linux-kernel - [PATCH 0/7] module: avoid userspace pressure on unwanted allocations

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: <20230329052954.3974542-1-mcgrof@kernel.org>
Date:   Tue, 28 Mar 2023 22:29:54 -0700
From:   Luis Chamberlain <mcgrof@...nel.org>
To:     david@...hat.com, patches@...ts.linux.dev,
        linux-modules@...r.kernel.org, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org, pmladek@...e.com,
        petr.pavlu@...e.com, prarit@...hat.com,
        torvalds@...ux-foundation.org, gregkh@...uxfoundation.org,
        rafael@...nel.org
Cc:     christophe.leroy@...roup.eu, tglx@...utronix.de,
        peterz@...radead.org, song@...nel.org, rppt@...nel.org,
        willy@...radead.org, vbabka@...e.cz, mhocko@...e.com,
        dave.hansen@...ux.intel.com, mcgrof@...nel.org
Subject: [PATCH 0/7] module: avoid userspace pressure on unwanted allocations

This patch set addresses a fix to the vmap allocation presure issues which
David Hildenbrand had reported last year in October. While at it,
I've simplified the kmod concurrency delimiter using Linus' suggestion,
and added debugfs stats to help us keep sane in doing analysis for memory
pressure issues on the finit_module() side of things. That should *also*
help do an empirical evaluation of module .text sizes *actually* present
on systems, given userspace makes it a bit tricky to get that right.

All this would not have been possible without stress-ng and Colin Ian King's
help to getting a modules ops in shape so to reproduce a situation only
reported so far on a system with over 400 CPUs.

I *think* the degugfs stats *should* probably be used to help identify
areas where we perhaps need *more work* to try to mitigate vmalloc()
space, as in the worst case we can end up using vmap space 3 times for
a single module, two just as big as the module, and if you are enabling
compression one with the compressed module size. That's significant memory
pressure on vmalloc() / vmap() space.

If you'd like to give this a spin this is available on my branch based on
modules-next 20230328-module-alloc-opts [2].

[0] https://lkml.kernel.org/r/20221013180518.217405-1-david@redhat.com
[1] https://github.com/ColinIanKing/stress-ng
[2] https://git.kernel.org/pub/scm/linux/kernel/git/mcgrof/linux.git/log/?h=20230328-module-alloc-opts

Luis Chamberlain (7):
  module: move finished_loading()
  module: extract patient module check into helper
  module: avoid allocation if module is already present and ready
  sempahore: add a helper for a concurrency limiter
  modules/kmod: replace implementation with a sempahore
  debugfs: add debugfs_create_atomic64_t for atomic64_t
  module: add debug stats to help identify memory pressure

 fs/debugfs/file.c         |  36 +++++++
 include/linux/debugfs.h   |   2 +
 include/linux/semaphore.h |   3 +
 kernel/module/Kconfig     |  32 ++++++
 kernel/module/Makefile    |   4 +
 kernel/module/debug.c     |  16 +++
 kernel/module/internal.h  |  35 +++++++
 kernel/module/kmod.c      |  26 ++---
 kernel/module/main.c      | 164 ++++++++++++++++++++-----------
 kernel/module/stats.c     | 200 ++++++++++++++++++++++++++++++++++++++
 kernel/module/tracking.c  |   7 +-
 11 files changed, 445 insertions(+), 80 deletions(-)
 create mode 100644 kernel/module/debug.c
 create mode 100644 kernel/module/stats.c

-- 
2.39.2