Simplify locking by moving the slab_add_sysfs after all locks have been
dropped. Eases the upcoming move to provide sysfs support for all
allocators.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

The slab aliasing logic causes some strange contortions in slub. So add
a call to deal with aliases to slab_common.c but disable it for other
slab allocators by providng stubs that fail to create aliases.

Full general support for aliases will require additional cleanup passes
and more standardization of fields in kmem_cache.
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Duping of the slabname has to be done by each slab. Moving this code to
slab_common avoids duplicate implementations.

With this patch we have common string handling for all slab allocators.
Strings passed to kmem_cache_create() are copied internally. Subsystems
can create temporary strings to create slab caches.

Slabs allocated in early states of bootstrap will never be freed (and
those can never be freed since they are essential to slab allocator
operations).  During bootstrap we therefore do not have to worry about
duping names.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

What is done there can be done in __kmem_cache_shutdown.

This affects RCU handling somewhat. On rcu free all slab allocators do
not refer to other management structures than the kmem_cache structure.
Therefore these other structures can be freed before the rcu deferred
free to the page allocator occurs.
Reviewed-by: NJoonsoo Kim <js1304@gmail.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

The freeing action is basically the same in all slab allocators.
Move to the common kmem_cache_destroy() function.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Reviewed-by: NJoonsoo Kim <js1304@gmail.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Make all allocators use the "kmem_cache" slabname for the "kmem_cache"
structure.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Reviewed-by: NJoonsoo Kim <js1304@gmail.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

kmem_cache_destroy does basically the same in all allocators.

Extract common code which is easy since we already have common mutex
handling.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Move the code to append the new kmem_cache to the list of slab caches to
the kmem_cache_create code in the shared code.

This is possible now since the acquisition of the mutex was moved into
kmem_cache_create().
Acked-by: NDavid Rientjes <rientjes@google.com>
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Reviewed-by: NJoonsoo Kim <js1304@gmail.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Instead of using s == NULL use an errorcode. This allows much more
detailed diagnostics as to what went wrong. As we add more functionality
from the slab allocators to the common kmem_cache_create() function we will
also add more error conditions.

Print the error code during the panic as well as in a warning if the module
can handle failure. The API for kmem_cache_create() currently does not allow
the returning of an error code. Return NULL but log the cause of the problem
in the syslog.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Acked-by: NDavid Rientjes <rientjes@google.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Do not use kmalloc() but kmem_cache_alloc() for the allocation
of the kmem_cache structures in slub.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Acked-by: NDavid Rientjes <rientjes@google.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Add additional debugging to check that the objects is actually from the cache
the caller claims. Doing so currently trips up some other debugging code. It
takes a lot to infer from that what was happening.
Reviewed-by: NGlauber Costa <glommer@parallels.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
[ penberg@kernel.org: Use pr_err() ]
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Without this patch we can get (many) kmem trace events
with call site at krealloc().

This happens because krealloc is calling __krealloc,
which performs the allocation through kmalloc_track_caller.

Since neither krealloc nor __krealloc are marked inline explicitly,
the caller can be traced as being krealloc, which clearly is not
the intended behavior.

This patch allows to get the real caller of krealloc, by creating
an always inlined function __do_krealloc, thus tracing the
call site accurately.
Acked-by: NChristoph Lameter <cl@linux.com>
Cc: Glauber Costa <glommer@parallels.com>
Signed-off-by: NEzequiel Garcia <elezegarcia@gmail.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

page_get_cache() isn't called from anything, so remove it.
Signed-off-by: NDavid Rientjes <rientjes@google.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

kmem_cache_create() does cache integrity checks when CONFIG_DEBUG_VM is
defined.  These checks interspersed with the regular code path has lead
to compile time warnings when compiled without CONFIG_DEBUG_VM defined.
Restructuring the code to move the integrity checks in to a new function
would eliminate the current compile warning problem and also will allow
for future changes to the debug only code to evolve without introducing
new warnings in the regular path.

This restructuring work is based on the discussion in the following
thread:

https://lkml.org/lkml/2012/7/13/424

[akpm@linux-foundation.org: fix build, cleanup]
Signed-off-by: NShuah Khan <shuah.khan@hp.com>
Acked-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

This reverts commit 455ce9eb. Andrew
sent a better version.
Signed-off-by: NPekka Enberg <penberg@kernel.org>

In current implementation, after unfreezing, we doesn't touch oldpage,
so it remain 'NOT NULL'. When we call this_cpu_cmpxchg()
with this old oldpage, this_cpu_cmpxchg() is mostly be failed.

