[PATCH v5 08/15] landlock: TCP network hooks implementation

Konstantin Meskhidze konstantin.meskhidze at huawei.com
Thu May 19 11:40:42 UTC 2022



5/17/2022 11:51 AM, Mickaël Salaün пишет:
> 
> On 16/05/2022 17:20, Konstantin Meskhidze wrote:
>> Support of socket_bind() and socket_connect() hooks.
>> Its possible to restrict binding and connecting of TCP
>> types of sockets to particular ports. Its just basic idea
>> how Landlock could support network confinement.
>>
>> Signed-off-by: Konstantin Meskhidze <konstantin.meskhidze at huawei.com>
>> ---
>>
>> Changes since v3:
>> * Split commit.
>> * Add SECURITY_NETWORK in config.
>> * Add IS_ENABLED(CONFIG_INET) if a kernel has no INET configuration.
>> * Add hook_socket_bind and hook_socket_connect hooks.
>>
>> Changes since v4:
>> * Factors out CONFIG_INET into make file.
>> * Refactoring check_socket_access().
>> * Adds helper get_port().
>> * Adds CONFIG_IPV6 in  get_port(), hook_socket_bind/connect
>> functions to support AF_INET6 family.
>> * Adds AF_UNSPEC family support in hook_socket_bind/connect
>> functions.
>> * Refactoring add_rule_net_service() and landlock_add_rule
>> syscall to support network rule inserting.
>> * Refactoring init_layer_masks() to support network rules.
>>
>> ---
>>   security/landlock/Kconfig    |   1 +
>>   security/landlock/Makefile   |   2 +
>>   security/landlock/net.c      | 159 +++++++++++++++++++++++++++++++++++
>>   security/landlock/net.h      |  25 ++++++
>>   security/landlock/ruleset.c  |  15 +++-
>>   security/landlock/setup.c    |   2 +
>>   security/landlock/syscalls.c |  63 ++++++++++++--
>>   7 files changed, 261 insertions(+), 6 deletions(-)
>>   create mode 100644 security/landlock/net.c
>>   create mode 100644 security/landlock/net.h
>>
>> diff --git a/security/landlock/Kconfig b/security/landlock/Kconfig
>> index 8e33c4e8ffb8..10c099097533 100644
>> --- a/security/landlock/Kconfig
>> +++ b/security/landlock/Kconfig
>> @@ -3,6 +3,7 @@
>>   config SECURITY_LANDLOCK
>>       bool "Landlock support"
>>       depends on SECURITY && !ARCH_EPHEMERAL_INODES
>> +    select SECURITY_NETWORK
>>       select SECURITY_PATH
>>       help
>>         Landlock is a sandboxing mechanism that enables processes to 
>> restrict
>> diff --git a/security/landlock/Makefile b/security/landlock/Makefile
>> index 7bbd2f413b3e..53d3c92ae22e 100644
>> --- a/security/landlock/Makefile
>> +++ b/security/landlock/Makefile
>> @@ -2,3 +2,5 @@ obj-$(CONFIG_SECURITY_LANDLOCK) := landlock.o
>>
>>   landlock-y := setup.o syscalls.o object.o ruleset.o \
>>       cred.o ptrace.o fs.o
>> +
>> +landlock-$(CONFIG_INET) += net.o
>> \ No newline at end of file
>> diff --git a/security/landlock/net.c b/security/landlock/net.c
>> new file mode 100644
>> index 000000000000..9302e5891991
>> --- /dev/null
>> +++ b/security/landlock/net.c
>> @@ -0,0 +1,159 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/*
>> + * Landlock LSM - Network management and hooks
>> + *
>> + * Copyright (C) 2022 Huawei Tech. Co., Ltd.
>> + */
>> +
>> +#include <linux/in.h>
>> +#include <linux/net.h>
>> +#include <linux/socket.h>
>> +#include <net/ipv6.h>
>> +
>> +#include "cred.h"
>> +#include "limits.h"
>> +#include "net.h"
>> +
>> +int landlock_append_net_rule(struct landlock_ruleset *const ruleset,
>> +                 u16 port, u32 access_rights)
>> +{
>> +    int err;
>> +
>> +    /* Transforms relative access rights to absolute ones. */
