summaryrefslogtreecommitdiffstats
path: root/net/nfc/hci/llc_shdlc.c
diff options
context:
space:
mode:
authorDaniel Baumann <daniel.baumann@progress-linux.org>2024-05-06 01:02:30 +0000
committerDaniel Baumann <daniel.baumann@progress-linux.org>2024-05-06 01:02:30 +0000
commit76cb841cb886eef6b3bee341a2266c76578724ad (patch)
treef5892e5ba6cc11949952a6ce4ecbe6d516d6ce58 /net/nfc/hci/llc_shdlc.c
parentInitial commit. (diff)
downloadlinux-76cb841cb886eef6b3bee341a2266c76578724ad.tar.xz
linux-76cb841cb886eef6b3bee341a2266c76578724ad.zip
Adding upstream version 4.19.249.upstream/4.19.249
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'net/nfc/hci/llc_shdlc.c')
-rw-r--r--net/nfc/hci/llc_shdlc.c846
1 files changed, 846 insertions, 0 deletions
diff --git a/net/nfc/hci/llc_shdlc.c b/net/nfc/hci/llc_shdlc.c
new file mode 100644
index 000000000..fe988936a
--- /dev/null
+++ b/net/nfc/hci/llc_shdlc.c
@@ -0,0 +1,846 @@
+/*
+ * shdlc Link Layer Control
+ *
+ * Copyright (C) 2012 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define pr_fmt(fmt) "shdlc: %s: " fmt, __func__
+
+#include <linux/types.h>
+#include <linux/sched.h>
+#include <linux/wait.h>
+#include <linux/slab.h>
+#include <linux/skbuff.h>
+
+#include "llc.h"
+
+enum shdlc_state {
+ SHDLC_DISCONNECTED = 0,
+ SHDLC_CONNECTING = 1,
+ SHDLC_NEGOTIATING = 2,
+ SHDLC_HALF_CONNECTED = 3,
+ SHDLC_CONNECTED = 4
+};
+
+struct llc_shdlc {
+ struct nfc_hci_dev *hdev;
+ xmit_to_drv_t xmit_to_drv;
+ rcv_to_hci_t rcv_to_hci;
+
+ struct mutex state_mutex;
+ enum shdlc_state state;
+ int hard_fault;
+
+ wait_queue_head_t *connect_wq;
+ int connect_tries;
+ int connect_result;
+ struct timer_list connect_timer;/* aka T3 in spec 10.6.1 */
+
+ u8 w; /* window size */
+ bool srej_support;
+
+ struct timer_list t1_timer; /* send ack timeout */
+ bool t1_active;
+
+ struct timer_list t2_timer; /* guard/retransmit timeout */
+ bool t2_active;
+
+ int ns; /* next seq num for send */
+ int nr; /* next expected seq num for receive */
+ int dnr; /* oldest sent unacked seq num */
+
+ struct sk_buff_head rcv_q;
+
+ struct sk_buff_head send_q;
+ bool rnr; /* other side is not ready to receive */
+
+ struct sk_buff_head ack_pending_q;
+
+ struct work_struct sm_work;
+
+ int tx_headroom;
+ int tx_tailroom;
+
+ llc_failure_t llc_failure;
+};
+
+#define SHDLC_LLC_HEAD_ROOM 2
+
+#define SHDLC_MAX_WINDOW 4
+#define SHDLC_SREJ_SUPPORT false
+
+#define SHDLC_CONTROL_HEAD_MASK 0xe0
+#define SHDLC_CONTROL_HEAD_I 0x80
+#define SHDLC_CONTROL_HEAD_I2 0xa0
+#define SHDLC_CONTROL_HEAD_S 0xc0
+#define SHDLC_CONTROL_HEAD_U 0xe0
+
+#define SHDLC_CONTROL_NS_MASK 0x38
+#define SHDLC_CONTROL_NR_MASK 0x07
+#define SHDLC_CONTROL_TYPE_MASK 0x18
+
+#define SHDLC_CONTROL_M_MASK 0x1f
+
+enum sframe_type {
+ S_FRAME_RR = 0x00,
+ S_FRAME_REJ = 0x01,
+ S_FRAME_RNR = 0x02,
+ S_FRAME_SREJ = 0x03
+};
+
+enum uframe_modifier {
+ U_FRAME_UA = 0x06,
+ U_FRAME_RSET = 0x19
+};
+
+#define SHDLC_CONNECT_VALUE_MS 5