We can change value of oldpage to NULL after unfreezing,
because unfreeze_partial() ensure that all the cpu partial slabs is removed
from cpu partial list. In this time, we could expect that
this_cpu_cmpxchg is mostly succeed.
Acked-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NJoonsoo Kim <js1304@gmail.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Only applies to scenarios where debugging is on:

Validation of slabs can currently occur while debugging
information is updated from the fast paths of the allocator.
This results in various races where we get false reports about
slab metadata not being in order.

This patch makes the fast paths take the node lock so that
serialization with slab validation will occur. Causes additional
slowdown in debug scenarios.
Reported-by: NWaiman Long <Waiman.Long@hp.com>
Signed-off-by: NChristoph Lameter <cl@linux.com>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Eliminate an ifdef and a label by moving all the CONFIG_DEBUG_VM checking
inside the locked region.

Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

page_get_cache() does not need to call compound_head(), as its unique
caller virt_to_slab() already makes sure to return a head page.

Additionally, removing the compound_head() call makes page_get_cache()
consistent with page_get_slab().
Signed-off-by: NMichel Lespinasse <walken@google.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Pekka Enberg <penberg@kernel.org>
Acked-by: NDavid Rientjes <rientjes@google.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

When freeing objects, the slub allocator will most of the time free
empty pages by calling __free_pages(). But high-order kmalloc will be
diposed by means of put_page() instead. It makes no sense to call
put_page() in kernel pages that are provided by the object allocators,
so we shouldn't be doing this ourselves. Aside from the consistency
change, we don't change the flow too much. put_page()'s would call its
dtor function, which is __free_pages. We also already do all of the
Compound page tests ourselves, and the Mlock test we lose don't really
matter.
Signed-off-by: NGlauber Costa <glommer@parallels.com>
Acked-by: NChristoph Lameter <cl@linux.com>
CC: David Rientjes <rientjes@google.com>
CC: Pekka Enberg <penberg@kernel.org>
Signed-off-by: NPekka Enberg <penberg@kernel.org>

Borislav Petkov reports that the new warning added in commit
88fdf75d ("mm: warn if pg_data_t isn't initialized with zero")
triggers for him, and it is the node_start_pfn field that has already
been initialized once.

The call trace looks like this:

  x86_64_start_kernel ->
    x86_64_start_reservations ->
    start_kernel ->
    setup_arch ->
    paging_init ->
    zone_sizes_init ->
    free_area_init_nodes ->
    free_area_init_node

and (with the warning replaced by debug output), Borislav sees

  On node 0 totalpages: 4193848
    DMA zone: 64 pages used for memmap
    DMA zone: 6 pages reserved
    DMA zone: 3890 pages, LIFO batch:0
    DMA32 zone: 16320 pages used for memmap
    DMA32 zone: 798464 pages, LIFO batch:31
    Normal zone: 52736 pages used for memmap
    Normal zone: 3322368 pages, LIFO batch:31
  free_area_init_node: pgdat->node_start_pfn: 4423680      <----
  On node 1 totalpages: 4194304
    Normal zone: 65536 pages used for memmap
    Normal zone: 4128768 pages, LIFO batch:31
  free_area_init_node: pgdat->node_start_pfn: 8617984      <----
  On node 2 totalpages: 4194304
    Normal zone: 65536 pages used for memmap
    Normal zone: 4128768 pages, LIFO batch:31
  free_area_init_node: pgdat->node_start_pfn: 12812288     <----
  On node 3 totalpages: 4194304
    Normal zone: 65536 pages used for memmap
    Normal zone: 4128768 pages, LIFO batch:31

so remove the bogus warning for now to avoid annoying people.  Minchan
Kim is looking at it.
Reported-by: NBorislav Petkov <bp@amd64.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

If a process creates a large hugetlbfs mapping that is eligible for page
table sharing and forks heavily with children some of whom fault and
others which destroy the mapping then it is possible for page tables to
get corrupted.  Some teardowns of the mapping encounter a "bad pmd" and
output a message to the kernel log.  The final teardown will trigger a
BUG_ON in mm/filemap.c.

