From: Linus Torvalds <torvalds@linux-foundation.org> To: Feng Tang <feng.tang@intel.com> Cc: Jiri Olsa <jolsa@redhat.com>, Peter Zijlstra <peterz@infradead.org>, kernel test robot <rong.a.chen@intel.com>, Ingo Molnar <mingo@kernel.org>, Vince Weaver <vincent.weaver@maine.edu>, Jiri Olsa <jolsa@kernel.org>, Alexander Shishkin <alexander.shishkin@linux.intel.com>, Arnaldo Carvalho de Melo <acme@kernel.org>, Arnaldo Carvalho de Melo <acme@redhat.com>, "Naveen N. Rao" <naveen.n.rao@linux.vnet.ibm.com>, Ravi Bangoria <ravi.bangoria@linux.ibm.com>, Stephane Eranian <eranian@google.com>, Thomas Gleixner <tglx@linutronix.de>, LKML <linux-kernel@vger.kernel.org>, lkp@lists.01.org, andi.kleen@intel.com, "Huang, Ying" <ying.huang@intel.com> Subject: Re: [LKP] Re: [perf/x86] 81ec3f3c4c: will-it-scale.per_process_ops -5.5% regression Date: Sun, 23 Feb 2020 09:37:06 -0800 [thread overview] Message-ID: <CAHk-=wjKFTzfDWjAAabHTZcityeLpHmEQRrKdTuk0f4GWcoohQ@mail.gmail.com> (raw) In-Reply-To: <20200223141147.GA53531@shbuild999.sh.intel.com> On Sun, Feb 23, 2020 at 6:11 AM Feng Tang <feng.tang@intel.com> wrote: > > I tried to use perf-c2c on one platform (not the one that show > the 5.5% regression), and found the main "hitm" points to the > "root_user" global data, as there is a task for each CPU doing > the signal stress test, and both __sigqueue_alloc() and > __sigqueue_free() will call get_user() and free_uid() to inc/dec > this root_user's refcount. What's around it for you? There might be that 'uidhash_lock' spinlock right next to it, and maybe that exacerbates the issue? > Then I added some alignement inside struct "user_struct" (for > "root_user"), then the -5.5% is gone, with a +2.6% instead. Do you actually need to align things inside the struct, or is it sufficient to just align the structure itself? IOW, is the cache conflicts _within_ the user_struct itself, or is it with some nearby data (like that uidhash_lock or whatever?) > One thing I don't understand is, this -5.5% only happens in > one 2 sockets, 96C/192T Cascadelake platform, as we've run > the same test on several different platforms. In therory, > the false sharing may also take effect? Is that the biggest machine you have access to? Maybe it just isn't noticeable with smaller core counts. A lot of conflict loads tend to have "exponential" behavior - when things get overloaded, performance plummets because it just makes things worse as everybody gets slower at that contention point and now it gets even more contended... Linus
WARNING: multiple messages have this Message-ID (diff)
From: Linus Torvalds <torvalds@linux-foundation.org> To: lkp@lists.01.org Subject: Re: [perf/x86] 81ec3f3c4c: will-it-scale.per_process_ops -5.5% regression Date: Sun, 23 Feb 2020 09:37:06 -0800 [thread overview] Message-ID: <CAHk-=wjKFTzfDWjAAabHTZcityeLpHmEQRrKdTuk0f4GWcoohQ@mail.gmail.com> (raw) In-Reply-To: <20200223141147.GA53531@shbuild999.sh.intel.com> [-- Attachment #1: Type: text/plain, Size: 1543 bytes --] On Sun, Feb 23, 2020 at 6:11 AM Feng Tang <feng.tang@intel.com> wrote: > > I tried to use perf-c2c on one platform (not the one that show > the 5.5% regression), and found the main "hitm" points to the > "root_user" global data, as there is a task for each CPU doing > the signal stress test, and both __sigqueue_alloc() and > __sigqueue_free() will call get_user() and free_uid() to inc/dec > this root_user's refcount. What's around it for you? There might be that 'uidhash_lock' spinlock right next to it, and maybe that exacerbates the issue? > Then I added some alignement inside struct "user_struct" (for > "root_user"), then the -5.5% is gone, with a +2.6% instead. Do you actually need to align things inside the struct, or is it