Ganapatrao Kulkarni reported that the LTP test cpuset01 in stress mode
triggers OOM killer in few seconds, despite lots of free memory. The
test attempts to repeatedly fault in memory in one process in a cpuset,
while changing allowed nodes of the cpuset between 0 and 1 in another
process.
The problem comes from insufficient protection against cpuset changes,
which can cause get_page_from_freelist() to consider all zones as
non-eligible due to nodemask and/or current->mems_allowed. This was
masked in the past by sufficient retries, but since commit 682a3385e7
("mm, page_alloc: inline the fast path of the zonelist iterator") we fix
the preferred_zoneref once, and don't iterate over the whole zonelist in
further attempts, thus the only eligible zones might be placed in the
zonelist before our starting point and we always miss them.
A previous patch fixed this problem for current->mems_allowed. However,
cpuset changes also update the task's mempolicy nodemask. The fix has
two parts. We have to repeat the preferred_zoneref search when we
detect cpuset update by way of seqcount, and we have to check the
seqcount before considering OOM.
[akpm@linux-foundation.org: fix typo in comment]
Link: http://lkml.kernel.org/r/20170120103843.24587-5-vbabka@suse.cz
Fixes: c33d6c06f6 ("mm, page_alloc: avoid looking up the first zone in a zonelist twice")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Ganapatrao Kulkarni <gpkulkarni@gmail.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is a preparation for the following patch to make review simpler.
While the primary motivation is a bug fix, this also simplifies the fast
path, although the moved code is only enabled when cpusets are in use.
Link: http://lkml.kernel.org/r/20170120103843.24587-4-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Ganapatrao Kulkarni <gpkulkarni@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Ganapatrao Kulkarni reported that the LTP test cpuset01 in stress mode
triggers OOM killer in few seconds, despite lots of free memory. The
test attempts to repeatedly fault in memory in one process in a cpuset,
while changing allowed nodes of the cpuset between 0 and 1 in another
process.
One possible cause is that in the fast path we find the preferred
zoneref according to current mems_allowed, so that it points to the
middle of the zonelist, skipping e.g. zones of node 1 completely. If
the mems_allowed is updated to contain only node 1, we never reach it in
the zonelist, and trigger OOM before checking the cpuset_mems_cookie.
This patch fixes the particular case by redoing the preferred zoneref
search if we switch back to the original nodemask. The condition is
also slightly changed so that when the last non-root cpuset is removed,
we don't miss it.
Note that this is not a full fix, and more patches will follow.
Link: http://lkml.kernel.org/r/20170120103843.24587-3-vbabka@suse.cz
Fixes: 682a3385e7 ("mm, page_alloc: inline the fast path of the zonelist iterator")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Ganapatrao Kulkarni <gpkulkarni@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "fix premature OOM regression in 4.7+ due to cpuset races".
This is v2 of my attempt to fix the recent report based on LTP cpuset
stress test [1]. The intention is to go to stable 4.9 LTSS with this,
as triggering repeated OOMs is not nice. That's why the patches try to
be not too intrusive.
Unfortunately why investigating I found that modifying the testcase to
use per-VMA policies instead of per-task policies will bring the OOM's
back, but that seems to be much older and harder to fix problem. I have
posted a RFC [2] but I believe that fixing the recent regressions has a
higher priority.
Longer-term we might try to think how to fix the cpuset mess in a better
and less error prone way. I was for example very surprised to learn,
that cpuset updates change not only task->mems_allowed, but also
nodemask of mempolicies. Until now I expected the parameter to
alloc_pages_nodemask() to be stable. I wonder why do we then treat
cpusets specially in get_page_from_freelist() and distinguish HARDWALL
etc, when there's unconditional intersection between mempolicy and
cpuset. I would expect the nodemask adjustment for saving overhead in
g_p_f(), but that clearly doesn't happen in the current form. So we
have both crazy complexity and overhead, AFAICS.
[1] https://lkml.kernel.org/r/CAFpQJXUq-JuEP=QPidy4p_=FN0rkH5Z-kfB4qBvsf6jMS87Edg@mail.gmail.com
[2] https://lkml.kernel.org/r/7c459f26-13a6-a817-e508-b65b903a8378@suse.cz
This patch (of 4):
Since commit c33d6c06f6 ("mm, page_alloc: avoid looking up the first
zone in a zonelist twice") we have a wrong check for NULL preferred_zone,
which can theoretically happen due to concurrent cpuset modification. We
check the zoneref pointer which is never NULL and we should check the zone
pointer. Also document this in first_zones_zonelist() comment per Michal
Hocko.
Fixes: c33d6c06f6 ("mm, page_alloc: avoid looking up the first zone in a zonelist twice")
Link: http://lkml.kernel.org/r/20170120103843.24587-2-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Ganapatrao Kulkarni <gpkulkarni@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since commit be97a41b29 ("mm/mempolicy.c: merge alloc_hugepage_vma to
alloc_pages_vma") alloc_pages_vma() can potentially free a mempolicy by
mpol_cond_put() before accessing the embedded nodemask by
__alloc_pages_nodemask(). The commit log says it's so "we can use a
single exit path within the function" but that's clearly wrong. We can
still do that when doing mpol_cond_put() after the allocation attempt.
Make sure the mempolicy is not freed prematurely, otherwise
__alloc_pages_nodemask() can end up using a bogus nodemask, which could
lead e.g. to premature OOM.
Fixes: be97a41b29 ("mm/mempolicy.c: merge alloc_hugepage_vma to alloc_pages_vma")
Link: http://lkml.kernel.org/r/20170118141124.8345-1-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: <stable@vger.kernel.org> [4.0+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When memory.move_charge_at_immigrate is enabled and precharges are
depleted during move, mem_cgroup_move_charge_pte_range() will attempt to
increase the size of the precharge.
Prevent precharges from ever looping by setting __GFP_NORETRY. This was
probably the intention of the GFP_KERNEL & ~__GFP_NORETRY, which is
pointless as written.
Fixes: 0029e19ebf ("mm: memcontrol: remove explicit OOM parameter in charge path")
Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1701130208510.69402@chino.kir.corp.google.com
Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 73e64c51af ("mm, compaction: allow compaction for GFP_NOFS
requests") changed compation to skip FS pages if not explicitly allowed
to touch them, but missed to update the CMA compact_control.
This leads to a very high isolation failure rate, crippling performance
of CMA even on a lightly loaded system. Re-allow CMA to compact FS
pages by setting the correct GFP flags, restoring CMA behavior and
performance to the kernel 4.9 level.
