linux-kernel - Re: [PATCH] driver/perf: Add PMU driver for the ARM DMC-620 memory controller.

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>] [day] [month] [year] [list]

Date:   Fri, 20 Mar 2020 07:59:04 +0000
From:   Will Deacon <will@...nel.org>
To:     Tuan Phan <tuanphan@...eremail.onmicrosoft.com>
Cc:     Mark Rutland <mark.rutland@....com>,
        Tuan Phan <tuanphan@...amperecomputing.com>,
        linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH] driver/perf: Add PMU driver for the ARM DMC-620 memory
 controller.

On Thu, Mar 19, 2020 at 12:03:43PM -0700, Tuan Phan wrote:
>    Please find my comments below.
> 
>      On Mar 19, 2020, at 8:16 AM, Mark Rutland <[1]mark.rutland@....com>
>      wrote:
>      Hi Tuan,
> 
>      On Tue, Mar 17, 2020 at 05:29:38PM -0700, Tuan Phan wrote:
> 
>        DMC-620 PMU supports total 10 counters which each is
>        independently programmable to different events and can
>        be started and stopped individually.
> 
>      Looking at the TRM for DMC-620, the PMU registers are not in a separate
>      frame from the other DMC control registers, and start at offset 0xA00
>      (AKA 2560). I would generally expect that access to the DMC control
>      registers was restricted to the secure world; is that not the case on
>      your platform?
> 
>      I ask because if those are not restricted, the Normal world could
>      potentially undermine the Secure world through this (e.g. playing with
>      training settings, messing with the physical memory map, injecting RAS
>      errors). Have you considered this?
> 
>    => Only PMU registers can be accessed within normal world. I only pass
>    PMU registers (offset 0xA00) to kernel so shouldn’t be problem.

Just a stylistic thing since there's been a bit of back-and-forth on this
patch, but please can you fix your email client configuration so that the
replies are threaded properly? '=>' is non-standard and it's pretty
hard to spot in a mass of C code. You can look at
Documentation/process/email-clients.rst for some information about
configuring common clients.

Apologies if this comes across as pedantic, but it makes a surprising
difference in how easy it is to keep up with the thread (at least to me).

Will