linux-kernel - Re: [PATCH 09/15] mm/hmm/devmem: device memory hotplug using ZONE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 28 Jul 2017 13:10:03 +0200
From:   Michal Hocko <mhocko@...nel.org>
To:     Jérôme Glisse <jglisse@...hat.com>
Cc:     akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
        linux-mm@...ck.org, John Hubbard <jhubbard@...dia.com>,
        Dan Williams <dan.j.williams@...el.com>,
        David Nellans <dnellans@...dia.com>,
        Evgeny Baskakov <ebaskakov@...dia.com>,
        Mark Hairgrove <mhairgrove@...dia.com>,
        Sherry Cheung <SCheung@...dia.com>,
        Subhash Gutti <sgutti@...dia.com>
Subject: Re: [PATCH 09/15] mm/hmm/devmem: device memory hotplug using
 ZONE_DEVICE v6

I haven't seen a newer version posted but the same comment applies on
your hmm-v25-4.9 git version from
git://people.freedesktop.org/~glisse/linux

On Wed 28-06-17 14:00:41, Jérôme Glisse wrote:
> This introduce a simple struct and associated helpers for device driver
> to use when hotpluging un-addressable device memory as ZONE_DEVICE. It
> will find a unuse physical address range and trigger memory hotplug for
> it which allocates and initialize struct page for the device memory.

Please document the hotplug semantic some more please (who is in charge,
what is the lifetime, userspace API to add/remove this memory if any
etc...).

I can see you call add_pages. Please document why arch_add_memory (like
devm_memremap_pages) is not used. You also never seem to online the
range which is in line with nvdim usage and it is OK. But then I fail to
understand why you need

[...]
> +	mem_hotplug_begin();
> +	ret = add_pages(nid, align_start >> PAGE_SHIFT,
> +			align_size >> PAGE_SHIFT, false);
> +	if (ret) {
> +		mem_hotplug_done();
> +		goto error_add_memory;
> +	}
> +	move_pfn_range_to_zone(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
> +				align_start >> PAGE_SHIFT,
> +				align_size >> PAGE_SHIFT);
> +	mem_hotplug_done();
> +
> +	for (pfn = devmem->pfn_first; pfn < devmem->pfn_last; pfn++) {
> +		struct page *page = pfn_to_page(pfn);
> +
> +		/*
> +		 * ZONE_DEVICE pages union ->lru with a ->pgmap back
> +		 * pointer.  It is a bug if a ZONE_DEVICE page is ever
> +		 * freed or placed on a driver-private list. Therefore,
> +		 * seed the storage with LIST_POISON* values.
> +		 */
> +		list_del(&page->lru);

this? The page is not on any list yet - it hasn't been added to the page
allocator.

> +		page->pgmap = &devmem->pagemap;
> +	}
> +	return 0;
-- 
Michal Hocko
SUSE Labs