Skip to contentSkip to conversation

AMD IOMMU updates for 2.6.28

GIT PULL Joerg Roedel 2 participants 11 messages Sep 19, 2008 Documentation

lore
Share & tools
lore
#0 lore raw

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code AMD IOMMU: implement lazy IO/TLB flushing
AMD IOMMU: add branch hints to completion wait checks AMD IOMMU: align alloc_coherent addresses properly
AMD IOMMU: add event buffer allocation
AMD IOMMU: save pci segment from ACPI tables
AMD IOMMU: save pci_dev instead of devid
AMD IOMMU: add MSI interrupt support
AMD IOMMU: add event handling code
AMD IOMMU: enable event logging
AMD IOMMU: allow IO page faults from devices
AMD IOMMU: add dma_supported callback
AMD IOMMU: don't assign preallocated protection domains to devices AMD IOMMU: some set_device_domain cleanups
AMD IOMMU: replace memset with __GFP_ZERO in alloc_coherent AMD IOMMU: simplify dma_mask_to_pages
AMD IOMMU: free domain bitmap with its allocation order AMD IOMMU: remove unnecessary cast to u64 in the init code AMD IOMMU: calculate IVHD size with a function
AMD IOMMU: use cmd_buf_size when freeing the command buffer add AMD IOMMU tree to MAINTAINERS file
AMD IOMMU: use coherent_dma_mask in alloc_coherent

 Documentation/kernel-parameters.txt       |    4 +
 Documentation/x86/x86_64/boot-options.txt |    2 -
 MAINTAINERS                               |    1 +
 arch/x86/Kconfig                          |    1 +
 arch/x86/kernel/amd_iommu.c               |  318 ++++++++++++++++++++++++-----
 arch/x86/kernel/amd_iommu_init.c          |  191 ++++++++++++++++--
 arch/x86/kernel/pci-dma.c                 |   13 ++
 arch/x86/kernel/pci-gart_64.c             |   13 --
 include/asm-x86/amd_iommu.h               |    3 +
 include/asm-x86/amd_iommu_types.h         |   58 +++++-
 include/asm-x86/iommu.h                   |    1 +
 11 files changed, 517 insertions(+), 88 deletions(-)
Signature · 6 lines
           |           AMD Saxony Limited Liability Company & Co. KG
 Operating |         Wilschdorfer Landstr. 101, 01109 Dresden, Germany
 System    |                  Register Court Dresden: HRA 4896
 Research  |              General Partner authorized to represent:
 Center    |             AMD Saxony LLC (Wilmington, Delaware, US)
           | General Manager of AMD Saxony LLC: Dr. Hans-R. Deppe, Thomas McCoy
#1 lore raw
Quoted 15 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

Quoted 7 lines

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

As I wrote, I don't think that this GART patch is the right approach.

#2 lore raw
Quoted 21 lines · FUJITA Tomonori
Quoted 18 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Quoted 6 lines
Quoted 4 lines

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

As I wrote, I don't think that this GART patch is the right approach.

I don't thinks its clean to leave nofullflush in GART code and move fullflush to generic code only. If we move something then both. But feel free to submit a patch that removes the nofullflush option. I would not disagree with that.

Joerg

Signature · 6 lines
           |           AMD Saxony Limited Liability Company & Co. KG
 Operating |         Wilschdorfer Landstr. 101, 01109 Dresden, Germany
 System    |                  Register Court Dresden: HRA 4896
 Research  |              General Partner authorized to represent:
 Center    |             AMD Saxony LLC (Wilmington, Delaware, US)
           | General Manager of AMD Saxony LLC: Dr. Hans-R. Deppe, Thomas McCoy
#3 lore raw
Quoted 27 lines · On Fri, 19 Sep 2008 12:39:23 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 26 lines · FUJITA Tomonori
Quoted 22 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

Quoted 4 lines
Quoted 1 line

As I wrote, I don't think that this GART patch is the right approach.

I don't thinks its clean to leave nofullflush in GART code and move fullflush to generic code only.

It's not clean for GART but clean for other IOMMUs.

Quoted 3 lines

If we move something then both. But feel
free to submit a patch that removes the nofullflush option. I would not disagree with that.

I'm not sure we can remove an option that we added. If we can, please remove it now. There is no point to make the meaningless option generic.

If you don't like it, please add 'nofullflush' option as AMD IOMMU option for now. As long as AMD IOMMU uses the same option name, we can clean up them later if necessary.

