Re: [PATCH 2/2] iommu/vt-d: Fix checks in pgtable_walk()

From: Baolu Lu
Date: Wed Oct 23 2024 - 03:41:53 EST


On 2024/10/23 13:16, Duan, Zhenzhong wrote:
-----Original Message-----
From: Baolu Lu<baolu.lu@xxxxxxxxxxxxxxx>
Sent: Wednesday, October 23, 2024 12:50 PM
Subject: Re: [PATCH 2/2] iommu/vt-d: Fix checks in pgtable_walk()

On 2024/10/22 17:50, Zhenzhong Duan wrote:
It's not accurate to dump super page as non-present page,
meanwhile bit7 in first level page table entry is PAT bit,
Can you please explain how the 'bit7 in first level page table entry' is
relevant to the changes made in this patch?
When iterate to level 1, it is leaf page table entry, bit 7 is PAT bit instead of PS bit.
dma_pte_superpage(pte) may return true, then " PTE not present at level 1" may
be printed out.

I see. Thank you!

If you have a new version, can you please make it a bit clearer? My
understanding is that dma_pte_superpage(pte) should not check against
the leaf page table entries, right?

Thanks,
baolu