2008-04-02 19:54:13 +02:00
|
|
|
/******************************************************************************
|
|
|
|
* Xen balloon driver - enables returning/claiming memory to/from Xen.
|
|
|
|
*
|
|
|
|
* Copyright (c) 2003, B Dragovic
|
|
|
|
* Copyright (c) 2003-2004, M Williamson, K Fraser
|
|
|
|
* Copyright (c) 2005 Dan M. Smith, IBM Corporation
|
2011-07-26 02:12:06 +02:00
|
|
|
* Copyright (c) 2010 Daniel Kiper
|
|
|
|
*
|
|
|
|
* Memory hotplug support was written by Daniel Kiper. Work on
|
|
|
|
* it was sponsored by Google under Google Summer of Code 2010
|
|
|
|
* program. Jeremy Fitzhardinge from Citrix was the mentor for
|
|
|
|
* this project.
|
2008-04-02 19:54:13 +02:00
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License version 2
|
|
|
|
* as published by the Free Software Foundation; or, when distributed
|
|
|
|
* separately from the Linux kernel or incorporated into other
|
|
|
|
* software packages, subject to the following license:
|
|
|
|
*
|
|
|
|
* Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
|
|
* of this source file (the "Software"), to deal in the Software without
|
|
|
|
* restriction, including without limitation the rights to use, copy, modify,
|
|
|
|
* merge, publish, distribute, sublicense, and/or sell copies of the Software,
|
|
|
|
* and to permit persons to whom the Software is furnished to do so, subject to
|
|
|
|
* the following conditions:
|
|
|
|
*
|
|
|
|
* The above copyright notice and this permission notice shall be included in
|
|
|
|
* all copies or substantial portions of the Software.
|
|
|
|
*
|
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
|
|
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
|
|
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
|
|
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
|
|
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
|
|
|
|
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
|
|
|
|
* IN THE SOFTWARE.
|
|
|
|
*/
|
|
|
|
|
2013-06-28 12:21:41 +02:00
|
|
|
#define pr_fmt(fmt) "xen:" KBUILD_MODNAME ": " fmt
|
|
|
|
|
2013-08-04 16:39:40 +02:00
|
|
|
#include <linux/cpu.h>
|
2008-04-02 19:54:13 +02:00
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/errno.h>
|
2011-07-03 22:20:57 +02:00
|
|
|
#include <linux/module.h>
|
2008-04-02 19:54:13 +02:00
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/bootmem.h>
|
|
|
|
#include <linux/pagemap.h>
|
|
|
|
#include <linux/highmem.h>
|
|
|
|
#include <linux/mutex.h>
|
|
|
|
#include <linux/list.h>
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 09:04:11 +01:00
|
|
|
#include <linux/gfp.h>
|
2011-07-26 02:12:06 +02:00
|
|
|
#include <linux/notifier.h>
|
|
|
|
#include <linux/memory.h>
|
|
|
|
#include <linux/memory_hotplug.h>
|
2013-08-04 16:39:40 +02:00
|
|
|
#include <linux/percpu-defs.h>
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
#include <asm/page.h>
|
|
|
|
#include <asm/pgalloc.h>
|
|
|
|
#include <asm/pgtable.h>
|
|
|
|
#include <asm/tlb.h>
|
|
|
|
|
2008-12-16 21:37:07 +01:00
|
|
|
#include <asm/xen/hypervisor.h>
|
|
|
|
#include <asm/xen/hypercall.h>
|
2009-10-07 00:11:14 +02:00
|
|
|
|
|
|
|
#include <xen/xen.h>
|
2008-12-16 21:37:07 +01:00
|
|
|
#include <xen/interface/xen.h>
|
2008-04-02 19:54:13 +02:00
|
|
|
#include <xen/interface/memory.h>
|
2011-03-14 16:29:37 +01:00
|
|
|
#include <xen/balloon.h>
|
2008-04-02 19:54:13 +02:00
|
|
|
#include <xen/features.h>
|
|
|
|
#include <xen/page.h>
|
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
/*
|
|
|
|
* balloon_process() state:
|
|
|
|
*
|
|
|
|
* BP_DONE: done or nothing to do,
|
|
|
|
* BP_EAGAIN: error, go to sleep,
|
|
|
|
* BP_ECANCELED: error, balloon operation canceled.
