From: Leonardo Bras <leonardo@linux.ibm.com>
To: linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org,
kvm-ppc@vger.kernel.org, linux-arch@vger.kernel.org,
linux-mm@kvack.org
Cc: Song Liu <songliubraving@fb.com>, Michal Hocko <mhocko@suse.com>,
"Peter Zijlstra (Intel)" <peterz@infradead.org>,
"Dmitry V. Levin" <ldv@altlinux.org>,
Keith Busch <keith.busch@intel.com>,
Paul Mackerras <paulus@samba.org>,
Christoph Lameter <cl@linux.com>, Ira Weiny <ira.weiny@intel.com>,
Thomas Gleixner <tglx@linutronix.de>,
Elena Reshetova <elena.reshetova@intel.com>,
Andrea Arcangeli <aarcange@redhat.com>,
Santosh Sivaraj <santosh@fossix.org>,
Davidlohr Bueso <dave@stgolabs.net>,
"Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com>,
Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>,
Mike Rapoport <rppt@linux.ibm.com>,
Jason Gunthorpe <jgg@ziepe.ca>,
Allison Randal <allison@lohutok.net>,
Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>,
Leonardo Bras <leonardo@linux.ibm.com>,
Alexey Dobriyan <adobriyan@gmail.com>,
Ingo Molnar <mingo@kernel.org>, Ralph Campbell <rcampbe>
Subject: [PATCH v5 00/11] Introduces new count-based method for tracking lockless pagetable walks
Date: Wed, 2 Oct 2019 22:33:14 -0300 [thread overview]
Message-ID: <20191003013325.2614-1-leonardo@linux.ibm.com> (raw)
If a process (qemu) with a lot of CPUs (128) try to munmap() a large
chunk of memory (496GB) mapped with THP, it takes an average of 275
seconds, which can cause a lot of problems to the load (in qemu case,
the guest will lock for this time).
Trying to find the source of this bug, I found out most of this time is
spent on serialize_against_pte_lookup(). This function will take a lot
of time in smp_call_function_many() if there is more than a couple CPUs
running the user process. Since it has to happen to all THP mapped, it
will take a very long time for large amounts of memory.
By the docs, serialize_against_pte_lookup() is needed in order to avoid
pmd_t to pte_t casting inside find_current_mm_pte(), or any lockless
pagetable walk, to happen concurrently with THP splitting/collapsing.
It does so by calling a do_nothing() on each CPU in mm->cpu_bitmap[],
after interrupts are re-enabled.
Since, interrupts are (usually) disabled during lockless pagetable
walk, and serialize_against_pte_lookup will only return after
interrupts are enabled, it is protected.
So, by what I could understand, if there is no lockless pagetable walk
running, there is no need to call serialize_against_pte_lookup().
So, to avoid the cost of running serialize_against_pte_lookup(), I
propose a counter that keeps track of how many find_current_mm_pte()
are currently running, and if there is none, just skip
smp_call_function_many().
The related functions are:
begin_lockless_pgtbl_walk()
Insert before starting any lockless pgtable walk
end_lockless_pgtbl_walk()
Insert after the end of any lockless pgtable walk
(Mostly after the ptep is last used)
running_lockless_pgtbl_walk()
Returns the number of lockless pgtable walks running
On my workload (qemu), I could see munmap's time reduction from 275
seconds to 418ms.
Also, I documented some lockless pagetable walks in which it's not
necessary to keep track, given they work on init_mm or guest pgd.
The patchset works by focusing all steps needed to begin/end lockless
pagetable walks on the above functions, and then adding the config option
to enable the tracking of these functions using the counting method.
Changes since v4:
Rebased on top of v5.4-rc1
Declared real generic functions instead of dummies
start_lockless_pgtbl_walk renamed to begin_lockless_pgtbl_walk
Interrupt {dis,en}able is now inside of {begin,end}_lockless_pgtbl_walk
Power implementation has option to not {dis,en}able interrupt
More documentation inside the funtions.
