2011-01-13 16:46:30 -07:00
|
|
|
= Transparent Hugepage Support =
|
|
|
|
|
|
|
|
== Objective ==
|
|
|
|
|
|
|
|
Performance critical computing applications dealing with large memory
|
|
|
|
working sets are already running on top of libhugetlbfs and in turn
|
|
|
|
hugetlbfs. Transparent Hugepage Support is an alternative means of
|
|
|
|
using huge pages for the backing of virtual memory with huge pages
|
|
|
|
that supports the automatic promotion and demotion of page sizes and
|
|
|
|
without the shortcomings of hugetlbfs.
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
Currently it only works for anonymous memory mappings and tmpfs/shmem.
|
|
|
|
But in the future it can expand to other filesystems.
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
The reason applications are running faster is because of two
|
|
|
|
factors. The first factor is almost completely irrelevant and it's not
|
|
|
|
of significant interest because it'll also have the downside of
|
|
|
|
requiring larger clear-page copy-page in page faults which is a
|
|
|
|
potentially negative effect. The first factor consists in taking a
|
|
|
|
single page fault for each 2M virtual region touched by userland (so
|
|
|
|
reducing the enter/exit kernel frequency by a 512 times factor). This
|
|
|
|
only matters the first time the memory is accessed for the lifetime of
|
|
|
|
a memory mapping. The second long lasting and much more important
|
|
|
|
factor will affect all subsequent accesses to the memory for the whole
|
|
|
|
runtime of the application. The second factor consist of two
|
|
|
|
components: 1) the TLB miss will run faster (especially with
|
|
|
|
virtualization using nested pagetables but almost always also on bare
|
|
|
|
metal without virtualization) and 2) a single TLB entry will be
|
|
|
|
mapping a much larger amount of virtual memory in turn reducing the
|
|
|
|
number of TLB misses. With virtualization and nested pagetables the
|
|
|
|
TLB can be mapped of larger size only if both KVM and the Linux guest
|
|
|
|
are using hugepages but a significant speedup already happens if only
|
|
|
|
one of the two is using hugepages just because of the fact the TLB
|
|
|
|
miss is going to run faster.
|
|
|
|
|
|
|
|
== Design ==
|
|
|
|
|
2016-01-15 17:54:30 -07:00
|
|
|
- "graceful fallback": mm components which don't have transparent hugepage
|
|
|
|
knowledge fall back to breaking huge pmd mapping into table of ptes and,
|
|
|
|
if necessary, split a transparent hugepage. Therefore these components
|
|
|
|
can continue working on the regular pages or regular pte mappings.
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
- if a hugepage allocation fails because of memory fragmentation,
|
|
|
|
regular pages should be gracefully allocated instead and mixed in
|
|
|
|
the same vma without any failure or significant delay and without
|
|
|
|
userland noticing
|
|
|
|
|
|
|
|
- if some task quits and more hugepages become available (either
|
|
|
|
immediately in the buddy or through the VM), guest physical memory
|
|
|
|
backed by regular pages should be relocated on hugepages
|
|
|
|
automatically (with khugepaged)
|
|
|
|
|
|
|
|
- it doesn't require memory reservation and in turn it uses hugepages
|
|
|
|
whenever possible (the only possible reservation here is kernelcore=
|
|
|
|
to avoid unmovable pages to fragment all the memory but such a tweak
|
|
|
|
is not specific to transparent hugepage support and it's a generic
|
|
|
|
feature that applies to all dynamic high order allocations in the
|
|
|
|
kernel)
|
|
|
|
|
|
|
|
Transparent Hugepage Support maximizes the usefulness of free memory
|
|
|
|
if compared to the reservation approach of hugetlbfs by allowing all
|
|
|
|
unused memory to be used as cache or other movable (or even unmovable
|
|
|
|
entities). It doesn't require reservation to prevent hugepage
|
|
|
|
allocation failures to be noticeable from userland. It allows paging
|
|
|
|
and all other advanced VM features to be available on the
|
|
|
|
hugepages. It requires no modifications for applications to take
|
|
|
|
advantage of it.
|
|
|
|
|
|
|
|
Applications however can be further optimized to take advantage of
|
|
|
|
this feature, like for example they've been optimized before to avoid
|
|
|
|
a flood of mmap system calls for every malloc(4k). Optimizing userland
|
|
|
|
is by far not mandatory and khugepaged already can take care of long
|
|
|
|
lived page allocations even for hugepage unaware applications that
|
|
|
|
deals with large amounts of memory.
|
|
|
|
|
|
|
|
In certain cases when hugepages are enabled system wide, application
|
|
|
|
may end up allocating more memory resources. An application may mmap a
|
|
|
|
large region but only touch 1 byte of it, in that case a 2M page might
|
|
|
|
be allocated instead of a 4k page for no good. This is why it's
|
|
|
|
possible to disable hugepages system-wide and to only have them inside
|
|
|
|
MADV_HUGEPAGE madvise regions.
|
|
|
|
|
|
|
|
Embedded systems should enable hugepages only inside madvise regions
|
|
|
|
to eliminate any risk of wasting any precious byte of memory and to
|
|
|
|
only run faster.
|
|
|
|
|
|
|
|
Applications that gets a lot of benefit from hugepages and that don't
|
|
|
|
risk to lose memory by using hugepages, should use
|
|
|
|
madvise(MADV_HUGEPAGE) on their critical mmapped regions.
|
|
|
|
|
|
|
|
== sysfs ==
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
Transparent Hugepage Support for anonymous memory can be entirely disabled
|
|
|
|
(mostly for debugging purposes) or only enabled inside MADV_HUGEPAGE
|
|
|
|
regions (to avoid the risk of consuming more memory resources) or enabled
|
|
|
|
system wide. This can be achieved with one of:
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
echo always >/sys/kernel/mm/transparent_hugepage/enabled
|
|
|
|
echo madvise >/sys/kernel/mm/transparent_hugepage/enabled
|
|
|
|
echo never >/sys/kernel/mm/transparent_hugepage/enabled
|
|
|
|
|
|
|
|
It's also possible to limit defrag efforts in the VM to generate
|
2016-07-26 16:26:40 -06:00
|
|
|
anonymous hugepages in case they're not immediately free to madvise
|
|
|
|
regions or to never try to defrag memory and simply fallback to regular
|
|
|
|
pages unless hugepages are immediately available. Clearly if we spend CPU
|
|
|
|
time to defrag memory, we would expect to gain even more by the fact we
|
|
|
|
use hugepages later instead of regular pages. This isn't always
|
2011-01-13 16:46:30 -07:00
|
|
|
guaranteed, but it may be more likely in case the allocation is for a
|
|
|
|
MADV_HUGEPAGE region.
|
|
|
|
|
|
|
|
echo always >/sys/kernel/mm/transparent_hugepage/defrag
|
mm: thp: set THP defrag by default to madvise and add a stall-free defrag option
THP defrag is enabled by default to direct reclaim/compact but not wake
kswapd in the event of a THP allocation failure. The problem is that
THP allocation requests potentially enter reclaim/compaction. This
potentially incurs a severe stall that is not guaranteed to be offset by
reduced TLB misses. While there has been considerable effort to reduce
the impact of reclaim/compaction, it is still a high cost and workloads
that should fit in memory fail to do so. Specifically, a simple
anon/file streaming workload will enter direct reclaim on NUMA at least
even though the working set size is 80% of RAM. It's been years and
it's time to throw in the towel.
