Hi everyone,
While I'm working with a tiered memory system e.g. CXL memory, I have
been facing migration overhead esp. tlb shootdown on promotion or
demotion between different tiers. Yeah.. most tlb shootdowns on
migration through hinting fault can be avoided thanks to Huang Ying's
work, commit 4d4b6d66db ("mm,unmap: avoid flushing tlb in batch if PTE
is inaccessible").
However, it's only for migration through hinting fault. I thought it'd
be much better if we have a general mechanism to reduce all the tlb
numbers that we can apply to any unmap code, that we normally believe
tlb flush should be followed.
I'm suggesting a new mechanism, LUF(Lazy Unmap Flush), that defers tlb
flush until folios that have been unmapped and freed, eventually get
allocated again. It's safe for folios that had been mapped read-only
and were unmapped, as long as the contents of the folios don't change
while staying in pcp or buddy so we can still read the data through the
stale tlb entries.
tlb flush can be defered when folios get unmapped as long as it
guarantees to perform tlb flush needed, before the folios actually
become used, of course, only if all the corresponding ptes don't have
write permission. Otherwise, the system will get messed up.
To achieve that, for the folios that map only to non-writable tlb
entries, prevent tlb flush during unmapping but perform it just before
the folios actually become used, out of buddy or pcp.
However, we should cancel the pending by LUF and perform the deferred
TLB flush right away when:
1. a writable pte is newly set through fault handler
2. a file is updated
3. kasan needs poisoning on free
4. the kernel wants to init pages on free
No matter what type of workload is used for performance evaluation, the
result would be positive thanks to the unconditional reduction of tlb
flushes, tlb misses and interrupts. For the test, I picked up one of
the most popular and heavy workload, llama.cpp that is a
LLM(Large Language Model) inference engine.
The result would depend on memory latency and how often reclaim runs,
which implies tlb miss overhead and how many times unmapping happens.
In my system, the result shows:
1. tlb shootdown interrupts are reduced about 97%.
2. The test program runtime is reduced about 4.5%.
The test environment and the result is like:
Machine: bare metal, x86_64, Intel(R) Xeon(R) Gold 6430
CPU: 1 socket 64 core with hyper thread on
Numa: 2 nodes (64 CPUs DRAM 42GB, no CPUs CXL expander 98GB)
Config: swap off, numa balancing tiering on, demotion enabled
The test set:
llama.cpp/main -m $(70G_model1) -p "who are you?" -s 1 -t 15 -n 20 &
llama.cpp/main -m $(70G_model2) -p "who are you?" -s 1 -t 15 -n 20 &
llama.cpp/main -m $(70G_model3) -p "who are you?" -s 1 -t 15 -n 20 &
wait
where -t: nr of threads, -s: seed used to make the runtime stable,
-n: nr of tokens that determines the runtime, -p: prompt to ask,
-m: LLM model to use.
Run the test set 5 times successively with caches dropped every run
via 'echo 3 > /proc/sys/vm/drop_caches'. Each inference prints its
runtime at the end of each.
