[PATCH 0/10] Replace _PAGE_NUMA with PAGE_NONE protections v2
Mel Gorman
mgorman at suse.de
Thu Nov 20 21:19:40 AEDT 2014
V1 failed while running under kvm-tools very quickly and a second report
indicated that it happens on bare metal as well. This version survived
an overnight run of trinity running under kvm-tools here but verification
from Sasha would be appreciated.
Changelog since V1
o ppc64 paranoia checks and clarifications (aneesh)
o Fix trinity regression (hopefully)
o Reduce unnecessary TLB flushes (mel)
Automatic NUMA balancing depends on being able to protect PTEs to trap a
fault and gather reference locality information. Very broadly speaking it
would mark PTEs as not present and use another bit to distinguish between
NUMA hinting faults and other types of faults. It was universally loved
by everybody and caused no problems whatsoever. That last sentence might
be a lie.
This series is very heavily based on patches from Linus and Aneesh to
replace the existing PTE/PMD NUMA helper functions with normal change
protections. I did alter and add parts of it but I consider them relatively
minor contributions. At their suggestion, acked-bys are in there but I've
no problem converting them to Signed-off-by if requested.
AFAIK, this has received no testing on ppc64 and I'm depending on Aneesh for
that. I tested trinity under kvm-tool and passed and ran a few other basic
tests. In most cases I'm leaving out detail as it's not that interesting.
specjbb single JVM: There was negligible performance difference in the
benchmark itself for short and long runs. However, system activity
is higher and interrupts are much higher over time -- possibly
TLB flushes. Migrations are also higher. Overall, this is more
overhead but considering the problems faced with the old approach
I think we just have to suck it up and find another way of reducing
the overhead.
specjbb multi JVM: Negligible performance difference to the actual benchmarm
but like the single JVM case, the system overhead is noticably
higher. Again, interrupts are a major factor.
autonumabench: This was all over the place and about all that can be
reasonably concluded is that it's different but not necessarily
better or worse.
autonumabench
3.18.0-rc4 3.18.0-rc4
vanilla protnone-v2r5
User NUMA01 32806.01 ( 0.00%) 20250.67 ( 38.27%)
User NUMA01_THEADLOCAL 23910.28 ( 0.00%) 22734.37 ( 4.92%)
User NUMA02 3176.85 ( 0.00%) 3082.68 ( 2.96%)
User NUMA02_SMT 1600.06 ( 0.00%) 1547.08 ( 3.31%)
System NUMA01 719.07 ( 0.00%) 1344.39 (-86.96%)
System NUMA01_THEADLOCAL 916.26 ( 0.00%) 180.90 ( 80.26%)
System NUMA02 20.92 ( 0.00%) 17.34 ( 17.11%)
System NUMA02_SMT 8.76 ( 0.00%) 7.24 ( 17.35%)
Elapsed NUMA01 728.27 ( 0.00%) 519.28 ( 28.70%)
Elapsed NUMA01_THEADLOCAL 589.15 ( 0.00%) 554.73 ( 5.84%)
Elapsed NUMA02 81.20 ( 0.00%) 81.72 ( -0.64%)
Elapsed NUMA02_SMT 80.49 ( 0.00%) 79.58 ( 1.13%)
CPU NUMA01 4603.00 ( 0.00%) 4158.00 ( 9.67%)
CPU NUMA01_THEADLOCAL 4213.00 ( 0.00%) 4130.00 ( 1.97%)
CPU NUMA02 3937.00 ( 0.00%) 3793.00 ( 3.66%)
CPU NUMA02_SMT 1998.00 ( 0.00%) 1952.00 ( 2.30%)
System CPU usage of NUMA01 is worse but it's an adverse workload on this
machine so I'm reluctant to conclude that it's a problem that matters. On
the other workloads that are sensible on this machine, system CPU usage
is great. Overall time to complete the benchmark is comparable
3.18.0-rc4 3.18.0-rc4
vanillaprotnone-v2r5
User 61493.38 47615.01
System 1665.17 1550.07
Elapsed 1480.79 1236.74
NUMA alloc hit 4739774 5328362
NUMA alloc miss 0 0
NUMA interleave hit 0 0
NUMA alloc local 4664980 5328351
NUMA base PTE updates 556489407 444119981
NUMA huge PMD updates 1086000 866680
NUMA page range updates 1112521407 887860141
NUMA hint faults 1538964 1242142
NUMA hint local faults 835871 814313
NUMA hint local percent 54 65
NUMA pages migrated 7329212 59883854
The NUMA pages migrated look terrible but when I looked at a graph of the
activity over time I see that the massive spike in migration activity was
during NUMA01. This correlates with high system CPU usage and could be simply
down to bad luck but any modifications that affect that workload would be
related to scan rates and migrations, not the protection mechanism. For
all other workloads, migration activity was comparable.
Overall, headline performance figures are comparable but the overhead
is higher, mostly in interrupts. To some extent, higher overhead from
this approach was anticipated but not to this degree. It's going to be
necessary to reduce this again with a separate series in the future. It's
still worth going ahead with this series though as it's likely to avoid
constant headaches with Xen and is probably easier to maintain.
arch/powerpc/include/asm/pgtable.h | 53 ++----------
arch/powerpc/include/asm/pte-common.h | 5 --
arch/powerpc/include/asm/pte-hash64.h | 6 --
arch/powerpc/kvm/book3s_hv_rm_mmu.c | 2 +-
arch/powerpc/mm/copro_fault.c | 8 +-
arch/powerpc/mm/fault.c | 25 ++----
arch/powerpc/mm/gup.c | 4 +-
arch/powerpc/mm/pgtable.c | 8 +-
arch/powerpc/mm/pgtable_64.c | 3 +-
arch/x86/include/asm/pgtable.h | 46 +++++-----
arch/x86/include/asm/pgtable_64.h | 5 --
arch/x86/include/asm/pgtable_types.h | 41 +--------
arch/x86/mm/gup.c | 4 +-
include/asm-generic/pgtable.h | 152 ++--------------------------------
include/linux/migrate.h | 4 -
include/linux/swapops.h | 2 +-
include/uapi/linux/mempolicy.h | 2 +-
mm/gup.c | 8 +-
mm/huge_memory.c | 50 ++++++-----
mm/memory.c | 18 ++--
mm/mempolicy.c | 2 +-
mm/migrate.c | 8 +-
mm/mprotect.c | 48 +++++------
mm/pgtable-generic.c | 2 -
24 files changed, 131 insertions(+), 375 deletions(-)
--
2.1.2
More information about the Linuxppc-dev
mailing list