User: Password:
|
|
Subscribe / Log in / New account

[bug] /etc/profile: line 30: /dev/null: Permission denied (Was: Re: [PATCH] Remove broken by design and by implementation devtmpfs maintenance disaster)

From:  Ingo Molnar <mingo-AT-elte.hu>
To:  "Eric W. Biederman" <ebiederm-AT-xmission.com>
Subject:  [bug] /etc/profile: line 30: /dev/null: Permission denied (Was: Re: [PATCH] Remove broken by design and by implementation devtmpfs maintenance disaster)
Date:  Thu, 17 Sep 2009 20:53:06 +0200
Cc:  Greg KH <greg-AT-kroah.com>, Linus Torvalds <torvalds-AT-linux-foundation.org>, linux-kernel-AT-vger.kernel.org, Kay Sievers <kay.sievers-AT-vrfy.org>
Archive-link:  Article, Thread


* Eric W. Biederman <ebiederm@xmission.com> wrote:

> Greg this code does not live up to the standards you have repeatedly 
> asserted are required for accepting core kernel code.  Neither you nor 
> Kay show any interest in fixing even the most trivial of bugs. Must 
> less discuss alternate solutions to the problem.

on a tangential note - i have no opinion yet in this discussion, but i 
found no pull request email on lkml and wanted to report a devtmpfs bug:

I've reproduced a bug with the following .config options:

  CONFIG_DEVTMPFS=y
  CONFIG_DEVTMPFS_MOUNT=y

/dev/null and /dev/zero are not read/writable to ordinary users, 
breaking normal bootup and login:

  earth4:~/tip> m
  PTY allocation request failed on channel 0
  /etc/profile: line 30: /dev/null: Permission denied
                                                   
  /etc/profile.d/colorls.sh: line 3: /dev/null: Permission denied
                        /etc/profile.d/colorls.sh: line 4: /dev/null: 
  Permission denied
                                                                                     
  /etc/profile.d/vim.sh: line 5: /dev/null: Permission denied
                                                     Connection to m 
  closed by remote host.
  Connection to m closed.

Regarding the wish to have in-kernel code: another solution which i have 
not seen mentioned would be to move udev into tools/udev/ instead. That 
would give a single, coherent user-space kernel-space combo.

The concept certainly works great for tools/perf/.

The system boots up fine with devtmpfs disabled:

  earth4:~/tip> m
  Last login: Fri Sep 18 02:48:19 2009 from sirius
  mercury:~> 

	Ingo


(Log in to post comments)


Copyright © 2009, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds