This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH v6 1/3] Tunables: Add tunables of spin count for pthread adaptive spin mutex
- From: kemi <kemi dot wang at intel dot com>
- To: "H.J. Lu" <hjl dot tools at gmail dot com>
- Cc: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>, Florian Weimer <fweimer at redhat dot com>, Rical Jason <rj at 2c3t dot io>, Carlos Donell <carlos at redhat dot com>, Glibc alpha <libc-alpha at sourceware dot org>, Dave Hansen <dave dot hansen at linux dot intel dot com>, Tim Chen <tim dot c dot chen at intel dot com>, Andi Kleen <andi dot kleen at intel dot com>, Ying Huang <ying dot huang at intel dot com>, Aaron Lu <aaron dot lu at intel dot com>, Lu Aubrey <aubrey dot li at intel dot com>
- Date: Wed, 4 Jul 2018 12:51:30 +0800
- Subject: Re: [PATCH v6 1/3] Tunables: Add tunables of spin count for pthread adaptive spin mutex
- References: <1530520046-18343-1-git-send-email-kemi.wang@intel.com> <CAMe9rOopnM=io7k9z4a2LzjuYZMtwH8xUuNbRMLcTTLzHBeCYA@mail.gmail.com> <CAMe9rOrtBocsbRtdG6M9=A=KQu8HbZ=1gj8=hJ=hZZnOiS3WJA@mail.gmail.com>
On 2018年07月03日 21:40, H.J. Lu wrote:
> On Tue, Jul 3, 2018 at 5:32 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Mon, Jul 2, 2018 at 1:27 AM, Kemi Wang <kemi.wang@intel.com> wrote:
>>> This patch does not have any functionality change, we only provide a spin
>>> count tunes for pthread adaptive spin mutex. The tunable
>>> glibc.mutex.spin_count tunes can be used by system administrator to squeeze
>>> system performance according to different hardware capabilities and
>>> workload characteristics.
>>>
>>> The maximum value of spin count is limited to 30000 to avoid the overflow
>>> of mutex->__data.__spins variable with the possible type of short in
>>> pthread_mutex_lock ().
>>>
>>> The default value of spin count is set to 100 with the reference to the
>>> previous number of times of spinning via trylock. This value would be
>>> architecture-specific and can be tuned with kinds of benchmarks to fit most
>>> cases in future.
>>>
>>> This is the preparation work for the next patch, in which the way of
>>> adaptive spin would be changed from an expensive cmpxchg to read while
>>> spinning.
>>>
>>> * elf/dl-tunables.list: Add glibc.mutex.spin_count entry.
>>> * manual/tunables.texi: Add glibc.mutex.spin_count description.
>>> * nptl/Makefile: Add pthread_mutex_conf.c for compilation.
>>> * nptl/pthread_mutex_conf.h: New file.
>>> * nptl/pthread_mutex_conf.c: New file.
>>> * nptl/nptl-init.c: Put mutex tunable initialization in pthread
>>> initialization.
>>>
>>> ChangeLog:
>>> V5->V6:
>>> a) Missing "pthread mutex tunables" entry in the menu of tunables.texi,
>>> add it.
>>>
>>> V4->V5
>>> a) Put mutex tunable (glibc.mutex.spin_count) initialization as part of
>>> overall pthread initialization, that would avoid the extra relocation,
>>> as suggested by Florian Weimer. Thanks for pointing it out!
>>> b) Move the READ_ONLY_SPIN macro definition from the third patch to
>>> this patch
>>>
>>> V3->V4
>>> a) Add comments in elf/dl-tunables.list
>>>
>>> V2->V3
>>> a) Polish the description of glibc.mutex.spin_count tunable with the
>>> help from Rical Jasan.
>>> b) Get rid of the TUNABLE_CALLBACK_FNDECL macros in
>>> pthread_mutex_conf.c, as suggested by Florian Weimer.
>>> c) Adjust the default value of spin count to 100 with the reference of
>>> the previous spinning way via trylock.
>>>
>>> V1->V2
>>> a) Renamed nptl/mutex-conf.h -> nptl/pthread_mutex_conf.h
>>> b) Renamed nptl/mutex-conf.c -> nptl/pthread_mutex_conf.c
>>> c) Change the Makefile to compile pthread_mutex_conf.c
>>> d) Modify the copyright "2013-2018" -> "2018" for new added files
>>> e) Fix the indentation issue (tab -> double space) in
>>> elf/dl-tunables.list
>>> f) Remove the env alias LD_SPIN_COUNT in elf/dl-tunables.list
>>> g) Fix the typo errors and refresh glibc.mutex.spin_count tunable
>>> description in manual/tunables.texi.
>>> h) Fix the indentation issue in nptl/pthread_mutex_conf.c
>>> i) Fix the indentation issue for nested preprocessor (add one space for
>>> each level)
>>>
>>> Suggested-by: Andi Kleen <andi.kleen@intel.com>
>>> Signed-off-by: Kemi Wang <kemi.wang@intel.com>
>
> Please take a look at
>
> https://github.com/hjl-tools/glibc/commits/hjl/spin/master
>
Reviewed. Thanks for refining, more clear!
>