|
|
|
|
*/
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
enum bp_state {
|
|
|
|
BP_DONE,
|
|
|
|
BP_EAGAIN,
|
|
|
|
BP_ECANCELED
|
2008-04-02 19:54:13 +02:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
static DEFINE_MUTEX(balloon_mutex);
|
|
|
|
|
2011-03-14 16:29:37 +01:00
|
|
|
struct balloon_stats balloon_stats;
|
|
|
|
EXPORT_SYMBOL_GPL(balloon_stats);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
/* We increase/decrease in batches which fit in a page */
|
2012-10-17 10:39:16 +02:00
|
|
|
static xen_pfn_t frame_list[PAGE_SIZE / sizeof(unsigned long)];
|
2013-08-04 16:39:40 +02:00
|
|
|
static DEFINE_PER_CPU(struct page *, balloon_scratch_page);
|
|
|
|
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
/* List of ballooned pages, threaded through the mem_map array. */
|
|
|
|
static LIST_HEAD(ballooned_pages);
|
|
|
|
|
|
|
|
/* Main work function, always executed in process context. */
|
|
|
|
static void balloon_process(struct work_struct *work);
|
2011-03-08 22:47:39 +01:00
|
|
|
static DECLARE_DELAYED_WORK(balloon_worker, balloon_process);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
/* When ballooning out (allocating memory to return to Xen) we don't really
|
|
|
|
want the kernel to try too hard since that can trigger the oom killer. */
|
|
|
|
#define GFP_BALLOON \
|
|
|
|
(GFP_HIGHUSER | __GFP_NOWARN | __GFP_NORETRY | __GFP_NOMEMALLOC)
|
|
|
|
|
|
|
|
static void scrub_page(struct page *page)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_XEN_SCRUB_PAGES
|
2008-11-17 18:35:00 +01:00
|
|
|
clear_highpage(page);
|
2008-04-02 19:54:13 +02:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
/* balloon_append: add the given page to the balloon. */
|
2010-09-01 00:01:16 +02:00
|
|
|
static void __balloon_append(struct page *page)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
|
|
|
/* Lowmem is re-populated first, so highmem pages go at list tail. */
|
|
|
|
if (PageHighMem(page)) {
|
|
|
|
list_add_tail(&page->lru, &ballooned_pages);
|
|
|
|
balloon_stats.balloon_high++;
|
|
|
|
} else {
|
|
|
|
list_add(&page->lru, &ballooned_pages);
|
|
|
|
balloon_stats.balloon_low++;
|
|
|
|
}
|
2010-09-01 00:01:16 +02:00
|
|
|
}
|
2009-07-30 23:54:36 +02:00
|
|
|
|
2010-09-01 00:01:16 +02:00
|
|
|
static void balloon_append(struct page *page)
|
|
|
|
{
|
|
|
|
__balloon_append(page);
|
2013-07-04 00:03:21 +02:00
|
|
|
adjust_managed_page_count(page, -1);
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
|
2011-03-15 15:23:57 +01:00
|
|
|
static struct page *balloon_retrieve(bool prefer_highmem)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
|
|
|
struct page *page;
|
|
|
|
|
|
|
|
if (list_empty(&ballooned_pages))
|
|
|
|
return NULL;
|
|
|
|
|
2011-03-15 15:23:57 +01:00
|
|
|
if (prefer_highmem)
|
|
|
|
page = list_entry(ballooned_pages.prev, struct page, lru);
|
|
|
|
else
|
|
|
|
page = list_entry(ballooned_pages.next, struct page, lru);
|
2008-04-02 19:54:13 +02:00
|
|
|
list_del(&page->lru);
|
|
|
|
|
2013-07-04 00:03:21 +02:00
|
|
|
if (PageHighMem(page))
|
2008-04-02 19:54:13 +02:00
|
|
|
balloon_stats.balloon_high--;
|
2013-07-04 00:03:21 +02:00
|
|
|
else
|
2008-04-02 19:54:13 +02:00
|
|
|
balloon_stats.balloon_low--;
|
|
|
|
|
2013-07-04 00:03:21 +02:00
|
|
|
adjust_managed_page_count(page, 1);
|
2009-07-30 23:54:36 +02:00
|
|
|
|
2008-04-02 19:54:13 +02:00
|
|
|
return page;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct page *balloon_next_page(struct page *page)
|
|
|
|
{
|
|
|
|
struct list_head *next = page->lru.next;
|
|
|
|
if (next == &ballooned_pages)
|
|
|
|
return NULL;
|
|
|
|
return list_entry(next, struct page, lru);
|
|
|
|
}
|
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
static enum bp_state update_schedule(enum bp_state state)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
2014-10-07 23:00:07 +02:00
|
|
|
if (state == BP_ECANCELED)
|
|
|
|
return BP_ECANCELED;
|
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
if (state == BP_DONE) {
|
|
|
|
balloon_stats.schedule_delay = 1;
|
|
|
|
balloon_stats.retry_count = 1;
|
|
|
|
return BP_DONE;
|
|
|
|
}
|
|
|
|
|
|
|
|
++balloon_stats.retry_count;
|
|
|
|
|
|
|
|
if (balloon_stats.max_retry_count != RETRY_UNLIMITED &&
|
|
|
|
balloon_stats.retry_count > balloon_stats.max_retry_count) {
|
|
|
|
balloon_stats.schedule_delay = 1;
|
|
|
|
balloon_stats.retry_count = 1;
|
|
|
|
return BP_ECANCELED;
|
|
|
|
}
|
|
|
|
|
|
|
|
balloon_stats.schedule_delay <<= 1;
|
|
|
|
|
|
|
|
if (balloon_stats.schedule_delay > balloon_stats.max_schedule_delay)
|
|
|
|
balloon_stats.schedule_delay = balloon_stats.max_schedule_delay;
|
|
|
|
|
|
|
|
return BP_EAGAIN;
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
|
|
|
|
2011-07-26 02:12:06 +02:00
|
|
|
#ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
|
|
|
|
static long current_credit(void)
|
|
|
|
{
|
|
|
|
return balloon_stats.target_pages - balloon_stats.current_pages -
|
|
|
|
balloon_stats.hotplug_pages;
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool balloon_is_inflated(void)
|
|
|
|
{
|
|
|
|
if (balloon_stats.balloon_low || balloon_stats.balloon_high ||
|
|
|
|
balloon_stats.balloon_hotplug)
|
|
|
|
return true;
|
|
|
|
else
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* reserve_additional_memory() adds memory region of size >= credit above
|
|
|
|
* max_pfn. New region is section aligned and size is modified to be multiple
|
|
|
|
* of section size. Those features allow optimal use of address space and
|
|
|
|
* establish proper alignment when this function is called first time after
|
|
|
|
* boot (last section not fully populated at boot time contains unused memory
|
|
|
|
* pages with PG_reserved bit not set; online_pages_range() does not allow page
|
|
|
|
* onlining in whole range if first onlined page does not have PG_reserved
|
|
|
|
* bit set). Real size of added memory is established at page onlining stage.
