linux-kernel - Re: [RFC] Make need_resched() return true when rcu_urgent

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] [day] [month] [year] [list]

Date:   Mon, 23 Jul 2018 05:22:31 -0700
From:   "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
To:     David Woodhouse <dwmw2@...radead.org>
Cc:     Christian Borntraeger <borntraeger@...ibm.com>,
        Peter Zijlstra <peterz@...radead.org>, mhillenb@...zon.de,
        linux-kernel <linux-kernel@...r.kernel.org>,
        kvm <kvm@...r.kernel.org>
Subject: Re: [RFC] Make need_resched() return true when rcu_urgent_qs
 requested

On Mon, Jul 23, 2018 at 10:08:59AM +0200, David Woodhouse wrote:
> On Thu, 2018-07-19 at 10:09 -0700, Paul E. McKenney wrote:
> > 
> > Of course, the real reason for the lack of fault on your part will not
> > because I believe I found the bug elsewhere, but instead because I will
> > be dropping your patch (and mine as well) on Frederic's advice.  ;-)
> 
> You're keeping the need_resched() one though?

Yes.  This the current commit in -rcu (which will change when I rebase
onto v4.19-rc1, if not earlier):

fcf0407e6e63 ("rcu: Make need_resched() respond to urgent RCU-QS needs")

> And we are still left with the fact that CONTEXT_TRACKING_FORCE is
> making the existing code in guest_enter_irqoff() do the wrong thing for
> !NO_HZ_FULL. But in fact the rcu_virt_note_context_switch() there is
> completely redundant now we fixed need_resched(), so can be dropped,
> leaving only the rcu_user_enter/exit calls for the NO_HZ_FULL case?

I am not yet convinced that we know exactly the right thing to be
doing for guest OSes for either value of NO_HZ_FULL, much less that
we are actually doing it.  ;-)

But what does your testing say?

							Thanx, Paul