Fixes: 73e64c51af (mm, compaction: allow compaction for GFP_NOFS requests)
Link: http://lkml.kernel.org/r/20170113115155.24335-1-l.stach@pengutronix.de
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently when trace is enabled (e.g. slub_debug=T,kmalloc-128 ) the
trace messages are mostly output at KERN_INFO. However the trace code
also calls print_section() to hexdump the head of a free object. This
is hard coded to use KERN_ERR, meaning the console is deluged with trace
messages even if we've asked for quiet.
Fix this the obvious way but adding a level parameter to
print_section(), allowing calls from the trace code to use the same
trace level as other trace messages.
Link: http://lkml.kernel.org/r/20170113154850.518-1-daniel.thompson@linaro.org
Signed-off-by: Daniel Thompson <daniel.thompson@linaro.org>
Acked-by: Christoph Lameter <cl@linux.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In commit 19be0eaffa ("mm: remove gup_flags FOLL_WRITE games from
__get_user_pages()"), the mm code was changed from unsetting FOLL_WRITE
after a COW was resolved to setting the (newly introduced) FOLL_COW
instead. Simultaneously, the check in gup.c was updated to still allow
writes with FOLL_FORCE set if FOLL_COW had also been set.
However, a similar check in huge_memory.c was forgotten. As a result,
remote memory writes to ro regions of memory backed by transparent huge
pages cause an infinite loop in the kernel (handle_mm_fault sets
FOLL_COW and returns 0 causing a retry, but follow_trans_huge_pmd bails
out immidiately because `(flags & FOLL_WRITE) && !pmd_write(*pmd)` is
true.
While in this state the process is stil SIGKILLable, but little else
works (e.g. no ptrace attach, no other signals). This is easily
reproduced with the following code (assuming thp are set to always):
#include <assert.h>
#include <fcntl.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <unistd.h>
#define TEST_SIZE 5 * 1024 * 1024
int main(void) {
int status;
pid_t child;
int fd = open("/proc/self/mem", O_RDWR);
void *addr = mmap(NULL, TEST_SIZE, PROT_READ,
MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
assert(addr != MAP_FAILED);
pid_t parent_pid = getpid();
if ((child = fork()) == 0) {
void *addr2 = mmap(NULL, TEST_SIZE, PROT_READ | PROT_WRITE,
MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
assert(addr2 != MAP_FAILED);
memset(addr2, 'a', TEST_SIZE);
pwrite(fd, addr2, TEST_SIZE, (uintptr_t)addr);
return 0;
}
assert(child == waitpid(child, &status, 0));
assert(WIFEXITED(status) && WEXITSTATUS(status) == 0);
return 0;
}
Fix this by updating follow_trans_huge_pmd in huge_memory.c analogously
to the update in gup.c in the original commit. The same pattern exists
in follow_devmap_pmd. However, we should not be able to reach that
check with FOLL_COW set, so add WARN_ONCE to make sure we notice if we
ever do.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/20170106015025.GA38411@juliacomputing.com
Signed-off-by: Keno Fischer <keno@juliacomputing.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Willy Tarreau <w@1wt.eu>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
online_{kernel|movable} is used to change the memory zone to
ZONE_{NORMAL|MOVABLE} and online the memory.
To check that memory zone can be changed, zone_can_shift() is used.
Currently the function returns minus integer value, plus integer
value and 0. When the function returns minus or plus integer value,
it means that the memory zone can be changed to ZONE_{NORNAL|MOVABLE}.
But when the function returns 0, there are two meanings.
One of the meanings is that the memory zone does not need to be changed.
For example, when memory is in ZONE_NORMAL and onlined by online_kernel
the memory zone does not need to be changed.
Another meaning is that the memory zone cannot be changed. When memory
is in ZONE_NORMAL and onlined by online_movable, the memory zone may
not be changed to ZONE_MOVALBE due to memory online limitation(see
Documentation/memory-hotplug.txt). In this case, memory must not be
onlined.
The patch changes the return type of zone_can_shift() so that memory
online operation fails when memory zone cannot be changed as follows:
Before applying patch:
# grep -A 35 "Node 2" /proc/zoneinfo
Node 2, zone Normal
<snip>
node_scanned 0
spanned 8388608
present 7864320
managed 7864320
# echo online_movable > memory4097/state
# grep -A 35 "Node 2" /proc/zoneinfo
Node 2, zone Normal
<snip>
node_scanned 0
spanned 8388608
present 8388608
managed 8388608
online_movable operation succeeded. But memory is onlined as
ZONE_NORMAL, not ZONE_MOVABLE.
After applying patch:
# grep -A 35 "Node 2" /proc/zoneinfo
Node 2, zone Normal
<snip>
node_scanned 0
spanned 8388608
present 7864320
managed 7864320
# echo online_movable > memory4097/state
bash: echo: write error: Invalid argument
# grep -A 35 "Node 2" /proc/zoneinfo
Node 2, zone Normal
<snip>
node_scanned 0
spanned 8388608
present 7864320
managed 7864320
online_movable operation failed because of failure of changing
the memory zone from ZONE_NORMAL to ZONE_MOVABLE
Fixes: df429ac039 ("memory-hotplug: more general validation of zone during online")
Link: http://lkml.kernel.org/r/2f9c3837-33d7-b6e5-59c0-6ca4372b2d84@gmail.com
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Reviewed-by: Reza Arbab <arbab@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
return_unused_surplus_pages() decrements the global reservation count,
and frees any unused surplus pages that were backing the reservation.
Commit 7848a4bf51 ("mm/hugetlb.c: add cond_resched_lock() in
return_unused_surplus_pages()") added a call to cond_resched_lock in the
loop freeing the pages.
As a result, the hugetlb_lock could be dropped, and someone else could
use the pages that will be freed in subsequent iterations of the loop.
This could result in inconsistent global hugetlb page state, application
api failures (such as mmap) failures or application crashes.
When dropping the lock in return_unused_surplus_pages, make sure that
the global reservation count (resv_huge_pages) remains sufficiently
large to prevent someone else from claiming pages about to be freed.
Analyzed by Paul Cassella.
