dmar_domain, 0, MAX_AGAW_PFN_WIDTH);
 	}
 
 	if (dmar_domain->nested_parent)
Signature · 1 line
2.34.1
#1 lore raw
Quoted 19 lines

-----Original Message-----
From: Lu Baolu <[email protected]>
Sent: Monday, April 15, 2024 9:39 AM
To: [email protected]
Cc: Tian, Kevin <[email protected]>; Liu, Yi L <[email protected]>; Jacob Pan <[email protected]>; Joerg Roedel <[email protected]>; Will Deacon <[email protected]>; Robin Murphy <[email protected]>; linux- [email protected]; Lu Baolu <[email protected]>
Subject: [PATCH v3 1/1] iommu/vt-d: Remove caching mode check before device TLB flush

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Quoted 7 lines

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

The current logic is if the caching mode is being used and a domain isn't using first level I/O page table, then flush-queue won't be used. Otherwise, the flush-queue can be enabled. See https://github.com/torvalds/linux/commit/257ec29074

In other words, if the caching mode is being used and a domain is using first level I/O page table, the flush-queue can be used for this domain to flush iotlb. Could the code change in this patch bring any performance impact to this case?

Regards,
-Tina

#2 lore raw
Quoted 26 lines · Tina
Quoted 19 lines

-----Original Message-----
From: Lu Baolu <[email protected]>
Sent: Monday, April 15, 2024 9:39 AM
To: [email protected]
Cc: Tian, Kevin <[email protected]>; Liu, Yi L <[email protected]>; Jacob Pan <[email protected]>; Joerg Roedel <[email protected]>; Will Deacon <[email protected]>; Robin Murphy <[email protected]>; linux- [email protected]; Lu Baolu <[email protected]>
Subject: [PATCH v3 1/1] iommu/vt-d: Remove caching mode check before device TLB flush

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Quoted 7 lines

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

Quoted 4 lines

The current logic is if the caching mode is being used and a domain isn't using first level I/O page table, then flush-queue won't be used. Otherwise, the flush-queue can be enabled. See https://github.com/torvalds/linux/commit/257ec29074

In other words, if the caching mode is being used and a domain is using first level I/O page table, the flush-queue can be used for this domain to flush iotlb. Could the code change in this patch bring any performance impact to this case?

This seems to have performance deduction in the nested translation case. The iommufd nested support bas been merged in 6.8, while the Qemu side is wip. So this performance deduction does not happen until Qemu is done. Should this also be considered as a performance regression? TBH. I doubt if it should be.

Signature · 2 lines
Regards,
Yi Liu
author maintainer
#3 lore raw
Quoted 30 lines · Yi Liu
Quoted 29 lines · Tina
Quoted 19 lines

-----Original Message-----
From: Lu Baolu <[email protected]>
Sent: Monday, April 15, 2024 9:39 AM
To: [email protected]
Cc: Tian, Kevin <[email protected]>; Liu, Yi L <[email protected]>; Jacob Pan <[email protected]>; Joerg Roedel <[email protected]>; Will Deacon <[email protected]>; Robin Murphy <[email protected]>; linux- [email protected]; Lu Baolu <[email protected]>
Subject: [PATCH v3 1/1] iommu/vt-d: Remove caching mode check before device TLB flush

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Quoted 7 lines

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

The current logic is if the caching mode is being used and a domain isn't using first level I/O page table, then flush-queue won't be used. Otherwise, the flush-queue can be enabled. See https://github.com/torvalds/linux/commit/257ec29074

Quoted 6 lines
Quoted 1 line

In other words, if the caching mode is being used and a domain is using first level I/O page table, the flush-queue can be used for this domain to flush iotlb. Could the code change in this patch bring any performance impact to this case?

This seems to have performance deduction in the nested translation case. The iommufd nested support bas been merged in 6.8, while the Qemu side is wip. So this performance deduction does not happen until Qemu is done. Should this also be considered as a performance regression? TBH.

Caching mode is irrelevant to first-stage and nesting translations. If the QEMU implementation still relies on caching mode for nesting support, it's already broken.

Best regards,
baolu

#4 lore raw
Quoted 33 lines · Baolu Lu
Quoted 32 lines · Yi Liu
Quoted 30 lines · Tina
Quoted 19 lines

-----Original Message-----
From: Lu Baolu <[email protected]>
Sent: Monday, April 15, 2024 9:39 AM
To: [email protected]
Cc: Tian, Kevin <[email protected]>; Liu, Yi L <[email protected]>; Jacob Pan <[email protected]>; Joerg Roedel <[email protected]>; Will Deacon <[email protected]>; Robin Murphy <[email protected]>; linux- [email protected]; Lu Baolu <[email protected]>
Subject: [PATCH v3 1/1] iommu/vt-d: Remove caching mode check before device TLB flush

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Quoted 7 lines

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

The current logic is if the caching mode is being used and a domain isn't using first level I/O page table, then flush-queue won't be used. Otherwise, the flush-queue can be enabled. See https://github.com/torvalds/linux/commit/257ec29074

In other words, if the caching mode is being used and a domain is using first level I/O page table, the flush-queue can be used for this domain to flush iotlb. Could the code change in this patch bring any performance impact to this case?

