Re: [PATCH v14 0/6] LSM: Multiple concurrent LSMs

[Casey Schaufler]

On 7/31/2013 7:48 PM, Balbir Singh wrote:
> On Thu, Jul 25, 2013 at 11:52 PM, Casey Schaufler
> <casey@xxxxxxxxxxxxxxxx> wrote:
>> Subject: [PATCH v14 0/6] LSM: Multiple concurrent LSMs
>>
>> Version 14 of this patchset is based on v3.10.
>> It required significant change from version 13 due to changes
>> in the audit code. It came out cleaner, especially in the changes
>> to NetLabel. This version supports all existing LSMs running
>> together at the same time. The combinations tested most completely
>> are:
>>
>>     apparmor,tomoyo,smack,yama  - Ubuntu
>>     apparmor,selinux,smack,yama - Fedora
>>
> Does this change the way one would develop a new LSM module? I presume
> it does not

The change that LSM developers need to be aware of is the security blob
abstraction. Instead of using cred->security, inode->i_security and the
like the code needs to use lsm_get_cred() and lsm_set_cred() and similar
functions.

>> I have been unable to figure out how to configure SELinux on
>> Ubuntu and TOMOYO on Fedora. That's the only reason the list
>> does not include all five LSMs at once. Combining LSMs that
>> use networking is tricky, but can be done. There are changes
>> coming from AppArmor that might make it even trickier, but
>> that's a problem for the future.
>>
>>
>> Change the infrastructure for Linux Security Modules (LSM)s from a
>> single vector of hook handlers to a list based method for handling
>> multiple concurrent modules. All combinations of existing LSMs
>> are supported.
>>
>> The "security=" boot option takes a comma separated list of LSMs,
>> registering them in the order presented. The LSM hooks will be
>> executed in the order registered. Hooks that return errors are
>> not short circuited. All hooks are called even if one of the LSM
>> hooks fails. The result returned will be that of the last LSM
>> hook that failed.
>>
> This is an important design trade-off. From my perspective I think you
> might want to revisit this, today it sounds like effective security ==
> all hooks process and allow the operation. In this world a lack of
> proper policy/setting can make hooks fail. I've not yet looked at the
> code, but you might want to revisit this.

The result of an LSM hook will be failure if any of the LSMs
indicates failure. The key here is that all of the LSM hooks
get called even if it's known that the overall result is failure.
This is done because many LSM hooks maintain internal state and
shortcutting can disrupt that.

>> All behavior from security/capability.c has been moved into
>> the hook handling.  The security/commoncap functions used
>> to get called from the LSM specific code. The handling of the
>> capability functions has been moved out of the LSMs and into the
>> hook handling.
>>
>> A level of indirection has been introduced in the handling of
>> security blobs. LSMs no longer access ->security fields directly,
>> instead they use an abstraction provided by lsm_[gs]et field
>> functions.
>>
>> The notion that "the security context" can be represented as a
>> single u32 "secid" does not scale to the case where multiple LSMs
>> want to provide "the security context". The XFRM and secmark
>> facilities appear unlikely to ever allow for more than the existing
>> 32 bit values. The NetLabel scheme might possibly be used to
>> represent more than one labeling scheme (CIPSO does allow for
>> multiple tags) although there is no plan to do so at this time.
>> The SO_PEERSEC scheme is capable of providing information from
>> multiple LSMs. Auditing can deal with multiple secids.
>>
>> The NetLabel, XFRM and secmark facilities are restricted to use
>> by one LSM at a time. The SO_PEERSEC facility can provide information
>> from multiple LSMs, but existing user space tools don't understand
>> that. The default behavior is to assign each of these facilities
>> to the first registered LSM that uses them. They can be configured
>> for use by any of the LSMs that provide hooks for them. SO_PEERSEC
>> can be configured to provide information from all of the LSMs that
>> provide hooks.
>>
>> The /proc/*/attr interfaces are given to one LSM. This can be
>> done by setting CONFIG_SECURITY_PRESENT. Additional interfaces
>> have been created in /proc/*/attr so that each LSM has its own
>> named interfaces. The name of the presenting LSM can be read from
>> /sys/kernel/security/present. The list of LSMs being used can be
>> read from /sys/kernel/security/lsm.
>>
>> A "security context" may now contrain information processed by
>> more than one LSM. The proper form of a security context identifies
>> the information it contains by LSM:
>>
>>     smack='Pop'selinux='system_u:object_r:etc_r:s0'
>>
>> A security context without the LSM identifying lsm='<text>' gets
>> passed through to all of the LSMs that use a security context. This
>> maintains compatability in the case where there is only one LSM
>> using the security context.
>>
>> Signed-off-by: Casey Schaufler <casey@xxxxxxxxxxxxxxxx>
> Balbir Singh
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/