Re: [RFC PATCH v2 4/5] mm: secretmem: use PMD-size pages to amortize direct map fragmentation

[Mike Rapoport <rppt@kernel.org>]

On Mon, Jul 13, 2020 at 02:05:05PM +0300, Kirill A. Shutemov wrote:
> On Mon, Jul 06, 2020 at 08:20:50PM +0300, Mike Rapoport wrote:
> > From: Mike Rapoport <rppt@linux.ibm.com>
> > 
> > Removing a PAGE_SIZE page from the direct map every time such page is
> > allocated for a secret memory mapping will cause severe fragmentation of
> > the direct map. This fragmentation can be reduced by using PMD-size pages
> > as a pool for small pages for secret memory mappings.
> > 
> > Add a gen_pool per secretmem inode and lazily populate this pool with
> > PMD-size pages.
> > 
> > Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
> > ---
> >  mm/secretmem.c | 107 ++++++++++++++++++++++++++++++++++++++++---------
> >  1 file changed, 88 insertions(+), 19 deletions(-)
> > 
> > diff --git a/mm/secretmem.c b/mm/secretmem.c
> > index df8f8c958cc2..c6fcf6d76951 100644
> > --- a/mm/secretmem.c
> > +++ b/mm/secretmem.c
> > @@ -5,6 +5,7 @@
> >  #include <linux/memfd.h>
> >  #include <linux/printk.h>
> >  #include <linux/pagemap.h>
> > +#include <linux/genalloc.h>
> >  #include <linux/pseudo_fs.h>
> >  #include <linux/set_memory.h>
> >  #include <linux/sched/signal.h>
> > @@ -23,24 +24,66 @@
> >  #define SECRETMEM_UNCACHED	0x2
> >  
> >  struct secretmem_ctx {
> > +	struct gen_pool *pool;
> >  	unsigned int mode;
> >  };
> >  
> > -static struct page *secretmem_alloc_page(gfp_t gfp)
> > +static int secretmem_pool_increase(struct secretmem_ctx *ctx, gfp_t gfp)
> >  {
> > -	/*
> > -	 * FIXME: use a cache of large pages to reduce the direct map
> > -	 * fragmentation
> > -	 */
> > -	return alloc_page(gfp);
> > +	unsigned long nr_pages = (1 << HPAGE_PMD_ORDER);
> > +	struct gen_pool *pool = ctx->pool;
> > +	unsigned long addr;
> > +	struct page *page;
> > +	int err;
> > +
> > +	page = alloc_pages(gfp, HPAGE_PMD_ORDER);
> > +	if (!page)
> > +		return -ENOMEM;
> > +
> > +	addr = (unsigned long)page_address(page);
> > +	split_page(page, HPAGE_PMD_ORDER);
> > +
> > +	err = gen_pool_add(pool, addr, HPAGE_PMD_SIZE, NUMA_NO_NODE);
> > +	if (err) {
> > +		__free_pages(page, HPAGE_PMD_ORDER);
> > +		return err;
> > +	}
> > +
> > +	__kernel_map_pages(page, nr_pages, 0);
> 
> It's worth nothing that unlike flush_tlb_kernel_range(),
> __kernel_map_pages() only flushed local TLB, so other CPU may still have
> access to the page. It's shouldn't be a blocker, but deserve a comment.

Sure.
 
> > +
> > +	return 0;
> > +}
> > +

-- 
Sincerely yours,
Mike.