Re: [PATCH] mm: Speed up mremap on large regions

[Joel Fernandes]

On Thu, Oct 11, 2018 at 11:11:11AM +0300, Kirill A. Shutemov wrote:
> On Thu, Oct 11, 2018 at 08:14:19AM +0300, Kirill A. Shutemov wrote:
> > On Wed, Oct 10, 2018 at 05:46:18PM -0700, Joel Fernandes wrote:
> > > diff --git a/arch/powerpc/include/asm/book3s/64/pgalloc.h b/arch/powerpc/include/asm/book3s/64/pgalloc.h
> > > index 391ed2c3b697..8a33f2044923 100644
> > > --- a/arch/powerpc/include/asm/book3s/64/pgalloc.h
> > > +++ b/arch/powerpc/include/asm/book3s/64/pgalloc.h
> > > @@ -192,14 +192,12 @@ static inline pgtable_t pmd_pgtable(pmd_t pmd)
> > >  	return (pgtable_t)pmd_page_vaddr(pmd);
> > >  }
> > >  
> > > -static inline pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
> > > -					  unsigned long address)
> > > +static inline pte_t *pte_alloc_one_kernel(struct mm_struct *mm)
> > >  {
> > >  	return (pte_t *)pte_fragment_alloc(mm, address, 1);
> > >  }
> > 
> > This is obviously broken.

I was actually aware of this, I was in the early stages of writing the script
but just shared the diff to give an idea the number of changes.

> I've checked pte_fragment_alloc() and it doesn't use the address too.
> We need to modify it too.

I rewrote the Coccinelle script and manually fixed pte_fragment_alloc in
that. I sent an update with you and Michal on CC, please check it. Thanks a
lot.

 - Joel