|
|
|
|
*/
|
|
|
|
|
|
|
|
static enum bp_state reserve_additional_memory(long credit)
|
|
|
|
{
|
|
|
|
int nid, rc;
|
|
|
|
u64 hotplug_start_paddr;
|
|
|
|
unsigned long balloon_hotplug = credit;
|
|
|
|
|
|
|
|
hotplug_start_paddr = PFN_PHYS(SECTION_ALIGN_UP(max_pfn));
|
|
|
|
balloon_hotplug = round_up(balloon_hotplug, PAGES_PER_SECTION);
|
|
|
|
nid = memory_add_physaddr_to_nid(hotplug_start_paddr);
|
|
|
|
|
|
|
|
rc = add_memory(nid, hotplug_start_paddr, balloon_hotplug << PAGE_SHIFT);
|
|
|
|
|
|
|
|
if (rc) {
|
2014-09-01 19:52:44 +02:00
|
|
|
pr_warn("Cannot add additional memory (%i)\n", rc);
|
|
|
|
return BP_ECANCELED;
|
2011-07-26 02:12:06 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
balloon_hotplug -= credit;
|
|
|
|
|
|
|
|
balloon_stats.hotplug_pages += credit;
|
|
|
|
balloon_stats.balloon_hotplug = balloon_hotplug;
|
|
|
|
|
|
|
|
return BP_DONE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void xen_online_page(struct page *page)
|
|
|
|
{
|
|
|
|
__online_page_set_limits(page);
|
|
|
|
|
|
|
|
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);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int xen_memory_notifier(struct notifier_block *nb, unsigned long val, void *v)
|
|
|
|
{
|
|
|
|
if (val == MEM_ONLINE)
|
|
|
|
schedule_delayed_work(&balloon_worker, 0);
|
|
|
|
|
|
|
|
return NOTIFY_OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct notifier_block xen_memory_nb = {
|
|
|
|
.notifier_call = xen_memory_notifier,
|
|
|
|
.priority = 0
|
|
|
|
};
|
|
|
|
#else
|
2011-03-28 11:34:10 +02:00
|
|
|
static long current_credit(void)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
2009-06-05 12:58:37 +02:00
|
|
|
unsigned long target = balloon_stats.target_pages;
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
target = min(target,
|
|
|
|
balloon_stats.current_pages +
|
|
|
|
balloon_stats.balloon_low +
|
|
|
|
balloon_stats.balloon_high);
|
|
|
|
|
2011-03-28 11:34:10 +02:00
|
|
|
return target - balloon_stats.current_pages;
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
|
|
|
|
2011-07-26 02:12:06 +02:00
|
|
|
static bool balloon_is_inflated(void)
|
|
|
|
{
|
|
|
|
if (balloon_stats.balloon_low || balloon_stats.balloon_high)
|
|
|
|
return true;
|
|
|
|
else
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static enum bp_state reserve_additional_memory(long credit)
|
|
|
|
{
|
|
|
|
balloon_stats.target_pages = balloon_stats.current_pages;
|
|
|
|
return BP_DONE;
|
|
|
|
}
|
|
|
|
#endif /* CONFIG_XEN_BALLOON_MEMORY_HOTPLUG */
|
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
static enum bp_state increase_reservation(unsigned long nr_pages)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
2011-03-08 22:48:24 +01:00
|
|
|
int rc;
|
2010-09-03 08:11:17 +02:00
|
|
|
unsigned long pfn, i;
|
2008-04-02 19:54:13 +02:00
|
|
|
struct page *page;
|
|
|
|
struct xen_memory_reservation reservation = {
|
|
|
|
.address_bits = 0,
|
|
|
|
.extent_order = 0,
|
|
|
|
.domid = DOMID_SELF
|
|
|
|
};
|
|
|
|
|
2011-07-26 02:12:06 +02:00
|
|
|
#ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
|
|
|
|
if (!balloon_stats.balloon_low && !balloon_stats.balloon_high) {
|
|
|
|
nr_pages = min(nr_pages, balloon_stats.balloon_hotplug);
|
|
|
|
balloon_stats.hotplug_pages += nr_pages;
|
|
|
|
balloon_stats.balloon_hotplug -= nr_pages;
|
|
|
|
return BP_DONE;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2008-04-02 19:54:13 +02:00
|
|
|
if (nr_pages > ARRAY_SIZE(frame_list))
|
|
|
|
nr_pages = ARRAY_SIZE(frame_list);
|
|
|
|
|
2013-11-13 13:59:58 +01:00
|
|
|
page = list_first_entry_or_null(&ballooned_pages, struct page, lru);
|
2008-04-02 19:54:13 +02:00
|
|
|
for (i = 0; i < nr_pages; i++) {
|
2011-03-08 22:48:24 +01:00
|
|
|
if (!page) {
|
|
|
|
nr_pages = i;
|
|
|
|
break;
|
|
|
|
}
|
2009-08-18 20:18:35 +02:00
|
|
|
frame_list[i] = page_to_pfn(page);
|
2008-04-02 19:54:13 +02:00
|
|
|
page = balloon_next_page(page);
|
|
|
|
}
|
|
|
|
|