1. Runtime from the output of llama.cpp:
BEFORE
------
llama_print_timings: total time = 883450.54 ms / 24 tokens
llama_print_timings: total time = 861665.91 ms / 24 tokens
llama_print_timings: total time = 898079.02 ms / 24 tokens
llama_print_timings: total time = 879897.69 ms / 24 tokens
llama_print_timings: total time = 892360.75 ms / 24 tokens
llama_print_timings: total time = 884587.85 ms / 24 tokens
llama_print_timings: total time = 861023.19 ms / 24 tokens
llama_print_timings: total time = 900022.18 ms / 24 tokens
llama_print_timings: total time = 878771.88 ms / 24 tokens
llama_print_timings: total time = 889027.98 ms / 24 tokens
llama_print_timings: total time = 880783.90 ms / 24 tokens
llama_print_timings: total time = 856475.29 ms / 24 tokens
llama_print_timings: total time = 896842.21 ms / 24 tokens
llama_print_timings: total time = 878883.53 ms / 24 tokens
llama_print_timings: total time = 890122.10 ms / 24 tokens
AFTER
-----
llama_print_timings: total time = 871060.86 ms / 24 tokens
llama_print_timings: total time = 825609.53 ms / 24 tokens
llama_print_timings: total time = 836854.81 ms / 24 tokens
llama_print_timings: total time = 843147.99 ms / 24 tokens
llama_print_timings: total time = 831426.65 ms / 24 tokens
llama_print_timings: total time = 873939.23 ms / 24 tokens
llama_print_timings: total time = 826127.69 ms / 24 tokens
llama_print_timings: total time = 835489.26 ms / 24 tokens
llama_print_timings: total time = 842589.62 ms / 24 tokens
llama_print_timings: total time = 833700.66 ms / 24 tokens
llama_print_timings: total time = 875996.19 ms / 24 tokens
llama_print_timings: total time = 826401.73 ms / 24 tokens
llama_print_timings: total time = 839341.28 ms / 24 tokens
llama_print_timings: total time = 841075.10 ms / 24 tokens
llama_print_timings: total time = 835136.41 ms / 24 tokens
2. tlb shootdowns from 'cat /proc/interrupts':
BEFORE
------
TLB:
80911532 93691786 100296251 111062810 109769109 109862429
108968588 119175230 115779676 118377498 119325266 120300143
124514185 116697222 121068466 118031913 122660681 117494403
121819907 116960596 120936335 117217061 118630217 122322724
119595577 111693298 119232201 120030377 115334687 113179982
118808254 116353592 140987367 137095516 131724276 139742240
136501150 130428761 127585535 132483981 133430250 133756207
131786710 126365824 129812539 133850040 131742690 125142213
128572830 132234350 131945922 128417707 133355434 129972846
126331823 134050849 133991626 121129038 124637283 132830916
126875507 122322440 125776487 124340278 TLB shootdowns
AFTER
-----
TLB:
2121206 2615108 2983494 2911950 3055086 3092672
3204894 3346082 3286744 3307310 3357296 3315940
3428034 3112596 3143325 3185551 3186493 3322314
3330523 3339663 3156064 3272070 3296309 3198962
3332662 3315870 3234467 3353240 3281234 3300666
3345452 3173097 4009196 3932215 3898735 3726531
3717982 3671726 3728788 3724613 3799147 3691764
3620630 3684655 3666688 3393974 3448651 3487593
3446357 3618418 3671920 3712949 3575264 3715385
3641513 3630897 3691047 3630690 3504933 3662647
3629926 3443044 3832970 3548813 TLB shootdowns
---
Changes from v10:
1. Rebase on akpm/mm.git mm-unstable as of May 28, 2024.
2. Cancel LUF on file_end_write() when updating a file.
(feedbacked by Dave Hansen)
3. Cancel LUF after every update_mmu_tlb*() in fault handler if
it's going to set the pte to writable.
4. Cancel LUF on freeing pages if kasan needs poisoning.
(feedbacked by David Hildenbrand)
5. Cancel LUF on freeing pages if want_init_on_free().
(feedbacked by David Hildenbrand)
6. Change test iteration from 10 times to 5 times.
7. Not include perf result. (I will add it if needed.)
8. Trivial optimization.
Changes from v9:
1. Expand the candidate to apply this mechanism:
BEFORE - The souce folios at any type of migration.
AFTER - Any folios that have been unmapped and freed.
2. Change the workload for test:
BEFORE - XSBench
AFTER - llama.cpp (one of the most popluar real workload)
3. Change the test environment:
BEFORE - qemu machine, too small DRAM(1GB), large remote mem
AFTER - bare metal, real CXL memory, practical memory size
4. Rename the mechanism from MIGRC(Migration Read Copy) to
LUF(Lazy Unmap Flush) to reflect the current version of the
mechanism can be applied not only to unmap during migration
but any unmap code e.g. unmap in shrink_folio_list().