>> +    access_rights |= LANDLOCK_MASK_ACCESS_NET &
>> +             ~landlock_get_net_access_mask(ruleset, 0);
>> +
>> +    BUILD_BUG_ON(sizeof(port) > sizeof(uintptr_t));
>> +    mutex_lock(&ruleset->lock);
>> +    err = landlock_insert_rule(ruleset, NULL, port,
>> +                access_rights, LANDLOCK_RULE_NET_SERVICE);
>> +    mutex_unlock(&ruleset->lock);
>> +
>> +    return err;
>> +}
>> +
>> +static int check_socket_access(const struct landlock_ruleset *const 
>> domain,
>> +                   u16 port, access_mask_t access_request)
>> +{
>> +    bool allowed = false;
>> +    layer_mask_t layer_masks[LANDLOCK_NUM_ACCESS_NET] = {};
>> +    const struct landlock_rule *rule;
>> +    access_mask_t handled_access;
>> +
>> +    if (WARN_ON_ONCE(!domain))
>> +        return 0;
>> +    if (WARN_ON_ONCE(domain->num_layers < 1))
>> +        return -EACCES;
>> +
>> +    rule = landlock_find_rule(domain, port,
>> +                    LANDLOCK_RULE_NET_SERVICE);
>> +
>> +    handled_access = init_layer_masks(domain, access_request,
>> +            &layer_masks, sizeof(layer_masks),
>> +            LANDLOCK_RULE_NET_SERVICE);
>> +    allowed = unmask_layers(rule, handled_access,
>> +            &layer_masks, ARRAY_SIZE(layer_masks));
>> +
>> +    return allowed ? 0 : -EACCES;
>> +}
>> +
>> +static u16 get_port(const struct sockaddr *const address)
>> +{
>> +    /* Gets port value in host byte order. */
>> +    switch (address->sa_family) {
>> +    case AF_UNSPEC:
> 
> Are you sure about that?
> 
> Please write a test for this case.
> 
   I checked the logic in Selinux LSM in a bind() hook and AF_UNSPEC
family is treated as AF_INET (only for INADDR_ANY address)  but we can 
discuss it.
> 
>> +    case AF_INET:
>> +    {
> 
> You don't need these braces (except if it is required by checkpatch.pl).
> 
   I got compilation error without braces:
...
error: a label can only be part of a statement and a declaration is not 
a statement
    67 |   const struct sockaddr_in *const sockaddr =
...
  so I added braces.
> 
>> +        const struct sockaddr_in *const sockaddr =
>> +                    (struct sockaddr_in *)address;
>> +        return ntohs(sockaddr->sin_port);
>> +    }
>> +#if IS_ENABLED(CONFIG_IPV6)
>> +    case AF_INET6:
>> +    {
>> +        const struct sockaddr_in6 *const sockaddr_ip6 =
>> +                    (struct sockaddr_in6 *)address;
>> +        return ntohs(sockaddr_ip6->sin6_port);
>> +    }
>> +#endif
>> +    }
> 
> You missed some part of my patch… We should not get the port for a 
> protocol we don't know, hence the WARN_ON_ONCE.
> 
  Sorry. I manually updated the code and missed that. I will fix it.
> 
>> +    return 0;
>> +}
>> +
>> +static int hook_socket_bind(struct socket *sock, struct sockaddr 
>> *address,
>> +                int addrlen)
>> +{
>> +    const struct landlock_ruleset *const dom =
>> +                        landlock_get_current_domain();
>> +
>> +    if (!dom)
>> +        return 0;
>> +
>> +    /* Check if it's a TCP socket */
>> +    if (sock->type != SOCK_STREAM)
>> +        return 0;
>> +
>> +    /* Get port value in host byte order */
> 
> I moved/removed this in my patch against v4 for a reason. Please, ask if 
> you don't understand or if you don't agree with something I said.
> 
   Sorry. I updated your changed manually and missed that.
   Will be fixed.
> 
>> +    switch (address->sa_family) {
>> +    case AF_UNSPEC:
> 
> Is this correct?