+#define SHDLC_T1_VALUE_MS(w) ((5 * w) / 4)
+#define SHDLC_T2_VALUE_MS 300
+
+#define SHDLC_DUMP_SKB(info, skb) \
+do { \
+ pr_debug("%s:\n", info); \
+ print_hex_dump(KERN_DEBUG, "shdlc: ", DUMP_PREFIX_OFFSET, \
+ 16, 1, skb->data, skb->len, 0); \
+} while (0)
+
+/* checks x < y <= z modulo 8 */
+static bool llc_shdlc_x_lt_y_lteq_z(int x, int y, int z)
+{
+ if (x < z)
+ return ((x < y) && (y <= z)) ? true : false;
+ else
+ return ((y > x) || (y <= z)) ? true : false;
+}
+
+/* checks x <= y < z modulo 8 */
+static bool llc_shdlc_x_lteq_y_lt_z(int x, int y, int z)
+{
+ if (x <= z)
+ return ((x <= y) && (y < z)) ? true : false;
+ else /* x > z -> z+8 > x */
+ return ((y >= x) || (y < z)) ? true : false;
+}
+
+static struct sk_buff *llc_shdlc_alloc_skb(struct llc_shdlc *shdlc,
+ int payload_len)
+{
+ struct sk_buff *skb;
+
+ skb = alloc_skb(shdlc->tx_headroom + SHDLC_LLC_HEAD_ROOM +
+ shdlc->tx_tailroom + payload_len, GFP_KERNEL);
+ if (skb)
+ skb_reserve(skb, shdlc->tx_headroom + SHDLC_LLC_HEAD_ROOM);
+
+ return skb;
+}
+
+/* immediately sends an S frame. */
+static int llc_shdlc_send_s_frame(struct llc_shdlc *shdlc,
+ enum sframe_type sframe_type, int nr)
+{
+ int r;
+ struct sk_buff *skb;
+
+ pr_debug("sframe_type=%d nr=%d\n", sframe_type, nr);
+
+ skb = llc_shdlc_alloc_skb(shdlc, 0);
+ if (skb == NULL)
+ return -ENOMEM;
+
+ *(u8 *)skb_push(skb, 1) = SHDLC_CONTROL_HEAD_S | (sframe_type << 3) | nr;
+
+ r = shdlc->xmit_to_drv(shdlc->hdev, skb);
+
+ kfree_skb(skb);
+
+ return r;
+}
+
+/* immediately sends an U frame. skb may contain optional payload */
+static int llc_shdlc_send_u_frame(struct llc_shdlc *shdlc,
+ struct sk_buff *skb,
+ enum uframe_modifier uframe_modifier)
+{
+ int r;
+
+ pr_debug("uframe_modifier=%d\n", uframe_modifier);
+
+ *(u8 *)skb_push(skb, 1) = SHDLC_CONTROL_HEAD_U | uframe_modifier;
+
+ r = shdlc->xmit_to_drv(shdlc->hdev, skb);
+
+ kfree_skb(skb);
+
+ return r;
+}
+
+/*
+ * Free ack_pending frames until y_nr - 1, and reset t2 according to
+ * the remaining oldest ack_pending frame sent time
+ */
+static void llc_shdlc_reset_t2(struct llc_shdlc *shdlc, int y_nr)
+{
+ struct sk_buff *skb;
+ int dnr = shdlc->dnr; /* MUST initially be < y_nr */
+
+ pr_debug("release ack pending up to frame %d excluded\n", y_nr);
+
+ while (dnr != y_nr) {
+ pr_debug("release ack pending frame %d\n", dnr);
+
+ skb = skb_dequeue(&shdlc->ack_pending_q);
+ kfree_skb(skb);
+
+ dnr = (dnr + 1) % 8;
+ }
+
+ if (skb_queue_empty(&shdlc->ack_pending_q)) {
+ if (shdlc->t2_active) {
+ del_timer_sync(&shdlc->t2_timer);
+ shdlc->t2_active = false;
+
+ pr_debug
+ ("All sent frames acked. Stopped T2(retransmit)\n");
+ }
+ } else {
+ skb = skb_peek(&shdlc->ack_pending_q);
+
+ mod_timer(&shdlc->t2_timer, *(unsigned long *)skb->cb +
+ msecs_to_jiffies(SHDLC_T2_VALUE_MS));
+ shdlc->t2_active = true;
+
+ pr_debug
+ ("Start T2(retransmit) for remaining unacked sent frames\n");
+ }
+}
+
+/*
+ * Receive validated frames from lower layer. skb contains HCI payload only.
