linux-kernel - Re: [PATCH v2 4/4] perf python: Add counting.py as example for counting perf events

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: <CAP-5=fX8xkO8jXqvgksTF_5m+gknhZ2p1NuPZ_Lb0fzbDMDJ=A@mail.gmail.com>
Date: Tue, 13 May 2025 14:47:52 -0700
From: Ian Rogers <irogers@...gle.com>
To: Arnaldo Carvalho de Melo <acme@...nel.org>
Cc: Gautam Menghani <gautam@...ux.ibm.com>, namhyung@...nel.org, peterz@...radead.org, 
	mingo@...hat.com, mark.rutland@....com, alexander.shishkin@...ux.intel.com, 
	jolsa@...nel.org, adrian.hunter@...el.com, kan.liang@...ux.intel.com, 
	linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org, 
	maddy@...ux.ibm.com
Subject: Re: [PATCH v2 4/4] perf python: Add counting.py as example for
 counting perf events

On Tue, May 13, 2025 at 2:16 PM Arnaldo Carvalho de Melo
<acme@...nel.org> wrote:
>
> On Tue, May 13, 2025 at 01:59:28PM -0700, Ian Rogers wrote:
> > On Tue, May 13, 2025 at 1:50 PM Arnaldo Carvalho de Melo
> > <acme@...nel.org> wrote:
> > [snip]
> > > Right, I like the effort he is making into having perf more usable in
> > > python, and I encourage him to think about the issues you raised so that
> > > we can come to some good abstractions.
> >
> > Thanks Arnaldo, can we be tolerant to API changes in the python from a
> > "regression" point-of-view? Like avoiding the notion of indices?
>
> But correct me if I am missing something, aren't indices only introduced
> with this new patchset?

Checking the code I think you're right. Unless you do something like
the range loop:
```
 for thread in range(len(thread_map)):
```
so I think we'd all prefer indices not to be a part of the python API.
On the C side we can get indices via helpers like perf_cpu_map__idx,
which will introduce O(log N) overhead - thread map is missing this
currently I believe. For compatibility a CPU and thread should remain
an int.

Thanks,
Ian

> - Arnaldo
>
> > Presumably such a fix would also need fixing in all the perf python
> > scripts, but the external users I worry about. My sense is the number
> > of external users is minimal, for example, toplev I don't believe is a
> > user [1].
> >
> > Ian
> >
> > [1] https://github.com/andikleen/pmu-tools