2008-05-27 00:31:14 +02:00
|
|
|
set_xen_guest_handle(reservation.extent_start, frame_list);
|
2008-07-25 01:28:00 +02:00
|
|
|
reservation.nr_extents = nr_pages;
|
|
|
|
rc = HYPERVISOR_memory_op(XENMEM_populate_physmap, &reservation);
|
2011-03-14 16:42:40 +01:00
|
|
|
if (rc <= 0)
|
2011-03-08 22:48:24 +01:00
|
|
|
return BP_EAGAIN;
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2009-06-05 12:58:37 +02:00
|
|
|
for (i = 0; i < rc; i++) {
|
2011-03-15 15:23:57 +01:00
|
|
|
page = balloon_retrieve(false);
|
2008-04-02 19:54:13 +02:00
|
|
|
BUG_ON(page == NULL);
|
|
|
|
|
|
|
|
pfn = page_to_pfn(page);
|
|
|
|
|
2012-10-03 13:17:50 +02:00
|
|
|
#ifdef CONFIG_XEN_HAVE_PVMMU
|
2013-12-11 17:58:42 +01:00
|
|
|
if (!xen_feature(XENFEAT_auto_translated_physmap)) {
|
|
|
|
set_phys_to_machine(pfn, frame_list[i]);
|
|
|
|
|
|
|
|
/* Link back into the page tables if not highmem. */
|
|
|
|
if (!PageHighMem(page)) {
|
|
|
|
int ret;
|
|
|
|
ret = HYPERVISOR_update_va_mapping(
|
|
|
|
(unsigned long)__va(pfn << PAGE_SHIFT),
|
|
|
|
mfn_pte(frame_list[i], PAGE_KERNEL),
|
|
|
|
0);
|
|
|
|
BUG_ON(ret);
|
|
|
|
}
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
2012-10-03 13:17:50 +02:00
|
|
|
#endif
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
/* Relinquish the page back to the allocator. */
|
2013-07-04 00:03:21 +02:00
|
|
|
__free_reserved_page(page);
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
|
|
|
|
2009-06-05 12:58:37 +02:00
|
|
|
balloon_stats.current_pages += rc;
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
return BP_DONE;
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
|
|
|
|
2011-03-15 15:23:57 +01:00
|
|
|
static enum bp_state decrease_reservation(unsigned long nr_pages, gfp_t gfp)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
2011-03-08 22:48:24 +01:00
|
|
|
enum bp_state state = BP_DONE;
|
2010-09-03 08:11:17 +02:00
|
|
|
unsigned long pfn, i;
|
2008-04-02 19:54:13 +02:00
|
|
|
struct page *page;
|
|
|
|
int ret;
|
|
|
|
struct xen_memory_reservation reservation = {
|
|
|
|
.address_bits = 0,
|
|
|
|
.extent_order = 0,
|
|
|
|
.domid = DOMID_SELF
|
|
|
|
};
|
|
|
|
|
2011-07-26 02:12:06 +02:00
|
|
|
#ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
|
|
|
|
if (balloon_stats.hotplug_pages) {
|
|
|
|
nr_pages = min(nr_pages, balloon_stats.hotplug_pages);
|
|
|
|
balloon_stats.hotplug_pages -= nr_pages;
|
|
|
|
balloon_stats.balloon_hotplug += nr_pages;
|
|
|
|
return BP_DONE;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2008-04-02 19:54:13 +02:00
|
|
|
if (nr_pages > ARRAY_SIZE(frame_list))
|
|
|
|
nr_pages = ARRAY_SIZE(frame_list);
|
|
|
|
|
|
|
|
for (i = 0; i < nr_pages; i++) {
|
2013-05-16 07:59:40 +02:00
|
|
|
page = alloc_page(gfp);
|
|
|
|
if (page == NULL) {
|
2008-04-02 19:54:13 +02:00
|
|
|
nr_pages = i;
|
2011-03-08 22:48:24 +01:00
|
|
|
state = BP_EAGAIN;
|
2008-04-02 19:54:13 +02:00
|
|
|
break;
|
|
|
|
}
|
2014-03-15 17:11:47 +01:00
|
|
|
scrub_page(page);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2014-03-15 17:11:47 +01:00
|
|
|
frame_list[i] = page_to_pfn(page);
|
|
|
|
}
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2014-03-15 17:11:47 +01:00
|
|
|
/*
|
|
|
|
* Ensure that ballooned highmem pages don't have kmaps.
|
|
|
|
*
|
|
|
|
* Do this before changing the p2m as kmap_flush_unused()
|
|
|
|
* reads PTEs to obtain pages (and hence needs the original
|
|
|
|
* p2m entry).
|
|
|
|
*/
|
|
|
|
kmap_flush_unused();
|
|
|
|
|
|
|
|
/* Update direct mapping, invalidate P2M, and add to balloon. */
|
|
|
|
for (i = 0; i < nr_pages; i++) {
|
|
|
|
pfn = frame_list[i];
|
|
|
|
frame_list[i] = pfn_to_mfn(pfn);
|
|
|
|
page = pfn_to_page(pfn);
|
2009-01-22 23:36:08 +01:00
|
|
|
|
2013-12-11 17:58:42 +01:00
|
|
|
#ifdef CONFIG_XEN_HAVE_PVMMU
|
2013-09-19 18:14:53 +02:00
|
|
|
/*
|
|
|
|
* Ballooned out frames are effectively replaced with
|
|
|
|
* a scratch frame. Ensure direct mappings and the
|
|
|
|
* p2m are consistent.