Actually I did'not find any issue with bind() using AF_UNSPEC family.
(instead of connect() one and we came to an agreement that it must be 
banned). So I checked athoner LSMs like Selinux and found out that
it treats bind() hook with AF_UNSPEC family as with AF_INET one.
But Selinux allows AF_UNSPEC family only for INADDR_ANY address and
I cant understand the reason now. So I let AF_UNSPEC family bind()
hook for any address, but we can discuss it. What do you think?
> 
> Please write a test for this case.
> 
  I have one. Please check it in connect_afunspec_with_restictions
  test:

...
  /* Enforces the ruleset. */
	enforce_ruleset(_metadata, ruleset_fd_1);

	/* Creates a server socket 1 */
	sockfd = create_socket(_metadata, false, false);
	ASSERT_LE(0, sockfd);

	/* Binds the socket 1 to address with port[0] with AF_UNSPEC
            family */
	self->addr4[0].sin_family = AF_UNSPEC;
	ASSERT_EQ(0, bind(sockfd, (struct sockaddr *)&self->addr4[0],
                           sizeof(self->addr4[0])));

	/* Makes connection to socket with port[0] */
	ASSERT_EQ(0, connect(sockfd, (struct sockaddr *)&self->addr4[0],
			     sizeof(self->addr4[0])));

	const int ruleset_fd_2 =
                              landlock_create_ruleset(&ruleset_attr_2,
					sizeof(ruleset_attr_2), 0);
	ASSERT_LE(0, ruleset_fd_2);

	/* Allows connect and bind operations to the port[0] socket */
	ASSERT_EQ(0, landlock_add_rule(ruleset_fd_2, 	
                                        LANDLOCK_RULE_NET_SERVICE,
				       &net_service_2, 0));
...
>> +    case AF_INET:
>> +#if IS_ENABLED(CONFIG_IPV6)
>> +    case AF_INET6:
>> +#endif
>> +        return check_socket_access(dom, get_port(address),
>> +                    LANDLOCK_ACCESS_NET_BIND_TCP);
>> +    default:
>> +        return 0;
>> +    }
>> +}
>> +
>> +static int hook_socket_connect(struct socket *sock, struct sockaddr 
>> *address,
>> +                int addrlen)
>> +{
>> +    const struct landlock_ruleset *const dom =
>> +                        landlock_get_current_domain();
>> +
>> +    if (!dom)
>> +        return 0;
>> +
>> +    /* Check if it's a TCP socket */
>> +    if (sock->type != SOCK_STREAM)
>> +        return 0;
>> +
>> +    /* Get port value in host byte order */
>> +    switch (address->sa_family) {
>> +    case AF_INET:
>> +#if IS_ENABLED(CONFIG_IPV6)
>> +    case AF_INET6:
>> +#endif
>> +        return check_socket_access(dom, get_port(address),
>> +                       LANDLOCK_ACCESS_NET_CONNECT_TCP);
>> +    case AF_UNSPEC:
>> +    {
>> +        u16 i;
>> +        /*
>> +         * If just in a layer a mask supports connect access,
>> +         * the socket_connect() hook with AF_UNSPEC family flag
>> +         * must be banned. This prevents from disconnecting already
>> +         * connected sockets.
>> +         */
>> +        for (i = 0; i < dom->num_layers; i++) {
>> +            if (landlock_get_net_access_mask(dom, i) &
>> +                LANDLOCK_ACCESS_NET_CONNECT_TCP)
>> +                return -EACCES;
>> +        }
>> +    }
>> +    }
>> +    return 0;
>> +}
>> +
>> +static struct security_hook_list landlock_hooks[] __lsm_ro_after_init 
>> = {
>> +    LSM_HOOK_INIT(socket_bind, hook_socket_bind),
>> +    LSM_HOOK_INIT(socket_connect, hook_socket_connect),
>> +};
>> +
>> +__init void landlock_add_net_hooks(void)
>> +{
>> +    security_add_hooks(landlock_hooks, ARRAY_SIZE(landlock_hooks),
>> +            LANDLOCK_NAME);
>> +}
>> diff --git a/security/landlock/net.h b/security/landlock/net.h
>> new file mode 100644
>> index 000000000000..da5ce8fa04cc
>> --- /dev/null
>> +++ b/security/landlock/net.h
>> @@ -0,0 +1,25 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/*
>> + * Landlock LSM - Network management and hooks
>> + *
>> + * Copyright (C) 2022 Huawei Tech. Co., Ltd.
>> + */
>> +
>> +#ifndef _SECURITY_LANDLOCK_NET_H
>> +#define _SECURITY_LANDLOCK_NET_H
>> +
>> +#include "common.h"
>> +#include "ruleset.h"
>> +#include "setup.h"
>> +
>> +#if IS_ENABLED(CONFIG_INET)
>> +__init void landlock_add_net_hooks(void);
>> +
>> +int landlock_append_net_rule(struct landlock_ruleset *const ruleset,
>> +                u16 port, u32 access_hierarchy);
>> +#else /* IS_ENABLED(CONFIG_INET) */
>> +static inline void landlock_add_net_hooks(void)
>> +{}
>> +#endif /* IS_ENABLED(CONFIG_INET) */
>> +
>> +#endif /* _SECURITY_LANDLOCK_NET_H */
>> diff --git a/security/landlock/ruleset.c b/security/landlock/ruleset.c
>> index ea9ecb3f471a..317cf98890f6 100644
>> --- a/security/landlock/ruleset.c
>> +++ b/security/landlock/ruleset.c
>> @@ -671,7 +671,7 @@ access_mask_t get_handled_accesses(
>>           }
>>           break;
>>       default:
>> -        break;
>> +        return 0;
> 
> Why?