#4 lore raw
Quoted 30 lines · FUJITA Tomonori
Quoted 27 lines · On Fri, 19 Sep 2008 12:39:23 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 26 lines · FUJITA Tomonori
Quoted 22 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

As I wrote, I don't think that this GART patch is the right approach.

I don't thinks its clean to leave nofullflush in GART code and move fullflush to generic code only.

It's not clean for GART but clean for other IOMMUs.

So its unclean anyway. It doesn't matter if for GART or for AMD IOMMU.

Quoted 7 lines
Quoted 3 lines

If we move something then both. But feel
free to submit a patch that removes the nofullflush option. I would not disagree with that.

I'm not sure we can remove an option that we added. If we can, please remove it now. There is no point to make the meaningless option generic.

I don't added this option. I just moved it to generic code from the GART. The nofullflush option has no meaning for GART and for AMD IOMMU. The only reason to keep it is if we want to change the default in the future. If you don't think we would change the default any time in the future feel free to remove it.

Joerg

Signature · 6 lines
           |           AMD Saxony Limited Liability Company & Co. KG
 Operating |         Wilschdorfer Landstr. 101, 01109 Dresden, Germany
 System    |                  Register Court Dresden: HRA 4896
 Research  |              General Partner authorized to represent:
 Center    |             AMD Saxony LLC (Wilmington, Delaware, US)
           | General Manager of AMD Saxony LLC: Dr. Hans-R. Deppe, Thomas McCoy
#5 lore raw
Quoted 44 lines · On Fri, 19 Sep 2008 13:02:40 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 30 lines · FUJITA Tomonori
Quoted 27 lines · On Fri, 19 Sep 2008 12:39:23 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 26 lines · FUJITA Tomonori
Quoted 22 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

As I wrote, I don't think that this GART patch is the right approach.

I don't thinks its clean to leave nofullflush in GART code and move fullflush to generic code only.

It's not clean for GART but clean for other IOMMUs.

So its unclean anyway. It doesn't matter if for GART or for AMD IOMMU.

Quoted 4 lines
Quoted 3 lines

If we move something then both. But feel
free to submit a patch that removes the nofullflush option. I would not disagree with that.

Quoted 5 lines
Quoted 3 lines

I'm not sure we can remove an option that we added. If we can, please remove it now. There is no point to make the meaningless option generic.

I don't added this option. I just moved it to generic code from the

You added it to the generic place. Moving it to the generic place means adding it to the generic place.

There is no difference in the end. Now 'nofullflush' is listed as the generic place.

Documentation/kernel-parameters.txt +4 -0
index c2e00ee..569527e 100644
@@ -888,6 +888,10 @@ and is between 256 and 4096 characters. It is defined in the file
 		nomerge
 		forcesac
 		soft
+		fullflush
+			Flush IO/TLB at every deallocation
+		nofullflush
+			Flush IO/TLB only when addresses are reused (default)
 
 

Please don't do that.

Quoted 3 lines

GART. The nofullflush option has no meaning for GART and for AMD IOMMU. The only reason to keep it is if we want to change the default in the future.

No, we can't change the default behavior. It's the user interface.

Quoted 2 lines

If you don't think we would change the default any time in the future feel free to remove it.

#6 lore raw
Quoted 64 lines · On Fri, 19 Sep 2008 20:23:50 +0900 FUJITA Tomonori…
Quoted 44 lines · On Fri, 19 Sep 2008 13:02:40 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 30 lines · FUJITA Tomonori
Quoted 27 lines · On Fri, 19 Sep 2008 12:39:23 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 26 lines · FUJITA Tomonori
Quoted 22 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

As I wrote, I don't think that this GART patch is the right approach.

I don't thinks its clean to leave nofullflush in GART code and move fullflush to generic code only.

It's not clean for GART but clean for other IOMMUs.

So its unclean anyway. It doesn't matter if for GART or for AMD IOMMU.

Quoted 4 lines
Quoted 3 lines

If we move something then both. But feel
free to submit a patch that removes the nofullflush option. I would not disagree with that.

Quoted 5 lines
Quoted 3 lines

I'm not sure we can remove an option that we added. If we can, please remove it now. There is no point to make the meaningless option generic.

I don't added this option. I just moved it to generic code from the

You added it to the generic place. Moving it to the generic place means adding it to the generic place.

There is no difference in the end. Now 'nofullflush' is listed as the generic place.