|
|
|
|
*/
|
2013-08-27 17:17:25 +02:00
|
|
|
if (!xen_feature(XENFEAT_auto_translated_physmap)) {
|
2013-12-11 17:58:42 +01:00
|
|
|
if (!PageHighMem(page)) {
|
2014-06-27 11:42:03 +02:00
|
|
|
struct page *scratch_page = get_balloon_scratch_page();
|
|
|
|
|
2013-12-11 17:58:42 +01:00
|
|
|
ret = HYPERVISOR_update_va_mapping(
|
|
|
|
(unsigned long)__va(pfn << PAGE_SHIFT),
|
|
|
|
pfn_pte(page_to_pfn(scratch_page),
|
|
|
|
PAGE_KERNEL_RO), 0);
|
|
|
|
BUG_ON(ret);
|
|
|
|
|
2014-06-27 11:42:03 +02:00
|
|
|
put_balloon_scratch_page();
|
|
|
|
}
|
|
|
|
__set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
|
2013-08-27 17:17:25 +02:00
|
|
|
}
|
2013-12-11 17:58:42 +01:00
|
|
|
#endif
|
2013-09-19 18:14:53 +02:00
|
|
|
|
2014-03-15 17:11:47 +01:00
|
|
|
balloon_append(page);
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
|
|
|
|
2013-09-19 18:14:53 +02:00
|
|
|
flush_tlb_all();
|
2013-09-11 19:45:44 +02:00
|
|
|
|
2008-05-27 00:31:14 +02:00
|
|
|
set_xen_guest_handle(reservation.extent_start, frame_list);
|
2008-04-02 19:54:13 +02:00
|
|
|
reservation.nr_extents = nr_pages;
|
|
|
|
ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
|
|
|
|
BUG_ON(ret != nr_pages);
|
|
|
|
|
|
|
|
balloon_stats.current_pages -= nr_pages;
|
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
return state;
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We avoid multiple worker processes conflicting via the balloon mutex.
|
|
|
|
* We may of course race updates of the target counts (which are protected
|
|
|
|
* by the balloon lock), or with changes to the Xen hard limit, but we will
|
|
|
|
* recover from these in time.
|
|
|
|
*/
|
|
|
|
static void balloon_process(struct work_struct *work)
|
|
|
|
{
|
2011-03-08 22:48:24 +01:00
|
|
|
enum bp_state state = BP_DONE;
|
2008-04-02 19:54:13 +02:00
|
|
|
long credit;
|
|
|
|
|
|
|
|
mutex_lock(&balloon_mutex);
|
|
|
|
|
|
|
|
do {
|
2011-03-28 11:34:10 +02:00
|
|
|
credit = current_credit();
|
2011-03-08 22:48:24 +01:00
|
|
|
|
2011-07-26 02:12:06 +02:00
|
|
|
if (credit > 0) {
|
|
|
|
if (balloon_is_inflated())
|
|
|
|
state = increase_reservation(credit);
|
|
|
|
else
|
|
|
|
state = reserve_additional_memory(credit);
|
|
|
|
}
|
2011-03-08 22:48:24 +01:00
|
|
|
|
2008-04-02 19:54:13 +02:00
|
|
|
if (credit < 0)
|
2011-03-15 15:23:57 +01:00
|
|
|
state = decrease_reservation(-credit, GFP_BALLOON);
|
2011-03-08 22:48:24 +01:00
|
|
|
|
|
|
|
state = update_schedule(state);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
#ifndef CONFIG_PREEMPT
|
|
|
|
if (need_resched())
|
|
|
|
schedule();
|
|
|
|
#endif
|
2011-03-08 22:48:24 +01:00
|
|
|
} while (credit && state == BP_DONE);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
/* Schedule more work if there is some still to be done. */
|
2011-03-08 22:48:24 +01:00
|
|
|
if (state == BP_EAGAIN)
|
|
|
|
schedule_delayed_work(&balloon_worker, balloon_stats.schedule_delay * HZ);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
mutex_unlock(&balloon_mutex);
|
|
|
|
}
|
|
|
|
|
2013-08-04 16:39:40 +02:00
|
|
|
struct page *get_balloon_scratch_page(void)
|
|
|
|
{
|
|
|
|
struct page *ret = get_cpu_var(balloon_scratch_page);
|
|
|
|
BUG_ON(ret == NULL);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
void put_balloon_scratch_page(void)
|
|
|
|
{
|
|
|
|
put_cpu_var(balloon_scratch_page);
|
|
|
|
}
|
|
|
|
|
2008-04-02 19:54:13 +02:00
|
|
|
/* Resets the Xen limit, sets new target, and kicks off processing. */
|
2011-03-14 16:29:37 +01:00
|
|
|
void balloon_set_new_target(unsigned long target)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
|
|
|
/* No need for lock. Not read-modify-write updates. */
|
|
|
|
balloon_stats.target_pages = target;
|
2011-03-08 22:47:39 +01:00
|
|
|
schedule_delayed_work(&balloon_worker, 0);
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
2011-03-14 16:29:37 +01:00
|
|
|
EXPORT_SYMBOL_GPL(balloon_set_new_target);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-03-15 15:23:57 +01:00
|
|
|
/**
|
|
|
|
* alloc_xenballooned_pages - get pages that have been ballooned out
|
|
|
|
* @nr_pages: Number of pages to get
