linux-kernel - Re: [patch V9 10/39] x86/entry: Provide helpers for execute on irqstack

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200609023259.GA3579@lca.pw>
Date:   Mon, 8 Jun 2020 22:32:59 -0400
From:   Qian Cai <cai@....pw>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Andy Lutomirski <luto@...nel.org>,
        Andrew Cooper <andrew.cooper3@...rix.com>,
        X86 ML <x86@...nel.org>,
        "Paul E. McKenney" <paulmck@...nel.org>,
        Alexandre Chartre <alexandre.chartre@...cle.com>,
        Frederic Weisbecker <frederic@...nel.org>,
        Paolo Bonzini <pbonzini@...hat.com>,
        Sean Christopherson <sean.j.christopherson@...el.com>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Petr Mladek <pmladek@...e.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Joel Fernandes <joel@...lfernandes.org>,
        Boris Ostrovsky <boris.ostrovsky@...cle.com>,
        Juergen Gross <jgross@...e.com>,
        Brian Gerst <brgerst@...il.com>,
        Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        Will Deacon <will@...nel.org>,
        Tom Lendacky <thomas.lendacky@....com>,
        Wei Liu <wei.liu@...nel.org>,
        Michael Kelley <mikelley@...rosoft.com>,
        Jason Chen CJ <jason.cj.chen@...el.com>,
        Zhao Yakui <yakui.zhao@...el.com>,
        Alexander Potapenko <glider@...gle.com>
Subject: Re: [patch V9 10/39] x86/entry: Provide helpers for execute on
 irqstack

On Tue, Jun 09, 2020 at 12:20:06AM +0200, Thomas Gleixner wrote:
> Qian,
> 
> can you please ensure that people who got cc'ed because the problem
> affects their subsystem are included on your replies even if you are
> replying to a different subthread?
> 
> I explicitely did:
> 
>      Cc:+ Alexander
> 
> at the very beginning of my reply:
> 
>    https://lore.kernel.org/r/87v9k3jdc6.fsf@nanos.tec.linutronix.de
> 
> to make you aware of that.
> 
> Yes, email sucks, but it sucks even more when people are careless.

Sorry, I will remeber that next time.

[]
> To get facts instead of useless loop theories, can you please apply the
> patch below, enable DEBUGFS and provide the output of
> 
>        /sys/kernel/debug/stackdepot/info
> 
> for a kernel before that change and after? Please read out that file at
> periodically roughly the same amounts of time after starting your test
> scenario.
> 
> Note, that I doubled the size of the stack depot so that we get real
> numbers and not the cutoff by the size limit. IOW, the warning should
> not trigger anymore. If it triggers nevertheless then the numbers will
> still tell us an interesting story.

Instead of running the whole testsuite, I just picked this single LTP
oom02 test which seems usually trigger it within the testsuite. Let me
know if this is insufficient (which indeed tell the big difference in
"Unique stacks"), and I am happy to run the whole things.

BAD: next-20200608
GOOD: next-20200528 (which does not include this series)

BAD (after boot)
# cat /sys/kernel/debug/stackdepot/info
Unique stacks: 33547
Depot index:   359
Depot offset:  6752

BAD (after oom02)
# cat /sys/kernel/debug/stackdepot/info
Unique stacks: 140476
Depot index:   2555
Depot offset:  9168

GOOD (after boot)
# cat /sys/kernel/debug/stackdepot/info
Unique stacks: 31112
Depot index:   317
Depot offset:  14384

GOOD (after oom02)
# cat /sys/kernel/debug/stackdepot/info
Unique stacks: 34176
Depot index:   354
Depot offset:  4032

BTW, I am happy to run another one using next-20200608 with just this
series reverted if you suspect there is something else going on between
those two trees.