Documentation/kernel-parameters.txt +0 -0
index c2e00ee..569527e 100644

Documentation/kernel-parameters.txt and is between 256 and 4096 characters. It is defined in the file · line 894

Quoted 7 lines
Documentation/kernel-parameters.txt +4 -0
@@ -888,6 +888,10 @@ and is between 256 and 4096 characters. It is defined in the file
 		nomerge
 		forcesac
 		soft
+		fullflush
+			Flush IO/TLB at every deallocation
+		nofullflush
+			Flush IO/TLB only when addresses are reused (default)

And you don't need to add 'fullflush' to the generic place too.

'fullflush' will be supported with only GART and AMD IOMMU. So adding the description of it to both GART and AMD IOMMU should be fine.

'fullflush' has the same meaning for both IOMMUs. That's nice consistency, I think.

#7 lore raw
Quoted 77 lines · FUJITA Tomonori
Quoted 64 lines · On Fri, 19 Sep 2008 20:23:50 +0900 FUJITA Tomonori…
Quoted 44 lines · On Fri, 19 Sep 2008 13:02:40 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 30 lines · FUJITA Tomonori
Quoted 27 lines · On Fri, 19 Sep 2008 12:39:23 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 26 lines · FUJITA Tomonori
Quoted 22 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

As I wrote, I don't think that this GART patch is the right approach.

I don't thinks its clean to leave nofullflush in GART code and move fullflush to generic code only.

It's not clean for GART but clean for other IOMMUs.

So its unclean anyway. It doesn't matter if for GART or for AMD IOMMU.

Quoted 4 lines
Quoted 3 lines

If we move something then both. But feel
free to submit a patch that removes the nofullflush option. I would not disagree with that.

Quoted 5 lines
Quoted 3 lines

I'm not sure we can remove an option that we added. If we can, please remove it now. There is no point to make the meaningless option generic.

I don't added this option. I just moved it to generic code from the

You added it to the generic place. Moving it to the generic place means adding it to the generic place.

There is no difference in the end. Now 'nofullflush' is listed as the generic place.

Documentation/kernel-parameters.txt +0 -0
index c2e00ee..569527e 100644

Documentation/kernel-parameters.txt and is between 256 and 4096 characters. It is defined in the file · line 894

Quoted 7 lines
Documentation/kernel-parameters.txt +4 -0
@@ -888,6 +888,10 @@ and is between 256 and 4096 characters. It is defined in the file
 		nomerge
 		forcesac
 		soft
+		fullflush
+			Flush IO/TLB at every deallocation
+		nofullflush
+			Flush IO/TLB only when addresses are reused (default)

And you don't need to add 'fullflush' to the generic place too.

Quoted 5 lines

'fullflush' will be supported with only GART and AMD IOMMU. So adding the description of it to both GART and AMD IOMMU should be fine.

'fullflush' has the same meaning for both IOMMUs. That's nice consistency, I think.

Huh? The whole point of this patch was to have a common option between IOMMUs to disable lazy IOTLB flushing. This was suggested by _you_ and the only reason I wrote this patch.
After this patch we can change other IOMMU implementations with lazy flushing to use that parameter too.

Joerg

Signature · 6 lines
           |           AMD Saxony Limited Liability Company & Co. KG
 Operating |         Wilschdorfer Landstr. 101, 01109 Dresden, Germany
 System    |                  Register Court Dresden: HRA 4896
 Research  |              General Partner authorized to represent:
 Center    |             AMD Saxony LLC (Wilmington, Delaware, US)
           | General Manager of AMD Saxony LLC: Dr. Hans-R. Deppe, Thomas McCoy
#8 lore raw
Quoted 81 lines · On Fri, 19 Sep 2008 13:52:59 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 80 lines · FUJITA Tomonori
Quoted 64 lines · On Fri, 19 Sep 2008 20:23:50 +0900 FUJITA Tomonori…
Quoted 44 lines · On Fri, 19 Sep 2008 13:02:40 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 30 lines · FUJITA Tomonori
Quoted 27 lines · On Fri, 19 Sep 2008 12:39:23 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 26 lines · FUJITA Tomonori
Quoted 22 lines · On Fri, 19 Sep 2008 12:07:11 +0200 Joerg Roedel <[email protected]> wrote:

Hi Ingo,

please pull the updates for the AMD IOMMU driver for the 2.6.28 merge window. Most of the patches had been sent to LKML for review and got comments. One patch touches the GART driver and was added after an objection with a patch previously in this series. Please pull.

