linux-kernel - Re: [PATCH R4 7/7] xen/balloon: Memory hotplug support for Xen balloon driver

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:	Tue, 08 Mar 2011 16:02:19 -0800
From:	Dave Hansen <dave@...ux.vnet.ibm.com>
To:	Daniel Kiper <dkiper@...-space.pl>
Cc:	ian.campbell@...rix.com, akpm@...ux-foundation.org,
	andi.kleen@...el.com, haicheng.li@...ux.intel.com,
	fengguang.wu@...el.com, jeremy@...p.org, konrad.wilk@...cle.com,
	dan.magenheimer@...cle.com, v.tolstov@...fip.ru, pasik@....fi,
	wdauchy@...il.com, rientjes@...gle.com,
	xen-devel@...ts.xensource.com, linux-kernel@...r.kernel.org,
	linux-mm@...ck.org
Subject: Re: [PATCH R4 7/7] xen/balloon: Memory hotplug support for Xen
 balloon driver

On Tue, 2011-03-08 at 22:50 +0100, Daniel Kiper wrote:
> +static enum bp_state reserve_additional_memory(long credit)
> +{
> +	int rc;
> +	unsigned long balloon_hotplug = credit;
> +
> +	balloon_hotplug <<= PAGE_SHIFT;
> +
> +	rc = add_virtual_memory((u64 *)&balloon_hotplug);

This would work if all 'unsigned long's were 64-bits.  It'll break on
32-bit kernels in a very bad way by overwriting 4 bytes of stack.

> +	if (rc) {
> +		pr_info("xen_balloon: %s: add_virtual_memory() failed: %i\n", __func__, rc);
> +		return BP_EAGAIN;
> +	}
> +
> +	balloon_hotplug >>= PAGE_SHIFT;
> +
> +	balloon_hotplug -= credit;
> +
> +	balloon_stats.hotplug_pages += credit;
> +	balloon_stats.balloon_hotplug = balloon_hotplug;
> +
> +	return BP_DONE;
> +}
> +
> +static int xen_online_page_notifier(struct notifier_block *nb, unsigned long val, void *v)
> +{
> +	struct page *page = v;
> +	unsigned long pfn = page_to_pfn(page);
> +
> +	if (pfn >= num_physpages)
> +		num_physpages = pfn + 1;
> +
> +	inc_totalhigh_pages();
> +
> +#ifdef CONFIG_FLATMEM
> +	max_mapnr = max(pfn, max_mapnr);
> +#endif

I really don't like that this is a direct copy of online_page() up to
this point.  They're already subtly different.  I'm also curious if this
breaks on 32-bit kernels because of the unconditional
inc_totalhigh_pages().

If it's done this way, I'd almost guarantee that the first time someone
fixes a bug or adds a generic feature in online_page() that Xen gets
missed.  

> +	mutex_lock(&balloon_mutex);
> +
> +	__balloon_append(page);
> +
> +	if (balloon_stats.hotplug_pages)
> +		--balloon_stats.hotplug_pages;
> +	else
> +		--balloon_stats.balloon_hotplug;
> +
> +	mutex_unlock(&balloon_mutex);
> +
> +	return NOTIFY_STOP;
> +}

I'm not a _huge_ fan of these notifier chains, but I guess it works.
However, if you're going to use these notifier chains, then we probably
should use them to full effect.  Have a notifier list like this:

	1. generic online_page()
	2. xen_online_page_notifier() (returns NOTIFY_STOP)
	3. free_online_page()

Where finish_online_page() does something like this:

finish_online_page(...)
{
        ClearPageReserved(page);
        init_page_count(page);
        __free_page(page);
}

These patches are definitely getting there.  Just another round or two,
and they should be ready to go.

-- Dave

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/