[PATCH 13/14] nvmet-tcp: enable TLS handshake upcall
Sagi Grimberg
sagi at grimberg.me
Mon Aug 7 01:51:38 PDT 2023
On 8/3/23 13:51, Hannes Reinecke wrote:
> Add functions to start the TLS handshake upcall when
> the TCP TSAS sectype is set to 'tls1.3' and add a config
> option NVME_TARGET_TCP_TLS.
>
> Signed-off-by: Hannes Reincke <hare at suse.de>
> ---
> drivers/nvme/target/Kconfig | 14 ++++
> drivers/nvme/target/configfs.c | 63 +++++++++++++++-
> drivers/nvme/target/nvmet.h | 1 +
> drivers/nvme/target/tcp.c | 133 +++++++++++++++++++++++++++++++--
> 4 files changed, 203 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/nvme/target/Kconfig b/drivers/nvme/target/Kconfig
> index 79fc64035ee3..15af2b4341f6 100644
> --- a/drivers/nvme/target/Kconfig
> +++ b/drivers/nvme/target/Kconfig
> @@ -84,6 +84,20 @@ config NVME_TARGET_TCP
>
> If unsure, say N.
>
> +config NVME_TARGET_TCP_TLS
> + bool "NVMe over Fabrics TCP target TLS encryption support"
> + depends on NVME_TARGET_TCP
> + select NVME_COMMON
> + select NVME_KEYRING
> + select NET_HANDSHAKE
> + help
> + Enables TLS encryption for the NVMe TCP target using the netlink handshake API.
> +
> + The TLS handshake daemon is availble at
> + https://github.com/oracle/ktls-utils.
> +
> + If unsure, say N.
> +
> config NVME_TARGET_AUTH
> bool "NVMe over Fabrics In-band Authentication support"
> depends on NVME_TARGET
> diff --git a/drivers/nvme/target/configfs.c b/drivers/nvme/target/configfs.c
> index 7f826ac8b75c..49b407702ad5 100644
> --- a/drivers/nvme/target/configfs.c
> +++ b/drivers/nvme/target/configfs.c
> @@ -15,6 +15,7 @@
> #ifdef CONFIG_NVME_TARGET_AUTH
> #include <linux/nvme-auth.h>
> #endif
> +#include <linux/nvme-keyring.h>
> #include <crypto/hash.h>
> #include <crypto/kpp.h>
>
> @@ -159,10 +160,15 @@ static const struct nvmet_type_name_map nvmet_addr_treq[] = {
> { NVMF_TREQ_NOT_REQUIRED, "not required" },
> };
>
> +static inline u8 nvmet_port_treq(struct nvmet_port *port)
> +{
> + return (port->disc_addr.treq & NVME_TREQ_SECURE_CHANNEL_MASK);
> +}
> +
> static ssize_t nvmet_addr_treq_show(struct config_item *item, char *page)
> {
> - u8 treq = to_nvmet_port(item)->disc_addr.treq &
> - NVME_TREQ_SECURE_CHANNEL_MASK;
> + struct nvmet_port *port = to_nvmet_port(item);
> + u8 treq = nvmet_port_treq(port);
> int i;
>
> for (i = 0; i < ARRAY_SIZE(nvmet_addr_treq); i++) {
> @@ -174,11 +180,16 @@ static ssize_t nvmet_addr_treq_show(struct config_item *item, char *page)
> return snprintf(page, PAGE_SIZE, "\n");
> }
>
> +static inline u8 nvmet_port_treq_mask(struct nvmet_port *port)
> +{
> + return (port->disc_addr.treq & ~NVME_TREQ_SECURE_CHANNEL_MASK);
> +}
> +
> static ssize_t nvmet_addr_treq_store(struct config_item *item,
> const char *page, size_t count)
> {
> struct nvmet_port *port = to_nvmet_port(item);
> - u8 treq = port->disc_addr.treq & ~NVME_TREQ_SECURE_CHANNEL_MASK;
> + u8 treq = nvmet_port_treq_mask(port);
> int i;
>
> if (nvmet_is_port_enabled(port, __func__))
> @@ -193,6 +204,23 @@ static ssize_t nvmet_addr_treq_store(struct config_item *item,
> return -EINVAL;
>
> found:
> + if (port->disc_addr.trtype == NVMF_TRTYPE_TCP) {
> + if (!IS_ENABLED(CONFIG_NVME_TARGET_TCP_TLS)) {
> + pr_err("TLS is not supported\n");
> + return -EINVAL;
> + }
> + if (!port->keyring) {
> + pr_err("TLS keyring not configured\n");
> + return -EINVAL;
> + }
> + if (port->disc_addr.tsas.tcp.sectype != NVMF_TCP_SECTYPE_TLS13) {
> + pr_warn("cannot change TREQ when TLS is not enabled\n");
> + return -EINVAL;
> + } else if (nvmet_addr_treq[i].type == NVMF_TREQ_NOT_SPECIFIED) {
> + pr_warn("cannot set TREQ to 'not specified' when TLS is enabled\n");
> + return -EINVAL;
