From: "Mickaël Salaün" <mic@digikod.net>
To: Konstantin Meskhidze <konstantin.meskhidze@huawei.com>
Cc: linux-security-module@vger.kernel.org, netdev@vger.kernel.org,
netfilter@vger.kernel.org, yusongping@huawei.com,
artem.kuzin@huawei.com
Subject: Re: [RFC PATCH 1/2] landlock: TCP network hooks implementation
Date: Tue, 8 Feb 2022 13:09:55 +0100 [thread overview]
Message-ID: <a08cd5ec-8bbd-3984-a189-f1e436d09ae8@digikod.net> (raw)
In-Reply-To: <cc023cc4-043b-c67e-1e6e-acf1eb18d155@huawei.com>
On 08/02/2022 08:55, Konstantin Meskhidze wrote:
>
>
> 2/7/2022 5:17 PM, Mickaël Salaün пишет:
>>
>> On 07/02/2022 14:09, Konstantin Meskhidze wrote:
>>>
>>>
>>> 2/1/2022 3:13 PM, Mickaël Salaün пишет:
>>>>
>>>> On 24/01/2022 09:02, Konstantin Meskhidze wrote:
>>>>> Support of socket_bind() and socket_connect() hooks.
>>>>> Current prototype can restrict binding and connecting of TCP
>>>>> types of sockets. Its just basic idea how Landlock could support
>>>>> network confinement.
>>>>>
>>>>> Changes:
>>>>> 1. Access masks array refactored into 1D one and changed
>>>>> to 32 bits. Filesystem masks occupy 16 lower bits and network
>>>>> masks reside in 16 upper bits.
>>>>> 2. Refactor API functions in ruleset.c:
>>>>> 1. Add void *object argument.
>>>>> 2. Add u16 rule_type argument.
>>>>> 3. Use two rb_trees in ruleset structure:
>>>>> 1. root_inode - for filesystem objects
>>>>> 2. root_net_port - for network port objects
>>>>
>>>> It's good to add a changelog, but they must not be in commit
>>>> messages that get copied by git am. Please use "---" to separate
>>>> this additionnal info (but not the Signed-off-by). Please also
>>>> include a version in the email subjects (this one should have been
>>>> "[RFC PATCH v3 1/2] landlock: …"), e.g. using git format-patch
>>>> --reroll-count=X .
>>>>
>>>> Please follow these rules:
>>>> https://www.kernel.org/doc/html/latest/process/submitting-patches.html
>>>> You can take some inspiration from this patch series:
>>>> https://lore.kernel.org/lkml/20210422154123.13086-1-mic@digikod.net/
>>>
>>> Ok. I will add patch vervison in next patch. So it will be "[RFC PATCH
>>> v4 ../..] landlock: ..."
>>> But the previous patches remain with no version, correct?
>>
>> Right, you can't change the subject of already sent emails. ;)
>
> Ok. But I can add previous patches like:
> v1:
> https://lore.kernel.org/linux-security-module/20211210072123.386713-1-konstantin.meskhidze@huawei.com
>
> v2:
> https://lore.kernel.org/netdev/20211228115212.703084-1-konstantin.meskhidze@huawei.com/
>
> v3: ....
>
> right ?
Absolutely! This is a good practice (and would be better in reverse order).
>> [...]
>>
>>>>> @@ -67,10 +76,11 @@ static void build_check_rule(void)
>>>>> }
>>>>> static struct landlock_rule *create_rule(
>>>>> - struct landlock_object *const object,
>>>>> + void *const object,
>>>>
>>>> Instead of shoehorning two different types into one (and then
>>>> loosing the typing), you should rename object to object_ptr and add
>>>> a new object_data argument. Only one of these should be set
>>>> according to the rule_type. However, if there is no special action
>>>> performed on one of these type (e.g. landlock_get_object), only one
>>>> uintptr_t argument should be enough.
>>>>
>>> Do you mean using 2 object arguments in create_rule():
>>>
>>> 1. create_rule( object_ptr = landlock_object , object_data = 0,
>>> ..., fs_rule_type);
>>> 2. create_rule( object_ptr = NULL , object_data = port, .... ,
>>> net_rule_type);
>>
>> Yes, and you can add a WARN_ON_ONCE() in these function to check that
>> only one argument is set (but object_data could be 0 in each case).
>> The landlock_get_object() function should only require an object_data
>> though.
>>
> Sorry. As you said in previous comment in landlock_get_object, only
> one uintptr_t argument should be enough. But, I did not get: "The
> landlock_get_object() function should only require an object_data
> though".
> uintptr_t is the only argument in landlock_get_object?
I was thinking about landlock_find_rule(), not landlock_get_object():
const struct landlock_rule *landlock_find_rule(
const struct landlock_ruleset *const ruleset,
const uintptr_t object_data)
>> [...]