|
|
|
|
* @pages: pages returned
|
2011-10-19 23:59:37 +02:00
|
|
|
* @highmem: allow highmem pages
|
2011-03-15 15:23:57 +01:00
|
|
|
* @return 0 on success, error otherwise
|
|
|
|
*/
|
2011-09-29 12:57:55 +02:00
|
|
|
int alloc_xenballooned_pages(int nr_pages, struct page **pages, bool highmem)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
2011-03-15 15:23:57 +01:00
|
|
|
int pgno = 0;
|
2011-07-26 13:15:59 +02:00
|
|
|
struct page *page;
|
2011-03-15 15:23:57 +01:00
|
|
|
mutex_lock(&balloon_mutex);
|
|
|
|
while (pgno < nr_pages) {
|
2011-09-29 12:57:55 +02:00
|
|
|
page = balloon_retrieve(highmem);
|
2011-10-19 23:59:37 +02:00
|
|
|
if (page && (highmem || !PageHighMem(page))) {
|
2011-03-15 15:23:57 +01:00
|
|
|
pages[pgno++] = page;
|
|
|
|
} else {
|
|
|
|
enum bp_state st;
|
2011-09-29 12:57:55 +02:00
|
|
|
if (page)
|
|
|
|
balloon_append(page);
|
|
|
|
st = decrease_reservation(nr_pages - pgno,
|
|
|
|
highmem ? GFP_HIGHUSER : GFP_USER);
|
2011-03-15 15:23:57 +01:00
|
|
|
if (st != BP_DONE)
|
|
|
|
goto out_undo;
|
|
|
|
}
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
2011-03-15 15:23:57 +01:00
|
|
|
mutex_unlock(&balloon_mutex);
|
|
|
|
return 0;
|
|
|
|
out_undo:
|
|
|
|
while (pgno)
|
|
|
|
balloon_append(pages[--pgno]);
|
|
|
|
/* Free the memory back to the kernel soon */
|
|
|
|
schedule_delayed_work(&balloon_worker, 0);
|
|
|
|
mutex_unlock(&balloon_mutex);
|
|
|
|
return -ENOMEM;
|
2008-04-02 19:54:13 +02:00
|
|
|
}
|
2011-03-15 15:23:57 +01:00
|
|
|
EXPORT_SYMBOL(alloc_xenballooned_pages);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-03-15 15:23:57 +01:00
|
|
|
/**
|
|
|
|
* free_xenballooned_pages - return pages retrieved with get_ballooned_pages
|
|
|
|
* @nr_pages: Number of pages
|
|
|
|
* @pages: pages to return
|
|
|
|
*/
|
2011-07-26 13:15:59 +02:00
|
|
|
void free_xenballooned_pages(int nr_pages, struct page **pages)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
2011-03-15 15:23:57 +01:00
|
|
|
int i;
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-03-15 15:23:57 +01:00
|
|
|
mutex_lock(&balloon_mutex);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-03-15 15:23:57 +01:00
|
|
|
for (i = 0; i < nr_pages; i++) {
|
|
|
|
if (pages[i])
|
|
|
|
balloon_append(pages[i]);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* The balloon may be too large now. Shrink it if needed. */
|
2011-03-28 11:34:10 +02:00
|
|
|
if (current_credit())
|
2011-03-15 15:23:57 +01:00
|
|
|
schedule_delayed_work(&balloon_worker, 0);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-03-15 15:23:57 +01:00
|
|
|
mutex_unlock(&balloon_mutex);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(free_xenballooned_pages);
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-09-28 18:46:34 +02:00
|
|
|
static void __init balloon_add_region(unsigned long start_pfn,
|
|
|
|
unsigned long pages)
|
2008-04-02 19:54:13 +02:00
|
|
|
{
|
2011-03-28 11:33:18 +02:00
|
|
|
unsigned long pfn, extra_pfn_end;
|
2008-04-02 19:54:13 +02:00
|
|
|
struct page *page;
|
|
|
|
|
2011-09-28 18:46:34 +02:00
|
|
|
/*
|
|
|
|
* If the amount of usable memory has been limited (e.g., with
|
|
|
|
* the 'mem' command line parameter), don't add pages beyond
|
|
|
|
* this limit.
|
|
|
|
*/
|
|
|
|
extra_pfn_end = min(max_pfn, start_pfn + pages);
|
|
|
|
|
|
|
|
for (pfn = start_pfn; pfn < extra_pfn_end; pfn++) {
|
|
|
|
page = pfn_to_page(pfn);
|
|
|
|
/* totalram_pages and totalhigh_pages do not
|
|
|
|
include the boot-time balloon extension, so
|
|
|
|
don't subtract from it. */
|
|
|
|
__balloon_append(page);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
xen, balloon: Fix CPU hotplug callback registration
Subsystems that want to register CPU hotplug callbacks, as well as perform
initialization for the CPUs that are already online, often do it as shown
below:
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
register_cpu_notifier(&foobar_cpu_notifier);
put_online_cpus();
This is wrong, since it is prone to ABBA deadlocks involving the
cpu_add_remove_lock and the cpu_hotplug.lock (when running concurrently
with CPU hotplug operations).