   You are right - this is useless code, cause anyway 0 is retured in 
default rule_type. Will be fixed.
> 
> 
>>       }
>>       return access_dom;
>>   }
>> @@ -763,6 +763,19 @@ access_mask_t init_layer_masks(const struct 
>> landlock_ruleset *const domain,
>>                   }
>>               }
>>               break;
>> +        case LANDLOCK_RULE_NET_SERVICE:
>> +            for_each_set_bit(access_bit, &access_req,
>> +                    LANDLOCK_NUM_ACCESS_NET) {
>> +                if (landlock_get_net_access_mask(domain,
>> +                                 layer_level) &
>> +                        BIT_ULL(access_bit)) {
>> +                    (*layer_masks)[access_bit] |=
>> +                        BIT_ULL(layer_level);
>> +                    handled_accesses |=
>> +                               BIT_ULL(access_bit);
>> +                }
>> +            }
>> +            break;
>>           default:
>>               return 0;
>>           }
>> diff --git a/security/landlock/setup.c b/security/landlock/setup.c
>> index f8e8e980454c..8059dc0b47d3 100644
>> --- a/security/landlock/setup.c
>> +++ b/security/landlock/setup.c
>> @@ -14,6 +14,7 @@
>>   #include "fs.h"
>>   #include "ptrace.h"
>>   #include "setup.h"
>> +#include "net.h"
>>
>>   bool landlock_initialized __lsm_ro_after_init = false;
>>
>> @@ -28,6 +29,7 @@ static int __init landlock_init(void)
>>       landlock_add_cred_hooks();
>>       landlock_add_ptrace_hooks();
>>       landlock_add_fs_hooks();
>> +    landlock_add_net_hooks();
>>       landlock_initialized = true;
>>       pr_info("Up and running.\n");
>>       return 0;
>> diff --git a/security/landlock/syscalls.c b/security/landlock/syscalls.c
>> index 812541f4e155..9454c6361011 100644
>> --- a/security/landlock/syscalls.c
>> +++ b/security/landlock/syscalls.c
>> @@ -29,6 +29,7 @@
>>   #include "cred.h"
>>   #include "fs.h"
>>   #include "limits.h"
>> +#include "net.h"
>>   #include "ruleset.h"
>>   #include "setup.h"
>>
>> @@ -74,7 +75,8 @@ static void build_check_abi(void)
>>   {
>>       struct landlock_ruleset_attr ruleset_attr;
>>       struct landlock_path_beneath_attr path_beneath_attr;
>> -    size_t ruleset_size, path_beneath_size;
>> +    struct landlock_net_service_attr net_service_attr;
>> +    size_t ruleset_size, path_beneath_size, net_service_size;
>>
>>       /*
>>        * For each user space ABI structures, first checks that there 
>> is no
>> @@ -90,6 +92,11 @@ static void build_check_abi(void)
>>       path_beneath_size += sizeof(path_beneath_attr.parent_fd);
>>       BUILD_BUG_ON(sizeof(path_beneath_attr) != path_beneath_size);
>>       BUILD_BUG_ON(sizeof(path_beneath_attr) != 12);
>> +
>> +    net_service_size = sizeof(net_service_attr.allowed_access);
>> +    net_service_size += sizeof(net_service_attr.port);
>> +    BUILD_BUG_ON(sizeof(net_service_attr) != net_service_size);
>> +    BUILD_BUG_ON(sizeof(net_service_attr) != 10);
>>   }
>>
>>   /* Ruleset handling */
>> @@ -299,9 +306,9 @@ static int add_rule_path_beneath(struct 
>> landlock_ruleset *const ruleset,
>>        * Informs about useless rule: empty allowed_access (i.e. deny 
>> rules)
>>        * are ignored in path walks.
>>        */
>> -    if (!path_beneath_attr.allowed_access) {
>> +    if (!path_beneath_attr.allowed_access)
> 
> Why?