Quoted 8 lines
Quoted 4 lines

This seems to have performance deduction in the nested translation case. The iommufd nested support bas been merged in 6.8, while the Qemu side is wip. So this performance deduction does not happen until Qemu is done. Should this also be considered as a performance regression? TBH.

Caching mode is irrelevant to first-stage and nesting translations. If the QEMU implementation still relies on caching mode for nesting support, it's already broken.

For first-stage, yes, caching mode is irrelevant. But for nesting, still
be relevant. Because it relies on caching mode to capture the mofifications
of the pasid entry. Without caching mode, vIOMMU device model within QEMU will miss the newly setup pasid entries, hence no way to setup nested translation at all.

Signature · 2 lines
Regards,
Yi Liu
#7 lore raw
In reply to Lu Baolu
Quoted 36 lines · Lu Baolu

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

---
  drivers/iommu/intel/iommu.c | 9 ++-------
  1 file changed, 2 insertions(+), 7 deletions(-)

Change log:
v3:
- It turned out that the removals don't change the driver's behavior, hence change it from a fix patch to a cleanup one.
- No functionality changes.
v2: https://lore.kernel.org/lkml/[email protected]/
- Squash two patches into a single one.
- No functionality changes.
v1: https://lore.kernel.org/linux-iommu/[email protected]/

drivers/iommu/intel/iommu.c +0 -0
index a7ecd90303dc..f0a67e9d9faf 100644
@@ -1501,11 +1501,7 @@ static void iommu_flush_iotlb_psi(struct intel_iommu *iommu,
  	else

drivers/iommu/intel/iommu.c static void iommu_flush_iotlb_psi(struct intel_iommu *iommu, · line 1506

Quoted 10 lines
drivers/iommu/intel/iommu.c +1 -5
  		__iommu_flush_iotlb_psi(iommu, did, pfn, pages, ih);
 
-	/*
-	 * In caching mode, changes of pages from non-present to present require
-	 * flush. However, device IOTLB doesn't need to be flushed in this case.
-	 */
-	if (!cap_caching_mode(iommu->cap) || !map)
+	if (!map)
  		iommu_flush_dev_iotlb(domain, addr, mask);
  }

Given devTLB flushing is irrelavent to CM, put iommu_flush_dev_iotlb() in iommu_flush_iotlb_psi() and called with CM checking context is not reasonable. the logic is buggy.

static void __mapping_notify_one(struct intel_iommu *iommu, struct dmar_domain *domain,
unsigned long pfn, unsigned int pages)

{
	/*
	 * It's a non-present to present mapping. Only flush if caching mode
	 * and second level.
	 */
	if (cap_caching_mode(iommu->cap) && !domain->use_first_level)
		iommu_flush_iotlb_psi(iommu, domain, pfn, pages, 0, 1);
	else
		iommu_flush_write_buffer(iommu);

then how about fold all CM checking logic in iommu_flush_iotlb_psi() or speperate iommu_flush_dev_iotlb() from iommu_flush_iotlb_psi() ?

Thanks,
Ethan

Quoted 11 lines
drivers/iommu/intel/iommu.c +1 -2
@@ -1579,8 +1575,7 @@ static void intel_flush_iotlb_all(struct iommu_domain *domain)
  			iommu->flush.flush_iotlb(iommu, did, 0, 0,
  						 DMA_TLB_DSI_FLUSH);
 
-		if (!cap_caching_mode(iommu->cap))
-			iommu_flush_dev_iotlb(dmar_domain, 0, MAX_AGAW_PFN_WIDTH);
+		iommu_flush_dev_iotlb(dmar_domain, 0, MAX_AGAW_PFN_WIDTH);
  	}
 
  	if (dmar_domain->nested_parent)
author maintainer
#8 lore raw
Quoted 67 lines · Ethan Zhao
Quoted 40 lines · Lu Baolu

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

---
  drivers/iommu/intel/iommu.c | 9 ++-------

  1 file changed, 2 insertions(+), 7 deletions(-)

Change log:
v3:
  - It turned out that the removals don't change the driver's behavior, hence change it from a fix patch to a cleanup one.
  - No functionality changes.
v2: https://lore.kernel.org/lkml/[email protected]/
  - Squash two patches into a single one.
  - No functionality changes.
v1: https://lore.kernel.org/linux-iommu/[email protected]/

drivers/iommu/intel/iommu.c +0 -0
index a7ecd90303dc..f0a67e9d9faf 100644
@@ -1501,11 +1501,7 @@ static void iommu_flush_iotlb_psi(struct 
 intel_iommu *iommu,

      else
          __iommu_flush_iotlb_psi(iommu, did, pfn, pages, ih); -    /*
-     * In caching mode, changes of pages from non-present to present

