mbox series

[0/4,PULL,REQUEST] iommu/vt-d: Fixes for v6.2-rc8

Message ID 20230216130816.151824-1-baolu.lu@linux.intel.com
Headers
Series iommu/vt-d: Fixes for v6.2-rc8 |

Message

Baolu Lu Feb. 16, 2023, 1:08 p.m. UTC 
  Hi Joerg,

Below iommu/vt-d fixes are queued for your fixes branch.

- Two performance optimizations
- Fix PASID directory pointer coherency
- Fix missed rollbacks in error path

Please consider it for the iommu/fixes branch.

Best regards,
Lu Baolu

Jacob Pan (2):
  iommu/vt-d: Avoid superfluous IOTLB tracking in lazy mode
  iommu/vt-d: Fix PASID directory pointer coherency

Lu Baolu (1):
  iommu/vt-d: Fix error handling in sva enable/disable paths

Tina Zhang (1):
  iommu/vt-d: Allow to use flush-queue when first level is default

 drivers/iommu/intel/iommu.c | 26 ++++++++++++++++++++------
 drivers/iommu/intel/pasid.c |  7 +++++++
 2 files changed, 27 insertions(+), 6 deletions(-)

Comments

Joerg Roedel Feb. 16, 2023, 1:42 p.m. UTC | #1 
Hi Baolu,

On Thu, Feb 16, 2023 at 09:08:12PM +0800, Lu Baolu wrote:
> Below iommu/vt-d fixes are queued for your fixes branch.
> 
> - Two performance optimizations
> - Fix PASID directory pointer coherency
> - Fix missed rollbacks in error path
> 
> Please consider it for the iommu/fixes branch.

So nothing of this seems really critical (e.g. fixes a regression that a
number of people are encountering). Especially the performance
optimizations do not qualify as fixes at this stage of the cycle. I will
queue them in the VT-d branch so that they go upstream in the next merge
window, unless you convince me otherwise.

Regards,

	Joerg
Baolu Lu Feb. 16, 2023, 1:45 p.m. UTC | #2 
On 2023/2/16 21:42, Joerg Roedel wrote:
> Hi Baolu,
> 
> On Thu, Feb 16, 2023 at 09:08:12PM +0800, Lu Baolu wrote:
>> Below iommu/vt-d fixes are queued for your fixes branch.
>>
>> - Two performance optimizations
>> - Fix PASID directory pointer coherency
>> - Fix missed rollbacks in error path
>>
>> Please consider it for the iommu/fixes branch.
> So nothing of this seems really critical (e.g. fixes a regression that a
> number of people are encountering). Especially the performance
> optimizations do not qualify as fixes at this stage of the cycle. I will
> queue them in the VT-d branch so that they go upstream in the next merge
> window, unless you convince me otherwise.

Yes. Nothing really critical. It's fine to put them in the vt-d branch.

Best regards,
baolu