This was reproduced in 3.4 but is known to have existed for a long time
and goes back at least as far as 2.6.37.  It was probably was introduced
in 2.6.20 by [39dde65c: shared page table for hugetlb page].  The messages
look like this;

[  ..........] Lots of bad pmd messages followed by this
[  127.164256] mm/memory.c:391: bad pmd ffff880412e04fe8(80000003de4000e7).
[  127.164257] mm/memory.c:391: bad pmd ffff880412e04ff0(80000003de6000e7).
[  127.164258] mm/memory.c:391: bad pmd ffff880412e04ff8(80000003de0000e7).
[  127.186778] ------------[ cut here ]------------
[  127.186781] kernel BUG at mm/filemap.c:134!
[  127.186782] invalid opcode: 0000 [#1] SMP
[  127.186783] CPU 7
[  127.186784] Modules linked in: af_packet cpufreq_conservative cpufreq_userspace cpufreq_powersave acpi_cpufreq mperf ext3 jbd dm_mod coretemp crc32c_intel usb_storage ghash_clmulni_intel aesni_intel i2c_i801 r8169 mii uas sr_mod cdrom sg iTCO_wdt iTCO_vendor_support shpchp serio_raw cryptd aes_x86_64 e1000e pci_hotplug dcdbas aes_generic container microcode ext4 mbcache jbd2 crc16 sd_mod crc_t10dif i915 drm_kms_helper drm i2c_algo_bit ehci_hcd ahci libahci usbcore rtc_cmos usb_common button i2c_core intel_agp video intel_gtt fan processor thermal thermal_sys hwmon ata_generic pata_atiixp libata scsi_mod
[  127.186801]
[  127.186802] Pid: 9017, comm: hugetlbfs-test Not tainted 3.4.0-autobuild #53 Dell Inc. OptiPlex 990/06D7TR
[  127.186804] RIP: 0010:[<ffffffff810ed6ce>]  [<ffffffff810ed6ce>] __delete_from_page_cache+0x15e/0x160
[  127.186809] RSP: 0000:ffff8804144b5c08  EFLAGS: 00010002
[  127.186810] RAX: 0000000000000001 RBX: ffffea000a5c9000 RCX: 00000000ffffffc0
[  127.186811] RDX: 0000000000000000 RSI: 0000000000000009 RDI: ffff88042dfdad00
[  127.186812] RBP: ffff8804144b5c18 R08: 0000000000000009 R09: 0000000000000003
[  127.186813] R10: 0000000000000000 R11: 000000000000002d R12: ffff880412ff83d8
[  127.186814] R13: ffff880412ff83d8 R14: 0000000000000000 R15: ffff880412ff83d8
[  127.186815] FS:  00007fe18ed2c700(0000) GS:ffff88042dce0000(0000) knlGS:0000000000000000
[  127.186816] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[  127.186817] CR2: 00007fe340000503 CR3: 0000000417a14000 CR4: 00000000000407e0
[  127.186818] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  127.186819] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[  127.186820] Process hugetlbfs-test (pid: 9017, threadinfo ffff8804144b4000, task ffff880417f803c0)
[  127.186821] Stack:
[  127.186822]  ffffea000a5c9000 0000000000000000 ffff8804144b5c48 ffffffff810ed83b
[  127.186824]  ffff8804144b5c48 000000000000138a 0000000000001387 ffff8804144b5c98
[  127.186825]  ffff8804144b5d48 ffffffff811bc925 ffff8804144b5cb8 0000000000000000
[  127.186827] Call Trace:
[  127.186829]  [<ffffffff810ed83b>] delete_from_page_cache+0x3b/0x80
[  127.186832]  [<ffffffff811bc925>] truncate_hugepages+0x115/0x220
[  127.186834]  [<ffffffff811bca43>] hugetlbfs_evict_inode+0x13/0x30
[  127.186837]  [<ffffffff811655c7>] evict+0xa7/0x1b0
[  127.186839]  [<ffffffff811657a3>] iput_final+0xd3/0x1f0
[  127.186840]  [<ffffffff811658f9>] iput+0x39/0x50
[  127.186842]  [<ffffffff81162708>] d_kill+0xf8/0x130
[  127.186843]  [<ffffffff81162812>] dput+0xd2/0x1a0
[  127.186845]  [<ffffffff8114e2d0>] __fput+0x170/0x230
[  127.186848]  [<ffffffff81236e0e>] ? rb_erase+0xce/0x150
[  127.186849]  [<ffffffff8114e3ad>] fput+0x1d/0x30
[  127.186851]  [<ffffffff81117db7>] remove_vma+0x37/0x80
[  127.186853]  [<ffffffff81119182>] do_munmap+0x2d2/0x360
[  127.186855]  [<ffffffff811cc639>] sys_shmdt+0xc9/0x170
[  127.186857]  [<ffffffff81410a39>] system_call_fastpath+0x16/0x1b
[  127.186858] Code: 0f 1f 44 00 00 48 8b 43 08 48 8b 00 48 8b 40 28 8b b0 40 03 00 00 85 f6 0f 88 df fe ff ff 48 89 df e8 e7 cb 05 00 e9 d2 fe ff ff <0f> 0b 55 83 e2 fd 48 89 e5 48 83 ec 30 48 89 5d d8 4c 89 65 e0
[  127.186868] RIP  [<ffffffff810ed6ce>] __delete_from_page_cache+0x15e/0x160
[  127.186870]  RSP <ffff8804144b5c08>
[  127.186871] ---[ end trace 7cbac5d1db69f426 ]---

