linux-kernel - Perf MMAP records differ from /proc/[X]/maps when chroot is used

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]

Message-ID: <20220601180404.GA2029@kbox>
Date:   Wed, 1 Jun 2022 11:04:04 -0700
From:   Beau Belgrave <beaub@...ux.microsoft.com>
To:     namhyung@...nel.org, jolsa@...nel.org, acme@...nel.org,
        peterz@...radead.org, mingo@...nel.org
Cc:     linux-trace-devel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Perf MMAP records differ from /proc/[X]/maps when chroot is used

I noticed a related conversation that occurred in the past here:
https://lore.kernel.org/lkml/20220202070828.143303-1-namhyung@kernel.org/

I hit this same issue, however, during debugging I noticed that when
chroot is used the full path including the chroot are in /proc/[X]/maps.
The perf MMAP records only give the relative path within the chroot.

This causes different views of a process when combining mmaps that existed
before perf_event_open() via /proc/[X]/maps and observed mmaps within the
perf_event_open() ring buffer.

I would like to know if this is how mmap records are wanted to be represented
long term via perf? It would be great if these mmap records aligned with how
/proc/[X]/maps are represented so chroot usage becomes (mostly) seamless.

Thanks,
-Beau