sufficient to just align the structure itself? IOW, is the cache conflicts _within_ the user_struct itself, or is it with some nearby data (like that uidhash_lock or whatever?) > One thing I don't understand is, this -5.5% only happens in > one 2 sockets, 96C/192T Cascadelake platform, as we've run > the same test on several different platforms. In therory, > the false sharing may also take effect? Is that the biggest machine you have access to? Maybe it just isn't noticeable with smaller core counts. A lot of conflict loads tend to have "exponential" behavior - when things get overloaded, performance plummets because it just makes things worse as everybody gets slower at that contention point and now it gets even more contended... Linus
next prev parent reply other threads:[~2020-02-23 17:37 UTC|newest] Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-02-05 12:32 [perf/x86] 81ec3f3c4c: will-it-scale.per_process_ops -5.5% regression kernel test robot 2020-02-05 12:32 ` kernel test robot 2020-02-05 12:58 ` Peter Zijlstra 2020-02-05 12:58 ` Peter Zijlstra 2020-02-06 3:04 ` [LKP] " Li, Philip 2020-02-06 3:04 ` Li, Philip 2020-02-21 8:03 ` [LKP] " Feng Tang 2020-02-21 8:03 ` Feng Tang 2020-02-21 10:58 ` [LKP] " Peter Zijlstra 2020-02-21 10:58 ` Peter Zijlstra 2020-02-21 13:20 ` [LKP] " Jiri Olsa 2020-02-21 13:20 ` Jiri Olsa 2020-02-23 14:11 ` [LKP] " Feng Tang 2020-02-23 14:11 ` Feng Tang 2020-02-23 17:37 ` Linus Torvalds [this message] 2020-02-23 17:37 ` Linus Torvalds 2020-02-24 0:33 ` [LKP] " Feng Tang 2020-02-24 0:33 ` Feng Tang 2020-02-24 1:06 ` [LKP] " Linus Torvalds 2020-02-24 1:06 ` Linus Torvalds 2020-02-24 1:58 ` [LKP] " Huang, Ying 2020-02-24 1:58 ` Huang, Ying 2020-02-24 2:19 ` [LKP] " Feng Tang 2020-02-24 2:19 ` Feng Tang 2020-02-24 13:20 ` [LKP] " Feng Tang 2020-02-24 13:20 ` Feng Tang 2020-02-24 19:24 ` [LKP] " Linus Torvalds 2020-02-24 19:24 ` Linus Torvalds 2020-02-24 19:42 ` [LKP] " Kleen, Andi 2020-02-24 19:42 ` Kleen, Andi 2020-02-24 20:09 ` [LKP] " Linus Torvalds 2020-02-24 20:09 ` Linus Torvalds 2020-02-24 20:47 ` [LKP] " Linus Torvalds 2020-02-24 20:47 ` Linus Torvalds 2020-02-24 21:20 ` [LKP] " Eric W. Biederman 2020-02-24 21:20 ` Eric W. Biederman 2020-02-24 21:43 ` [LKP] " Linus Torvalds 2020-02-24 21:43 ` Linus Torvalds 2020-02-24 21:59 ` [LKP] " Eric W. Biederman 2020-02-24 21:59 ` Eric W. Biederman 2020-02-24 22:12 ` [LKP] " Linus Torvalds 2020-02-24 22:12 ` Linus Torvalds 2020-02-25 2:57 ` [LKP] " Feng Tang 2020-02-25 2:57 ` Feng Tang 2020-02-25 3:15 ` [LKP] " Linus Torvalds 2020-02-25 3:15 ` Linus Torvalds 2020-02-25 4:53 ` [LKP] " Feng Tang 2020-02-25 4:53 ` Feng Tang 2020-02-23 19:36 ` [LKP] " Jiri Olsa 2020-02-23 19:36 ` Jiri Olsa 2020-02-24 1:14 ` Feng Tang 2020-02-21 18:05 ` [LKP] " Kleen, Andi 2020-02-21 18:05 ` Kleen, Andi 2020-02-22 12:43 ` [LKP] " Feng Tang 2020-02-22 12:43 ` Feng Tang 2020-02-22 17:08 ` [LKP] " Kleen, Andi 2020-02-22 17:08 ` Kleen, Andi
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='CAHk-=wjKFTzfDWjAAabHTZcityeLpHmEQRrKdTuk0f4GWcoohQ@mail.gmail.com' \ --to=torvalds@linux-foundation.org \ --cc=acme@kernel.org \ --cc=acme@redhat.com \ --cc=alexander.shishkin@linux.intel.com \ --cc=andi.kleen@intel.com \ --cc=eranian@google.com \ --cc=feng.tang@intel.com \ --cc=jolsa@kernel.org \ --cc=jolsa@redhat.com \ --cc=linux-kernel@vger.kernel.org \ --cc=lkp@lists.01.org \ --cc=mingo@kernel.org \ --cc=naveen.n.rao@linux.vnet.ibm.com \ --cc=peterz@infradead.org \ --cc=ravi.bangoria@linux.ibm.com \ --cc=rong.a.chen@intel.com \ --cc=tglx@linutronix.de \ --cc=vincent.weaver@maine.edu \ --cc=ying.huang@intel.com \ /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: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.