[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <48467663-5CD6-49C5-B43F-9FA0887D0575@oracle.com>
Date: Wed, 27 Apr 2022 16:12:41 +0000
From: Eric Snowberg <eric.snowberg@...cle.com>
To: Mimi Zohar <zohar@...ux.ibm.com>
CC: "dmitry.kasatkin@...il.com" <dmitry.kasatkin@...il.com>,
"jmorris@...ei.org" <jmorris@...ei.org>,
"serge@...lyn.com" <serge@...lyn.com>,
"linux-integrity@...r.kernel.org" <linux-integrity@...r.kernel.org>,
"linux-security-module@...r.kernel.org"
<linux-security-module@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] integrity: Allow ima_appraise bootparam to be set when SB
is enabled
> On Apr 26, 2022, at 12:18 PM, Mimi Zohar <zohar@...ux.ibm.com> wrote:
>
> On Mon, 2022-04-25 at 18:21 -0400, Eric Snowberg wrote:
>> The IMA_APPRAISE_BOOTPARM config allows enabling different "ima_appraise="
>> modes (log, fix, enforce) to be configured at boot time. When booting
>> with Secure Boot enabled, all modes are ignored except enforce. To use
>> log or fix, Secure Boot must be disabled.
>>
>> With a policy such as:
>>
>> appraise func=BPRM_CHECK appraise_type=imasig
>>
>> A user may just want to audit signature validation. Not all users
>> are interested in full enforcement and find the audit log appropriate
>> for their use case.
>>
>> Add a new IMA_APPRAISE_SB_BOOTPARAM config allowing "ima_appraise="
>> to work when Secure Boot is enabled.
>>
>> Signed-off-by: Eric Snowberg <eric.snowberg@...cle.com>
>
> Since the IMA architecture specific policy rules were first
> upstreamed, either enabling IMA_APPRAISE_BOOTPARAM or IMA_ARCH_POLICY
> was permitted, but not both.
I don’t see code preventing this and just created a config with both of them
enabled. Is this an assumption everyone is supposed to understand?
> This Kconfig negates the assumptions on
> which the CONFIG_IMA_ARCH_POLICY and the ima_appraise_signature() are
> based without any indication of the ramifications. This impacts the
> kexec file syscall lockdown LSM assumptions as well.
I will fix this in the next round
> A fuller, more complete explanation for needing "log" mode when secure
> boot is enabled is required.
and add a more thorough explanation. Thanks.
Powered by blists - more mailing lists