First, this patch defines THP defrag as follows;
madvise: A failed allocation will direct reclaim/compact if the application requests it
never: Neither reclaim/compact nor wake kswapd
defer: A failed allocation will wake kswapd/kcompactd
always: A failed allocation will direct reclaim/compact (historical behaviour)
khugepaged defrag will enter direct/reclaim but not wake kswapd.
Next it sets the default defrag option to be "madvise" to only enter
direct reclaim/compaction for applications that specifically requested
it.
Lastly, it removes a check from the page allocator slowpath that is
related to __GFP_THISNODE to allow "defer" to work. The callers that
really cares are slub/slab and they are updated accordingly. The slab
one may be surprising because it also corrects a comment as kswapd was
never woken up by that path.
This means that a THP fault will no longer stall for most applications
by default and the ideal for most users that get THP if they are
immediately available. There are still options for users that prefer a
stall at startup of a new application by either restoring historical
behaviour with "always" or pick a half-way point with "defer" where
kswapd does some of the work in the background and wakes kcompactd if
necessary. THP defrag for khugepaged remains enabled and will enter
direct/reclaim but no wakeup kswapd or kcompactd.
After this patch a THP allocation failure will quickly fallback and rely
on khugepaged to recover the situation at some time in the future. In
some cases, this will reduce THP usage but the benefit of THP is hard to
measure and not a universal win where as a stall to reclaim/compaction
is definitely measurable and can be painful.
The first test for this is using "usemem" to read a large file and write
a large anonymous mapping (to avoid the zero page) multiple times. The
total size of the mappings is 80% of RAM and the benchmark simply
measures how long it takes to complete. It uses multiple threads to see
if that is a factor. On UMA, the performance is almost identical so is
not reported but on NUMA, we see this
usemem
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Amean System-1 102.86 ( 0.00%) 46.81 ( 54.50%)
Amean System-4 37.85 ( 0.00%) 34.02 ( 10.12%)
Amean System-7 48.12 ( 0.00%) 46.89 ( 2.56%)
Amean System-12 51.98 ( 0.00%) 56.96 ( -9.57%)
Amean System-21 80.16 ( 0.00%) 79.05 ( 1.39%)
Amean System-30 110.71 ( 0.00%) 107.17 ( 3.20%)
Amean System-48 127.98 ( 0.00%) 124.83 ( 2.46%)
Amean Elapsd-1 185.84 ( 0.00%) 105.51 ( 43.23%)
Amean Elapsd-4 26.19 ( 0.00%) 25.58 ( 2.33%)
Amean Elapsd-7 21.65 ( 0.00%) 21.62 ( 0.16%)
Amean Elapsd-12 18.58 ( 0.00%) 17.94 ( 3.43%)
Amean Elapsd-21 17.53 ( 0.00%) 16.60 ( 5.33%)
Amean Elapsd-30 17.45 ( 0.00%) 17.13 ( 1.84%)
Amean Elapsd-48 15.40 ( 0.00%) 15.27 ( 0.82%)
For a single thread, the benchmark completes 43.23% faster with this
patch applied with smaller benefits as the thread increases. Similar,
notice the large reduction in most cases in system CPU usage. The
overall CPU time is
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
User 10357.65 10438.33
System 3988.88 3543.94
Elapsed 2203.01 1634.41
Which is substantial. Now, the reclaim figures
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
Minor Faults 128458477 278352931
Major Faults 2174976 225
Swap Ins 16904701 0
Swap Outs 17359627 0
Allocation stalls 43611 0
DMA allocs 0 0
DMA32 allocs 19832646 19448017
Normal allocs 614488453 580941839
Movable allocs 0 0
Direct pages scanned 24163800 0
Kswapd pages scanned 0 0
Kswapd pages reclaimed 0 0
Direct pages reclaimed 20691346 0
Compaction stalls 42263 0
Compaction success 938 0
Compaction failures 41325 0
This patch eliminates almost all swapping and direct reclaim activity.
There is still overhead but it's from NUMA balancing which does not
identify that it's pointless trying to do anything with this workload.
I also tried the thpscale benchmark which forces a corner case where
compaction can be used heavily and measures the latency of whether base
or huge pages were used
thpscale Fault Latencies
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Amean fault-base-1 5288.84 ( 0.00%) 2817.12 ( 46.73%)
Amean fault-base-3 6365.53 ( 0.00%) 3499.11 ( 45.03%)
Amean fault-base-5 6526.19 ( 0.00%) 4363.06 ( 33.15%)
Amean fault-base-7 7142.25 ( 0.00%) 4858.08 ( 31.98%)
Amean fault-base-12 13827.64 ( 0.00%) 10292.11 ( 25.57%)
Amean fault-base-18 18235.07 ( 0.00%) 13788.84 ( 24.38%)
Amean fault-base-24 21597.80 ( 0.00%) 24388.03 (-12.92%)
Amean fault-base-30 26754.15 ( 0.00%) 19700.55 ( 26.36%)
Amean fault-base-32 26784.94 ( 0.00%) 19513.57 ( 27.15%)
Amean fault-huge-1 4223.96 ( 0.00%) 2178.57 ( 48.42%)
Amean fault-huge-3 2194.77 ( 0.00%) 2149.74 ( 2.05%)
Amean fault-huge-5 2569.60 ( 0.00%) 2346.95 ( 8.66%)
Amean fault-huge-7 3612.69 ( 0.00%) 2997.70 ( 17.02%)
Amean fault-huge-12 3301.75 ( 0.00%) 6727.02 (-103.74%)
Amean fault-huge-18 6696.47 ( 0.00%) 6685.72 ( 0.16%)
Amean fault-huge-24 8000.72 ( 0.00%) 9311.43 (-16.38%)
Amean fault-huge-30 13305.55 ( 0.00%) 9750.45 ( 26.72%)
Amean fault-huge-32 9981.71 ( 0.00%) 10316.06 ( -3.35%)
The average time to fault pages is substantially reduced in the majority
of caseds but with the obvious caveat that fewer THPs are actually used
in this adverse workload
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Percentage huge-1 0.71 ( 0.00%) 14.04 (1865.22%)
Percentage huge-3 10.77 ( 0.00%) 33.05 (206.85%)
Percentage huge-5 60.39 ( 0.00%) 38.51 (-36.23%)
Percentage huge-7 45.97 ( 0.00%) 34.57 (-24.79%)
Percentage huge-12 68.12 ( 0.00%) 40.07 (-41.17%)
Percentage huge-18 64.93 ( 0.00%) 47.82 (-26.35%)
Percentage huge-24 62.69 ( 0.00%) 44.23 (-29.44%)
Percentage huge-30 43.49 ( 0.00%) 55.38 ( 27.34%)
Percentage huge-32 50.72 ( 0.00%) 51.90 ( 2.35%)
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
Minor Faults 37429143 47564000
Major Faults 1916 1558
Swap Ins 1466 1079
Swap Outs 2936863 149626
Allocation stalls 62510 3
DMA allocs 0 0
DMA32 allocs 6566458 6401314
Normal allocs 216361697 216538171
Movable allocs 0 0
Direct pages scanned 25977580 17998
Kswapd pages scanned 0 3638931
Kswapd pages reclaimed 0 207236
Direct pages reclaimed 8833714 88
Compaction stalls 103349 5
Compaction success 270 4
Compaction failures 103079 1
Note again that while this does swap as it's an aggressive workload, the
direct relcim activity and allocation stalls is substantially reduced.