Fixes: 7848a4bf51 ("mm/hugetlb.c: add cond_resched_lock() in return_unused_surplus_pages()")
Link: http://lkml.kernel.org/r/1483991767-6879-1-git-send-email-mike.kravetz@oracle.com
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reported-by: Paul Cassella <cassella@cray.com>
Suggested-by: Michal Hocko <mhocko@kernel.org>
Cc: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: <stable@vger.kernel.org> [3.15+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch fixes a bug in the freelist randomization code. When a high
random number is used, the freelist will contain duplicate entries. It
will result in different allocations sharing the same chunk.
It will result in odd behaviours and crashes. It should be uncommon but
it depends on the machines. We saw it happening more often on some
machines (every few hours of running tests).
Fixes: c7ce4f60ac ("mm: SLAB freelist randomization")
Link: http://lkml.kernel.org/r/20170103181908.143178-1-thgarnie@google.com
Signed-off-by: John Sperbeck <jsperbeck@google.com>
Signed-off-by: Thomas Garnier <thgarnie@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
During developemnt for zram-swap asynchronous writeback, I found strange
corruption of compressed page, resulting in:
Modules linked in: zram(E)
CPU: 3 PID: 1520 Comm: zramd-1 Tainted: G E 4.8.0-mm1-00320-ge0d4894c9c38-dirty #3274
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
task: ffff88007620b840 task.stack: ffff880078090000
RIP: set_freeobj.part.43+0x1c/0x1f
RSP: 0018:ffff880078093ca8 EFLAGS: 00010246
RAX: 0000000000000018 RBX: ffff880076798d88 RCX: ffffffff81c408c8
RDX: 0000000000000018 RSI: 0000000000000000 RDI: 0000000000000246
RBP: ffff880078093cb0 R08: 0000000000000000 R09: 0000000000000000
R10: ffff88005bc43030 R11: 0000000000001df3 R12: ffff880076798d88
R13: 000000000005bc43 R14: ffff88007819d1b8 R15: 0000000000000001
FS: 0000000000000000(0000) GS:ffff88007e380000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fc934048f20 CR3: 0000000077b01000 CR4: 00000000000406e0
Call Trace:
obj_malloc+0x22b/0x260
zs_malloc+0x1e4/0x580
zram_bvec_rw+0x4cd/0x830 [zram]
page_requests_rw+0x9c/0x130 [zram]
zram_thread+0xe6/0x173 [zram]
kthread+0xca/0xe0
ret_from_fork+0x25/0x30
With investigation, it reveals currently stable page doesn't support
anonymous page. IOW, reuse_swap_page can reuse the page without waiting
writeback completion so it can overwrite page zram is compressing.
Unfortunately, zram has used per-cpu stream feature from v4.7.
It aims for increasing cache hit ratio of scratch buffer for
compressing. Downside of that approach is that zram should ask
memory space for compressed page in per-cpu context which requires
stricted gfp flag which could be failed. If so, it retries to
allocate memory space out of per-cpu context so it could get memory
this time and compress the data again, copies it to the memory space.
In this scenario, zram assumes the data should never be changed
but it is not true unless stable page supports. So, If the data is
changed under us, zram can make buffer overrun because second
compression size could be bigger than one we got in previous trial
and blindly, copy bigger size object to smaller buffer which is
buffer overrun. The overrun breaks zsmalloc free object chaining
so system goes crash like above.
I think below is same problem.
https://bugzilla.suse.com/show_bug.cgi?id=997574
Unfortunately, reuse_swap_page should be atomic so that we cannot wait on
writeback in there so the approach in this patch is simply return false if
we found it needs stable page. Although it increases memory footprint
temporarily, it happens rarely and it should be reclaimed easily althoug
it happened. Also, It would be better than waiting of IO completion,
which is critial path for application latency.
Fixes: da9556a236 ("zram: user per-cpu compression streams")
Link: http://lkml.kernel.org/r/20161120233015.GA14113@bbox
Link: http://lkml.kernel.org/r/1482366980-3782-2-git-send-email-minchan@kernel.org
Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Darrick J. Wong <darrick.wong@oracle.com>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Hyeoncheol Lee <cheol.lee@lge.com>
Cc: <yjay.kim@lge.com>
Cc: Sangseok Lee <sangseok.lee@lge.com>
Cc: <stable@vger.kernel.org> [4.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch does two things.
First it goes through and renames the __page_frag prefixed functions to
__page_frag_cache so that we can be clear that we are draining or
refilling the cache, not the frags themselves.
Second we drop the order parameter from __page_frag_cache_drain since we
don't actually need to pass it since all fragments are either order 0 or
must be a compound page.
Link: http://lkml.kernel.org/r/20170104023954.13451.5678.stgit@localhost.localdomain
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "Page fragment updates", v4.
This patch series takes care of a few cleanups for the page fragments
API.
First we do some renames so that things are much more consistent. First
we move the page_frag_ portion of the name to the front of the functions
names. Secondly we split out the cache specific functions from the
other page fragment functions by adding the word "cache" to the name.
Finally I added a bit of documentation that will hopefully help to
explain some of this. I plan to revisit this later as we get things
more ironed out in the near future with the changes planned for the DMA
setup to support eXpress Data Path.
This patch (of 3):
This patch renames the page frag functions to be more consistent with
other APIs. Specifically we place the name page_frag first in the name
and then have either an alloc or free call name that we append as the
suffix. This makes it a bit clearer in terms of naming.
In addition we drop the leading double underscores since we are
technically no longer a backing interface and instead the front end that
is called from the networking APIs.
Link: http://lkml.kernel.org/r/20170104023854.13451.67390.stgit@localhost.localdomain
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Nils Holland and Klaus Ethgen have reported unexpected OOM killer
invocations with 32b kernel starting with 4.8 kernels
kworker/u4:5 invoked oom-killer: gfp_mask=0x2400840(GFP_NOFS|__GFP_NOFAIL), nodemask=0, order=0, oom_score_adj=0
kworker/u4:5 cpuset=/ mems_allowed=0
CPU: 1 PID: 2603 Comm: kworker/u4:5 Not tainted 4.9.0-gentoo #2
[...]
