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-next>] [day] [month] [year] [list]
Message-Id: <20231102104429.025835330@infradead.org>
Date:   Thu, 02 Nov 2023 11:44:29 +0100
From:   Peter Zijlstra <peterz@...radead.org>
To:     Linus Torvalds <torvalds@...uxfoundation.org>
Cc:     x86@...nel.org, linux-kernel@...r.kernel.org, peterz@...radead.org,
        Jonathan Cameron <jic23@...nel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Oleg Nesterov <oleg@...hat.com>,
        Eric Biederman <ebiederm@...ssion.com>
Subject: [PATCH 0/2] cleanup: Conditional locking support

Hi,

Two patches, the first extending the cleanup/guard stuff to better handle
conditional locks, mutex_trylock(), mutex_lock_interruptible() etc.. And the
second a small ptrace conversion that I've been using as a test-case.

The normal scoped_guard() is changed to simply skip the body when a conditional
lock fails to acquire. A new scoped_cond_guard() is added that takes an extra
statement argument to provide an explicit 'fail' action.

The ptrace patch has:

  scoped_cond_guard (mutex_intr, return -ERESTARTNOINTR,
		     &task->signal->cred_guard_mutex) {
    ...
  }
  ...
  return 0;

Where if the lock acquire fails, it does 'return -ERESTARTNOINTR'.

The crazy perf thing then becomes:

  scoped_cond_guard (rwsem_read_intr, goto no_lock,
                     task ? &task->signal->exec_update_lock : NULL) {
    if (0) {
no_lock:
      if (task)
        return -EINTR;
    }
    ... body with or without lock ...
  }


Specifically, that thing needs the lock when there is a task, but otherwise
needs to still do the body without the lock.

IIO also wanted something along these lines, although they have a custom
'trylock' thing.

Barring objections, I'm planning to merge this into tip/locking/cleanup which
I'll merge into tip/locking/core (and tip/perf/core when times comes).

My plan is to post the perf patches in 3 batches of roughly 10 patches each,
the simpler first and the more crazy ones (including the above) last.

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