The bug is a race and not always easy to reproduce.  To reproduce it I was
doing the following on a single socket I7-based machine with 16G of RAM.

$ hugeadm --pool-pages-max DEFAULT:13G
$ echo $((18*1048576*1024)) > /proc/sys/kernel/shmmax
$ echo $((18*1048576*1024)) > /proc/sys/kernel/shmall
$ for i in `seq 1 9000`; do ./hugetlbfs-test; done

On my particular machine, it usually triggers within 10 minutes but
enabling debug options can change the timing such that it never hits.
Once the bug is triggered, the machine is in trouble and needs to be
rebooted.  The machine will respond but processes accessing proc like "ps
aux" will hang due to the BUG_ON.  shutdown will also hang and needs a
hard reset or a sysrq-b.

The basic problem is a race between page table sharing and teardown.  For
the most part page table sharing depends on i_mmap_mutex.  In some cases,
it is also taking the mm->page_table_lock for the PTE updates but with
shared page tables, it is the i_mmap_mutex that is more important.

Unfortunately it appears to be also insufficient. Consider the following
situation

Process A					Process B
---------					---------
hugetlb_fault					shmdt
  						LockWrite(mmap_sem)
    						  do_munmap
						    unmap_region
						      unmap_vmas
						        unmap_single_vma
						          unmap_hugepage_range
      						            Lock(i_mmap_mutex)
							    Lock(mm->page_table_lock)
							    huge_pmd_unshare/unmap tables <--- (1)
							    Unlock(mm->page_table_lock)
      						            Unlock(i_mmap_mutex)
  huge_pte_alloc				      ...
    Lock(i_mmap_mutex)				      ...
    vma_prio_walk, find svma, spte		      ...
    Lock(mm->page_table_lock)			      ...
    share spte					      ...
    Unlock(mm->page_table_lock)			      ...
    Unlock(i_mmap_mutex)			      ...
  hugetlb_no_page									  <--- (2)
						      free_pgtables
						        unlink_file_vma
							hugetlb_free_pgd_range
						    remove_vma_list

In this scenario, it is possible for Process A to share page tables with
Process B that is trying to tear them down.  The i_mmap_mutex on its own
does not prevent Process A walking Process B's page tables.  At (1) above,
the page tables are not shared yet so it unmaps the PMDs.  Process A sets
up page table sharing and at (2) faults a new entry.  Process B then trips
up on it in free_pgtables.

This patch fixes the problem by adding a new function
__unmap_hugepage_range_final that is only called when the VMA is about to
be destroyed.  This function clears VM_MAYSHARE during
unmap_hugepage_range() under the i_mmap_mutex.  This makes the VMA
ineligible for sharing and avoids the race.  Superficially this looks like
it would then be vunerable to truncate and madvise issues but hugetlbfs
has its own truncate handlers so does not use unmap_mapping_range() and
does not support madvise(DONTNEED).

This should be treated as a -stable candidate if it is merged.

Test program is as follows. The test case was mostly written by Michal
Hocko with a few minor changes to reproduce this bug.

