On Tue, May 05, 2026 at 11:35:06AM +1200, Tao Liu wrote:
Hi Huang,
Lianbo may be unavailable for the time being, and sorry for my late
response. I have rethought your patchset. Please see my comments:
For this patch, I still think the THIS_KERNEL_VERSION() macro isn't a
good choice, and I have made my point in the previous discussion.
To illustrate my approach:
Linux source $ git log -L :folio_order:include/linux/mm.h
...
mm: free up a word in the first tail page
mm: reimplement folio_order() and folio_nr_pages()
mm: Add folio_test_pmd_mappable()
Each kernel commit represents an update to folio_order() that you'd
like to backport to crash. And each one belongs to a parent patchset,
e.g the "mm: Add folio_test_pmd_mappable()" one belongs to patchset
[1], so a struct/enum modification within the patchset can be choose
as an indicator, e.g. struct folio_batch is introduced within this
patchset, this might be a good candidate. BTW, if you do this, please
attach the patchset link with the indicator, so in the future people
will know where the indicator was chosen from, to improve
maintainability.
Okay, no problem.
I can follow this method.
Thanks
Huang Shijie