Quoted 8 lines

require
-     * flush. However, device IOTLB doesn't need to be flushed in this case.
-     */
-    if (!cap_caching_mode(iommu->cap) || !map)
+    if (!map)
          iommu_flush_dev_iotlb(domain, addr, mask);
  }

Given devTLB flushing is irrelavent to CM, put iommu_flush_dev_iotlb() in iommu_flush_iotlb_psi() and called with CM checking context is not reasonable. the logic is buggy.

static void __mapping_notify_one(struct intel_iommu *iommu, struct dmar_domain *domain,
                 unsigned long pfn, unsigned int pages)
{
    /*
     * It's a non-present to present mapping. Only flush if caching mode
     * and second level.
     */
    if (cap_caching_mode(iommu->cap) && !domain->use_first_level) iommu_flush_iotlb_psi(iommu, domain, pfn, pages, 0, 1);
    else
        iommu_flush_write_buffer(iommu);

Quoted 2 lines

then how about fold all CM checking logic in iommu_flush_iotlb_psi() or speperate iommu_flush_dev_iotlb() from iommu_flush_iotlb_psi() ?

I am refactoring the code with a new series.

https://lore.kernel.org/linux-iommu/[email protected]/

Best regards,
baolu

author maintainer
#5 lore raw
In reply to Yi Liu
Quoted 39 lines · Yi Liu
Quoted 38 lines · Baolu Lu
Quoted 36 lines · Yi Liu
Quoted 26 lines · Tina
Quoted 19 lines

-----Original Message-----
From: Lu Baolu <[email protected]>
Sent: Monday, April 15, 2024 9:39 AM
To: [email protected]
Cc: Tian, Kevin <[email protected]>; Liu, Yi L <[email protected]>; Jacob Pan <[email protected]>; Joerg Roedel <[email protected]>; Will Deacon <[email protected]>; Robin Murphy <[email protected]>; linux- [email protected]; Lu Baolu <[email protected]>
Subject: [PATCH v3 1/1] iommu/vt-d: Remove caching mode check before device TLB flush

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Quoted 7 lines

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

Quoted 4 lines

The current logic is if the caching mode is being used and a domain isn't using first level I/O page table, then flush-queue won't be used. Otherwise, the flush-queue can be enabled. See https://github.com/torvalds/linux/commit/257ec29074

In other words, if the caching mode is being used and a domain is using first level I/O page table, the flush-queue can be used for this domain to flush iotlb. Could the code change in this patch bring any performance impact to this case?

This seems to have performance deduction in the nested translation case. The iommufd nested support bas been merged in 6.8, while the Qemu side is wip. So this performance deduction does not happen until Qemu is done. Should this also be considered as a performance regression? TBH.

Quoted 8 lines
Quoted 3 lines

Caching mode is irrelevant to first-stage and nesting translations. If the QEMU implementation still relies on caching mode for nesting support, it's already broken.

For first-stage, yes, caching mode is irrelevant. But for nesting, still
be relevant. Because it relies on caching mode to capture the mofifications
of the pasid entry. Without caching mode, vIOMMU device model within QEMU will miss the newly setup pasid entries, hence no way to setup nested translation at all.

Okay, it's reasonable that nesting translation still uses second-stage. But the removed checks in this patch don't impact the pasid table change capture, right?

Best regards,
baolu

#9 lore raw
In reply to Baolu Lu
Quoted 68 lines · Baolu Lu
Quoted 67 lines · Ethan Zhao
Quoted 40 lines · Lu Baolu

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

---
  drivers/iommu/intel/iommu.c | 9 ++-------

  1 file changed, 2 insertions(+), 7 deletions(-)

Change log:
v3:
  - It turned out that the removals don't change the driver's behavior, hence change it from a fix patch to a cleanup one.
  - No functionality changes.
v2: https://lore.kernel.org/lkml/[email protected]/
  - Squash two patches into a single one.
  - No functionality changes.
v1: https://lore.kernel.org/linux-iommu/[email protected]/

drivers/iommu/intel/iommu.c +0 -0
index a7ecd90303dc..f0a67e9d9faf 100644
@@ -1501,11 +1501,7 @@ static void iommu_flush_iotlb_psi(struct 
 intel_iommu *iommu,

      else
          __iommu_flush_iotlb_psi(iommu, did, pfn, pages, ih); -    /*
-     * In caching mode, changes of pages from non-present to