Mem-Info:
active_anon:58685 inactive_anon:90 isolated_anon:0
active_file:274324 inactive_file:281962 isolated_file:0
unevictable:0 dirty:649 writeback:0 unstable:0
slab_reclaimable:40662 slab_unreclaimable:17754
mapped:7382 shmem:202 pagetables:351 bounce:0
free:206736 free_pcp:332 free_cma:0
Node 0 active_anon:234740kB inactive_anon:360kB active_file:1097296kB inactive_file:1127848kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:29528kB dirty:2596kB writeback:0kB shmem:0kB shmem_thp: 0kB shmem_pmdmapped: 184320kB anon_thp: 808kB writeback_tmp:0kB unstable:0kB pages_scanned:0 all_unreclaimable? no
DMA free:3952kB min:788kB low:984kB high:1180kB active_anon:0kB inactive_anon:0kB active_file:7316kB inactive_file:0kB unevictable:0kB writepending:96kB present:15992kB managed:15916kB mlocked:0kB slab_reclaimable:3200kB slab_unreclaimable:1408kB kernel_stack:0kB pagetables:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB
lowmem_reserve[]: 0 813 3474 3474
Normal free:41332kB min:41368kB low:51708kB high:62048kB active_anon:0kB inactive_anon:0kB active_file:532748kB inactive_file:44kB unevictable:0kB writepending:24kB present:897016kB managed:836248kB mlocked:0kB slab_reclaimable:159448kB slab_unreclaimable:69608kB kernel_stack:1112kB pagetables:1404kB bounce:0kB free_pcp:528kB local_pcp:340kB free_cma:0kB
lowmem_reserve[]: 0 0 21292 21292
HighMem free:781660kB min:512kB low:34356kB high:68200kB active_anon:234740kB inactive_anon:360kB active_file:557232kB inactive_file:1127804kB unevictable:0kB writepending:2592kB present:2725384kB managed:2725384kB mlocked:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:0kB bounce:0kB free_pcp:800kB local_pcp:608kB free_cma:0kB
the oom killer is clearly pre-mature because there there is still a lot
of page cache in the zone Normal which should satisfy this lowmem
request. Further debugging has shown that the reclaim cannot make any
forward progress because the page cache is hidden in the active list
which doesn't get rotated because inactive_list_is_low is not memcg
aware.
The code simply subtracts per-zone highmem counters from the respective
memcg's lru sizes which doesn't make any sense. We can simply end up
always seeing the resulting active and inactive counts 0 and return
false. This issue is not limited to 32b kernels but in practice the
effect on systems without CONFIG_HIGHMEM would be much harder to notice
because we do not invoke the OOM killer for allocations requests
targeting < ZONE_NORMAL.
Fix the issue by tracking per zone lru page counts in mem_cgroup_per_node
and subtract per-memcg highmem counts when memcg is enabled. Introduce
helper lruvec_zone_lru_size which redirects to either zone counters or
mem_cgroup_get_zone_lru_size when appropriate.
We are losing empty LRU but non-zero lru size detection introduced by
ca707239e8 ("mm: update_lru_size warn and reset bad lru_size") because
of the inherent zone vs. node discrepancy.
Fixes: f8d1a31163 ("mm: consider whether to decivate based on eligible zones inactive ratio")
Link: http://lkml.kernel.org/r/20170104100825.3729-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: Nils Holland <nholland@tisys.org>
Tested-by: Nils Holland <nholland@tisys.org>
Reported-by: Klaus Ethgen <Klaus@Ethgen.de>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: <stable@vger.kernel.org> [4.8+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The VM_BUG_ON() check in move_freepages() checks whether the node id of
a page matches the node id of its zone. However, it does this before
having checked whether the struct page pointer refers to a valid struct
page to begin with. This is guaranteed in most cases, but may not be
the case if CONFIG_HOLES_IN_ZONE=y.
So reorder the VM_BUG_ON() with the pfn_valid_within() check.
Link: http://lkml.kernel.org/r/1481706707-6211-2-git-send-email-ard.biesheuvel@linaro.org
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Hanjun Guo <hanjun.guo@linaro.org>
Cc: Yisheng Xie <xieyisheng1@huawei.com>
Cc: Robert Richter <rrichter@cavium.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Andreas reported [1] made a test in jemalloc hang in THP mode in arm64:
http://lkml.kernel.org/r/mvmmvfy37g1.fsf@hawking.suse.de
The problem is currently page fault handler doesn't supports dirty bit
emulation of pmd for non-HW dirty-bit architecture so that application
stucks until VM marked the pmd dirty.
How the emulation work depends on the architecture. In case of arm64,
when it set up pte firstly, it sets pte PTE_RDONLY to get a chance to
mark the pte dirty via triggering page fault when store access happens.
Once the page fault occurs, VM marks the pmd dirty and arch code for
setting pmd will clear PTE_RDONLY for application to proceed.
IOW, if VM doesn't mark the pmd dirty, application hangs forever by
repeated fault(i.e., store op but the pmd is PTE_RDONLY).
This patch enables pmd dirty-bit emulation for those architectures.
[1] b8d3c4c300, mm/huge_memory.c: don't split THP page when MADV_FREE syscall is called
Fixes: b8d3c4c300 ("mm/huge_memory.c: don't split THP page when MADV_FREE syscall is called")
Link: http://lkml.kernel.org/r/1482506098-6149-1-git-send-email-minchan@kernel.org
Signed-off-by: Minchan Kim <minchan@kernel.org>
Reported-by: Andreas Schwab <schwab@suse.de>
Tested-by: Andreas Schwab <schwab@suse.de>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Jason Evans <je@fb.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: <stable@vger.kernel.org> [4.5+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The flag was introduced by commit 78afd5612d ("mm: add
__GFP_OTHER_NODE flag") to allow proper accounting of remote node
allocations done by kernel daemons on behalf of a process - e.g.
khugepaged.
After "mm: fix remote numa hits statistics" we do not need and actually
use the flag so we can safely remove it because all allocations which
are satisfied from their "home" node are accounted properly.
[mhocko@suse.com: fix build]
Link: http://lkml.kernel.org/r/20170106122225.GK5556@dhcp22.suse.cz
Link: http://lkml.kernel.org/r/20170102153057.9451-3-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Taku Izumi <izumi.taku@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jia He has noticed that commit b9f00e147f ("mm, page_alloc: reduce
branches in zone_statistics") has an unintentional side effect that
remote node allocation requests are accounted as NUMA_MISS rathat than
NUMA_HIT and NUMA_OTHER if such a request doesn't use __GFP_OTHER_NODE.
There are many of these potentially because the flag is used very rarely
while we have many users of __alloc_pages_node.
Fix this by simply ignoring __GFP_OTHER_NODE (it can be removed in a
follow up patch) and treat all allocations that were satisfied from the
preferred zone's node as NUMA_HITS because this is the same node we
requested the allocation from in most cases. If this is not the local
node then we just account it as NUMA_OTHER rather than NUMA_LOCAL.