There is some kswapd activity but ftrace showed that the kswapd activity
was due to normal wakeups from 4K pages being allocated.
Compaction-related stalls and activity are almost eliminated.
I also tried the stutter benchmark. For this, I do not have figures for
NUMA but it's something that does impact UMA so I'll report what is
available
stutter
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Min mmap 7.3571 ( 0.00%) 7.3438 ( 0.18%)
1st-qrtle mmap 7.5278 ( 0.00%) 17.9200 (-138.05%)
2nd-qrtle mmap 7.6818 ( 0.00%) 21.6055 (-181.25%)
3rd-qrtle mmap 11.0889 ( 0.00%) 21.8881 (-97.39%)
Max-90% mmap 27.8978 ( 0.00%) 22.1632 ( 20.56%)
Max-93% mmap 28.3202 ( 0.00%) 22.3044 ( 21.24%)
Max-95% mmap 28.5600 ( 0.00%) 22.4580 ( 21.37%)
Max-99% mmap 29.6032 ( 0.00%) 25.5216 ( 13.79%)
Max mmap 4109.7289 ( 0.00%) 4813.9832 (-17.14%)
Mean mmap 12.4474 ( 0.00%) 19.3027 (-55.07%)
This benchmark is trying to fault an anonymous mapping while there is a
heavy IO load -- a scenario that desktop users used to complain about
frequently. This shows a mix because the ideal case of mapping with THP
is not hit as often. However, note that 99% of the mappings complete
13.79% faster. The CPU usage here is particularly interesting
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
User 67.50 0.99
System 1327.88 91.30
Elapsed 2079.00 2128.98
And once again we look at the reclaim figures
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
Minor Faults 335241922 1314582827
Major Faults 715 819
Swap Ins 0 0
Swap Outs 0 0
Allocation stalls 532723 0
DMA allocs 0 0
DMA32 allocs 1822364341 1177950222
Normal allocs 1815640808 1517844854
Movable allocs 0 0
Direct pages scanned 21892772 0
Kswapd pages scanned 20015890 41879484
Kswapd pages reclaimed 19961986 41822072
Direct pages reclaimed 21892741 0
Compaction stalls 1065755 0
Compaction success 514 0
Compaction failures 1065241 0
Allocation stalls and all direct reclaim activity is eliminated as well
as compaction-related stalls.
THP gives impressive gains in some cases but only if they are quickly
available. We're not going to reach the point where they are completely
free so lets take the costs out of the fast paths finally and defer the
cost to kswapd, kcompactd and khugepaged where it belongs.
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 15:19:23 -06:00
|
|
|
echo defer >/sys/kernel/mm/transparent_hugepage/defrag
|
2011-01-13 16:46:30 -07:00
|
|
|
echo madvise >/sys/kernel/mm/transparent_hugepage/defrag
|
|
|
|
echo never >/sys/kernel/mm/transparent_hugepage/defrag
|
|
|
|
|
mm: thp: set THP defrag by default to madvise and add a stall-free defrag option
THP defrag is enabled by default to direct reclaim/compact but not wake
kswapd in the event of a THP allocation failure. The problem is that
THP allocation requests potentially enter reclaim/compaction. This
potentially incurs a severe stall that is not guaranteed to be offset by
reduced TLB misses. While there has been considerable effort to reduce
the impact of reclaim/compaction, it is still a high cost and workloads
that should fit in memory fail to do so. Specifically, a simple
anon/file streaming workload will enter direct reclaim on NUMA at least
even though the working set size is 80% of RAM. It's been years and
it's time to throw in the towel.
First, this patch defines THP defrag as follows;
madvise: A failed allocation will direct reclaim/compact if the application requests it
never: Neither reclaim/compact nor wake kswapd
defer: A failed allocation will wake kswapd/kcompactd
always: A failed allocation will direct reclaim/compact (historical behaviour)
khugepaged defrag will enter direct/reclaim but not wake kswapd.
Next it sets the default defrag option to be "madvise" to only enter
direct reclaim/compaction for applications that specifically requested
it.
Lastly, it removes a check from the page allocator slowpath that is
related to __GFP_THISNODE to allow "defer" to work. The callers that
really cares are slub/slab and they are updated accordingly. The slab
one may be surprising because it also corrects a comment as kswapd was
never woken up by that path.
This means that a THP fault will no longer stall for most applications
by default and the ideal for most users that get THP if they are
immediately available. There are still options for users that prefer a
stall at startup of a new application by either restoring historical
behaviour with "always" or pick a half-way point with "defer" where
kswapd does some of the work in the background and wakes kcompactd if
necessary. THP defrag for khugepaged remains enabled and will enter
direct/reclaim but no wakeup kswapd or kcompactd.
After this patch a THP allocation failure will quickly fallback and rely
on khugepaged to recover the situation at some time in the future. In
some cases, this will reduce THP usage but the benefit of THP is hard to
measure and not a universal win where as a stall to reclaim/compaction
is definitely measurable and can be painful.