  Unnecessary braces for single statement blocks.
  I got a warning by scripts/checkpatch.pl:

   WARNING: braces {} are not necessary for single statement blocks
#309: FILE: security/landlock/syscalls.c:309:
+       if (!path_beneath_attr.allowed_access) {
+               return -ENOMSG;
+       }

> 
> 
>>           return -ENOMSG;
>> -    }
>> +
>>       /*
>>        * Checks that allowed_access matches the @ruleset constraints
>>        * (ruleset->access_masks[0] is automatically upgraded to 64-bits).
>> @@ -323,13 +330,54 @@ static int add_rule_path_beneath(struct 
>> landlock_ruleset *const ruleset,
>>       return err;
>>   }
>>
>> +static int add_rule_net_service(struct landlock_ruleset *ruleset,
>> +                const void *const rule_attr)
>> +{
>> +#if IS_ENABLED(CONFIG_INET)
>> +    struct landlock_net_service_attr net_service_attr;
>> +    int res;
>> +    u32 mask;
>> +
>> +    /* Copies raw user space buffer, only one type for now. */
>> +    res = copy_from_user(&net_service_attr, rule_attr,
>> +            sizeof(net_service_attr));
>> +    if (res)
>> +        return -EFAULT;
>> +
>> +    /*
>> +     * Informs about useless rule: empty allowed_access (i.e. deny 
>> rules)
>> +     * are ignored by network actions
>> +     */
>> +    if (!net_service_attr.allowed_access)
>> +        return -ENOMSG;
>> +
>> +    /*
>> +     * Checks that allowed_access matches the @ruleset constraints
>> +     * (ruleset->access_masks[0] is automatically upgraded to 64-bits).
>> +     */
>> +    mask = landlock_get_net_access_mask(ruleset, 0);
>> +    if ((net_service_attr.allowed_access | mask) != mask)
>> +        return -EINVAL;
>> +
>> +    /* Denies inserting a rule with port 0 */
>> +    if (net_service_attr.port == 0)
>> +        return -EINVAL;
>> +
>> +    /* Imports the new rule. */
>> +    return landlock_append_net_rule(ruleset, net_service_attr.port,
>> +                       net_service_attr.allowed_access);
>> +#else /* IS_ENABLED(CONFIG_INET) */
>> +    return -EAFNOSUPPORT;
>> +#endif /* IS_ENABLED(CONFIG_INET) */
>> +}
>> +
>>   /**
>>    * sys_landlock_add_rule - Add a new rule to a ruleset
>>    *
>>    * @ruleset_fd: File descriptor tied to the ruleset that should be 
>> extended
>>    *        with the new rule.
>> - * @rule_type: Identify the structure type pointed to by @rule_attr 
>> (only
>> - *             LANDLOCK_RULE_PATH_BENEATH for now).
>> + * @rule_type: Identify the structure type pointed to by @rule_attr:
>> + *             LANDLOCK_RULE_PATH_BENEATH or LANDLOCK_RULE_NET_SERVICE.
>>    * @rule_attr: Pointer to a rule (only of type &struct
>>    *             landlock_path_beneath_attr for now).
>>    * @flags: Must be 0.
>> @@ -340,6 +388,8 @@ static int add_rule_path_beneath(struct 
>> landlock_ruleset *const ruleset,
>>    * Possible returned errors are:
>>    *
>>    * - EOPNOTSUPP: Landlock is supported by the kernel but disabled at 
>> boot time;
>> + * - EAFNOSUPPORT: @rule_type is LANDLOCK_RULE_NET_SERVICE but TCP/IP 
>> is not
>> + *   supported by the running kernel;
>>    * - EINVAL: @flags is not 0, or inconsistent access in the rule (i.e.
>>    *   &landlock_path_beneath_attr.allowed_access is not a subset of 
>> the rule's
>>    *   accesses);
>> @@ -375,6 +425,9 @@ SYSCALL_DEFINE4(landlock_add_rule,
>>       case LANDLOCK_RULE_PATH_BENEATH:
>>           err = add_rule_path_beneath(ruleset, rule_attr);
>>           break;
>> +    case LANDLOCK_RULE_NET_SERVICE:
>> +        err = add_rule_net_service(ruleset, rule_attr);
>> +        break;
>>       default:
>>           err = -EINVAL;
>>           break;
>> -- 
>> 2.25.1
>>
> .



More information about the Linux-security-module-archive mailing list