The following changes since commit 74546a8cd9a4e2c26a15a534f0be2f9cc08ae675:
Ingo Molnar (1):
Revert "fix warning in: "x86: HPET_MSI Initialise per-cpu HPET timers""

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git iommu-updates-2.6.28

FUJITA Tomonori (1):
AMD IOMMU: avoid unnecessary low zone allocation in alloc_coherent

Joerg Roedel (24):
AMD IOMMU: check for invalid device pointers
AMD IOMMU: move TLB flushing to the map/unmap helper functions
x86: move GART TLB flushing options to generic code

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

As I wrote, I don't think that this GART patch is the right approach.

I don't thinks its clean to leave nofullflush in GART code and move fullflush to generic code only.

It's not clean for GART but clean for other IOMMUs.

So its unclean anyway. It doesn't matter if for GART or for AMD IOMMU.

Quoted 4 lines
Quoted 3 lines

If we move something then both. But feel
free to submit a patch that removes the nofullflush option. I would not disagree with that.

Quoted 5 lines
Quoted 3 lines

I'm not sure we can remove an option that we added. If we can, please remove it now. There is no point to make the meaningless option generic.

I don't added this option. I just moved it to generic code from the

You added it to the generic place. Moving it to the generic place means adding it to the generic place.

There is no difference in the end. Now 'nofullflush' is listed as the generic place.

Documentation/kernel-parameters.txt +0 -0
index c2e00ee..569527e 100644

Documentation/kernel-parameters.txt and is between 256 and 4096 characters. It is defined in the file · line 894

Quoted 7 lines
Documentation/kernel-parameters.txt +4 -0
@@ -888,6 +888,10 @@ and is between 256 and 4096 characters. It is defined in the file
 		nomerge
 		forcesac
 		soft
+		fullflush
+			Flush IO/TLB at every deallocation
+		nofullflush
+			Flush IO/TLB only when addresses are reused (default)

And you don't need to add 'fullflush' to the generic place too.

'fullflush' will be supported with only GART and AMD IOMMU. So adding the description of it to both GART and AMD IOMMU should be fine.

Quoted 6 lines
Quoted 2 lines

'fullflush' has the same meaning for both IOMMUs. That's nice consistency, I think.

Huh? The whole point of this patch was to have a common option between IOMMUs to disable lazy IOTLB flushing. This was suggested by _you_ and the only reason I wrote this patch.

You misunderstand what I meant. I'm sorry if my explanation is not clear.

Quoted 2 lines

After this patch we can change other IOMMU implementations with lazy flushing to use that parameter too.

I'm not sure that Calgary wants to support such option. It always uses lazy flushing.

What I don't like is that there is no consistency about the option name for lazy flushing. It doesn't mean that we move the option to the generic place.

Here's my first reply:

=
Would it be nice to have consistency of IOMMU parameters?

VT-d also has the kernel-boot option for this lazy flushing trick though VT-d 'strict' option is more vague than 'unmap_flush' =

What I meant that using the option name 'strict' that VT-d uses for lazy flushing for AMD IOMMU would be better than introducing a new option name, "unmap_flush" for AMD IOMMU though I don't think that 'strict' is the good name.

Seems 'fullflush' is better than 'strict'. So I think that it's better to use 'fullflush' for AMD IMMU rather introducing a new name, 'unmap_flush'. But again, it doesn't mean that 'fullflush' moves to the generic place.

#9 lore raw
Quoted 49 lines · FUJITA Tomonori
Quoted 9 lines · On Fri, 19 Sep 2008 13:52:59 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 8 lines · FUJITA Tomonori

On Fri, 19 Sep 2008 20:23:50 +0900
FUJITA Tomonori <[email protected]> wrote:

And you don't need to add 'fullflush' to the generic place too.

'fullflush' will be supported with only GART and AMD IOMMU. So adding the description of it to both GART and AMD IOMMU should be fine.

Quoted 6 lines
Quoted 2 lines

'fullflush' has the same meaning for both IOMMUs. That's nice consistency, I think.

Huh? The whole point of this patch was to have a common option between IOMMUs to disable lazy IOTLB flushing. This was suggested by _you_ and the only reason I wrote this patch.

You misunderstand what I meant. I'm sorry if my explanation is not clear.

Quoted 2 lines

After this patch we can change other IOMMU implementations with lazy flushing to use that parameter too.

I'm not sure that Calgary wants to support such option. It always uses lazy flushing.