The first test for this is using "usemem" to read a large file and write
a large anonymous mapping (to avoid the zero page) multiple times. The
total size of the mappings is 80% of RAM and the benchmark simply
measures how long it takes to complete. It uses multiple threads to see
if that is a factor. On UMA, the performance is almost identical so is
not reported but on NUMA, we see this
usemem
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Amean System-1 102.86 ( 0.00%) 46.81 ( 54.50%)
Amean System-4 37.85 ( 0.00%) 34.02 ( 10.12%)
Amean System-7 48.12 ( 0.00%) 46.89 ( 2.56%)
Amean System-12 51.98 ( 0.00%) 56.96 ( -9.57%)
Amean System-21 80.16 ( 0.00%) 79.05 ( 1.39%)
Amean System-30 110.71 ( 0.00%) 107.17 ( 3.20%)
Amean System-48 127.98 ( 0.00%) 124.83 ( 2.46%)
Amean Elapsd-1 185.84 ( 0.00%) 105.51 ( 43.23%)
Amean Elapsd-4 26.19 ( 0.00%) 25.58 ( 2.33%)
Amean Elapsd-7 21.65 ( 0.00%) 21.62 ( 0.16%)
Amean Elapsd-12 18.58 ( 0.00%) 17.94 ( 3.43%)
Amean Elapsd-21 17.53 ( 0.00%) 16.60 ( 5.33%)
Amean Elapsd-30 17.45 ( 0.00%) 17.13 ( 1.84%)
Amean Elapsd-48 15.40 ( 0.00%) 15.27 ( 0.82%)
For a single thread, the benchmark completes 43.23% faster with this
patch applied with smaller benefits as the thread increases. Similar,
notice the large reduction in most cases in system CPU usage. The
overall CPU time is
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
User 10357.65 10438.33
System 3988.88 3543.94
Elapsed 2203.01 1634.41
Which is substantial. Now, the reclaim figures
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
Minor Faults 128458477 278352931
Major Faults 2174976 225
Swap Ins 16904701 0
Swap Outs 17359627 0
Allocation stalls 43611 0
DMA allocs 0 0
DMA32 allocs 19832646 19448017
Normal allocs 614488453 580941839
Movable allocs 0 0
Direct pages scanned 24163800 0
Kswapd pages scanned 0 0
Kswapd pages reclaimed 0 0
Direct pages reclaimed 20691346 0
Compaction stalls 42263 0
Compaction success 938 0
Compaction failures 41325 0
This patch eliminates almost all swapping and direct reclaim activity.
There is still overhead but it's from NUMA balancing which does not
identify that it's pointless trying to do anything with this workload.
I also tried the thpscale benchmark which forces a corner case where
compaction can be used heavily and measures the latency of whether base
or huge pages were used
thpscale Fault Latencies
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Amean fault-base-1 5288.84 ( 0.00%) 2817.12 ( 46.73%)
Amean fault-base-3 6365.53 ( 0.00%) 3499.11 ( 45.03%)
Amean fault-base-5 6526.19 ( 0.00%) 4363.06 ( 33.15%)
Amean fault-base-7 7142.25 ( 0.00%) 4858.08 ( 31.98%)
Amean fault-base-12 13827.64 ( 0.00%) 10292.11 ( 25.57%)
Amean fault-base-18 18235.07 ( 0.00%) 13788.84 ( 24.38%)
Amean fault-base-24 21597.80 ( 0.00%) 24388.03 (-12.92%)
Amean fault-base-30 26754.15 ( 0.00%) 19700.55 ( 26.36%)
Amean fault-base-32 26784.94 ( 0.00%) 19513.57 ( 27.15%)
Amean fault-huge-1 4223.96 ( 0.00%) 2178.57 ( 48.42%)
Amean fault-huge-3 2194.77 ( 0.00%) 2149.74 ( 2.05%)
Amean fault-huge-5 2569.60 ( 0.00%) 2346.95 ( 8.66%)
Amean fault-huge-7 3612.69 ( 0.00%) 2997.70 ( 17.02%)
Amean fault-huge-12 3301.75 ( 0.00%) 6727.02 (-103.74%)
Amean fault-huge-18 6696.47 ( 0.00%) 6685.72 ( 0.16%)
Amean fault-huge-24 8000.72 ( 0.00%) 9311.43 (-16.38%)
Amean fault-huge-30 13305.55 ( 0.00%) 9750.45 ( 26.72%)
Amean fault-huge-32 9981.71 ( 0.00%) 10316.06 ( -3.35%)
The average time to fault pages is substantially reduced in the majority
of caseds but with the obvious caveat that fewer THPs are actually used
in this adverse workload
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Percentage huge-1 0.71 ( 0.00%) 14.04 (1865.22%)
Percentage huge-3 10.77 ( 0.00%) 33.05 (206.85%)
Percentage huge-5 60.39 ( 0.00%) 38.51 (-36.23%)
Percentage huge-7 45.97 ( 0.00%) 34.57 (-24.79%)
Percentage huge-12 68.12 ( 0.00%) 40.07 (-41.17%)
Percentage huge-18 64.93 ( 0.00%) 47.82 (-26.35%)
Percentage huge-24 62.69 ( 0.00%) 44.23 (-29.44%)
Percentage huge-30 43.49 ( 0.00%) 55.38 ( 27.34%)
Percentage huge-32 50.72 ( 0.00%) 51.90 ( 2.35%)
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
Minor Faults 37429143 47564000
Major Faults 1916 1558
Swap Ins 1466 1079
Swap Outs 2936863 149626
Allocation stalls 62510 3
DMA allocs 0 0
DMA32 allocs 6566458 6401314
Normal allocs 216361697 216538171
Movable allocs 0 0
Direct pages scanned 25977580 17998
Kswapd pages scanned 0 3638931
Kswapd pages reclaimed 0 207236
Direct pages reclaimed 8833714 88
Compaction stalls 103349 5
Compaction success 270 4
Compaction failures 103079 1
Note again that while this does swap as it's an aggressive workload, the
direct relcim activity and allocation stalls is substantially reduced.
There is some kswapd activity but ftrace showed that the kswapd activity
was due to normal wakeups from 4K pages being allocated.
Compaction-related stalls and activity are almost eliminated.