> + }
> + }
> treq |= nvmet_addr_treq[i].type;
> port->disc_addr.treq = treq;
> return count;
> @@ -371,6 +399,7 @@ static ssize_t nvmet_addr_tsas_store(struct config_item *item,
> const char *page, size_t count)
> {
> struct nvmet_port *port = to_nvmet_port(item);
> + u8 treq = nvmet_port_treq_mask(port);
> int i;
>
> if (nvmet_is_port_enabled(port, __func__))
> @@ -379,6 +408,15 @@ static ssize_t nvmet_addr_tsas_store(struct config_item *item,
> if (port->disc_addr.trtype != NVMF_TRTYPE_TCP)
> return -EINVAL;
>
> + if (!IS_ENABLED(CONFIG_NVME_TARGET_TCP_TLS)) {
> + pr_err("TLS is not supported\n");
> + return -EINVAL;
> + }
> + if (!port->keyring) {
> + pr_err("TLS keyring not configured\n");
> + return -EINVAL;
> + }
> +
> for (i = 0; i < ARRAY_SIZE(nvmet_addr_tsas_tcp); i++) {
> if (sysfs_streq(page, nvmet_addr_tsas_tcp[i].name))
> goto found;
> @@ -389,6 +427,16 @@ static ssize_t nvmet_addr_tsas_store(struct config_item *item,
>
> found:
> nvmet_port_init_tsas_tcp(port, nvmet_addr_tsas_tcp[i].type);
> + if (nvmet_addr_tsas_tcp[i].type == NVMF_TCP_SECTYPE_TLS13) {
> + if (nvmet_port_treq(port) == NVMF_TREQ_NOT_SPECIFIED)
> + treq |= NVMF_TREQ_REQUIRED;
> + else
> + treq |= nvmet_port_treq(port);
> + } else {
> + /* Set to 'not specified' if TLS is not enabled */
> + treq |= NVMF_TREQ_NOT_SPECIFIED;
> + }
> + port->disc_addr.treq = treq;
> return count;
> }
>
Can the treq/tsas be split from the actual nvmet-tcp upcall addition?
> @@ -1795,6 +1843,7 @@ static void nvmet_port_release(struct config_item *item)
> flush_workqueue(nvmet_wq);
> list_del(&port->global_entry);
>
> + key_put(port->keyring);
> kfree(port->ana_state);
> kfree(port);
> }
> @@ -1844,6 +1893,14 @@ static struct config_group *nvmet_ports_make(struct config_group *group,
> return ERR_PTR(-ENOMEM);
> }
>
> + if (nvme_keyring_id()) {
> + port->keyring = key_lookup(nvme_keyring_id());
> + if (IS_ERR(port->keyring)) {
> + pr_warn("NVMe keyring not available, disabling TLS\n");
> + port->keyring = NULL;
> + }
> + }
> +
Nice.
> for (i = 1; i <= NVMET_MAX_ANAGRPS; i++) {
> if (i == NVMET_DEFAULT_ANA_GRPID)
> port->ana_state[1] = NVME_ANA_OPTIMIZED;
> diff --git a/drivers/nvme/target/nvmet.h b/drivers/nvme/target/nvmet.h
> index 8cfd60f3b564..7f9ae53c1df5 100644
> --- a/drivers/nvme/target/nvmet.h
> +++ b/drivers/nvme/target/nvmet.h
> @@ -158,6 +158,7 @@ struct nvmet_port {
> struct config_group ana_groups_group;
> struct nvmet_ana_group ana_default_group;
> enum nvme_ana_state *ana_state;
> + struct key *keyring;
> void *priv;
> bool enabled;
> int inline_data_size;
> diff --git a/drivers/nvme/target/tcp.c b/drivers/nvme/target/tcp.c
> index fdc351f591a4..7279c994abd6 100644
> --- a/drivers/nvme/target/tcp.c
> +++ b/drivers/nvme/target/tcp.c
> @@ -8,9 +8,13 @@
> #include <linux/init.h>
> #include <linux/slab.h>
> #include <linux/err.h>
> +#include <linux/key.h>
> #include <linux/nvme-tcp.h>
> +#include <linux/nvme-keyring.h>
> #include <net/sock.h>
> #include <net/tcp.h>
> +#include <net/tls.h>
> +#include <net/handshake.h>
> #include <linux/inet.h>
> #include <linux/llist.h>
> #include <crypto/hash.h>
> @@ -66,6 +70,16 @@ device_param_cb(idle_poll_period_usecs, &set_param_ops,
> MODULE_PARM_DESC(idle_poll_period_usecs,
> "nvmet tcp io_work poll till idle time period in usecs: Default 0");
>
> +#ifdef CONFIG_NVME_TARGET_TCP_TLS
> +/*
> + * TLS handshake timeout
> + */
> +static int tls_handshake_timeout = 30;
> +module_param(tls_handshake_timeout, int, 0644);
> +MODULE_PARM_DESC(tls_handshake_timeout,
> + "nvme TLS handshake timeout in seconds (default 30)");
> +#endif
On the host it is 10 and here 30? what is the source of the assymmetry?