What I don't like is that there is no consistency about the option name for lazy flushing. It doesn't mean that we move the option to the generic place.

Here's my first reply:

=
Would it be nice to have consistency of IOMMU parameters?

VT-d also has the kernel-boot option for this lazy flushing trick though VT-d 'strict' option is more vague than 'unmap_flush' =

What I meant that using the option name 'strict' that VT-d uses for lazy flushing for AMD IOMMU would be better than introducing a new option name, "unmap_flush" for AMD IOMMU though I don't think that 'strict' is the good name.

Quoted 4 lines

Seems 'fullflush' is better than 'strict'. So I think that it's better to use 'fullflush' for AMD IMMU rather introducing a new name, 'unmap_flush'. But again, it doesn't mean that 'fullflush' moves to the generic place.

Ok, so now we have fullflush, which makes sense for all x86 hardware IOMMUs except maybe Calgary (from what I know about Calgary fullflush can be implemented there only with a abyssal performance penalty nobody is willing to pay). So it makes sense to have the option in the generic place. But we can surely add a comment that it does not affect Calgary to the kernel documentation (and change VT-d to use that parameter too). But the AMD IOMMU update chain is not the right place for making big changes to other IOMMUs.

Joerg

Signature · 6 lines
           |           AMD Saxony Limited Liability Company & Co. KG
 Operating |         Wilschdorfer Landstr. 101, 01109 Dresden, Germany
 System    |                  Register Court Dresden: HRA 4896
 Research  |              General Partner authorized to represent:
 Center    |             AMD Saxony LLC (Wilmington, Delaware, US)
           | General Manager of AMD Saxony LLC: Dr. Hans-R. Deppe, Thomas McCoy
#10 lore raw
Quoted 55 lines · On Fri, 19 Sep 2008 14:34:26 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 49 lines · FUJITA Tomonori
Quoted 9 lines · On Fri, 19 Sep 2008 13:52:59 +0200 Joerg Roedel <[email protected]> wrote:
Quoted 8 lines · FUJITA Tomonori

On Fri, 19 Sep 2008 20:23:50 +0900
FUJITA Tomonori <[email protected]> wrote:

And you don't need to add 'fullflush' to the generic place too.

'fullflush' will be supported with only GART and AMD IOMMU. So adding the description of it to both GART and AMD IOMMU should be fine.

Quoted 6 lines
Quoted 2 lines

'fullflush' has the same meaning for both IOMMUs. That's nice consistency, I think.

Huh? The whole point of this patch was to have a common option between IOMMUs to disable lazy IOTLB flushing. This was suggested by _you_ and the only reason I wrote this patch.

You misunderstand what I meant. I'm sorry if my explanation is not clear.

Quoted 2 lines

After this patch we can change other IOMMU implementations with lazy flushing to use that parameter too.

I'm not sure that Calgary wants to support such option. It always uses lazy flushing.

What I don't like is that there is no consistency about the option name for lazy flushing. It doesn't mean that we move the option to the generic place.

Here's my first reply:

=
Would it be nice to have consistency of IOMMU parameters?

VT-d also has the kernel-boot option for this lazy flushing trick though VT-d 'strict' option is more vague than 'unmap_flush' =

What I meant that using the option name 'strict' that VT-d uses for lazy flushing for AMD IOMMU would be better than introducing a new option name, "unmap_flush" for AMD IOMMU though I don't think that 'strict' is the good name.

Quoted 4 lines

Seems 'fullflush' is better than 'strict'. So I think that it's better to use 'fullflush' for AMD IMMU rather introducing a new name, 'unmap_flush'. But again, it doesn't mean that 'fullflush' moves to the generic place.

Quoted 5 lines

Ok, so now we have fullflush, which makes sense for all x86 hardware IOMMUs except maybe Calgary (from what I know about Calgary fullflush can be implemented there only with a abyssal performance penalty nobody is willing to pay). So it makes sense to have the option in the generic place.

Not sure. It would make sense but maybe not.

But it's the different topic. Making a generic option that affects all the IOMMUs is an important issue for everyone. All the parties need to discuss it and agree.

Quoted 4 lines

But we can surely add a comment that it does not affect Calgary to the kernel documentation (and change VT-d to use that parameter too). But the AMD IOMMU update chain is not the right place for making big changes to other IOMMUs.

Yes, that's the point. We should not have such important change in AMD IOMMU updates.

So please just make 'fullflush' as AMD IOMMU's option.