Re: [PATCH 1/2] mm/vmalloc: Add interfaces to free unused page table

[Kani, Toshi]

On Wed, 2018-03-07 at 14:54 -0800, Andrew Morton wrote:
> On Wed,  7 Mar 2018 11:32:26 -0700 Toshi Kani <toshi.kani@xxxxxxx> wrote:
> 
> > On architectures with CONFIG_HAVE_ARCH_HUGE_VMAP set, ioremap()
> > may create pud/pmd mappings.  Kernel panic was observed on arm64
> > systems with Cortex-A75 in the following steps as described by
> > Hanjun Guo.
> > 
> > 1. ioremap a 4K size, valid page table will build,
> > 2. iounmap it, pte0 will set to 0;
> > 3. ioremap the same address with 2M size, pgd/pmd is unchanged,
> >    then set the a new value for pmd;
> > 4. pte0 is leaked;
> > 5. CPU may meet exception because the old pmd is still in TLB,
> >    which will lead to kernel panic.
> > 
> > This panic is not reproducible on x86.  INVLPG, called from iounmap,
> > purges all levels of entries associated with purged address on x86.
> > x86 still has memory leak.
> > 
> > Add two interfaces, pud_free_pmd_page() and pmd_free_pte_page(),
> > which clear a given pud/pmd entry and free up a page for the lower
> > level entries.
> > 
> > This patch implements their stub functions on x86 and arm64, which
> > work as workaround.
> > 
> > index 004abf9ebf12..942f4fa341f1 100644
> > --- a/arch/x86/mm/pgtable.c
> > +++ b/arch/x86/mm/pgtable.c
> > @@ -702,4 +702,24 @@ int pmd_clear_huge(pmd_t *pmd)
> >  
> >  	return 0;
> >  }
> > +
> > +/**
> > + * pud_free_pmd_page - clear pud entry and free pmd page
> > + *
> > + * Returns 1 on success and 0 on failure (pud not cleared).
> > + */
> > +int pud_free_pmd_page(pud_t *pud)
> > +{
> > +	return pud_none(*pud);
> > +}
> > +
> > +/**
> > + * pmd_free_pte_page - clear pmd entry and free pte page
> > + *
> > + * Returns 1 on success and 0 on failure (pmd not cleared).
> > + */
> > +int pmd_free_pte_page(pmd_t *pmd)
> > +{
> > +	return pmd_none(*pmd);
> > +}
> 
> Are these functions well named?  I mean, the comment says "clear pud
> entry and free pmd page" but the implementatin does neither of those
> things.  The name implies that the function frees a pmd_page but the
> callsites use the function as a way of querying state.

This patch 1/2 only implements stubs.  Patch 2/2 implements what is
described here.

> Also, as this fixes an arm64 kernel panic, should we be using
> cc:stable?

Right.

Thanks,
-Toshi