linux-kernel - Re: [PATCH v2 03/13] ANDROID: binder: add support for RT prio inheritance.

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: <CAB0TPYFuqs36reMTS7FsJg+62Dk7YYuK2b2UzjzpoHx8OfbfqA@mail.gmail.com>
Date:   Thu, 31 Aug 2017 14:00:54 +0200
From:   Martijn Coenen <maco@...roid.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Greg KH <gregkh@...uxfoundation.org>,
        John Stultz <john.stultz@...aro.org>,
        Todd Kjos <tkjos@...gle.com>,
        Arve Hjønnevåg <arve@...roid.com>,
        Amit Pundir <amit.pundir@...aro.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Christoph Hellwig <hch@....de>,
        LKML <linux-kernel@...r.kernel.org>, devel@...verdev.osuosl.org,
        Martijn Coenen <maco@...gle.com>,
        Iliyan Malchev <malchev@...gle.com>,
        Colin Cross <ccross@...roid.com>
Subject: Re: [PATCH v2 03/13] ANDROID: binder: add support for RT prio inheritance.

On Thu, Aug 31, 2017 at 1:32 PM, Peter Zijlstra <peterz@...radead.org> wrote:
> AFAIK people are actively working on fixing that.

SCHED_DEADLINE was definitely looked at in the past. We certainly
don't use it on our own devices in Android Oreo, and I am not aware of
any current plans to use it. But the Android org is big, so I may
simply not be aware. In either case I'd be happy to support it if
there is a need for it.

>
> I still have to look at the actual patches, but it all sounds very
> dodgy. Probably won't have time until after LPC.

Appreciate if you could take a look! In particular I'm curious why
this approach is considered "dodgy", or "engineered sideways" as
Thomas pointed out earlier. From what I can tell the reason why we
want to change task priorities (potentially from a different task) is
understood. So is the main concern using sched_setscheduler()
directly, instead of having a prio "modifier" much like rt_mutex has?
We (and our partners) are using these patches without issues, though
it does come with the caveat that threads in the binder threadpool
should not be changing their own priority from userspace for this to
work correctly (though they shouldn't have been doing that before
these patches either).

Either way, it sounds like this discussion will take a bit longer, so
I will pull out the few unrelated patches from this series and post
them separately.

Thanks,
Martijn