==== CUT HERE ====

static size_t huge_page_size = (2UL << 20);
static size_t nr_huge_page_A = 512;
static size_t nr_huge_page_B = 5632;

unsigned int get_random(unsigned int max)
{
	struct timeval tv;

	gettimeofday(&tv, NULL);
	srandom(tv.tv_usec);
	return random() % max;
}

static void play(void *addr, size_t size)
{
	unsigned char *start = addr,
		      *end = start + size,
		      *a;
	start += get_random(size/2);

	/* we could itterate on huge pages but let's give it more time. */
	for (a = start; a < end; a += 4096)
		*a = 0;
}

int main(int argc, char **argv)
{
	key_t key = IPC_PRIVATE;
	size_t sizeA = nr_huge_page_A * huge_page_size;
	size_t sizeB = nr_huge_page_B * huge_page_size;
	int shmidA, shmidB;
	void *addrA = NULL, *addrB = NULL;
	int nr_children = 300, n = 0;

	if ((shmidA = shmget(key, sizeA, IPC_CREAT|SHM_HUGETLB|0660)) == -1) {
		perror("shmget:");
		return 1;
	}

	if ((addrA = shmat(shmidA, addrA, SHM_R|SHM_W)) == (void *)-1UL) {
		perror("shmat");
		return 1;
	}
	if ((shmidB = shmget(key, sizeB, IPC_CREAT|SHM_HUGETLB|0660)) == -1) {
		perror("shmget:");
		return 1;
	}

	if ((addrB = shmat(shmidB, addrB, SHM_R|SHM_W)) == (void *)-1UL) {
		perror("shmat");
		return 1;
	}

fork_child:
	switch(fork()) {
		case 0:
			switch (n%3) {
			case 0:
				play(addrA, sizeA);
				break;
			case 1:
				play(addrB, sizeB);
				break;
			case 2:
				break;
			}
			break;
		case -1:
			perror("fork:");
			break;
		default:
			if (++n < nr_children)
				goto fork_child;
			play(addrA, sizeA);
			break;
	}
	shmdt(addrA);
	shmdt(addrB);
	do {
		wait(NULL);
	} while (--n > 0);
	shmctl(shmidA, IPC_RMID, NULL);
	shmctl(shmidB, IPC_RMID, NULL);
	return 0;
}

[akpm@linux-foundation.org: name the declaration's args, fix CONFIG_HUGETLBFS=n build]
Signed-off-by: NHugh Dickins <hughd@google.com>
Reviewed-by: NMichal Hocko <mhocko@suse.cz>
Signed-off-by: NMel Gorman <mgorman@suse.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

When tmpfs has the interleave memory policy, it always starts allocating
for each file from node 0 at offset 0.  When there are many small files,
the lower nodes fill up disproportionately.

This patch spreads out node usage by starting files at nodes other than 0,
by using the inode number to bias the starting node for interleave.
Signed-off-by: NNathan Zimmer <nzimmer@sgi.com>
Signed-off-by: NHugh Dickins <hughd@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

pg_data_t is zeroed before reaching free_area_init_core(), so remove the
now unnecessary initializations.
Signed-off-by: NMinchan Kim <minchan@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Warn if memory-hotplug/boot code doesn't initialize pg_data_t with zero
when it is allocated.  Arch code and memory hotplug already initiailize
pg_data_t.  So this warning should never happen.  I select fields randomly
near the beginning, middle and end of pg_data_t for checking.

This patch isn't for performance but for removing initialization code
which is necessary to add whenever we adds new field to pg_data_t or zone.

Firstly, Andrew suggested clearing out of pg_data_t in MM core part but
Tejun doesn't like it because in the future, some archs can initialize
some fields in arch code and pass them into general MM part so blindly
clearing it out in mm core part would be very annoying.
Signed-off-by: NMinchan Kim <minchan@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

I noticed in a multi-process parallel files reading benchmark I ran on a 8
socket machine, throughput slowed down by a factor of 8 when I ran the
benchmark within a cgroup container.  I traced the problem to the
following code path (see below) when we are trying to reclaim memory from
file cache.  The res_counter_uncharge function is called on every page
that's reclaimed and created heavy lock contention.  The patch below
allows the reclaimed pages to be uncharged from the resource counter in
batch and recovered the regression.