> +
> #define NVMET_TCP_RECV_BUDGET 8
> #define NVMET_TCP_SEND_BUDGET 8
> #define NVMET_TCP_IO_WORK_BUDGET 64
> @@ -122,6 +136,7 @@ struct nvmet_tcp_cmd {
>
> enum nvmet_tcp_queue_state {
> NVMET_TCP_Q_CONNECTING,
> + NVMET_TCP_Q_TLS_HANDSHAKE,
> NVMET_TCP_Q_LIVE,
> NVMET_TCP_Q_DISCONNECTING,
> };
> @@ -154,6 +169,8 @@ struct nvmet_tcp_queue {
> bool data_digest;
> struct ahash_request *snd_hash;
> struct ahash_request *rcv_hash;
> + key_serial_t tls_pskid;
> + struct delayed_work tls_handshake_work;
>
> unsigned long poll_end;
>
> @@ -1285,12 +1302,12 @@ static int nvmet_tcp_try_recv(struct nvmet_tcp_queue *queue,
>
> static void nvmet_tcp_schedule_release_queue(struct nvmet_tcp_queue *queue)
> {
> - spin_lock(&queue->state_lock);
> + spin_lock_irq(&queue->state_lock);
Where is this lock taken in irq context that needs disabling irq?
> if (queue->state != NVMET_TCP_Q_DISCONNECTING) {
> queue->state = NVMET_TCP_Q_DISCONNECTING;
> queue_work(nvmet_wq, &queue->release_work);
> }
> - spin_unlock(&queue->state_lock);
> + spin_unlock_irq(&queue->state_lock);
> }
>
> static inline void nvmet_tcp_arm_queue_deadline(struct nvmet_tcp_queue *queue)
> @@ -1512,8 +1529,12 @@ static void nvmet_tcp_data_ready(struct sock *sk)
>
> read_lock_bh(&sk->sk_callback_lock);
> queue = sk->sk_user_data;
> - if (likely(queue))
> - queue_work_on(queue_cpu(queue), nvmet_tcp_wq, &queue->io_work);
> + if (queue->data_ready)
> + queue->data_ready(sk);
> + if (likely(queue) &&
> + queue->state != NVMET_TCP_Q_TLS_HANDSHAKE)
> + queue_work_on(queue_cpu(queue), nvmet_tcp_wq,
> + &queue->io_work);
> read_unlock_bh(&sk->sk_callback_lock);
> }
>
> @@ -1621,6 +1642,85 @@ static int nvmet_tcp_set_queue_sock(struct nvmet_tcp_queue *queue)
> return ret;
> }
>
> +#ifdef CONFIG_NVME_TARGET_TCP_TLS
> +static void nvmet_tcp_tls_queue_reset(struct nvmet_tcp_queue *queue)
> +{
> + spin_lock_irq(&queue->state_lock);
> + if (queue->state != NVMET_TCP_Q_TLS_HANDSHAKE) {
> + pr_warn("queue %d: TLS handshake already completed\n",
> + queue->idx);
> + spin_unlock_irq(&queue->state_lock);
> + return;
trigger fatal error here?
> + }
> + queue->state = NVMET_TCP_Q_CONNECTING;
> + spin_unlock_irq(&queue->state_lock);
> +
> + pr_debug("queue %d: resetting queue callbacks after TLS handshake\n",
> + queue->idx);
> + /*
> + * Set callbacks after handshake; TLS implementation
> + * might have changed the socket callbacks.
> + */
> + nvmet_tcp_set_queue_sock(queue);
> +}
> +
> +static void nvmet_tcp_tls_handshake_done(void *data, int status,
> + key_serial_t peerid)
lets call peerid psk_id throughout.