Some irq maks variables renamed
Removed some proxy mm_structs
Few typos fixed
Changes since v3:
Explain (comments) why some lockless pgtbl walks don't need
local_irq_disable (real mode + MSR_EE=0)
Explain (comments) places where counting method is not needed (guest pgd,
which is not touched by THP)
Fixes some misplaced local_irq_restore()
Link: http://patchwork.ozlabs.org/project/linuxppc-dev/list/?series=132417
Changes since v2:
Rebased to v5.3
Adds support on __get_user_pages_fast
Adds usage decription to *_lockless_pgtbl_walk()
Better style to dummy functions
Link: http://patchwork.ozlabs.org/project/linuxppc-dev/list/?series=131839
Changes since v1:
Isolated atomic operations in functions *_lockless_pgtbl_walk()
Fixed behavior of decrementing before last ptep was used
Link: http://patchwork.ozlabs.org/patch/1163093/
Leonardo Bras (11):
asm-generic/pgtable: Adds generic functions to monitor lockless
pgtable walks
powerpc/mm: Adds counting method to monitor lockless pgtable walks
mm/gup: Applies counting method to monitor gup_pgd_range
powerpc/mce_power: Applies counting method to monitor lockless pgtbl
walks
powerpc/perf: Applies counting method to monitor lockless pgtbl walks
powerpc/mm/book3s64/hash: Applies counting method to monitor lockless
pgtbl walks
powerpc/kvm/e500: Applies counting method to monitor lockless pgtbl
walks
powerpc/kvm/book3s_hv: Applies counting method to monitor lockless
pgtbl walks
powerpc/kvm/book3s_64: Applies counting method to monitor lockless
pgtbl walks
mm/Kconfig: Adds config option to track lockless pagetable walks
powerpc/mm/book3s64/pgtable: Uses counting method to skip serializing
arch/powerpc/include/asm/book3s/64/pgtable.h | 9 ++
arch/powerpc/kernel/mce_power.c | 6 +-
arch/powerpc/kvm/book3s_64_mmu_hv.c | 6 +-
arch/powerpc/kvm/book3s_64_mmu_radix.c | 34 +++++-
arch/powerpc/kvm/book3s_64_vio_hv.c | 7 +-
arch/powerpc/kvm/book3s_hv_nested.c | 22 +++-
arch/powerpc/kvm/book3s_hv_rm_mmu.c | 32 ++---
arch/powerpc/kvm/e500_mmu_host.c | 9 +-
arch/powerpc/mm/book3s64/hash_tlb.c | 6 +-
arch/powerpc/mm/book3s64/hash_utils.c | 27 +++--
arch/powerpc/mm/book3s64/pgtable.c | 120 ++++++++++++++++++-
arch/powerpc/perf/callchain.c | 6 +-
include/asm-generic/pgtable.h | 58 +++++++++
include/linux/mm_types.h | 11 ++
kernel/fork.c | 3 +
mm/Kconfig | 11 ++
mm/gup.c | 10 +-
17 files changed, 325 insertions(+), 52 deletions(-)
--
2.20.1
next reply other threads:[~2019-10-03 1:33 UTC|newest]
Thread overview: 45+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-03 1:33 Leonardo Bras [this message]
2019-10-03 1:33 ` [PATCH v5 01/11] asm-generic/pgtable: Adds generic functions to monitor lockless pgtable walks Leonardo Bras
2019-10-03 7:11 ` Peter Zijlstra
2019-10-03 11:51 ` Peter Zijlstra
2019-10-03 20:40 ` John Hubbard
2019-10-04 11:24 ` Peter Zijlstra
2019-10-03 21:24 ` Leonardo Bras
2019-10-03 21:24 ` Leonardo Bras
2019-10-04 11:28 ` Peter Zijlstra
2019-10-04 11:28 ` Peter Zijlstra
2019-10-09 18:09 ` Leonardo Bras
2019-10-09 18:09 ` Leonardo Bras
2019-10-05 8:35 ` Aneesh Kumar K.V
2019-10-08 14:47 ` Kirill A. Shutemov
2019-10-03 1:33 ` [PATCH v5 02/11] powerpc/mm: Adds counting method " Leonardo Bras
2019-10-08 15:11 ` Christopher Lameter
2019-10-08 17:13 ` Leonardo Bras
2019-10-08 17:43 ` Christopher Lameter
2019-10-08 18:02 ` Leonardo Bras
2019-10-08 18:27 ` Christopher Lameter
2019-10-03 1:33 ` [PATCH v5 03/11] mm/gup: Applies counting method to monitor gup_pgd_range Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 04/11] powerpc/mce_power: Applies counting method to monitor lockless pgtbl walks Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 05/11] powerpc/perf: " Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 06/11] powerpc/mm/book3s64/hash: " Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 07/11] powerpc/kvm/e500: " Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 08/11] powerpc/kvm/book3s_hv: " Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 09/11] powerpc/kvm/book3s_64: " Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 10/11] mm/Kconfig: Adds config option to track lockless pagetable walks Leonardo Bras
2019-10-03 2:08 ` Qian Cai
2019-10-03 19:04 ` Leonardo Bras
2019-10-03 19:08 ` Leonardo Bras
2019-10-03 7:44 ` Peter Zijlstra
2019-10-03 20:40 ` Leonardo Bras
2019-10-03 20:40 ` Leonardo Bras
2019-10-03 1:33 ` [PATCH v5 11/11] powerpc/mm/book3s64/pgtable: Uses counting method to skip serializing Leonardo Bras
2019-10-03 7:29 ` [PATCH v5 00/11] Introduces new count-based method for tracking lockless pagetable walks Peter Zijlstra
2019-10-03 20:36 ` Leonardo Bras
2019-10-03 20:36 ` Leonardo Bras
2019-10-03 20:49 ` John Hubbard
2019-10-03 21:38 ` Leonardo Bras
2019-10-03 21:38 ` Leonardo Bras
2019-10-04 11:42 ` Peter Zijlstra
2019-10-04 11:42 ` Peter Zijlstra
2019-10-04 12:57 ` Peter Zijlstra
2019-10-04 12:57 ` Peter Zijlstra
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20191003013325.2614-1-leonardo@linux.ibm.com \
--to=leonardo@linux.ibm.com \
--cc=aarcange@redhat.com \
--cc=adobriyan@gmail.com \
--cc=allison@lohutok.net \
--cc=aneesh.kumar@linux.ibm.com \
--cc=b.zolnierkie@samsung.com \
--cc=cl@linux.com \
--cc=dave@stgolabs.net \
--cc=elena.reshetova@intel.com \
--cc=ira.weiny@intel.com \
--cc=jgg@ziepe.ca \
--cc=keith.busch@intel.com \
--cc=kvm-ppc@vger.kernel.org \
--cc=ldv@altlinux.org \
--cc=linux-arch@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=mahesh@linux.vnet.ibm.com \
--cc=mhocko@suse.com \
--cc=mingo@kernel.org \
--cc=paulus@samba.org \
--cc=peterz@infradead.org \
--cc=rppt@linux.ibm.com \
--cc=santosh@fossix.org \
--cc=songliubraving@fb.com \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).