One downsize would be that an allocation request for a node which is
outside of the mempolicy nodemask would be reported as a hit which is a
bit weird but that was the case before b9f00e147f already.
Fixes: b9f00e147f ("mm, page_alloc: reduce branches in zone_statistics")
Link: http://lkml.kernel.org/r/20170102153057.9451-2-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: Jia He <hejianet@gmail.com>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz> # with cbmc[1] superpowers
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Taku Izumi <izumi.taku@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently dax_mapping_entry_mkclean() fails to clean and write protect
the pmd_t of a DAX PMD entry during an *sync operation. This can result
in data loss in the following sequence:
1) mmap write to DAX PMD, dirtying PMD radix tree entry and making the
pmd_t dirty and writeable
2) fsync, flushing out PMD data and cleaning the radix tree entry. We
currently fail to mark the pmd_t as clean and write protected.
3) more mmap writes to the PMD. These don't cause any page faults since
the pmd_t is dirty and writeable. The radix tree entry remains clean.
4) fsync, which fails to flush the dirty PMD data because the radix tree
entry was clean.
5) crash - dirty data that should have been fsync'd as part of 4) could
still have been in the processor cache, and is lost.
Fix this by marking the pmd_t clean and write protected in
dax_mapping_entry_mkclean(), which is called as part of the fsync
operation 2). This will cause the writes in step 3) above to generate
page faults where we'll re-dirty the PMD radix tree entry, resulting in
flushes in the fsync that happens in step 4).
Fixes: 4b4bb46d00 ("dax: clear dirty entry tags on cache flush")
Link: http://lkml.kernel.org/r/1482272586-21177-3-git-send-email-ross.zwisler@linux.intel.com
Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "Write protect DAX PMDs in *sync path".
Currently dax_mapping_entry_mkclean() fails to clean and write protect
the pmd_t of a DAX PMD entry during an *sync operation. This can result
in data loss, as detailed in patch 2.
This series is based on Dan's "libnvdimm-pending" branch, which is the
current home for Jan's "dax: Page invalidation fixes" series. You can
find a working tree here:
https://git.kernel.org/cgit/linux/kernel/git/zwisler/linux.git/log/?h=dax_pmd_clean
This patch (of 2):
Similar to follow_pte(), follow_pte_pmd() allows either a PTE leaf or a
huge page PMD leaf to be found and returned.
Link: http://lkml.kernel.org/r/1482272586-21177-2-git-send-email-ross.zwisler@linux.intel.com
Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Suggested-by: Dave Hansen <dave.hansen@intel.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With THP page cache, when trying to build a huge page from regular pte
pages, we just clear the pmd entry. We will take another fault and at
that point we will find the huge page in the radix tree, thereby using
the huge page to complete the page fault
The second fault path will allocate the needed pgtable_t page for archs
like ppc64. So no need to deposit the same in collapse path.
Depositing them in the collapse path resulting in a pgtable_t memory
leak also giving errors like
BUG: non-zero nr_ptes on freeing mm: 3
Fixes: 953c66c2b2 ("mm: THP page cache support for ppc64")
Link: http://lkml.kernel.org/r/20161212163428.6780-2-aneesh.kumar@linux.vnet.ibm.com
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently in DAX if we have three read faults on the same hole address we
can end up with the following:
Thread 0 Thread 1 Thread 2
-------- -------- --------
dax_iomap_fault
grab_mapping_entry
lock_slot
<locks empty DAX entry>
dax_iomap_fault
grab_mapping_entry
get_unlocked_mapping_entry
<sleeps on empty DAX entry>
dax_iomap_fault
grab_mapping_entry
get_unlocked_mapping_entry
<sleeps on empty DAX entry>
dax_load_hole
find_or_create_page
...
page_cache_tree_insert
dax_wake_mapping_entry_waiter
<wakes one sleeper>
__radix_tree_replace
<swaps empty DAX entry with 4k zero page>
<wakes>
get_page
lock_page
...
put_locked_mapping_entry
unlock_page
put_page
<sleeps forever on the DAX
wait queue>
The crux of the problem is that once we insert a 4k zero page, all
locking from then on is done in terms of that 4k zero page and any
additional threads sleeping on the empty DAX entry will never be woken.
Fix this by waking all sleepers when we replace the DAX radix tree entry
with a 4k zero page. This will allow all sleeping threads to
successfully transition from locking based on the DAX empty entry to
locking on the 4k zero page.
With the test case reported by Xiong this happens very regularly in my
test setup, with some runs resulting in 9+ threads in this deadlocked
state. With this fix I've been able to run that same test dozens of
times in a loop without issue.
Fixes: ac401cc782 ("dax: New fault locking")
Link: http://lkml.kernel.org/r/1483479365-13607-1-git-send-email-ross.zwisler@linux.intel.com
Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reported-by: Xiong Zhou <xzhou@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: <stable@vger.kernel.org> [4.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Several people report seeing warnings about inconsistent radix tree
nodes followed by crashes in the workingset code, which all looked like
use-after-free access from the shadow node shrinker.
Dave Jones managed to reproduce the issue with a debug patch applied,
which confirmed that the radix tree shrinking indeed frees shadow nodes
while they are still linked to the shadow LRU:
WARNING: CPU: 2 PID: 53 at lib/radix-tree.c:643 delete_node+0x1e4/0x200
CPU: 2 PID: 53 Comm: kswapd0 Not tainted 4.10.0-rc2-think+ #3
Call Trace:
delete_node+0x1e4/0x200
__radix_tree_delete_node+0xd/0x10
shadow_lru_isolate+0xe6/0x220
__list_lru_walk_one.isra.4+0x9b/0x190
list_lru_walk_one+0x23/0x30
scan_shadow_nodes+0x2e/0x40
shrink_slab.part.44+0x23d/0x5d0
shrink_node+0x22c/0x330
kswapd+0x392/0x8f0
This is the WARN_ON_ONCE(!list_empty(&node->private_list)) placed in the
inlined radix_tree_shrink().
The problem is with 14b468791f ("mm: workingset: move shadow entry
tracking to radix tree exceptional tracking"), which passes an update
callback into the radix tree to link and unlink shadow leaf nodes when
tree entries change, but forgot to pass the callback when reclaiming a
shadow node.
While the reclaimed shadow node itself is unlinked by the shrinker, its
deletion from the tree can cause the left-most leaf node in the tree to
be shrunk. If that happens to be a shadow node as well, we don't unlink
it from the LRU as we should.