I also tried the stutter benchmark. For this, I do not have figures for
NUMA but it's something that does impact UMA so I'll report what is
available
stutter
4.4.0 4.4.0
kcompactd-v1r1 nodefrag-v1r3
Min mmap 7.3571 ( 0.00%) 7.3438 ( 0.18%)
1st-qrtle mmap 7.5278 ( 0.00%) 17.9200 (-138.05%)
2nd-qrtle mmap 7.6818 ( 0.00%) 21.6055 (-181.25%)
3rd-qrtle mmap 11.0889 ( 0.00%) 21.8881 (-97.39%)
Max-90% mmap 27.8978 ( 0.00%) 22.1632 ( 20.56%)
Max-93% mmap 28.3202 ( 0.00%) 22.3044 ( 21.24%)
Max-95% mmap 28.5600 ( 0.00%) 22.4580 ( 21.37%)
Max-99% mmap 29.6032 ( 0.00%) 25.5216 ( 13.79%)
Max mmap 4109.7289 ( 0.00%) 4813.9832 (-17.14%)
Mean mmap 12.4474 ( 0.00%) 19.3027 (-55.07%)
This benchmark is trying to fault an anonymous mapping while there is a
heavy IO load -- a scenario that desktop users used to complain about
frequently. This shows a mix because the ideal case of mapping with THP
is not hit as often. However, note that 99% of the mappings complete
13.79% faster. The CPU usage here is particularly interesting
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
User 67.50 0.99
System 1327.88 91.30
Elapsed 2079.00 2128.98
And once again we look at the reclaim figures
4.4.0 4.4.0
kcompactd-v1r1nodefrag-v1r3
Minor Faults 335241922 1314582827
Major Faults 715 819
Swap Ins 0 0
Swap Outs 0 0
Allocation stalls 532723 0
DMA allocs 0 0
DMA32 allocs 1822364341 1177950222
Normal allocs 1815640808 1517844854
Movable allocs 0 0
Direct pages scanned 21892772 0
Kswapd pages scanned 20015890 41879484
Kswapd pages reclaimed 19961986 41822072
Direct pages reclaimed 21892741 0
Compaction stalls 1065755 0
Compaction success 514 0
Compaction failures 1065241 0
Allocation stalls and all direct reclaim activity is eliminated as well
as compaction-related stalls.
THP gives impressive gains in some cases but only if they are quickly
available. We're not going to reach the point where they are completely
free so lets take the costs out of the fast paths finally and defer the
cost to kswapd, kcompactd and khugepaged where it belongs.
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-17 15:19:23 -06:00
|
|
|
"always" means that an application requesting THP will stall on allocation
|
|
|
|
failure and directly reclaim pages and compact memory in an effort to
|
|
|
|
allocate a THP immediately. This may be desirable for virtual machines
|
|
|
|
that benefit heavily from THP use and are willing to delay the VM start
|
|
|
|
to utilise them.
|
|
|
|
|
|
|
|
"defer" means that an application will wake kswapd in the background
|
|
|
|
to reclaim pages and wake kcompact to compact memory so that THP is
|
|
|
|
available in the near future. It's the responsibility of khugepaged
|
|
|
|
to then install the THP pages later.
|
|
|
|
|
|
|
|
"madvise" will enter direct reclaim like "always" but only for regions
|
|
|
|
that are have used madvise(MADV_HUGEPAGE). This is the default behaviour.
|
|
|
|
|
|
|
|
"never" should be self-explanatory.
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
By default kernel tries to use huge zero page on read page fault to
|
|
|
|
anonymous mapping. It's possible to disable huge zero page by writing 0
|
|
|
|
or enable it back by writing 1:
|
2012-12-12 14:51:12 -07:00
|
|
|
|
2013-07-08 17:00:16 -06:00
|
|
|
echo 0 >/sys/kernel/mm/transparent_hugepage/use_zero_page
|
|
|
|
echo 1 >/sys/kernel/mm/transparent_hugepage/use_zero_page
|
2012-12-12 14:51:12 -07:00
|
|
|
|
2011-01-13 16:46:30 -07:00
|
|
|
khugepaged will be automatically started when
|
|
|
|
transparent_hugepage/enabled is set to "always" or "madvise, and it'll
|
|
|
|
be automatically shutdown if it's set to "never".
|
|
|
|
|
|
|
|
khugepaged runs usually at low frequency so while one may not want to
|
|
|
|
invoke defrag algorithms synchronously during the page faults, it
|
|
|
|
should be worth invoking defrag at least in khugepaged. However it's
|
2011-09-22 15:11:38 -06:00
|
|
|
also possible to disable defrag in khugepaged by writing 0 or enable
|
|
|
|
defrag in khugepaged by writing 1:
|
2011-01-13 16:46:30 -07:00
|
|
|
|
2011-09-22 15:11:38 -06:00
|
|
|
echo 0 >/sys/kernel/mm/transparent_hugepage/khugepaged/defrag
|
|
|
|
echo 1 >/sys/kernel/mm/transparent_hugepage/khugepaged/defrag
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
You can also control how many pages khugepaged should scan at each
|
|
|
|
pass:
|
|
|
|
|
|
|
|
/sys/kernel/mm/transparent_hugepage/khugepaged/pages_to_scan
|
|
|
|
|
|
|
|
and how many milliseconds to wait in khugepaged between each pass (you
|
|
|
|
can set this to 0 to run khugepaged at 100% utilization of one core):
|
|
|
|
|
|
|
|
/sys/kernel/mm/transparent_hugepage/khugepaged/scan_sleep_millisecs
|
|
|
|
|
|
|
|
and how many milliseconds to wait in khugepaged if there's an hugepage
|
|
|
|
allocation failure to throttle the next allocation attempt.
|
|
|
|
|
|
|
|
/sys/kernel/mm/transparent_hugepage/khugepaged/alloc_sleep_millisecs
|
|
|
|
|
|
|
|
The khugepaged progress can be seen in the number of pages collapsed:
|
|
|
|
|
|
|
|
/sys/kernel/mm/transparent_hugepage/khugepaged/pages_collapsed
|
|
|
|
|
|
|
|
for each pass:
|
|
|
|
|
|
|
|
/sys/kernel/mm/transparent_hugepage/khugepaged/full_scans
|
|
|
|
|
2015-02-26 14:34:36 -07:00
|
|
|
max_ptes_none specifies how many extra small pages (that are
|
|
|
|
not already mapped) can be allocated when collapsing a group
|
|
|
|
of small pages into one large page.
|
|
|
|
|
|
|
|
/sys/kernel/mm/transparent_hugepage/khugepaged/max_ptes_none
|
|
|
|
|
|
|
|
A higher value leads to use additional memory for programs.
|
|
|
|
A lower value leads to gain less thp performance. Value of
|
|
|
|
max_ptes_none can waste cpu time very little, you can
|
|
|
|
ignore it.
|
|
|
|
|
2015-11-05 19:47:32 -07:00
|
|
|
max_ptes_swap specifies how many pages can be brought in from
|
|
|
|
swap when collapsing a group of pages into a transparent huge page.
|
|
|
|
|
|
|
|
/sys/kernel/mm/transparent_hugepage/khugepaged/max_ptes_swap
|
|
|
|
|
|
|
|
A higher value can cause excessive swap IO and waste
|
|
|
|
memory. A lower value can prevent THPs from being
|
|
|
|
collapsed, resulting fewer pages being collapsed into
|
|
|
|
THPs, and lower memory access performance.
|
|
|
|
|
2011-01-13 16:46:30 -07:00
|
|
|
== Boot parameter ==
|
|
|
|
|
|
|
|
You can change the sysfs boot time defaults of Transparent Hugepage
|
|
|
|
Support by passing the parameter "transparent_hugepage=always" or
|
|
|
|
"transparent_hugepage=madvise" or "transparent_hugepage=never"
|
|
|
|
(without "") to the kernel command line.
