linux-kernel - Re: [PATCH] lib: add module unload support to sort tests

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]

Date:   Tue, 19 Dec 2017 14:51:36 -0800
From:   Andrew Morton <akpm@...ux-foundation.org>
To:     Pravin Shedge <pravin.shedge4linux@...il.com>
Cc:     fkostenzer@...e.at, andriy.shevchenko@...ux.intel.com,
        geert@...ux-m68k.org, paul.gortmaker@...driver.com,
        linux-kernel@...r.kernel.org, Jessica Yu <jeyu@...nel.org>,
        Rusty Russell <rusty@...tcorp.com.au>
Subject: Re: [PATCH] lib: add module unload support to sort tests

On Tue, 19 Dec 2017 23:10:00 +0530 Pravin Shedge <pravin.shedge4linux@...il.com> wrote:

> >
> > If so, why do you think we shiould alter lib/test_sort.c to behave in
> > this atypical fashion?
> 
> If test case is going affects only at boot time or at module load
> time, it's smart decision to unload module
> automatically on successful completion.

OK.

I think it does make sense for a lib/text_*.ko type module to unload
itself after successful completion of the test.  However:

- returning a fake error code from the module's module_init() is a
  daft way of doing that.  We should find a way to let the
  module_init() handler tell do_init_module() "I succeeded, but please
  unload me anyway".  So the initial sys_init_module() call doesn't say
  "it failed".  Could create a new, kernel-internal errno for this and
  have do_init_module() rewrite that to 0.

  Maybe.  A sys_init_module() caller's expectation is that if the
  syscall succeeded then the module is now loaded.

  Well.  Except for the test_*.ko modules, which are special.

- Changing any test module so that it now auto-unloads on success is
  a non-back-compat change.  The practical effect of which will be very
  minor: a subsequent rmmod finds that the module isn't there.

I'm not sure what to do, really.  Does any of this matter much at all?