> +{
> + struct nvmet_tcp_queue *queue = data;
> +
> + pr_debug("queue %d: TLS handshake done, key %x, status %d\n",
> + queue->idx, peerid, status);
> + if (!status) {
> + spin_lock_irq(&queue->state_lock);
> + queue->tls_pskid = peerid;
> + spin_unlock_irq(&queue->state_lock);
> + }
> + cancel_delayed_work_sync(&queue->tls_handshake_work);
Hmm, the cancel_delayed_work_sync is scary.
What happens if it ran and already scheduled a release (which
already ran and completed)?
> + if (status)
> + nvmet_tcp_schedule_release_queue(queue);
> + else
> + nvmet_tcp_tls_queue_reset(queue);
What I think you need is an atomic state that one or the
other access, and then you are fine with a normal async
cancel of the delayed_work.
> +}
> +
> +static void nvmet_tcp_tls_handshake_timeout_work(struct work_struct *w)
> +{
> + struct nvmet_tcp_queue *queue = container_of(to_delayed_work(w),
> + struct nvmet_tcp_queue, tls_handshake_work);
> +
> + pr_debug("queue %d: TLS handshake timeout\n", queue->idx);
> + nvmet_tcp_schedule_release_queue(queue);
> +}
> +
> +static int nvmet_tcp_tls_handshake(struct nvmet_tcp_queue *queue)
> +{
> + int ret = -EOPNOTSUPP;
> + struct tls_handshake_args args;
> +
> + if (queue->state != NVMET_TCP_Q_TLS_HANDSHAKE) {
> + pr_warn("cannot start TLS in state %d\n", queue->state);
> + return -EINVAL;
> + }
> +
> + pr_debug("queue %d: TLS ServerHello\n", queue->idx);
> + memset(&args, 0, sizeof(args));
> + args.ta_sock = queue->sock;
> + args.ta_done = nvmet_tcp_tls_handshake_done;
> + args.ta_data = queue;
> + args.ta_keyring = key_serial(queue->port->nport->keyring);
> + args.ta_timeout_ms = tls_handshake_timeout * 2 * 1024;
* 2 * 1024 ? I didn't know we have 2048 ms in a second...
> +
> + ret = tls_server_hello_psk(&args, GFP_KERNEL);
> + if (ret) {
> + pr_err("failed to start TLS, err=%d\n", ret);
> + } else {
> + queue_delayed_work(nvmet_wq, &queue->tls_handshake_work,
> + tls_handshake_timeout * HZ);
> + }
> + return ret;
> +}
> +#endif
> +
> static void nvmet_tcp_alloc_queue(struct nvmet_tcp_port *port,
> struct socket *newsock)
> {
> @@ -1638,7 +1738,11 @@ static void nvmet_tcp_alloc_queue(struct nvmet_tcp_port *port,
> queue->port = port;
> queue->nr_cmds = 0;
> spin_lock_init(&queue->state_lock);
> - queue->state = NVMET_TCP_Q_CONNECTING;
> + if (queue->port->nport->disc_addr.tsas.tcp.sectype ==
> + NVMF_TCP_SECTYPE_TLS13)
> + queue->state = NVMET_TCP_Q_TLS_HANDSHAKE;
> + else
> + queue->state = NVMET_TCP_Q_CONNECTING;
> INIT_LIST_HEAD(&queue->free_list);
> init_llist_head(&queue->resp_list);
> INIT_LIST_HEAD(&queue->resp_send_list);
> @@ -1669,6 +1773,25 @@ static void nvmet_tcp_alloc_queue(struct nvmet_tcp_port *port,
> list_add_tail(&queue->queue_list, &nvmet_tcp_queue_list);
> mutex_unlock(&nvmet_tcp_queue_mutex);
>
> +#ifdef CONFIG_NVME_TARGET_TCP_TLS
maybe the ifdef can be avoided with stubs?
> + INIT_DELAYED_WORK(&queue->tls_handshake_work,
> + nvmet_tcp_tls_handshake_timeout_work);
> + if (queue->state == NVMET_TCP_Q_TLS_HANDSHAKE) {
> + struct sock *sk = queue->sock->sk;
> +
> + /* Restore the default callbacks before starting upcall */
> + read_lock_bh(&sk->sk_callback_lock);
> + sk->sk_user_data = NULL;
> + sk->sk_data_ready = port->data_ready;
> + read_unlock_bh(&sk->sk_callback_lock);
> + if (!nvmet_tcp_tls_handshake(queue))
> + return;
> +
> + /* TLS handshake failed, terminate the connection */
> + goto out_destroy_sq;
> + }
> +#endif
> +
> ret = nvmet_tcp_set_queue_sock(queue);
> if (ret)
> goto out_destroy_sq;
More information about the Linux-nvme
mailing list