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
== Hugepages in tmpfs/shmem ==
|
|
|
|
|
|
|
|
You can control hugepage allocation policy in tmpfs with mount option
|
|
|
|
"huge=". It can have following values:
|
|
|
|
|
|
|
|
- "always":
|
|
|
|
Attempt to allocate huge pages every time we need a new page;
|
|
|
|
|
|
|
|
- "never":
|
|
|
|
Do not allocate huge pages;
|
|
|
|
|
|
|
|
- "within_size":
|
|
|
|
Only allocate huge page if it will be fully within i_size.
|
|
|
|
Also respect fadvise()/madvise() hints;
|
|
|
|
|
|
|
|
- "advise:
|
|
|
|
Only allocate huge pages if requested with fadvise()/madvise();
|
|
|
|
|
|
|
|
The default policy is "never".
|
|
|
|
|
|
|
|
"mount -o remount,huge= /mountpoint" works fine after mount: remounting
|
|
|
|
huge=never will not attempt to break up huge pages at all, just stop more
|
|
|
|
from being allocated.
|
|
|
|
|
|
|
|
There's also sysfs knob to control hugepage allocation policy for internal
|
|
|
|
shmem mount: /sys/kernel/mm/transparent_hugepage/shmem_enabled. The mount
|
|
|
|
is used for SysV SHM, memfds, shared anonymous mmaps (of /dev/zero or
|
|
|
|
MAP_ANONYMOUS), GPU drivers' DRM objects, Ashmem.
|
|
|
|
|
|
|
|
In addition to policies listed above, shmem_enabled allows two further
|
|
|
|
values:
|
|
|
|
|
|
|
|
- "deny":
|
|
|
|
For use in emergencies, to force the huge option off from
|
|
|
|
all mounts;
|
|
|
|
- "force":
|
|
|
|
Force the huge option on for all - very useful for testing;
|
|
|
|
|
2011-01-13 16:46:30 -07:00
|
|
|
== Need of application restart ==
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
The transparent_hugepage/enabled values and tmpfs mount option only affect
|
|
|
|
future behavior. So to make them effective you need to restart any
|
|
|
|
application that could have been using hugepages. This also applies to the
|
|
|
|
regions registered in khugepaged.
|
2011-01-13 16:46:30 -07:00
|
|
|
|
2012-05-29 16:06:45 -06:00
|
|
|
== Monitoring usage ==
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
The number of anonymous transparent huge pages currently used by the
|
|
|
|
system is available by reading the AnonHugePages field in /proc/meminfo.
|
|
|
|
To identify what applications are using anonymous transparent huge pages,
|
|
|
|
it is necessary to read /proc/PID/smaps and count the AnonHugePages fields
|
|
|
|
for each mapping.
|
|
|
|
|
|
|
|
The number of file transparent huge pages mapped to userspace is available
|
|
|
|
by reading ShmemPmdMapped and ShmemHugePages fields in /proc/meminfo.
|
|
|
|
To identify what applications are mapping file transparent huge pages, it
|
|
|
|
is necessary to read /proc/PID/smaps and count the FileHugeMapped fields
|
|
|
|
for each mapping.
|
|
|
|
|
|
|
|
Note that reading the smaps file is expensive and reading it
|
|
|
|
frequently will incur overhead.
|
2012-05-29 16:06:45 -06:00
|
|
|
|
|
|
|
There are a number of counters in /proc/vmstat that may be used to
|
|
|
|
monitor how successfully the system is providing huge pages for use.
|
|
|
|
|
|
|
|
thp_fault_alloc is incremented every time a huge page is successfully
|
|
|
|
allocated to handle a page fault. This applies to both the
|
|
|
|
first time a page is faulted and for COW faults.
|
|
|
|
|
|
|
|
thp_collapse_alloc is incremented by khugepaged when it has found
|
|
|
|
a range of pages to collapse into one huge page and has
|
|
|
|
successfully allocated a new huge page to store the data.
|
|
|
|
|
|
|
|
thp_fault_fallback is incremented if a page fault fails to allocate
|
|
|
|
a huge page and instead falls back to using small pages.
|
|
|
|
|
|
|
|
thp_collapse_alloc_failed is incremented if khugepaged found a range
|
|
|
|
of pages that should be collapsed into one huge page but failed
|
|
|
|
the allocation.
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
thp_file_alloc is incremented every time a file huge page is successfully
|
|
|
|
i allocated.
|
|
|
|
|
|
|
|
thp_file_mapped is incremented every time a file huge page is mapped into
|
|
|
|
user address space.
|
|
|
|
|
2016-01-15 17:54:30 -07:00
|
|
|
thp_split_page is incremented every time a huge page is split into base
|
2012-05-29 16:06:45 -06:00
|
|
|
pages. This can happen for a variety of reasons but a common
|
|
|
|
reason is that a huge page is old and is being reclaimed.
|
2016-01-15 17:54:30 -07:00
|
|
|
This action implies splitting all PMD the page mapped with.
|
|
|
|
|
|
|
|
thp_split_page_failed is is incremented if kernel fails to split huge
|
|
|
|
page. This can happen if the page was pinned by somebody.
|
|
|
|
|
2016-03-17 15:18:45 -06:00
|
|
|
thp_deferred_split_page is incremented when a huge page is put onto split
|
|
|
|
queue. This happens when a huge page is partially unmapped and
|
|
|
|
splitting it would free up some memory. Pages on split queue are
|
|
|
|
going to be split under memory pressure.
|
|
|
|
|
2016-01-15 17:54:30 -07:00
|
|
|
thp_split_pmd is incremented every time a PMD split into table of PTEs.
|
|
|
|
This can happen, for instance, when application calls mprotect() or
|
|
|
|
munmap() on part of huge page. It doesn't split huge page, only
|
|
|
|
page table entry.
|
2012-05-29 16:06:45 -06:00
|
|
|
|
2012-12-12 14:51:09 -07:00
|
|
|
thp_zero_page_alloc is incremented every time a huge zero page is
|
|
|
|
successfully allocated. It includes allocations which where
|
|
|
|
dropped due race with other allocation. Note, it doesn't count
|
|
|
|
every map of the huge zero page, only its allocation.
