linux-kernel - Re: [PATCH 1/4] coresight/etm4x: disallow altering config via sysfs while enabled

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: <Z4AJ0kc0DxVwhLlu@e129823.arm.com>
Date: Thu, 9 Jan 2025 17:39:30 +0000
From: Yeoreum Yun <yeoreum.yun@....com>
To: Suzuki K Poulose <suzuki.poulose@....com>
Cc: mike.leach@...aro.org, james.clark@...aro.org,
	alexander.shishkin@...ux.intel.com, coresight@...ts.linaro.org,
	linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/4] coresight/etm4x: disallow altering config via sysfs
 while enabled

Hi Suzuki,

> On 09/01/2025 12:01, Yeoreum Yun wrote:
> > Hi Suzuki,
> >
> > > Is it not better to have separate "configs" for perf and sysfs ?
> > > And etmX driver can populate the "running" config, based on the
> > > mode specific config. That way, the configs can be updated
> > > independently without affecting the running config or the perf one.
> > >
> >
> > That was i've tried but I've accepted Mike's opinion that
> > it's enough to check whether CS_MODE_DISABLED via coresight_get_mode()
> > in *_store().
> >
> > "the .._store functions in sysfs should use coresight_get_mode() to ensure
> > this is set to CS_MODE_DISABLED before altering the config,
> > which ensures that the trace system is inactive.
> > We don't' really care about reading the config if trace is running."
>
> There are two issues with that :
>
> 1. Sprinkling the get_mode call in each sysfs stor function doesn't look
> good to me.
>
> 2. Someone preparing for a sysfs session must not be prevented from doing so
> just because there is a perf session running.
>
>
> Suzuki
>

But, when separate the config, it doesn't show anymore the current
configuration set by perf.
I'm not sure this is okay.
IMHO, If perf is enabled, since the configuration show the "perf",
I think prohabit to modify config via sysfs while PERF_ENABLE seems valid.

and about lossing session, I think this is up to user.
That means to use sysfs, user shouldn't use perf to prevent loss
its configuration.

Am I missing something?

Thanks