[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <6ec16f9d-b4f4-bb85-3496-be110fa68f6b@linux.microsoft.com>
Date: Mon, 2 Dec 2019 10:39:25 -0800
From: Lakshmi Ramasubramanian <nramas@...ux.microsoft.com>
To: Mimi Zohar <zohar@...ux.ibm.com>, linux-integrity@...r.kernel.org
Cc: eric.snowberg@...cle.com, dhowells@...hat.com,
matthewgarrett@...gle.com, sashal@...nel.org,
jamorris@...ux.microsoft.com, linux-kernel@...r.kernel.org,
keyrings@...r.kernel.org, Janne Karhunen <janne.karhunen@...il.com>
Subject: Re: [PATCH v0 1/2] IMA: Defined queue functions
On 12/2/19 10:00 AM, Mimi Zohar wrote:
>
> ima_update_policy() is called from multiple places. Initially, it is
> called before a custom policy has been loaded. The call to
> ima_process_queued_keys_for_measurement() needs to be moved to within
> the test, otherwise it runs the risk of dropping "key" measurements.
static const struct file_operations ima_measure_policy_ops = {
.release = ima_release_policy,
};
ima_update_policy() is called from ima_release_policy() function.
On my test machine I have the IMA policy in /etc/ima/ima-policy file.
When IMA policy is setup from this file, I see ima_release_policy()
called (which in turn calls ima_update_policy()).
How can I have ima_update_policy() called before a custom policy is loaded?
If CONFIG_IMA_WRITE_POLICY is enabled, IMA policy can be updated at
runtime - which would invoke ima_update_policy().
Is there any other way ima_update_policy() can get called?
> All the queued keys need to be processed at the same time. Afterwards
> the queue should be deleted. Unfortunately, the current queue locking
> assumes ima_process_queued_keys_for_measurement() is called multiple
> times.
When ima_process_queued_keys_for_measurement() is called the first time,
the flag ima_process_keys_for_measurement is set to true and all queued
keys are processed at that time.
The queue is not used after this point - In the IMA hook the key is
processed immediately when ima_process_keys_for_measurement is set to true.
ima_process_queued_keys_for_measurement() can be called multiple times,
but on 2nd and subsequent calls it will be a NOP.
>
> Perhaps using the RCU method of walking lists would help. I need to
> think about it some more.
>
> Mimi
Please let me know how using RCU method would help.
thanks,
-lakshmi
Powered by blists - more mailing lists