|
|
|
|
|
|
|
|
thp_zero_page_alloc_failed is incremented if kernel fails to allocate
|
|
|
|
huge zero page and falls back to using small pages.
|
|
|
|
|
2012-05-29 16:06:45 -06:00
|
|
|
As the system ages, allocating huge pages may be expensive as the
|
|
|
|
system uses memory compaction to copy data around memory to free a
|
|
|
|
huge page for use. There are some counters in /proc/vmstat to help
|
|
|
|
monitor this overhead.
|
|
|
|
|
|
|
|
compact_stall is incremented every time a process stalls to run
|
|
|
|
memory compaction so that a huge page is free for use.
|
|
|
|
|
|
|
|
compact_success is incremented if the system compacted memory and
|
|
|
|
freed a huge page for use.
|
|
|
|
|
|
|
|
compact_fail is incremented if the system tries to compact memory
|
|
|
|
but failed.
|
|
|
|
|
|
|
|
compact_pages_moved is incremented each time a page is moved. If
|
|
|
|
this value is increasing rapidly, it implies that the system
|
|
|
|
is copying a lot of data to satisfy the huge page allocation.
|
|
|
|
It is possible that the cost of copying exceeds any savings
|
|
|
|
from reduced TLB misses.
|
|
|
|
|
|
|
|
compact_pagemigrate_failed is incremented when the underlying mechanism
|
|
|
|
for moving a page failed.
|
|
|
|
|
|
|
|
compact_blocks_moved is incremented each time memory compaction examines
|
|
|
|
a huge page aligned range of pages.
|
|
|
|
|
|
|
|
It is possible to establish how long the stalls were using the function
|
|
|
|
tracer to record how long was spent in __alloc_pages_nodemask and
|
|
|
|
using the mm_page_alloc tracepoint to identify which allocations were
|
|
|
|
for huge pages.
|
|
|
|
|
2011-01-13 16:46:30 -07:00
|
|
|
== get_user_pages and follow_page ==
|
|
|
|
|
|
|
|
get_user_pages and follow_page if run on a hugepage, will return the
|
|
|
|
head or tail pages as usual (exactly as they would do on
|
|
|
|
hugetlbfs). Most gup users will only care about the actual physical
|
|
|
|
address of the page and its temporary pinning to release after the I/O
|
|
|
|
is complete, so they won't ever notice the fact the page is huge. But
|
|
|
|
if any driver is going to mangle over the page structure of the tail
|
|
|
|
page (like for checking page->mapping or other bits that are relevant
|
|
|
|
for the head page and not the tail page), it should be updated to jump
|
2016-01-15 17:54:30 -07:00
|
|
|
to check head page instead. Taking reference on any head/tail page would
|
|
|
|
prevent page from being split by anyone.
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
NOTE: these aren't new constraints to the GUP API, and they match the
|
|
|
|
same constrains that applies to hugetlbfs too, so any driver capable
|
|
|
|
of handling GUP on hugetlbfs will also work fine on transparent
|
|
|
|
hugepage backed mappings.
|
|
|
|
|
|
|
|
In case you can't handle compound pages if they're returned by
|
|
|
|
follow_page, the FOLL_SPLIT bit can be specified as parameter to
|
|
|
|
follow_page, so that it will split the hugepages before returning
|
|
|
|
them. Migration for example passes FOLL_SPLIT as parameter to
|
|
|
|
follow_page because it's not hugepage aware and in fact it can't work
|
|
|
|
at all on hugetlbfs (but it instead works fine on transparent
|
|
|
|
hugepages thanks to FOLL_SPLIT). migration simply can't deal with
|
|
|
|
hugepages being returned (as it's not only checking the pfn of the
|
|
|
|
page and pinning it during the copy but it pretends to migrate the
|
|
|
|
memory in regular page sizes and with regular pte/pmd mappings).
|
|
|
|
|
|
|
|
== Optimizing the applications ==
|
|
|
|
|
|
|
|
To be guaranteed that the kernel will map a 2M page immediately in any
|
|
|
|
memory region, the mmap region has to be hugepage naturally
|
|
|
|
aligned. posix_memalign() can provide that guarantee.
|
|
|
|
|
|
|
|
== Hugetlbfs ==
|
|
|
|
|
|
|
|
You can use hugetlbfs on a kernel that has transparent hugepage
|
|
|
|
support enabled just fine as always. No difference can be noted in
|
|
|
|
hugetlbfs other than there will be less overall fragmentation. All
|
|
|
|
usual features belonging to hugetlbfs are preserved and
|
|
|
|
unaffected. libhugetlbfs will also work fine as usual.
|
|
|
|
|
|
|
|
== Graceful fallback ==
|
|
|
|
|
2016-05-20 17:58:07 -06:00
|
|
|
Code walking pagetables but unaware about huge pmds can simply call
|
2016-01-15 17:54:30 -07:00
|
|
|
split_huge_pmd(vma, pmd, addr) where the pmd is the one returned by
|
2011-01-13 16:46:30 -07:00
|
|
|
pmd_offset. It's trivial to make the code transparent hugepage aware
|
2016-01-15 17:54:30 -07:00
|
|
|
by just grepping for "pmd_offset" and adding split_huge_pmd where
|
2011-01-13 16:46:30 -07:00
|
|
|
missing after pmd_offset returns the pmd. Thanks to the graceful
|
|
|
|
fallback design, with a one liner change, you can avoid to write
|
|
|
|
hundred if not thousand of lines of complex code to make your code
|
|
|
|
hugepage aware.
|
|
|
|
|
|
|
|
If you're not walking pagetables but you run into a physical hugepage
|
|
|
|
but you can't handle it natively in your code, you can split it by
|
|
|
|
calling split_huge_page(page). This is what the Linux VM does before
|
2016-01-15 17:54:30 -07:00
|
|
|
it tries to swapout the hugepage for example. split_huge_page() can fail
|
|
|
|
if the page is pinned and you must handle this correctly.
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
Example to make mremap.c transparent hugepage aware with a one liner
|
|
|
|
change:
|
|
|
|
|
|
|
|
diff --git a/mm/mremap.c b/mm/mremap.c
|
|
|
|
--- a/mm/mremap.c
|
|
|
|
+++ b/mm/mremap.c
|
|
|
|
@@ -41,6 +41,7 @@ static pmd_t *get_old_pmd(struct mm_stru
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
pmd = pmd_offset(pud, addr);
|
2016-01-15 17:54:30 -07:00
|
|
|
+ split_huge_pmd(vma, pmd, addr);
|
2011-01-13 16:46:30 -07:00
|
|
|
if (pmd_none_or_clear_bad(pmd))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
== Locking in hugepage aware code ==
|
|
|
|
|
|
|
|
We want as much code as possible hugepage aware, as calling
|
2016-01-15 17:54:30 -07:00
|
|
|
split_huge_page() or split_huge_pmd() has a cost.
