On Thu, 27 Feb 2025 14:48:17 +1100 Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
> Hi all,
>
> On Thu, 27 Feb 2025 14:44:10 +1100 Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
> >
> > The following commit is also in the mm tree as a different commit (but
> > the same patch):
> >
> > a37259732a7d ("x86/mm: Make MMU_GATHER_RCU_TABLE_FREE unconditional")
> >
> > This is commit
> >
> > a30104ede395 ("x86/mm: make MMU_GATHER_RCU_TABLE_FREE unconditional")
> >
> > in the mm-unstable branch of the mm tree.
> >
> > This is already causing a conflct in arch/x86/kernel/paravirt.c due commit
> >
> > f2c5c2105827 ("x86/mm: Remove pv_ops.mmu.tlb_remove_table call")
> >
> > in the tip tree (where I just used the tip tree version).
yes, I duplicated that match in mm.git so it can carry the series
"remove tlb_remove_page_ptdesc()".
> And another in arch/x86/mm/pgtable.c due to commit
>
> 530c12f84d2c ("x86: pgtable: convert to use tlb_remove_ptdesc()")
>
> in the tip tree (where I again just used the tip tree version).
Really, I'd mildly prefer that subsystem maintainers not cherrypick
patches from the middle of a series. An acked-by would be preferred.
I can understand the desire to test a patch within the subsystem's
tree, but that can (should?) be done by testing linux-next overall.
Whatever. I'll retain "x86: pgtable: convert to use
tlb_remove_ptdesc()" and shall figure it out within the merge window.
