From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756905Ab2LOALt (ORCPT ); Fri, 14 Dec 2012 19:11:49 -0500 Received: from out01.mta.xmission.com ([166.70.13.231]:55522 "EHLO out01.mta.xmission.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756713Ab2LOALq (ORCPT ); Fri, 14 Dec 2012 19:11:46 -0500 From: ebiederm@xmission.com (Eric W. Biederman) To: "Serge E. Hallyn" Cc: Linux Containers , linux-security-module@vger.kernel.org, linux-kernel@vger.kernel.org, Andy Lutomirski , David Howells References: <87txroxpgq.fsf@xmission.com> <87bodwxpcg.fsf@xmission.com> <20121215000338.GC13659@mail.hallyn.com> Date: Fri, 14 Dec 2012 16:11:38 -0800 In-Reply-To: <20121215000338.GC13659@mail.hallyn.com> (Serge E. Hallyn's message of "Sat, 15 Dec 2012 00:03:38 +0000") Message-ID: <87r4msrx6t.fsf@xmission.com> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.1 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain X-XM-AID: U2FsdGVkX1/j1J0aN+YOu787ooFSfCYtTwRtf1a1YJo= X-SA-Exim-Connect-IP: 98.207.153.68 X-SA-Exim-Mail-From: ebiederm@xmission.com X-Spam-Report: * -1.0 ALL_TRUSTED Passed through trusted hosts only via SMTP * 1.5 XMNoVowels Alpha-numberic number with no vowels * 0.1 XMSubLong Long Subject * 0.0 T_TM2_M_HEADER_IN_MSG BODY: T_TM2_M_HEADER_IN_MSG * -3.0 BAYES_00 BODY: Bayes spam probability is 0 to 1% * [score: 0.0059] * -0.0 DCC_CHECK_NEGATIVE Not listed in DCC * [sa01 1397; Body=1 Fuz1=1 Fuz2=1] * 0.0 T_TooManySym_01 4+ unique symbols in subject * 0.0 T_XMDrugObfuBody_08 obfuscated drug references X-Spam-DCC: XMission; sa01 1397; Body=1 Fuz1=1 Fuz2=1 X-Spam-Combo: ;"Serge E. Hallyn" X-Spam-Relay-Country: Subject: Re: [PATCH 3/4] userns: Add a more complete capability subset test to commit_creds X-SA-Exim-Version: 4.2.1 (built Sun, 08 Jan 2012 03:05:19 +0000) X-SA-Exim-Scanned: Yes (on in02.mta.xmission.com) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org "Serge E. Hallyn" writes: > Quoting Eric W. Biederman (ebiederm@xmission.com): >> >> When unsharing a user namespace we reduce our credentials to just what >> can be done in that user namespace. This is a subset of the credentials >> we previously had. Teach commit_creds to recognize this is a subset >> of the credentials we have had before and don't clear the dumpability flag. >> >> This allows an unprivileged program to do: >> unshare(CLONE_NEWUSER); >> fd = open("/proc/self/uid_map", O_RDWR); >> >> Where previously opening the uid_map writable would fail because >> the the task had been made non-dumpable. >> >> Signed-off-by: "Eric W. Biederman" > > Acked-by: Serge Hallyn > >> --- >> kernel/cred.c | 26 +++++++++++++++++++++++++- >> 1 files changed, 25 insertions(+), 1 deletions(-) >> >> diff --git a/kernel/cred.c b/kernel/cred.c >> index 48cea3d..993a7ea41 100644 >> --- a/kernel/cred.c >> +++ b/kernel/cred.c >> @@ -455,6 +455,30 @@ error_put: >> return ret; >> } >> > > Do you think we need to warn that this can only be used for > commit_creds? (i.e. if someone tried ot use this in some > other context, the 'creds are subset of target ns is a child > of current_ns' assumption would be wrong) This function should be a general test valid at any time. Except that I forgot the bit of the test that asks is the original cred the owner of the subset user namespace. I will respin this patch. As a small segway this property that unshare(CLONE_NEWUSER) results in a subset of the capabilities a process already had is a very important property to make it possible to reason about user namespaces. Maintaining this property is the reason behind the choices I made in fixing cap_capable. >> +static bool cred_cap_issubset(const struct cred *set, const struct cred *subset) >> +{ >> + const struct user_namespace *set_ns = set->user_ns; >> + const struct user_namespace *subset_ns = subset->user_ns; >> + >> + /* If the two credentials are in the same user namespace see if >> + * the capabilities of subset are a subset of set. >> + */ >> + if (set_ns == subset_ns) >> + return cap_issubset(subset->cap_permitted, set->cap_permitted); >> + >> + /* The credentials are in a different user namespaces > > This can only happen during setns and CLONE_NEWUSER right? Right. This can only happen during setns, unshare(CLONE_NEWUSER), and possibly during clone. Otherwise we are not changing the user namespace. However for clarity and robustness I don't want the code to rely on that. >> + * therefore one is a subset of the other only if a set is an >> + * ancestor of subset. >> + */ > >> + while (subset_ns != &init_user_ns) { >> + if (set_ns == subset_ns->parent) >> + return true; >> + subset_ns = subset_ns->parent; >> + } >> + >> + return false; >> +} >> + >> /** >> * commit_creds - Install new credentials upon the current task >> * @new: The credentials to be assigned >> @@ -493,7 +517,7 @@ int commit_creds(struct cred *new) >> !gid_eq(old->egid, new->egid) || >> !uid_eq(old->fsuid, new->fsuid) || >> !gid_eq(old->fsgid, new->fsgid) || >> - !cap_issubset(new->cap_permitted, old->cap_permitted)) { >> + !cred_cap_issubset(old, new)) { >> if (task->mm) >> set_dumpable(task->mm, suid_dumpable); >> task->pdeath_signal = 0; >> -- >> 1.7.5.4