Consider this tree, where the s are shadow entries:
root->rnode
|
[0 n]
| |
[s ] [sssss]
Now the shadow node shrinker reclaims the rightmost leaf node through
the shadow node LRU:
root->rnode
|
[0 ]
|
[s ]
Because the parent of the deleted node is the first level below the
root and has only one child in the left-most slot, the intermediate
level is shrunk and the node containing the single shadow is put in
its place:
root->rnode
|
[s ]
The shrinker again sees a single left-most slot in a first level node
and thus decides to store the shadow in root->rnode directly and free
the node - which is a leaf node on the shadow node LRU.
root->rnode
|
s
Without the update callback, the freed node remains on the shadow LRU,
where it causes later shrinker runs to crash.
Pass the node updater callback into __radix_tree_delete_node() in case
the deletion causes the left-most branch in the tree to collapse too.
Also add warnings when linked nodes are freed right away, rather than
wait for the use-after-free when the list is scanned much later.
Fixes: 14b468791f ("mm: workingset: move shadow entry tracking to radix tree exceptional tracking")
Reported-by: Dave Chinner <david@fromorbit.com>
Reported-by: Hugh Dickins <hughd@google.com>
Reported-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-and-tested-by: Dave Jones <davej@codemonkey.org.uk>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Chris Leech <cleech@redhat.com>
Cc: Lee Duncan <lduncan@suse.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <mawilcox@linuxonhyperv.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4.10-rc loadtest (even on x86, and even without THPCache) fails with
"fork: Cannot allocate memory" or some such; and /proc/meminfo shows
PageTables growing.
Commit 953c66c2b2 ("mm: THP page cache support for ppc64") that got
merged in rc1 removed the freeing of an unused preallocated pagetable
after do_fault_around() has called map_pages().
This is usually a good optimization, so that the followup doesn't have
to reallocate one; but it's not sufficient to shift the freeing into
alloc_set_pte(), since there are failure cases (most commonly
VM_FAULT_RETRY) which never reach finish_fault().
Check and free it at the outer level in do_fault(), then we don't need
to worry in alloc_set_pte(), and can restore that to how it was (I
cannot find any reason to pte_free() under lock as it was doing).
And fix a separate pagetable leak, or crash, introduced by the same
change, that could only show up on some ppc64: why does do_set_pmd()'s
failure case attempt to withdraw a pagetable when it never deposited
one, at the same time overwriting (so leaking) the vmf->prealloc_pte?
Residue of an earlier implementation, perhaps? Delete it.
Fixes: 953c66c2b2 ("mm: THP page cache support for ppc64")
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull DAX updates from Dan Williams:
"The completion of Jan's DAX work for 4.10.
As I mentioned in the libnvdimm-for-4.10 pull request, these are some
final fixes for the DAX dirty-cacheline-tracking invalidation work
that was merged through the -mm, ext4, and xfs trees in -rc1. These
patches were prepared prior to the merge window, but we waited for
4.10-rc1 to have a stable merge base after all the prerequisites were
merged.
Quoting Jan on the overall changes in these patches:
"So I'd like all these 6 patches to go for rc2. The first three
patches fix invalidation of exceptional DAX entries (a bug which
is there for a long time) - without these patches data loss can
occur on power failure even though user called fsync(2). The other
three patches change locking of DAX faults so that ->iomap_begin()
is called in a more relaxed locking context and we are safe to
start a transaction there for ext4"
These have received a build success notification from the kbuild
robot, and pass the latest libnvdimm unit tests. There have not been
any -next releases since -rc1, so they have not appeared there"
* 'libnvdimm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
ext4: Simplify DAX fault path
dax: Call ->iomap_begin without entry lock during dax fault
dax: Finish fault completely when loading holes
dax: Avoid page invalidation races and unnecessary radix tree traversals
mm: Invalidate DAX radix tree entries only if appropriate
ext2: Return BH_New buffers for zeroed blocks
mm/filemap.c: In function 'clear_bit_unlock_is_negative_byte':
mm/filemap.c:933:9: error: too few arguments to function 'test_bit'
return test_bit(PG_waiters);
^~~~~~~~
Fixes: b91e1302ad ('mm: optimize PageWaiters bit use for unlock_page()')
Signed-off-by: Olof Johansson <olof@lixom.net>
Brown-paper-bag-by: Linus Torvalds <dummy@duh.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In commit 6290602709 ("mm: add PageWaiters indicating tasks are
waiting for a page bit") Nick Piggin made our page locking no longer
unconditionally touch the hashed page waitqueue, which not only helps
performance in general, but is particularly helpful on NUMA machines
where the hashed wait queues can bounce around a lot.
However, the "clear lock bit atomically and then test the waiters bit"
sequence turns out to be much more expensive than it needs to be,
because you get a nasty stall when trying to access the same word that
just got updated atomically.
On architectures where locking is done with LL/SC, this would be trivial
to fix with a new primitive that clears one bit and tests another
atomically, but that ends up not working on x86, where the only atomic
operations that return the result end up being cmpxchg and xadd. The
atomic bit operations return the old value of the same bit we changed,
not the value of an unrelated bit.
On x86, we could put the lock bit in the high bit of the byte, and use
"xadd" with that bit (where the overflow ends up not touching other
bits), and look at the other bits of the result. However, an even
simpler model is to just use a regular atomic "and" to clear the lock
bit, and then the sign bit in eflags will indicate the resulting state
of the unrelated bit #7.
So by moving the PageWaiters bit up to bit #7, we can atomically clear
the lock bit and test the waiters bit on x86 too. And architectures
with LL/SC (which is all the usual RISC suspects), the particular bit
doesn't matter, so they are fine with this approach too.
This avoids the extra access to the same atomic word, and thus avoids
the costly stall at page unlock time.
The only downside is that the interface ends up being a bit odd and
specialized: clear a bit in a byte, and test the sign bit. Nick doesn't
love the resulting name of the new primitive, but I'd rather make the
name be descriptive and very clear about the limitation imposed by
trying to work across all relevant architectures than make it be some
generic thing that doesn't make the odd semantics explicit.
So this introduces the new architecture primitive
clear_bit_unlock_is_negative_byte();
and adds the trivial implementation for x86. We have a generic
non-optimized fallback (that just does a "clear_bit()"+"test_bit(7)"
combination) which can be overridden by any architecture that can do
better. According to Nick, Power has the same hickup x86 has, for
example, but some other architectures may not even care.
