Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210123011704.1901835-1-joel@joelfernandes.org>
 <20210123011704.1901835-3-joel@joelfernandes.org> <0e91838e-4cca-4c3b-cb36-226c098f36c9@oracle.com>
 <YDTFWZPdmrDuYd91@hirez.programming.kicks-ass.net> <e1ee6187-77a7-dbf2-3e14-adba48460f5b@oracle.com>
In-Reply-To: <e1ee6187-77a7-dbf2-3e14-adba48460f5b@oracle.com>
From:   Josh Don <joshdon@google.com>
Date:   Tue, 23 Feb 2021 21:15:59 -0800
Message-ID: <CABk29NvX9_RxpZ71ihR7Y_Nhpg0TpBfdXzehptO52VuwOmS2Ww@mail.gmail.com>
Subject: Re: [PATCH v10 2/5] sched: CGroup tagging interface for core scheduling
To:     Chris Hyser <chris.hyser@oracle.com>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        "Joel Fernandes (Google)" <joel@joelfernandes.org>,
        Nishanth Aravamudan <naravamudan@digitalocean.com>,
        Julien Desfossez <jdesfossez@digitalocean.com>,
        Tim Chen <tim.c.chen@linux.intel.com>,
        Vineeth Pillai <viremana@linux.microsoft.com>,
        Aaron Lu <aaron.lwe@gmail.com>,
        Aubrey Li <aubrey.intel@gmail.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        linux-kernel <linux-kernel@vger.kernel.org>, mingo@kernel.org,
        torvalds@linux-foundation.org, fweisbec@gmail.com,
        Kees Cook <keescook@chromium.org>,
        Greg Kerr <kerrnel@google.com>, Phil Auld <pauld@redhat.com>,
        Valentin Schneider <valentin.schneider@arm.com>,
        Mel Gorman <mgorman@techsingularity.net>,
        Pawan Gupta <pawan.kumar.gupta@linux.intel.com>,
        Paolo Bonzini <pbonzini@redhat.com>, vineeth@bitbyteword.org,
        Chen Yu <yu.c.chen@intel.com>,
        Christian Brauner <christian.brauner@ubuntu.com>,
        Agata Gruza <agata.gruza@intel.com>,
        Antonio Gomez Iglesias <antonio.gomez.iglesias@intel.com>,
        graf@amazon.com, konrad.wilk@oracle.com, dfaggioli@suse.com,
        Paul Turner <pjt@google.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Patrick Bellasi <derkling@google.com>, benbjiang@tencent.com,
        Alexandre Chartre <alexandre.chartre@oracle.com>,
        James.Bottomley@hansenpartnership.com, OWeisse@umich.edu,
        Dhaval Giani <dhaval.giani@oracle.com>,
        Junaid Shahid <junaids@google.com>,
        Jesse Barnes <jsbarnes@google.com>,
        Ben Segall <bsegall@google.com>, Hao Luo <haoluo@google.com>,
        Tom Lendacky <thomas.lendacky@amd.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Feb 23, 2021 at 11:26 AM Chris Hyser <chris.hyser@oracle.com> wrote:
>
> On 2/23/21 4:05 AM, Peter Zijlstra wrote:
> > On Mon, Feb 22, 2021 at 11:00:37PM -0500, Chris Hyser wrote:
> >> On 1/22/21 8:17 PM, Joel Fernandes (Google) wrote:
> >> While trying to test the new prctl() code I'm working on, I ran into a bug I
> >> chased back into this v10 code. Under a fair amount of stress, when the
> >> function __sched_core_update_cookie() is ultimately called from
> >> sched_core_fork(), the system deadlocks or otherwise non-visibly crashes.
> >> I've not had much success figuring out why/what. I'm running with LOCKDEP on
> >> and seeing no complaints. Duplicating it only requires setting a cookie on a
> >> task and forking a bunch of threads ... all of which then want to update
> >> their cookie.
> >
> > Can you share the code and reproducer?
>
> Attached is a tarball with c code (source) and scripts. Just run ./setup_bug which will compile the source and start a
> bash with a cs cookie. Then run ./show_bug which dumps the cookie and then fires off some processes and threads. Note
> the cs_clone command is not doing any core sched prctls for this test (not needed and currently coded for a diff prctl
> interface). It just creates processes and threads. I see this hang almost instantly.
>
> Josh, I did verify that this occurs on Joel's coresched tree both with and w/o the kprot patch and that should exactly
> correspond to these patches.
>
> -chrish
>

I think I've gotten to the root of this. In the fork code, our cases
for inheriting task_cookie are inverted for CLONE_THREAD vs
!CLONE_THREAD. As a result, we are creating a new cookie per-thread,
rather than inheriting from the parent. Now this is actually ok; I'm
not observing a scalability problem with creating this many cookies.
However, it means that overall throughput of your binary is cut in
~half, since none of the threads can share a core. Note that I never
saw an indefinite deadlock, just ~2x runtime for your binary vs the
control. I've verified that both a) manually hardcoding all threads to
be able to share regardless of cookie, and b) using a machine with 6
cores instead of 2, both allow your binary to complete in the same
amount of time as without the new API.