>>
>>>>> @@ -317,47 +331,91 @@ SYSCALL_DEFINE4(landlock_add_rule,
>>>>> if (flags)
>>>>> return -EINVAL;
>>>>> - if (rule_type != LANDLOCK_RULE_PATH_BENEATH)
>>>>> + if ((rule_type != LANDLOCK_RULE_PATH_BENEATH) &&
>>>>> + (rule_type != LANDLOCK_RULE_NET_SERVICE))
>>>>
>>>> Please replace with a switch/case.
>>>
>>> Ok. I got it.
>>>>
>>>>
>>>>> return -EINVAL;
>>>>> - /* Copies raw user space buffer, only one type for now. */
>>>>> - res = copy_from_user(&path_beneath_attr, rule_attr,
>>>>> - sizeof(path_beneath_attr));
>>>>> - if (res)
>>>>> - return -EFAULT;
>>>>> -
>>>>> - /* Gets and checks the ruleset. */
>>>>> - ruleset = get_ruleset_from_fd(ruleset_fd, FMODE_CAN_WRITE);
>>>>> - if (IS_ERR(ruleset))
>>>>> - return PTR_ERR(ruleset);
>>>>> -
>>>>> - /*
>>>>> - * Informs about useless rule: empty allowed_access (i.e. deny
>>>>> rules)
>>>>> - * are ignored in path walks.
>>>>> - */
>>>>> - if (!path_beneath_attr.allowed_access) {
>>>>> - err = -ENOMSG;
>>>>> - goto out_put_ruleset;
>>>>> - }
>>>>> - /*
>>>>> - * Checks that allowed_access matches the @ruleset constraints
>>>>> - * (ruleset->fs_access_masks[0] is automatically upgraded to
>>>>> 64-bits).
>>>>> - */
>>>>> - if ((path_beneath_attr.allowed_access |
>>>>> ruleset->fs_access_masks[0]) !=
>>>>> - ruleset->fs_access_masks[0]) {
>>>>> - err = -EINVAL;
>>>>> - goto out_put_ruleset;
>>>>> + switch (rule_type) {
>>>>> + case LANDLOCK_RULE_PATH_BENEATH:
>>>>> + /* Copies raw user space buffer, for fs rule type. */
>>>>> + res = copy_from_user(&path_beneath_attr, rule_attr,
>>>>> + sizeof(path_beneath_attr));
>>>>> + if (res)
>>>>> + return -EFAULT;
>>>>> + break;
>>>>> +
>>>>> + case LANDLOCK_RULE_NET_SERVICE:
>>>>> + /* Copies raw user space buffer, for net rule type. */
>>>>> + res = copy_from_user(&net_service_attr, rule_attr,
>>>>> + sizeof(net_service_attr));
>>>>> + if (res)
>>>>> + return -EFAULT;
>>>>> + break;
>>>>> }
>>>>> - /* Gets and checks the new rule. */
>>>>> - err = get_path_from_fd(path_beneath_attr.parent_fd, &path);
>>>>> - if (err)
>>>>> - goto out_put_ruleset;
>>>>> + if (rule_type == LANDLOCK_RULE_PATH_BENEATH) {
>>>>> + /* Gets and checks the ruleset. */
>>>>> + ruleset = get_ruleset_from_fd(ruleset_fd, FMODE_CAN_WRITE);
>>>>> + if (IS_ERR(ruleset))
>>>>> + return PTR_ERR(ruleset);
>>>>> +
>>>>> + /*
>>>>> + * Informs about useless rule: empty allowed_access (i.e.
>>>>> deny rules)
>>>>> + * are ignored in path walks.
>>>>> + */
>>>>> + if (!path_beneath_attr.allowed_access) {
>>>>> + err = -ENOMSG;
>>>>> + goto out_put_ruleset;
>>>>> + }
>>>>> + /*
>>>>> + * Checks that allowed_access matches the @ruleset
>>>>> constraints
>>>>> + * (ruleset->access_masks[0] is automatically upgraded to
>>>>> 64-bits).
>>>>> + */
>>>>> + if ((path_beneath_attr.allowed_access |
>>>>> ruleset->access_masks[0]) !=
>>>>> + ruleset->access_masks[0]) {
>>>>> + err = -EINVAL;
>>>>> + goto out_put_ruleset;
>>>>> + }
>>>>> +
>>>>> + /* Gets and checks the new rule. */
>>>>> + err = get_path_from_fd(path_beneath_attr.parent_fd, &path);
>>>>> + if (err)
>>>>> + goto out_put_ruleset;
>>>>> +
>>>>> + /* Imports the new rule. */
>>>>> + err = landlock_append_fs_rule(ruleset, &path,
>>>>> + path_beneath_attr.allowed_access);
>>>>> + path_put(&path);
>>>>> + }
>>>>> - /* Imports the new rule. */
>>>>> - err = landlock_append_fs_rule(ruleset, &path,
>>>>> - path_beneath_attr.allowed_access);
>>>>> - path_put(&path);
>>>>> + if (rule_type == LANDLOCK_RULE_NET_SERVICE) {
>>>>> + /* Gets and checks the ruleset. */
>>>>> + ruleset = get_ruleset_from_fd(ruleset_fd, FMODE_CAN_WRITE);
>>>>
>>>> You need to factor out more code.