Tim

     40.67%           usemem  [kernel.kallsyms]                   [k] _raw_spin_lock
                      |
                      --- _raw_spin_lock
                         |
                         |--92.61%-- res_counter_uncharge
                         |          |
                         |          |--100.00%-- __mem_cgroup_uncharge_common
                         |          |          |
                         |          |          |--100.00%-- mem_cgroup_uncharge_cache_page
                         |          |          |          __remove_mapping
                         |          |          |          shrink_page_list
                         |          |          |          shrink_inactive_list
                         |          |          |          shrink_mem_cgroup_zone
                         |          |          |          shrink_zone
                         |          |          |          do_try_to_free_pages
                         |          |          |          try_to_free_pages
                         |          |          |          __alloc_pages_nodemask
                         |          |          |          alloc_pages_current
Signed-off-by: NTim Chen <tim.c.chen@linux.intel.com>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

sparse_index_init() uses the index_init_lock spinlock to protect root
mem_section assignment.  The lock is not necessary anymore because the
function is called only during boot (during paging init which is executed
only from a single CPU) and from the hotplug code (by add_memory() via
arch_add_memory()) which uses mem_hotplug_mutex.

The lock was introduced by 28ae55c9 ("sparsemem extreme: hotplug
preparation") and sparse_index_init() was used only during boot at that
time.

Later when the hotplug code (and add_memory()) was introduced there was no
synchronization so it was possible to online more sections from the same
root probably (though I am not 100% sure about that).  The first
synchronization has been added by 6ad696d2 ("mm: allow memory hotplug and
hibernation in the same kernel") which was later replaced by the
mem_hotplug_mutex - 20d6c96b ("mem-hotplug: introduce
{un}lock_memory_hotplug()").

Let's remove the lock as it is not needed and it makes the code more
confusing.

[mhocko@suse.cz: changelog]
Signed-off-by: NGavin Shan <shangw@linux.vnet.ibm.com>
Reviewed-by: NMichal Hocko <mhocko@suse.cz>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

__section_nr() was implemented to retrieve the corresponding memory
section number according to its descriptor.  It's possible that the
specified memory section descriptor doesn't exist in the global array.  So
add more checking on that and report an error for a wrong case.
Signed-off-by: NGavin Shan <shangw@linux.vnet.ibm.com>
Acked-by: NDavid Rientjes <rientjes@google.com>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

With CONFIG_SPARSEMEM_EXTREME, the two levels of memory section
descriptors are allocated from slab or bootmem.  When allocating from
slab, let slab/bootmem allocator clear the memory chunk.  We needn't clear
it explicitly.
Signed-off-by: NGavin Shan <shangw@linux.vnet.ibm.com>
Reviewed-by: NMichal Hocko <mhocko@suse.cz>
Acked-by: NDavid Rientjes <rientjes@google.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add a mem_cgroup_from_css() helper to replace open-coded invokations of
container_of().  To clarify the code and to add a little more type safety.

[akpm@linux-foundation.org: fix extensive breakage]
Signed-off-by: NWanpeng Li <liwanp@linux.vnet.ibm.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Gavin Shan <shangw@linux.vnet.ibm.com>
Cc: Wanpeng Li <liwanp@linux.vnet.ibm.com>
Cc: Gavin Shan <shangw@linux.vnet.ibm.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The may_enter_fs test turns out to be too restrictive: though I saw no
problem with it when testing on 3.5-rc6, it very soon OOMed when I tested
on 3.5-rc6-mm1.  I don't know what the difference there is, perhaps I just
slightly changed the way I started off the testing: dd if=/dev/zero
of=/mnt/temp bs=1M count=1024; rm -f /mnt/temp; sync repeatedly, in 20M
memory.limit_in_bytes cgroup to ext4 on USB stick.

ext4 (and gfs2 and xfs) turn out to allocate new pages for writing with
AOP_FLAG_NOFS: that seems a little worrying, and it's unclear to me why
the transaction needs to be started even before allocating pagecache
memory.  But it may not be worth worrying about these days: if direct
reclaim avoids FS writeback, does __GFP_FS now mean anything?

Anyway, we insisted on the may_enter_fs test to avoid hangs with the loop
device; but since that also masks off __GFP_IO, we can test for __GFP_IO
directly, ignoring may_enter_fs and __GFP_FS.

But even so, the test still OOMs sometimes: when originally testing on
3.5-rc6, it OOMed about one time in five or ten; when testing just now on
3.5-rc6-mm1, it OOMed on the first iteration.