5. Fix build error for riscv. (feedbacked by kernel test bot)
6. Supplement commit messages to describe what this mechanism is
for, especially in the patches for arch code. (feedbacked by
Thomas Gleixner)
7. Clean up some trivial things.
Changes from v8:
1. Rebase on akpm/mm.git mm-unstable as of April 18, 2024.
2. Supplement comments and commit message.
3. Change the candidate to apply migrc mechanism:
BEFORE - The source folios at demotion and promotion.
AFTER - The souce folios at any type of migration.
4. Change how migrc mechanism works:
BEFORE - Reduce tlb flushes by deferring folio_free() for
source folios during demotion and promotion.
AFTER - Reduce tlb flushes by deferring tlb flush until they
actually become used, out of pcp or buddy. The
current version of migrc does *not* defer calling
folio_free() but let it go as it is as the same as
vanilla kernel, with the folios marked kind of 'need
to tlb flush'. And then handle the flush when the
page exits from pcp or buddy so as to prevent
changing vm stats e.g. free pages.
Changes from v7:
1. Rewrite cover letter to explain what 'migrc' mechasism is.
(feedbacked by Andrew Morton)
2. Supplement the commit message of a patch 'mm: Add APIs to
free a folio directly to the buddy bypassing pcp'.
(feedbacked by Andrew Morton)
Changes from v6:
1. Fix build errors in case of
CONFIG_ARCH_WANT_BATCHED_UNMAP_tlb_FLUSH disabled by moving
migrc_flush_{start,end}() calls from arch code to
try_to_unmap_flush() in mm/rmap.c.
Changes from v5:
1. Fix build errors in case of CONFIG_MIGRATION disabled or
CONFIG_HWPOISON_INJECT moduled. (feedbacked by kernel test
bot and Raymond Jay Golo)
2. Organize migrc code with two kconfigs, CONFIG_MIGRATION and
CONFIG_ARCH_WANT_BATCHED_UNMAP_tlb_FLUSH.
Changes from v4:
1. Rebase on v6.7.
2. Fix build errors in arm64 that is doing nothing for tlb flush
but has CONFIG_ARCH_WANT_BATCHED_UNMAP_tlb_FLUSH. (reported
by kernel test robot)
3. Don't use any page flag. So the system would give up migrc
mechanism more often but it's okay. The final improvement is
good enough.
4. Instead, optimize full tlb flush(arch_tlbbatch_flush()) by
avoiding redundant CPUs from tlb flush.
Changes from v3:
1. Don't use the kconfig, CONFIG_MIGRC, and remove sysctl knob,
migrc_enable. (feedbacked by Nadav)
2. Remove the optimization skipping CPUs that have already
performed tlb flushes needed by any reason when performing
tlb flushes by migrc because I can't tell the performance
difference between w/ the optimization and w/o that.
(feedbacked by Nadav)
3. Minimize arch-specific code. While at it, move all the migrc
declarations and inline functions from include/linux/mm.h to
mm/internal.h (feedbacked by Dave Hansen, Nadav)
4. Separate a part making migrc paused when the system is in
high memory pressure to another patch. (feedbacked by Nadav)
5. Rename:
a. arch_tlbbatch_clean() to arch_tlbbatch_clear(),
b. tlb_ubc_nowr to tlb_ubc_ro,
c. migrc_try_flush_free_folios() to migrc_flush_free_folios(),
d. migrc_stop to migrc_pause.
(feedbacked by Nadav)
6. Use ->lru list_head instead of introducing a new llist_head.
(feedbacked by Nadav)
7. Use non-atomic operations of page-flag when it's safe.
(feedbacked by Nadav)
8. Use stack instead of keeping a pointer of 'struct migrc_req'
in struct task, which is for manipulating it locally.
