Re: [PATCH v14 6/6] iommu: Remove mode argument from iommu_set_dma_strict()

From: Robin Murphy
Date: Tue Jun 22 2021 - 18:25:24 EST

On 2021-06-21 15:32, Lu Baolu wrote:
Hi Robin,

On 2021/6/21 19:59, Robin Murphy wrote:
On 2021-06-21 11:34, John Garry wrote:
On 21/06/2021 11:00, Lu Baolu wrote:
void iommu_set_dma_strict(bool force)
          if (force == true)
         iommu_dma_strict = true;
     else if (!(iommu_cmd_line & IOMMU_CMD_LINE_STRICT))
         iommu_dma_strict = true;

So we would use iommu_set_dma_strict(true) for a) and b), but iommu_set_dma_strict(false) for c).

Yes. We need to distinguish the "must" and "nice-to-have" cases of
setting strict mode.

Then I am not sure what you want to do with the accompanying print for c). It was:
"IOMMU batching is disabled due to virtualization"

And now is from this series:
"IOMMU batching disallowed due to virtualization"

Using iommu_get_dma_strict(domain) is not appropriate here to know the current mode (so we know whether to print).

Note that this change would mean that the current series would require non-trivial rework, which would be unfortunate so late in the cycle.

This patch series looks good to me and I have added by reviewed-by.
Probably we could make another patch series to improve it so that the
kernel optimization should not override the user setting.

On a personal level I would be happy with that approach, but I think it's better to not start changing things right away in a follow-up series.

So how about we add this patch (which replaces 6/6 "iommu: Remove mode argument from iommu_set_dma_strict()")?

Robin, any opinion?

For me it boils down to whether there are any realistic workloads where non-strict mode *would* still perform better under virtualisation. The

At present, we see that strict mode has better performance in the
virtualization environment because it will make the shadow page table
management more efficient. When the hardware supports nested
translation, we may have to re-evaluate this since there's no need for
a shadowing page table anymore.

I guess I was assuming that in most cases, proper nested mode could look distinct enough that we'd be able to treat it differently in the first place. For instance, if it's handing guest tables directly to the hardware, would the host have any reason to still set the "caching mode" ID bit?