Quoted 8 lines

present require
-     * flush. However, device IOTLB doesn't need to be flushed in this case.
-     */
-    if (!cap_caching_mode(iommu->cap) || !map)
+    if (!map)
          iommu_flush_dev_iotlb(domain, addr, mask);
  }

Given devTLB flushing is irrelavent to CM, put iommu_flush_dev_iotlb() in iommu_flush_iotlb_psi() and called with CM checking context is not reasonable. the logic is buggy.

static void __mapping_notify_one(struct intel_iommu *iommu, struct dmar_domain *domain, unsigned long pfn, unsigned int pages)
{
     /*
      * It's a non-present to present mapping. Only flush if caching mode
      * and second level.
      */
     if (cap_caching_mode(iommu->cap) && !domain->use_first_level) iommu_flush_iotlb_psi(iommu, domain, pfn, pages, 0, 1);
     else
         iommu_flush_write_buffer(iommu);

Quoted 7 lines
Quoted 2 lines

then how about fold all CM checking logic in iommu_flush_iotlb_psi() or speperate iommu_flush_dev_iotlb() from iommu_flush_iotlb_psi() ?

I am refactoring the code with a new series.

https://lore.kernel.org/linux-iommu/[email protected]/

Great, thx.

Quoted 3 lines

Best regards,
baolu

#6 lore raw
In reply to Baolu Lu
Quoted 44 lines · Baolu Lu
Quoted 43 lines · Yi Liu
Quoted 41 lines · Baolu Lu
Quoted 30 lines · Yi Liu
Quoted 29 lines · Tina
Quoted 19 lines

-----Original Message-----
From: Lu Baolu <[email protected]>
Sent: Monday, April 15, 2024 9:39 AM
To: [email protected]
Cc: Tian, Kevin <[email protected]>; Liu, Yi L <[email protected]>; Jacob Pan <[email protected]>; Joerg Roedel <[email protected]>; Will Deacon <[email protected]>; Robin Murphy <[email protected]>; linux- [email protected]; Lu Baolu <[email protected]>
Subject: [PATCH v3 1/1] iommu/vt-d: Remove caching mode check before device TLB flush

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Quoted 7 lines

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

The current logic is if the caching mode is being used and a domain isn't using first level I/O page table, then flush-queue won't be used. Otherwise, the flush-queue can be enabled. See https://github.com/torvalds/linux/commit/257ec29074

Quoted 6 lines
Quoted 1 line

In other words, if the caching mode is being used and a domain is using first level I/O page table, the flush-queue can be used for this domain to flush iotlb. Could the code change in this patch bring any performance impact to this case?

This seems to have performance deduction in the nested translation case. The iommufd nested support bas been merged in 6.8, while the Qemu side is wip. So this performance deduction does not happen until Qemu is done. Should this also be considered as a performance regression? TBH.

Caching mode is irrelevant to first-stage and nesting translations. If the QEMU implementation still relies on caching mode for nesting support, it's already broken.

Quoted 6 lines
Quoted 4 lines

For first-stage, yes, caching mode is irrelevant. But for nesting, still
be relevant. Because it relies on caching mode to capture the mofifications
of the pasid entry. Without caching mode, vIOMMU device model within QEMU will miss the newly setup pasid entries, hence no way to setup nested translation at all.

Okay, it's reasonable that nesting translation still uses second-stage.

the key point is not nesting uses second-stage, it is because of the guest pasid entry capturing :)

Quoted 2 lines

But the removed checks in this patch don't impact the pasid table change capture, right?

yes. The removal in this patch is irrelevant to it.

So Tina's concern makes sense. But I don't know if it should be considered as a performance regression since nesting is not really fully ready (QEMU side is not there yet).

BTW. I'm not sure if you have considered to correct the other caching mode checks before dev-TLB invalidation. e.g. the CM check in drivers/iommu/intel/pasid.c.

Signature · 2 lines
Regards,
Yi Liu
author maintainer
#10 lore raw
In reply to Lu Baolu
Quoted 14 lines · Lu Baolu

The Caching Mode (CM) of the Intel IOMMU indicates if the hardware implementation caches not-present or erroneous translation-structure entries except for the first-stage translation. The caching mode is irrelevant to the device TLB, therefore there is no need to check it before a device TLB invalidation operation.

Remove two caching mode checks before device TLB invalidation in the driver. The removal of these checks doesn't change the driver's behavior in critical map/unmap paths. Hence, there is no functionality or performance impact, especially since commit <29b32839725f> ("iommu/vt-d:
Do not use flush-queue when caching-mode is on") has already disabled flush-queue for caching mode. Therefore, caching mode will never call intel_flush_iotlb_all().

Quoted 5 lines
---
  drivers/iommu/intel/iommu.c | 9 ++-------
  1 file changed, 2 insertions(+), 7 deletions(-)

Patch has been queued for iommu/vt-d.

Best regards,
baolu