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]
Message-ID: <20240903085454.GR4723@noisy.programming.kicks-ass.net>
Date: Tue, 3 Sep 2024 10:54:54 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Namhyung Kim <namhyung@...nel.org>
Cc: Ingo Molnar <mingo@...nel.org>, Kan Liang <kan.liang@...ux.intel.com>,
	Mark Rutland <mark.rutland@....com>,
	Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
	Arnaldo Carvalho de Melo <acme@...nel.org>,
	LKML <linux-kernel@...r.kernel.org>,
	Stephane Eranian <eranian@...gle.com>,
	Ravi Bangoria <ravi.bangoria@....com>
Subject: Re: [RFC/PATCH 4/4] perf/x86: Relax privilege filter restriction on
 AMD IBS

On Mon, Sep 02, 2024 at 10:30:19AM -0700, Namhyung Kim wrote:
> On Mon, Sep 2, 2024 at 2:12 AM Peter Zijlstra <peterz@...radead.org> wrote:
> >
> > On Fri, Aug 30, 2024 at 04:29:10PM -0700, Namhyung Kim wrote:
> > > While IBS is available for per-thread profiling, still regular users
> > > cannot open an event due to the default paranoid setting (2) which
> > > doesn't allow unprivileged users to get kernel samples.  That means
> > > it needs to set exclude_kernel bit in the attribute but IBS driver
> > > would reject it since it has PERF_PMU_CAP_NO_EXCLUDE.  This is not what
> > > we want and I've been getting requests to fix this issue.
> > >
> > > This should be done in the hardware, but until we get the HW fix we may
> > > allow exclude_{kernel,user} in the attribute and silently drop the
> > > samples in the PMU IRQ handler.  It won't guarantee the sampling
> > > frequency or even it'd miss some with fixed period too.  Not ideal,
> > > but that'd still be helpful to regular users.
> >
> > Urgh.... this is really rather bad. And I'm sure a bunch of people are
> > going to be spending a lot of time trying to figure out why their
> > results don't make sense.
> 
> I agree it can be confusing but there are use cases where regular users
> want IBS information like memory data source, data address and so on.

Sure, but I'm a bit worried about users not being aware of this
trickery. This makes IBS events that have exclude_kernel=1 behave
significantly different from those that don't have it.

At the very least you should kill the IBS forward in amd_pmu_hw_config()
when this is active. But perhaps we should also emit a one time
KERN_INFO message when such an event gets created?

> Also I realized that software events like cpu-clock use the same logic to
> discard samples by privilege mode already.

Right, but everybody expects the software things to suck :-) And they
always suck, unconditionally.

While the IBS thing only sucks when you use exclude_[user,kernel]
things. Stealth suckage is bad and enrages people.

> > I realize that having entry hooks to disable/enable the counters is also
> > not going to happen, this has a ton of problems too.
> 
> Do you mean kernel/user mode change hook?  I guess it'd be too costly.

Yep, insanely expensive :-)

> > Also, that PMU passthrough patch set has guest hooks, so you can
> > actually do the exclude_host/guest nonsense with those, right?
> 
> Oh.. this patch is about exclude_user/kernel not host/guest.  Anyway
> it'd be great if IBS could support the guest hooks and allow the exclude
> bits.

Right, but since your other patchset about disabling exclude_guest
because IBS don't support it I figured I'd mention that it would be
fairly simple to fix.

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