Re: [PATCH v3 2/3] selftests/mm: test userspace MFR for HugeTLB hugepage

[William Roche]

On 2/3/26 20:23, Jiaqi Yan wrote:
> Test the userspace memory failure recovery (MFR) policy for HugeTLB:
>
> 1. Create a memfd backed by HugeTLB and had MFD_MF_KEEP_UE_MAPPED set.
>
> 2. Allocate and map 4 hugepages to the process.
>
> 3. Create sub-threads to MADV_HWPOISON inner addresses of the 1st hugepage.
>
> 4. Check if the process gets correct SIGBUS for each poisoned raw page.
>
> 5. Check if all memory are still accessible and content valid.
>
> 6. Check if the poisoned hugepage is dealt with after memfd released.
>
> Two configurables in the test:
>
> - hugepage_size: size of the hugepage, 1G or 2M.
>
> - nr_hwp_pages: number of pages within the 1st hugepage to MADV_HWPOISON.
In this version, you are introducing this new test argument
"nr_hwp_pages" to indicate how many of the pre-defined offsets we want
to poison inside the hugepage (between 1 and 8).
But is there any advantage to give the choice to the user instead of
testing them all ?

As a suggestion, should we have this test program setting or verifying
the minimal number of hugepages of the right type, instead of relying on
the user to set them manually ?
And at the end, should we try to unpoison the impacted pages ? So that
the lab machine where the tests run can continue to use all its memory ?

Thanks for your feedback,
William.