All these optimizations mean that my page locking stress-test (which is
just executing a lot of small short-lived shell scripts: "make test" in
the git source tree) no longer makes our page locking look horribly bad.
Before all these optimizations, just the unlock_page() costs were just
over 3% of all CPU overhead on "make test". After this, it's down to
0.66%, so just a quarter of the cost it used to be.
(The difference on NUMA is bigger, but there this micro-optimization is
likely less noticeable, since the big issue on NUMA was not the accesses
to 'struct page', but the waitqueue accesses that were already removed
by Nick's earlier commit).
Acked-by: Nick Piggin <npiggin@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Bob Peterson <rpeterso@redhat.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Andrew Lutomirski <luto@kernel.org>
Cc: Andreas Gruenbacher <agruenba@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently invalidate_inode_pages2_range() and invalidate_mapping_pages()
just delete all exceptional radix tree entries they find. For DAX this
is not desirable as we track cache dirtiness in these entries and when
they are evicted, we may not flush caches although it is necessary. This
can for example manifest when we write to the same block both via mmap
and via write(2) (to different offsets) and fsync(2) then does not
properly flush CPU caches when modification via write(2) was the last
one.
Create appropriate DAX functions to handle invalidation of DAX entries
for invalidate_inode_pages2_range() and invalidate_mapping_pages() and
wire them up into the corresponding mm functions.
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Add a new page flag, PageWaiters, to indicate the page waitqueue has
tasks waiting. This can be tested rather than testing waitqueue_active
which requires another cacheline load.
This bit is always set when the page has tasks on page_waitqueue(page),
and is set and cleared under the waitqueue lock. It may be set when
there are no tasks on the waitqueue, which will cause a harmless extra
wakeup check that will clears the bit.
The generic bit-waitqueue infrastructure is no longer used for pages.
Instead, waitqueues are used directly with a custom key type. The
generic code was not flexible enough to have PageWaiters manipulation
under the waitqueue lock (which simplifies concurrency).
This improves the performance of page lock intensive microbenchmarks by
2-3%.
Putting two bits in the same word opens the opportunity to remove the
memory barrier between clearing the lock bit and testing the waiters
bit, after some work on the arch primitives (e.g., ensuring memory
operand widths match and cover both bits).
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Bob Peterson <rpeterso@redhat.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Andrew Lutomirski <luto@kernel.org>
Cc: Andreas Gruenbacher <agruenba@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A page is not added to the swap cache without being swap backed,
so PageSwapBacked mappings can use PG_owner_priv_1 for PageSwapCache.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Bob Peterson <rpeterso@redhat.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Andrew Lutomirski <luto@kernel.org>
Cc: Andreas Gruenbacher <agruenba@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This was entirely automated, using the script by Al:
PATT='^[[:blank:]]*#[[:blank:]]*include[[:blank:]]*<asm/uaccess.h>'
sed -i -e "s!$PATT!#include <linux/uaccess.h>!" \
$(git grep -l "$PATT"|grep -v ^include/linux/uaccess.h)
to do the replacement at the end of the merge window.
Requested-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When FADV_DONTNEED cannot drop all pages in the range, it observes that
some pages might still be on per-cpu LRU caches after recent
instantiation and so initiates remote calls to all CPUs to flush their
local caches. However, in most cases, the fadvise happens from the same
context that instantiated the pages, and any pre-LRU pages in the
specified range are most likely sitting on the local CPU's LRU cache,
and so in many cases this results in unnecessary remote calls, which, in
a loaded system, can hold up the fadvise() call significantly.
[ I didn't record it in the extreme case we observed at Facebook,
unfortunately. We had a slow-to-respond system and noticed it
lru_add_drain_all() leading the profile during fadvise calls. This
patch came out of thinking about the code and how we commonly call
FADV_DONTNEED.
FWIW, I wrote a silly directory tree walker/searcher that recurses
through /usr to read and FADV_DONTNEED each file it finds. On a 2
socket 40 ht machine, over 1% is spent in lru_add_drain_all(). With
the patch, that cost is gone; the local drain cost shows at 0.09%. ]
Try to avoid the remote call by flushing the local LRU cache before even
attempting to invalidate anything. It's a cheap operation, and the
local LRU cache is the most likely to hold any pre-LRU pages in the
specified fadvise range.
Link: http://lkml.kernel.org/r/20161214210017.GA1465@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull partial readlink cleanups from Miklos Szeredi.
This is the uncontroversial part of the readlink cleanup patch-set that
simplifies the default readlink handling.
Miklos and Al are still discussing the rest of the series.
* git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
vfs: make generic_readlink() static
vfs: remove ".readlink = generic_readlink" assignments
vfs: default to generic_readlink()
vfs: replace calling i_op->readlink with vfs_readlink()
proc/self: use generic_readlink
ecryptfs: use vfs_get_link()
bad_inode: add missing i_op initializers
Merge more updates from Andrew Morton:
- a few misc things
- kexec updates
- DMA-mapping updates to better support networking DMA operations
- IPC updates
- various MM changes to improve DAX fault handling
- lots of radix-tree changes, mainly to the test suite. All leading up
to reimplementing the IDA/IDR code to be a wrapper layer over the
radix-tree. However the final trigger-pulling patch is held off for
4.11.
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (114 commits)
radix tree test suite: delete unused rcupdate.c
radix tree test suite: add new tag check
radix-tree: ensure counts are initialised
radix tree test suite: cache recently freed objects
radix tree test suite: add some more functionality
idr: reduce the number of bits per level from 8 to 6
rxrpc: abstract away knowledge of IDR internals
tpm: use idr_find(), not idr_find_slowpath()
idr: add ida_is_empty
radix tree test suite: check multiorder iteration
radix-tree: fix replacement for multiorder entries
radix-tree: add radix_tree_split_preload()
radix-tree: add radix_tree_split
radix-tree: add radix_tree_join
radix-tree: delete radix_tree_range_tag_if_tagged()
radix-tree: delete radix_tree_locate_item()
radix-tree: improve multiorder iterators
btrfs: fix race in btrfs_free_dummy_fs_info()
radix-tree: improve dump output
radix-tree: make radix_tree_find_next_bit more useful
...