The xen balloon driver doesn't take get/put_online_cpus() around this code,
but that is also buggy, since it can miss CPU hotplug events in between the
initialization and callback registration:
for_each_online_cpu(cpu)
init_cpu(cpu);
^
| Race window; Can miss CPU hotplug events here.
v
register_cpu_notifier(&foobar_cpu_notifier);
Interestingly, the balloon code in xen can simply be reorganized as shown
below, to have a race-free method to register hotplug callbacks, without even
taking get/put_online_cpus(). This is because the initialization performed for
already online CPUs is exactly the same as that performed for CPUs that come
online later. Moreover, the code has checks in place to avoid double
initialization.
register_cpu_notifier(&foobar_cpu_notifier);
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
put_online_cpus();
A hotplug operation that occurs between registering the notifier and calling
get_online_cpus(), won't disrupt anything, because the code takes care to
perform the memory allocations only once.
So reorganize the balloon code in xen this way to fix the issues with CPU
hotplug callback registration.
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Cc: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-10 21:41:45 +01:00
|
|
|
static int alloc_balloon_scratch_page(int cpu)
|
|
|
|
{
|
|
|
|
if (per_cpu(balloon_scratch_page, cpu) != NULL)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
per_cpu(balloon_scratch_page, cpu) = alloc_page(GFP_KERNEL);
|
|
|
|
if (per_cpu(balloon_scratch_page, cpu) == NULL) {
|
|
|
|
pr_warn("Failed to allocate balloon_scratch_page for cpu %d\n", cpu);
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2013-11-08 17:21:51 +01:00
|
|
|
static int balloon_cpu_notify(struct notifier_block *self,
|
2013-08-04 16:39:40 +02:00
|
|
|
unsigned long action, void *hcpu)
|
|
|
|
{
|
|
|
|
int cpu = (long)hcpu;
|
|
|
|
switch (action) {
|
|
|
|
case CPU_UP_PREPARE:
|
xen, balloon: Fix CPU hotplug callback registration
Subsystems that want to register CPU hotplug callbacks, as well as perform
initialization for the CPUs that are already online, often do it as shown
below:
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
register_cpu_notifier(&foobar_cpu_notifier);
put_online_cpus();
This is wrong, since it is prone to ABBA deadlocks involving the
cpu_add_remove_lock and the cpu_hotplug.lock (when running concurrently
with CPU hotplug operations).
The xen balloon driver doesn't take get/put_online_cpus() around this code,
but that is also buggy, since it can miss CPU hotplug events in between the
initialization and callback registration:
for_each_online_cpu(cpu)
init_cpu(cpu);
^
| Race window; Can miss CPU hotplug events here.
v
register_cpu_notifier(&foobar_cpu_notifier);
Interestingly, the balloon code in xen can simply be reorganized as shown
below, to have a race-free method to register hotplug callbacks, without even
taking get/put_online_cpus(). This is because the initialization performed for
already online CPUs is exactly the same as that performed for CPUs that come
online later. Moreover, the code has checks in place to avoid double
initialization.
register_cpu_notifier(&foobar_cpu_notifier);
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
put_online_cpus();
A hotplug operation that occurs between registering the notifier and calling
get_online_cpus(), won't disrupt anything, because the code takes care to
perform the memory allocations only once.
So reorganize the balloon code in xen this way to fix the issues with CPU
hotplug callback registration.
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Cc: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-10 21:41:45 +01:00
|
|
|
if (alloc_balloon_scratch_page(cpu))
|
2013-08-04 16:39:40 +02:00
|
|
|
return NOTIFY_BAD;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return NOTIFY_OK;
|
|
|
|
}
|
|
|
|
|
2013-11-08 17:21:51 +01:00
|
|
|
static struct notifier_block balloon_cpu_notifier = {
|
2013-08-04 16:39:40 +02:00
|
|
|
.notifier_call = balloon_cpu_notify,
|
|
|
|
};
|
|
|
|
|
2011-09-28 18:46:34 +02:00
|
|
|
static int __init balloon_init(void)
|
|
|
|
{
|
2013-08-04 16:39:40 +02:00
|
|
|
int i, cpu;
|
2011-09-28 18:46:34 +02:00
|
|
|
|
2010-12-02 18:55:05 +01:00
|
|
|
if (!xen_domain())
|
2008-04-02 19:54:13 +02:00
|
|
|
return -ENODEV;
|
|
|
|
|
2013-12-11 17:58:42 +01:00
|
|
|
if (!xen_feature(XENFEAT_auto_translated_physmap)) {
|
xen, balloon: Fix CPU hotplug callback registration
Subsystems that want to register CPU hotplug callbacks, as well as perform
initialization for the CPUs that are already online, often do it as shown
below:
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
register_cpu_notifier(&foobar_cpu_notifier);
put_online_cpus();
This is wrong, since it is prone to ABBA deadlocks involving the
cpu_add_remove_lock and the cpu_hotplug.lock (when running concurrently
with CPU hotplug operations).
The xen balloon driver doesn't take get/put_online_cpus() around this code,
but that is also buggy, since it can miss CPU hotplug events in between the
initialization and callback registration:
for_each_online_cpu(cpu)
init_cpu(cpu);
^
| Race window; Can miss CPU hotplug events here.
v
register_cpu_notifier(&foobar_cpu_notifier);
Interestingly, the balloon code in xen can simply be reorganized as shown
below, to have a race-free method to register hotplug callbacks, without even
taking get/put_online_cpus(). This is because the initialization performed for
already online CPUs is exactly the same as that performed for CPUs that come
online later. Moreover, the code has checks in place to avoid double
initialization.
register_cpu_notifier(&foobar_cpu_notifier);
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
put_online_cpus();
A hotplug operation that occurs between registering the notifier and calling
get_online_cpus(), won't disrupt anything, because the code takes care to
perform the memory allocations only once.
