Re: [patch v2 4/4] mm, mempool: poison elements backed by page allocator

From: Andrey Ryabinin <a.ryabinin@samsung.com>
To: David Rientjes <rientjes@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Dave Kleikamp <shaggy@kernel.org>, Christoph Hellwig <hch@lst.de>,
	Sebastian Ott <sebott@linux.vnet.ibm.com>,
	Mikulas Patocka <mpatocka@redhat.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	LKML <linux-kernel@vger.kernel.org>,
	"linux-mm@kvack.org" <linux-mm@kvack.org>,
	jfs-discussion@lists.sourceforge.net
Subject: Re: [patch v2 4/4] mm, mempool: poison elements backed by page allocator
Date: Tue, 31 Mar 2015 14:33:47 +0300	[thread overview]
Message-ID: <551A861B.7020701@samsung.com> (raw)
In-Reply-To: <alpine.DEB.2.10.1503261542060.16259@chino.kir.corp.google.com>

On 03/27/2015 01:50 AM, David Rientjes wrote:
> We don't have a need to set PAGE_EXT_DEBUG_POISON on these pages sitting 
> in the reserved pool, nor do we have a need to do kmap_atomic() since it's 
> already mapped and must be mapped to be on the reserved pool, which is 
> handled by mempool_free().
> 

Hmm.. I just realized that this statement might be wrong.
Why pages has to be mapped to be on reserved pool?
mempool could be used for highmem pages and there is no need to kmap()
until these pages will be used.

drbd (drivers/block/drbd) already uses mempool for highmem pages:

static int drbd_create_mempools(void)
{
....
	drbd_md_io_page_pool = mempool_create_page_pool(DRBD_MIN_POOL_PAGES, 0);
....
}

static void bm_page_io_async(struct drbd_bm_aio_ctx *ctx, int page_nr) __must_hold(local)
{
....
		page = mempool_alloc(drbd_md_io_page_pool, __GFP_HIGHMEM|__GFP_WAIT);
		copy_highpage(page, b->bm_pages[page_nr]);