This is an exceptionally complicated function with just one caller
(tag_pages_for_writeback). We devote a large portion of the runtime of
the test suite to testing this one function which has one caller. By
introducing the new function radix_tree_iter_tag_set(), we can eliminate
all of the complexity while keeping the performance. The caller can now
use a fairly standard radix_tree_for_each() loop, and it doesn't need to
worry about tricksy things like 'start' wrapping.
The test suite continues to spend a large amount of time investigating
this function, but now it's testing the underlying primitives such as
radix_tree_iter_resume() and the radix_tree_for_each_tagged() iterator
which are also used by other parts of the kernel.
Link: http://lkml.kernel.org/r/1480369871-5271-57-git-send-email-mawilcox@linuxonhyperv.com
Signed-off-by: Matthew Wilcox <willy@infradead.org>
Tested-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This rather complicated function can be better implemented as an
iterator. It has only one caller, so move the functionality to the only
place that needs it. Update the test suite to follow the same pattern.
Link: http://lkml.kernel.org/r/1480369871-5271-56-git-send-email-mawilcox@linuxonhyperv.com
Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com>
Acked-by: Konstantin Khlebnikov <koct9i@gmail.com>
Tested-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This fixes several interlinked problems with the iterators in the
presence of multiorder entries.
1. radix_tree_iter_next() would only advance by one slot, which would
result in the iterators returning the same entry more than once if
there were sibling entries.
2. radix_tree_next_slot() could return an internal pointer instead of
a user pointer if a tagged multiorder entry was immediately followed by
an entry of lower order.
3. radix_tree_next_slot() expanded to a lot more code than it used to
when multiorder support was compiled in. And I wasn't comfortable with
entry_to_node() being in a header file.
Fixing radix_tree_iter_next() for the presence of sibling entries
necessarily involves examining the contents of the radix tree, so we now
need to pass 'slot' to radix_tree_iter_next(), and we need to change the
calling convention so it is called *before* dropping the lock which
protects the tree. Also rename it to radix_tree_iter_resume(), as some
people thought it was necessary to call radix_tree_iter_next() each time
around the loop.
radix_tree_next_slot() becomes closer to how it looked before multiorder
support was introduced. It only checks to see if the next entry in the
chunk is a sibling entry or a pointer to a node; this should be rare
enough that handling this case out of line is not a performance impact
(and such impact is amortised by the fact that the entry we just
processed was a multiorder entry). Also, radix_tree_next_slot() used to
force a new chunk lookup for untagged entries, which is more expensive
than the out of line sibling entry skipping.
Link: http://lkml.kernel.org/r/1480369871-5271-55-git-send-email-mawilcox@linuxonhyperv.com
Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com>
Tested-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently PTE gets updated in wp_pfn_shared() after dax_pfn_mkwrite()
has released corresponding radix tree entry lock. When we want to
writeprotect PTE on cache flush, we need PTE modification to happen
under radix tree entry lock to ensure consistent updates of PTE and
radix tree (standard faults use page lock to ensure this consistency).
So move update of PTE bit into dax_pfn_mkwrite().
Link: http://lkml.kernel.org/r/1479460644-25076-20-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
DAX will need to implement its own version of page_check_address(). To
avoid duplicating page table walking code, export follow_pte() which
does what we need.
Link: http://lkml.kernel.org/r/1479460644-25076-18-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently finish_mkwrite_fault() returns 0 when PTE got changed before
we acquired PTE lock and VM_FAULT_WRITE when we succeeded in modifying
the PTE. This is somewhat confusing since 0 generally means success, it
is also inconsistent with finish_fault() which returns 0 on success.
Change finish_mkwrite_fault() to return 0 on success and VM_FAULT_NOPAGE
when PTE changed. Practically, there should be no behavioral difference
since we bail out from the fault the same way regardless whether we
return 0, VM_FAULT_NOPAGE, or VM_FAULT_WRITE. Also note that
VM_FAULT_WRITE has no effect for shared mappings since the only two
places that check it - KSM and GUP - care about private mappings only.
Generally the meaning of VM_FAULT_WRITE for shared mappings is not well
defined and we should probably clean that up.
Link: http://lkml.kernel.org/r/1479460644-25076-17-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Provide a helper function for finishing write faults due to PTE being
read-only. The helper will be used by DAX to avoid the need of
complicating generic MM code with DAX locking specifics.
Link: http://lkml.kernel.org/r/1479460644-25076-16-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
wp_page_reuse() handles write shared faults which is needed only in
wp_page_shared(). Move the handling only into that location to make
wp_page_reuse() simpler and avoid a strange situation when we sometimes
pass in locked page, sometimes unlocked etc.
Link: http://lkml.kernel.org/r/1479460644-25076-15-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
So far we set vmf->page during WP faults only when we needed to pass it
to the ->page_mkwrite handler. Set it in all the cases now and use that
instead of passing page pointer explicitly around.
Link: http://lkml.kernel.org/r/1479460644-25076-14-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We will need more information in the ->page_mkwrite() helper for DAX to
be able to fully finish faults there. Pass vm_fault structure to
do_page_mkwrite() and use it there so that information propagates
properly from upper layers.
Link: http://lkml.kernel.org/r/1479460644-25076-13-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently we duplicate handling of shared write faults in
wp_page_reuse() and do_shared_fault(). Factor them out into a common
function.
Link: http://lkml.kernel.org/r/1479460644-25076-12-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Move final handling of COW faults from generic code into DAX fault
handler. That way generic code doesn't have to be aware of
peculiarities of DAX locking so remove that knowledge and make locking
functions private to fs/dax.c.
Link: http://lkml.kernel.org/r/1479460644-25076-11-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Introduce finish_fault() as a helper function for finishing page faults.
It is rather thin wrapper around alloc_set_pte() but since we'd want to
call this from DAX code or filesystems, it is still useful to avoid some
boilerplate code.
Link: http://lkml.kernel.org/r/1479460644-25076-10-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "dax: Clear dirty bits after flushing caches", v5.
Patchset to clear dirty bits from radix tree of DAX inodes when caches
for corresponding pfns have been flushed. In principle, these patches
enable handlers to easily update PTEs and do other work necessary to
finish the fault without duplicating the functionality present in the
generic code. I'd like to thank Kirill and Ross for reviews of the
series!
This patch (of 20):
To allow full handling of COW faults add memcg field to struct vm_fault
and a return value of ->fault() handler meaning that COW fault is fully
handled and memcg charge must not be canceled. This will allow us to
remove knowledge about special DAX locking from the generic fault code.
Link: http://lkml.kernel.org/r/1479460644-25076-9-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
