linux-kernel - Re: [GIT PULL] scheduler fixes

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: <20191117102952.GA56088@gmail.com>
Date:   Sun, 17 Nov 2019 11:29:52 +0100
From:   Ingo Molnar <mingo@...nel.org>
To:     Valentin Schneider <valentin.schneider@....com>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        linux-kernel@...r.kernel.org,
        Peter Zijlstra <a.p.zijlstra@...llo.nl>,
        Thomas Gleixner <tglx@...utronix.de>,
        Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [GIT PULL] scheduler fixes


* Valentin Schneider <valentin.schneider@....com> wrote:

> On 17/11/2019 09:45, Ingo Molnar wrote:
> > I've picked v2 up instead. I suspect it's not really consequential as 
> > enums don't really get truncated by compilers, right? Is there any other 
> > negative runtime side effect possible from the imprecise enum/uint 
> > typing?
> > 
> 
> AFAIUI the requirement for the enum type is that it has to be an int type that
> covers all its values, so I could see some funky optimization (e.g. check the
> returned value is < 512 but it's assumed the type for the enum is 8 bits so
> this becomes always true). Then again we don't have any explicit check on
> those returned values, plus they fit in 11 bits, so as you say it's
> mostly likely inconsequential (and I didn't see any compile diff).

Yeah, so unless there's evidence of there being a nonzero chance of this 
being misbuilt I'd gravitate towards doing this via via sched/core, 
especially so late in the cycle.

> My "worry" wasn't really about this patch, it was more about the 
> following one - it didn't like the idea of merging an unneeded patch 
> (with a Fixes: tag on top of it).

Yeah, agreed - should be fixed now.

> >>>       sched/topology, cpuset: Account for housekeeping CPUs to avoid empty cpumasks
> >>
> >> And this one is no longer needed, as Michal & I understood (IOW the fix in
> >> rc6 is sufficient), see:
> >>
> >>   c425c5cb-ba8a-e5f6-d91c-5479779cfb7a@....com
> > 
> > Ok.
> > 
> > I'm inclined to just reduce sched/urgent back to these three fixes:
> > 
> >   6e1ff0773f49: sched/uclamp: Fix incorrect condition
> >   b90f7c9d2198: sched/pelt: Fix update of blocked PELT ordering
> >   ff51ff84d82a: sched/core: Avoid spurious lock dependencies
> > 
> > and apply v2 of the uclamp_id type fix to sched/core. This would reduce 
> > the risks of a Sunday pull request ...
> > 
> 
> This sounds good to me. Sorry for the hassle.

No hassle at all - thanks for catching these!

Thanks,

	Ingo