This residual problem comes from an accumulation of pages under ordinary
writeback, not marked PageReclaim, so rightly not causing the memcg check
to wait on their writeback: these too can prevent shrink_page_list() from
freeing any pages, so many times that memcg reclaim fails and OOMs.

Deal with these in the same way as direct reclaim now deals with dirty FS
pages: mark them PageReclaim.  It is appropriate to rotate these to tail
of list when writepage completes, but more importantly, the PageReclaim
flag makes memcg reclaim wait on them if encountered again.  Increment
NR_VMSCAN_IMMEDIATE?  That's arguable: I chose not.

Setting PageReclaim here may occasionally race with end_page_writeback()
clearing it: lru_deactivate_fn() already faced the same race, and
correctly concluded that the window is small and the issue non-critical.

With these changes, the test runs indefinitely without OOMing on ext4,
ext3 and ext2: I'll move on to test with other filesystems later.

Trivia: invert conditions for a clearer block without an else, and goto
keep_locked to do the unlock_page.
Signed-off-by: NHugh Dickins <hughd@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujtisu.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Ying Han <yinghan@google.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: <stable@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The current implementation of dirty pages throttling is not memcg aware
which makes it easy to have memcg LRUs full of dirty pages.  Without
throttling, these LRUs can be scanned faster than the rate of writeback,
leading to memcg OOM conditions when the hard limit is small.

This patch fixes the problem by throttling the allocating process
(possibly a writer) during the hard limit reclaim by waiting on
PageReclaim pages.  We are waiting only for PageReclaim pages because
those are the pages that made one full round over LRU and that means that
the writeback is much slower than scanning.

The solution is far from being ideal - long term solution is memcg aware
dirty throttling - but it is meant to be a band aid until we have a real
fix.  We are seeing this happening during nightly backups which are placed
into containers to prevent from eviction of the real working set.

The change affects only memcg reclaim and only when we encounter
PageReclaim pages which is a signal that the reclaim doesn't catch up on
with the writers so somebody should be throttled.  This could be
potentially unfair because it could be somebody else from the group who
gets throttled on behalf of the writer but as writers need to allocate as
well and they allocate in higher rate the probability that only innocent
processes would be penalized is not that high.

I have tested this change by a simple dd copying /dev/zero to tmpfs or
ext3 running under small memcg (1G copy under 5M, 60M, 300M and 2G
containers) and dd got killed by OOM killer every time.  With the patch I
could run the dd with the same size under 5M controller without any OOM.
The issue is more visible with slower devices for output.

* With the patch
================
* tmpfs size=2G
---------------
$ vim cgroup_cache_oom_test.sh
$ ./cgroup_cache_oom_test.sh 5M
using Limit 5M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 30.4049 s, 34.5 MB/s
$ ./cgroup_cache_oom_test.sh 60M
using Limit 60M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 31.4561 s, 33.3 MB/s
$ ./cgroup_cache_oom_test.sh 300M
using Limit 300M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 20.4618 s, 51.2 MB/s
$ ./cgroup_cache_oom_test.sh 2G
using Limit 2G for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 1.42172 s, 738 MB/s

* ext3
------
$ ./cgroup_cache_oom_test.sh 5M
using Limit 5M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 27.9547 s, 37.5 MB/s
$ ./cgroup_cache_oom_test.sh 60M
using Limit 60M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 30.3221 s, 34.6 MB/s
$ ./cgroup_cache_oom_test.sh 300M
using Limit 300M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 24.5764 s, 42.7 MB/s
$ ./cgroup_cache_oom_test.sh 2G
using Limit 2G for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 3.35828 s, 312 MB/s

* Without the patch
===================
* tmpfs size=2G
---------------
$ ./cgroup_cache_oom_test.sh 5M
using Limit 5M for group
./cgroup_cache_oom_test.sh: line 46:  4668 Killed                  dd if=/dev/zero of=$OUT/zero bs=1M count=$count
$ ./cgroup_cache_oom_test.sh 60M
using Limit 60M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 25.4989 s, 41.1 MB/s
$ ./cgroup_cache_oom_test.sh 300M
using Limit 300M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 24.3928 s, 43.0 MB/s
$ ./cgroup_cache_oom_test.sh 2G
using Limit 2G for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 1.49797 s, 700 MB/s