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
To make pagetable walks huge pmd aware, all you need to do is to call
|
|
|
|
pmd_trans_huge() on the pmd returned by pmd_offset. You must hold the
|
|
|
|
mmap_sem in read (or write) mode to be sure an huge pmd cannot be
|
|
|
|
created from under you by khugepaged (khugepaged collapse_huge_page
|
|
|
|
takes the mmap_sem in write mode in addition to the anon_vma lock). If
|
|
|
|
pmd_trans_huge returns false, you just fallback in the old code
|
|
|
|
paths. If instead pmd_trans_huge returns true, you have to take the
|
2016-01-15 17:54:30 -07:00
|
|
|
page table lock (pmd_lock()) and re-run pmd_trans_huge. Taking the
|
|
|
|
page table lock will prevent the huge pmd to be converted into a
|
|
|
|
regular pmd from under you (split_huge_pmd can run in parallel to the
|
2011-01-13 16:46:30 -07:00
|
|
|
pagetable walk). If the second pmd_trans_huge returns false, you
|
2016-01-15 17:54:30 -07:00
|
|
|
should just drop the page table lock and fallback to the old code as
|
|
|
|
before. Otherwise you can proceed to process the huge pmd and the
|
|
|
|
hugepage natively. Once finished you can drop the page table lock.
|
|
|
|
|
|
|
|
== Refcounts and transparent huge pages ==
|
|
|
|
|
|
|
|
Refcounting on THP is mostly consistent with refcounting on other compound
|
|
|
|
pages:
|
|
|
|
|
2016-05-19 18:10:49 -06:00
|
|
|
- get_page()/put_page() and GUP operate in head page's ->_refcount.
|
2016-01-15 17:54:30 -07:00
|
|
|
|
2016-05-19 18:10:49 -06:00
|
|
|
- ->_refcount in tail pages is always zero: get_page_unless_zero() never
|
2016-01-15 17:54:30 -07:00
|
|
|
succeed on tail pages.
|
|
|
|
|
|
|
|
- map/unmap of the pages with PTE entry increment/decrement ->_mapcount
|
|
|
|
on relevant sub-page of the compound page.
|
|
|
|
|
|
|
|
- map/unmap of the whole compound page accounted in compound_mapcount
|
2016-07-26 16:26:40 -06:00
|
|
|
(stored in first tail page). For file huge pages, we also increment
|
|
|
|
->_mapcount of all sub-pages in order to have race-free detection of
|
|
|
|
last unmap of subpages.
|
2016-01-15 17:54:30 -07:00
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
PageDoubleMap() indicates that the page is *possibly* mapped with PTEs.
|
|
|
|
|
|
|
|
For anonymous pages PageDoubleMap() also indicates ->_mapcount in all
|
|
|
|
subpages is offset up by one. This additional reference is required to
|
|
|
|
get race-free detection of unmap of subpages when we have them mapped with
|
|
|
|
both PMDs and PTEs.
|
2016-01-15 17:54:30 -07:00
|
|
|
|
|
|
|
This is optimization required to lower overhead of per-subpage mapcount
|
|
|
|
tracking. The alternative is alter ->_mapcount in all subpages on each
|
|
|
|
map/unmap of the whole compound page.
|
|
|
|
|
2016-07-26 16:26:40 -06:00
|
|
|
For anonymous pages, we set PG_double_map when a PMD of the page got split
|
|
|
|
for the first time, but still have PMD mapping. The additional references
|
|
|
|
go away with last compound_mapcount.
|
|
|
|
|
|
|
|
File pages get PG_double_map set on first map of the page with PTE and
|
|
|
|
goes away when the page gets evicted from page cache.
|
2011-01-13 16:46:30 -07:00
|
|
|
|
|
|
|
split_huge_page internally has to distribute the refcounts in the head
|
2016-01-15 17:54:30 -07:00
|
|
|
page to the tail pages before clearing all PG_head/tail bits from the page
|
|
|
|
structures. It can be done easily for refcounts taken by page table
|
|
|
|
entries. But we don't have enough information on how to distribute any
|
|
|
|
additional pins (i.e. from get_user_pages). split_huge_page() fails any
|
|
|
|
requests to split pinned huge page: it expects page count to be equal to
|
|
|
|
sum of mapcount of all sub-pages plus one (split_huge_page caller must
|
|
|
|
have reference for head page).
|
|
|
|
|
2016-05-19 18:10:49 -06:00
|
|
|
split_huge_page uses migration entries to stabilize page->_refcount and
|
2016-07-26 16:26:40 -06:00
|
|
|
page->_mapcount of anonymous pages. File pages just got unmapped.
|
2016-01-15 17:54:30 -07:00
|
|
|
|
|
|
|
We safe against physical memory scanners too: the only legitimate way
|
|
|
|
scanner can get reference to a page is get_page_unless_zero().
|
|
|
|
|
2016-05-20 17:58:07 -06:00
|
|
|
All tail pages have zero ->_refcount until atomic_add(). This prevents the
|
|
|
|
scanner from getting a reference to the tail page up to that point. After the
|
|
|
|
atomic_add() we don't care about the ->_refcount value. We already known how
|
|
|
|
many references should be uncharged from the head page.
|
2016-01-15 17:54:30 -07:00
|
|
|
|
|
|
|
For head page get_page_unless_zero() will succeed and we don't mind. It's
|
|
|
|
clear where reference should go after split: it will stay on head page.
|
|
|
|
|
|
|
|
Note that split_huge_pmd() doesn't have any limitation on refcounting:
|
|
|
|
pmd can be split at any point and never fails.
|
|
|
|
|
|
|
|
== Partial unmap and deferred_split_huge_page() ==
|
|
|
|
|
|
|
|
Unmapping part of THP (with munmap() or other way) is not going to free
|
|
|
|
memory immediately. Instead, we detect that a subpage of THP is not in use
|
|
|
|
in page_remove_rmap() and queue the THP for splitting if memory pressure
|
|
|
|
comes. Splitting will free up unused subpages.
|
|
|
|
|
|
|
|
Splitting the page right away is not an option due to locking context in
|
|
|
|
the place where we can detect partial unmap. It's also might be
|
|
|
|
counterproductive since in many cases partial unmap unmap happens during
|
|
|
|
exit(2) if an THP crosses VMA boundary.
|
|
|
|
|
|
|
|
Function deferred_split_huge_page() is used to queue page for splitting.
|
|
|
|
The splitting itself will happen when we get memory pressure via shrinker
|
|
|
|
interface.
|