So reorganize the balloon code in xen this way to fix the issues with CPU
hotplug callback registration.
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Cc: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-10 21:41:45 +01:00
|
|
|
register_cpu_notifier(&balloon_cpu_notifier);
|
|
|
|
|
|
|
|
get_online_cpus();
|
|
|
|
for_each_online_cpu(cpu) {
|
|
|
|
if (alloc_balloon_scratch_page(cpu)) {
|
|
|
|
put_online_cpus();
|
|
|
|
unregister_cpu_notifier(&balloon_cpu_notifier);
|
2013-12-11 17:58:42 +01:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
2013-08-04 16:39:40 +02:00
|
|
|
}
|
xen, balloon: Fix CPU hotplug callback registration
Subsystems that want to register CPU hotplug callbacks, as well as perform
initialization for the CPUs that are already online, often do it as shown
below:
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
register_cpu_notifier(&foobar_cpu_notifier);
put_online_cpus();
This is wrong, since it is prone to ABBA deadlocks involving the
cpu_add_remove_lock and the cpu_hotplug.lock (when running concurrently
with CPU hotplug operations).
The xen balloon driver doesn't take get/put_online_cpus() around this code,
but that is also buggy, since it can miss CPU hotplug events in between the
initialization and callback registration:
for_each_online_cpu(cpu)
init_cpu(cpu);
^
| Race window; Can miss CPU hotplug events here.
v
register_cpu_notifier(&foobar_cpu_notifier);
Interestingly, the balloon code in xen can simply be reorganized as shown
below, to have a race-free method to register hotplug callbacks, without even
taking get/put_online_cpus(). This is because the initialization performed for
already online CPUs is exactly the same as that performed for CPUs that come
online later. Moreover, the code has checks in place to avoid double
initialization.
register_cpu_notifier(&foobar_cpu_notifier);
get_online_cpus();
for_each_online_cpu(cpu)
init_cpu(cpu);
put_online_cpus();
A hotplug operation that occurs between registering the notifier and calling
get_online_cpus(), won't disrupt anything, because the code takes care to
perform the memory allocations only once.
So reorganize the balloon code in xen this way to fix the issues with CPU
hotplug callback registration.
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Cc: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-10 21:41:45 +01:00
|
|
|
put_online_cpus();
|
2013-08-04 16:39:40 +02:00
|
|
|
}
|
|
|
|
|
2013-06-28 12:21:41 +02:00
|
|
|
pr_info("Initialising balloon driver\n");
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-09-28 18:46:32 +02:00
|
|
|
balloon_stats.current_pages = xen_pv_domain()
|
|
|
|
? min(xen_start_info->nr_pages - xen_released_pages, max_pfn)
|
2013-11-06 21:37:40 +01:00
|
|
|
: get_num_physpages();
|
2008-04-02 19:54:13 +02:00
|
|
|
balloon_stats.target_pages = balloon_stats.current_pages;
|
|
|
|
balloon_stats.balloon_low = 0;
|
|
|
|
balloon_stats.balloon_high = 0;
|
|
|
|
|
2011-03-08 22:48:24 +01:00
|
|
|
balloon_stats.schedule_delay = 1;
|
|
|
|
balloon_stats.max_schedule_delay = 32;
|
|
|
|
balloon_stats.retry_count = 1;
|
2011-03-14 16:42:40 +01:00
|
|
|
balloon_stats.max_retry_count = RETRY_UNLIMITED;
|
2008-04-02 19:54:13 +02:00
|
|
|
|
2011-07-26 02:12:06 +02:00
|
|
|
#ifdef CONFIG_XEN_BALLOON_MEMORY_HOTPLUG
|
|
|
|
balloon_stats.hotplug_pages = 0;
|
|
|
|
balloon_stats.balloon_hotplug = 0;
|
|
|
|
|
|
|
|
set_online_page_callback(&xen_online_page);
|
|
|
|
register_memory_notifier(&xen_memory_nb);
|
|
|
|
#endif
|
|
|
|
|
2010-12-03 00:30:06 +01:00
|
|
|
/*
|
2011-09-28 18:46:33 +02:00
|
|
|
* Initialize the balloon with pages from the extra memory
|
2011-09-28 18:46:34 +02:00
|
|
|
* regions (see arch/x86/xen/setup.c).
|
2010-12-03 00:30:06 +01:00
|
|
|
*/
|
2011-09-28 18:46:34 +02:00
|
|
|
for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++)
|
|
|
|
if (xen_extra_mem[i].size)
|
|
|
|
balloon_add_region(PFN_UP(xen_extra_mem[i].start),
|
|
|
|
PFN_DOWN(xen_extra_mem[i].size));
|
2008-04-02 19:54:13 +02:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
subsys_initcall(balloon_init);
|
|
|
|
|
2013-08-04 16:39:40 +02:00
|
|
|
static int __init balloon_clear(void)
|
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
|
|
|
|
for_each_possible_cpu(cpu)
|
|
|
|
per_cpu(balloon_scratch_page, cpu) = NULL;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
early_initcall(balloon_clear);
|
|
|
|
|
2008-04-02 19:54:13 +02:00
|
|
|
MODULE_LICENSE("GPL");
|