* ext3
------
$ ./cgroup_cache_oom_test.sh 5M
using Limit 5M for group
./cgroup_cache_oom_test.sh: line 46:  4689 Killed                  dd if=/dev/zero of=$OUT/zero bs=1M count=$count
$ ./cgroup_cache_oom_test.sh 60M
using Limit 60M for group
./cgroup_cache_oom_test.sh: line 46:  4692 Killed                  dd if=/dev/zero of=$OUT/zero bs=1M count=$count
$ ./cgroup_cache_oom_test.sh 300M
using Limit 300M for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 20.248 s, 51.8 MB/s
$ ./cgroup_cache_oom_test.sh 2G
using Limit 2G for group
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 2.85201 s, 368 MB/s

[akpm@linux-foundation.org: tweak changelog, reordered the test to optimize for CONFIG_CGROUP_MEM_RES_CTLR=n]
[hughd@google.com: fix deadlock with loop driver]
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujtisu.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Ying Han <yinghan@google.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Reviewed-by: NMel Gorman <mgorman@suse.de>
Acked-by: NJohannes Weiner <hannes@cmpxchg.org>
Reviewed-by: NFengguang Wu <fengguang.wu@intel.com>
Signed-off-by: NMichal Hocko <mhocko@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

mmu_notifier_release() is called when the process is exiting.  It will
delete all the mmu notifiers.  But at this time the page belonging to the
process is still present in page tables and is present on the LRU list, so
this race will happen:

      CPU 0                 CPU 1
mmu_notifier_release:    try_to_unmap:
   hlist_del_init_rcu(&mn->hlist);
                            ptep_clear_flush_notify:
                                  mmu nofifler not found
                            free page  !!!!!!
                            /*
                             * At the point, the page has been
                             * freed, but it is still mapped in
                             * the secondary MMU.
                             */

  mn->ops->release(mn, mm);

Then the box is not stable and sometimes we can get this bug:

[  738.075923] BUG: Bad page state in process migrate-perf  pfn:03bec
[  738.075931] page:ffffea00000efb00 count:0 mapcount:0 mapping:          (null) index:0x8076
[  738.075936] page flags: 0x20000000000014(referenced|dirty)

The same issue is present in mmu_notifier_unregister().

We can call ->release before deleting the notifier to ensure the page has
been unmapped from the secondary MMU before it is freed.
Signed-off-by: NXiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Avi Kivity <avi@redhat.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

shmem knows for sure that the page is in swap cache when attempting to
charge a page, because the cache charge entry function has a check for it.
Only anon pages may be removed from swap cache already when trying to
charge their swapin.

Adjust the comment, though: '4969c119 mm: fix swapin race condition' added
a stable PageSwapCache check under the page lock in the do_swap_page()
before calling the memory controller, so it's unuse_pte()'s pte_same()
that may fail.
Signed-off-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Only anon and shmem pages in the swap cache are attempted to be charged
multiple times, from every swap pte fault or from shmem_unuse().  No other
pages require checking PageCgroupUsed().

Charging pages in the swap cache is also serialized by the page lock, and
since both the try_charge and commit_charge are called under the same page
lock section, the PageCgroupUsed() check might as well happen before the
counter charging, let alone reclaim.
Signed-off-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

When shmem is charged upon swapin, it does not need to check twice whether
the memory controller is enabled.

Also, shmem pages do not have to be checked for everything that regular
anon pages have to be checked for, so let shmem use the internal version
directly and allow future patches to move around checks that are only
required when swapping in anon pages.
Signed-off-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

It does not matter to __mem_cgroup_try_charge() if the passed mm is NULL
or init_mm, it will charge the root memcg in either case.

Also fix up the comment in __mem_cgroup_try_charge() that claimed the
init_mm would be charged when no mm was passed.  It's not really
incorrect, but confusing.  Clarify that the root memcg is charged in this
case.
Signed-off-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

shmem page charges have not needed a separate charge type to tell them
from regular file pages since 08e552c6 ("memcg: synchronized LRU").
Signed-off-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Charging cache pages may require swapin in the shmem case.  Save the
forward declaration and just move the swapin functions above the cache
charging functions.
Signed-off-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Only anon pages that are uncharged at the time of the last page table
mapping vanishing may be in swapcache.

When shmem pages, file pages, swap-freed anon pages, or just migrated
pages are uncharged, they are known for sure to be not in swapcache.
Signed-off-by: NJohannes Weiner <hannes@cmpxchg.org>
Acked-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: NMichal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>