>>>
>>> Sorry. I did not get you here. Please could you explain more
>>> detailed?
>>
>> Instead of duplicating similar function calls (e.g.
>> get_ruleset_from_fd) or operations, try to use one switch statement
>> where you put the checks that are different (you can move the
>> copy_from_user(&path_beneath_attr...) call). It may be a good idea to
>> split this function into 3: one handling each rule_attr, which enables
>> to not mix different attr types in the same function. A standalone
>> patch should be refactoring the code to add and use a new function
>> add_rule_path_beneath(ruleset, rule_attr) (only need the "landlock_"
>> prefix for exported functions).
>
> Sorry again. Still don't get the point. What function do you suggetst
> to split in 3? Can you please give detailed template of these
> functions and the logic?
You can split SYSCALL_DEFINE4(landlock_add_rule) in 3:
- a lighten version of SYSCALL_DEFINE4(landlock_add_rule) containing
switch cases for rule_type (almost what you did but with the
get_ruleset_from_fd moved before);
- a new add_rule_path_beneath(ruleset, rule_attr) which will be called
by the switch case;
- a new add_rule_net_service(ruleset, rule_attr) which will be called by
the switch case.
next prev parent reply other threads:[~2022-02-08 13:15 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-01-24 8:02 [RFC PATCH 0/2] landlock network implementation cover letter Konstantin Meskhidze
2022-01-24 8:02 ` [RFC PATCH 1/2] landlock: TCP network hooks implementation Konstantin Meskhidze
2022-01-25 14:17 ` Willem de Bruijn
2022-01-26 8:05 ` Konstantin Meskhidze
2022-01-26 14:15 ` Willem de Bruijn
2022-01-29 3:12 ` Konstantin Meskhidze
2022-01-31 17:14 ` Willem de Bruijn
2022-02-01 12:33 ` Mickaël Salaün
2022-02-07 2:31 ` Konstantin Meskhidze
2022-02-07 16:00 ` Willem de Bruijn
2022-02-07 16:17 ` Willem de Bruijn
2022-02-10 2:05 ` Konstantin Meskhidze
2022-02-10 2:04 ` Konstantin Meskhidze
2022-02-01 12:28 ` Mickaël Salaün
2022-02-07 2:35 ` Konstantin Meskhidze
2022-02-01 12:13 ` Mickaël Salaün
2022-02-07 13:09 ` Konstantin Meskhidze
2022-02-07 14:17 ` Mickaël Salaün
2022-02-08 7:55 ` Konstantin Meskhidze
2022-02-08 12:09 ` Mickaël Salaün [this message]
2022-02-09 3:06 ` Konstantin Meskhidze
2022-01-24 8:02 ` [RFC PATCH 2/2] landlock: selftests for bind and connect hooks Konstantin Meskhidze
2022-02-01 18:31 ` Mickaël Salaün
2022-02-07 7:11 ` Konstantin Meskhidze
2022-02-07 12:49 ` Mickaël Salaün
2022-02-08 3:01 ` Konstantin Meskhidze
2022-02-08 12:17 ` Mickaël Salaün
2022-02-09 3:03 ` Konstantin Meskhidze
2022-02-10 10:16 ` Mickaël Salaün
2022-02-24 3:18 ` Konstantin Meskhidze
2022-02-24 9:55 ` Mickaël Salaün
2022-02-24 12:03 ` Konstantin Meskhidze
2022-02-24 14:15 ` Mickaël Salaün
2022-02-25 2:44 ` Konstantin Meskhidze
2022-02-01 17:53 ` [RFC PATCH 0/2] landlock network implementation cover letter Mickaël Salaün
2022-02-07 13:18 ` Konstantin Meskhidze
2022-02-07 13:35 ` Mickaël Salaün
2022-02-08 3:53 ` Konstantin Meskhidze
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=a08cd5ec-8bbd-3984-a189-f1e436d09ae8@digikod.net \
--to=mic@digikod.net \
--cc=artem.kuzin@huawei.com \
--cc=konstantin.meskhidze@huawei.com \
--cc=linux-security-module@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=netfilter@vger.kernel.org \
--cc=yusongping@huawei.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).