(feedbacked by Nadav)
9. Replace a lot of simple functions to inline functions placed
in a header, mm/internal.h. (feedbacked by Nadav)
10. Add additional sufficient comments. (feedbacked by Nadav)
11. Remove a lot of wrapper functions. (feedbacked by Nadav)
Changes from RFC v2:
1. Remove additional occupation in struct page. To do that,
unioned with lru field for migrc's list and added a page
flag. I know page flag is a thing that we don't like to add
but no choice because migrc should distinguish folios under
migrc's control from others. Instead, I force migrc to be
used only on 64 bit system to mitigate you guys from getting
angry.
2. Remove meaningless internal object allocator that I
introduced to minimize impact onto the system. However, a ton
of tests showed there was no difference.
3. Stop migrc from working when the system is in high memory
pressure like about to perform direct reclaim. At the
condition where the swap mechanism is heavily used, I found
the system suffered from regression without this control.
4. Exclude folios that pte_dirty() == true from migrc's interest
so that migrc can work simpler.
5. Combine several patches that work tightly coupled to one.
6. Add sufficient comments for better review.
7. Manage migrc's request in per-node manner (from globally).
8. Add tlb miss improvement in commit message.
9. Test with more CPUs(4 -> 16) to see bigger improvement.
Changes from RFC:
1. Fix a bug triggered when a destination folio at the previous
migration becomes a source folio at the next migration,
before the folio gets handled properly so that the folio can
play with another migration. There was inconsistency in the
folio's state. Fixed it.
2. Split the patch set into more pieces so that the folks can
review better. (Feedbacked by Nadav Amit)
3. Fix a wrong usage of barrier e.g. smp_mb__after_atomic().
(Feedbacked by Nadav Amit)
4. Tried to add sufficient comments to explain the patch set
better. (Feedbacked by Nadav Amit)
Byungchul Park (12):
x86/tlb: add APIs manipulating tlb batch's arch data
arm64: tlbflush: add APIs manipulating tlb batch's arch data
riscv, tlb: add APIs manipulating tlb batch's arch data
x86/tlb, riscv/tlb, mm/rmap: separate arch_tlbbatch_clear() out of
arch_tlbbatch_flush()
mm: buddy: make room for a new variable, ugen, in struct page
mm: add folio_put_ugen() to deliver unmap generation number to pcp or
buddy
mm: add a parameter, unmap generation number, to free_unref_folios()
mm/rmap: recognize read-only tlb entries during batched tlb flush
mm: implement LUF(Lazy Unmap Flush) defering tlb flush when folios get
unmapped
mm: separate move/undo parts from migrate_pages_batch()
mm, migrate: apply luf mechanism to unmapping during migration
mm, vmscan: apply luf mechanism to unmapping during folio reclaim
arch/arm64/include/asm/tlbflush.h | 18 ++
arch/riscv/include/asm/tlbflush.h | 21 ++
arch/riscv/mm/tlbflush.c | 1 -
arch/x86/include/asm/tlbflush.h | 18 ++
arch/x86/mm/tlb.c | 2 -
include/linux/fs.h | 6 +
include/linux/mm.h | 22 ++
include/linux/mm_types.h | 48 +++-
include/linux/rmap.h | 7 +-
include/linux/sched.h | 11 +
mm/compaction.c | 12 +-
mm/internal.h | 115 +++++++++-
mm/memory.c | 39 +++-
mm/migrate.c | 184 ++++++++++------
mm/page_alloc.c | 174 ++++++++++++---
mm/page_isolation.c | 6 +
mm/page_reporting.c | 10 +
mm/rmap.c | 352 +++++++++++++++++++++++++++++-
mm/swap.c | 18 +-
mm/vmscan.c | 29 ++-
20 files changed, 959 insertions(+), 134 deletions(-)
base-commit: b610f75d19a34b488021b9a4d2e3bd1cf34fc200
--
2.17.1
