linux-kernel - Re: [PATCH 1/3] clockevents: Retry programming min delta up to 10 times

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]

Date:   Thu, 19 Oct 2017 17:11:28 +0200 (CEST)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Martin Schwidefsky <schwidefsky@...ibm.com>
cc:     Matt Redfearn <matt.redfearn@...s.com>,
        Daniel Lezcano <daniel.lezcano@...aro.org>,
        James Hogan <james.hogan@...s.com>, linux-mips@...ux-mips.org,
        James Hogan <jhogan@...nel.org>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/3] clockevents: Retry programming min delta up to 10
 times

On Thu, 19 Oct 2017, Martin Schwidefsky wrote:

> On Thu, 19 Oct 2017 15:29:28 +0200 (CEST)
> Thomas Gleixner <tglx@...utronix.de> wrote:
> 
> > On Thu, 19 Oct 2017, Matt Redfearn wrote:
> > > On 19/10/17 13:43, Thomas Gleixner wrote:  
> > > > 	delta = 0;
> > > > 	for (i = 0; i < 10; i++) {
> > > > 		delta += dev->min_delta_ns;
> > > > 		dev->next_event = ktime_add_ns(ktime_get(), delta);
> > > > 		clc = .....
> > > > 	   	.....
> > > > 
> > > > That makes it more likely to succeed fast. Hmm?  
> > > 
> > > That will set the target time to increasing multiples of min_delta_ns in the
> > > future, right?  
> > 
> > Yes, but without fiddling with min_delta_ns itself.
> 
> Grumpf, more extra code for yet another piece of broken hardware
> I guess.

and virtualization. Oh wait.. the virt is the ultimate reference for broken
hardware...

> > > Sure, it should make it succeed faster - I'll make it like
> > > that. Are you OK with the arbitrarily chosen 10 retries?  
> > 
> > I lost my crystalball so I have to trust yours :)
> 
> The alternative implementation would be to do the retries in
> the clockevent driver itself. Then that particular driver can
> choose the correct number of retries, no?

There is no correct number ever. All you can do is set an upper limit.

Thanks,

	tglx