linux-kernel - Re: [PATCH 2/6] perf: Support branch events logging

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: <d53d4df7-d0b8-2fbc-4912-f89686b5d931@linux.intel.com>
Date:   Fri, 14 Apr 2023 15:47:29 -0700
From:   Andi Kleen <ak@...ux.intel.com>
To:     Peter Zijlstra <peterz@...radead.org>,
        "Liang, Kan" <kan.liang@...ux.intel.com>
Cc:     mingo@...hat.com, acme@...nel.org, linux-kernel@...r.kernel.org,
        mark.rutland@....com, alexander.shishkin@...ux.intel.com,
        jolsa@...nel.org, namhyung@...nel.org, irogers@...gle.com,
        adrian.hunter@...el.com, eranian@...gle.com
Subject: Re: [PATCH 2/6] perf: Support branch events logging


> Yeah, don't do this. There is no guarantee what so ever you'll get any
> of those events in the 0-3 range.


The kernel can simply force to 0-3 if LBR is enabled and the feature 
too. It's in Kan's patch

and it isn't particularly complicated.

>
> You really *must* make then a group such that perf knows what events to
> associated with the LBR event and constain them to the 0-3 range of
> PMCs.
>
> If you want multiplexing, simply create multiple groups with an LBR
> event in them.


Well if you force groups then you require user space or a user which 
understands all the constraints

to create groups. I thought one of the basic ideas of perf was to be 
able to abstract those things.

There are tools today (e.g. the perf builtin metrics[1] and I believe 
others) that don't have enough

knowledge to set up real groups and rely on the kernel.  While they 
could be fixed it's a lot of work

(I do it in pmu-tools, and it's quite hairy code)


-Andi


[1] Given they are are not supported by perf record yet, but since perf 
script already supports

evaluating them they could be at some point, and then it would make 
sense to use them

with this feature too.