+ * Handle according to algorithm at spec:10.8.2
+ */
+static void llc_shdlc_rcv_i_frame(struct llc_shdlc *shdlc,
+ struct sk_buff *skb, int ns, int nr)
+{
+ int x_ns = ns;
+ int y_nr = nr;
+
+ pr_debug("recvd I-frame %d, remote waiting frame %d\n", ns, nr);
+
+ if (shdlc->state != SHDLC_CONNECTED)
+ goto exit;
+
+ if (x_ns != shdlc->nr) {
+ llc_shdlc_send_s_frame(shdlc, S_FRAME_REJ, shdlc->nr);
+ goto exit;
+ }
+
+ if (shdlc->t1_active == false) {
+ shdlc->t1_active = true;
+ mod_timer(&shdlc->t1_timer, jiffies +
+ msecs_to_jiffies(SHDLC_T1_VALUE_MS(shdlc->w)));
+ pr_debug("(re)Start T1(send ack)\n");
+ }
+
+ if (skb->len) {
+ shdlc->rcv_to_hci(shdlc->hdev, skb);
+ skb = NULL;
+ }
+
+ shdlc->nr = (shdlc->nr + 1) % 8;
+
+ if (llc_shdlc_x_lt_y_lteq_z(shdlc->dnr, y_nr, shdlc->ns)) {
+ llc_shdlc_reset_t2(shdlc, y_nr);
+
+ shdlc->dnr = y_nr;
+ }
+
+exit:
+ kfree_skb(skb);
+}
+
+static void llc_shdlc_rcv_ack(struct llc_shdlc *shdlc, int y_nr)
+{
+ pr_debug("remote acked up to frame %d excluded\n", y_nr);
+
+ if (llc_shdlc_x_lt_y_lteq_z(shdlc->dnr, y_nr, shdlc->ns)) {
+ llc_shdlc_reset_t2(shdlc, y_nr);
+ shdlc->dnr = y_nr;
+ }
+}
+
+static void llc_shdlc_requeue_ack_pending(struct llc_shdlc *shdlc)
+{
+ struct sk_buff *skb;
+
+ pr_debug("ns reset to %d\n", shdlc->dnr);
+
+ while ((skb = skb_dequeue_tail(&shdlc->ack_pending_q))) {
+ skb_pull(skb, 1); /* remove control field */
+ skb_queue_head(&shdlc->send_q, skb);
+ }
+ shdlc->ns = shdlc->dnr;
+}
+
+static void llc_shdlc_rcv_rej(struct llc_shdlc *shdlc, int y_nr)
+{
+ struct sk_buff *skb;
+
+ pr_debug("remote asks retransmission from frame %d\n", y_nr);
+
+ if (llc_shdlc_x_lteq_y_lt_z(shdlc->dnr, y_nr, shdlc->ns)) {
+ if (shdlc->t2_active) {
+ del_timer_sync(&shdlc->t2_timer);
+ shdlc->t2_active = false;
+ pr_debug("Stopped T2(retransmit)\n");
+ }
+
+ if (shdlc->dnr != y_nr) {
+ while ((shdlc->dnr = ((shdlc->dnr + 1) % 8)) != y_nr) {
+ skb = skb_dequeue(&shdlc->ack_pending_q);
+ kfree_skb(skb);
+ }
+ }
+
+ llc_shdlc_requeue_ack_pending(shdlc);
+ }
+}
+
+/* See spec RR:10.8.3 REJ:10.8.4 */
+static void llc_shdlc_rcv_s_frame(struct llc_shdlc *shdlc,
+ enum sframe_type s_frame_type, int nr)
+{
+ struct sk_buff *skb;
+
+ if (shdlc->state != SHDLC_CONNECTED)
+ return;
+
+ switch (s_frame_type) {
+ case S_FRAME_RR:
+ llc_shdlc_rcv_ack(shdlc, nr);
+ if (shdlc->rnr == true) { /* see SHDLC 10.7.7 */
+ shdlc->rnr = false;
+ if (shdlc->send_q.qlen == 0) {
+ skb = llc_shdlc_alloc_skb(shdlc, 0);
+ if (skb)
+ skb_queue_tail(&shdlc->send_q, skb);
+ }
+ }
+ break;
+ case S_FRAME_REJ:
+ llc_shdlc_rcv_rej(shdlc, nr);
+ break;
+ case S_FRAME_RNR:
+ llc_shdlc_rcv_ack(shdlc, nr);
+ shdlc->rnr = true;
+ break;
+ default:
+ break;
+ }
+}
+
+static void llc_shdlc_connect_complete(struct llc_shdlc *shdlc, int r)
+{
+ pr_debug("result=%d\n", r);
+
+ del_timer_sync(&shdlc->connect_timer);
+
+ if (r == 0) {
+ shdlc->ns = 0;
+ shdlc->nr = 0;
+ shdlc->dnr = 0;
+
+ shdlc->state = SHDLC_HALF_CONNECTED;
+ } else {
+ shdlc->state = SHDLC_DISCONNECTED;
+ }
+
+ shdlc->connect_result = r;
+
+ wake_up(shdlc->connect_wq);
+}
+
+static int llc_shdlc_connect_initiate(struct llc_shdlc *shdlc)
+{
+ struct sk_buff *skb;
+
+ pr_debug("\n");
+
+ skb = llc_shdlc_alloc_skb(shdlc, 2);
+ if (skb == NULL)
+ return -ENOMEM;
+
+ skb_put_u8(skb, SHDLC_MAX_WINDOW);
+ skb_put_u8(skb, SHDLC_SREJ_SUPPORT ? 1 : 0);
+
+ return llc_shdlc_send_u_frame(shdlc, skb, U_FRAME_RSET);
+}
+
+static int llc_shdlc_connect_send_ua(struct llc_shdlc *shdlc)
+{
+ struct sk_buff *skb;
+
+ pr_debug("\n");
+
+ skb = llc_shdlc_alloc_skb(shdlc, 0);
+ if (skb == NULL)
+ return -ENOMEM;
+
+ return llc_shdlc_send_u_frame(shdlc, skb, U_FRAME_UA);
+}
+
+static void llc_shdlc_rcv_u_frame(struct llc_shdlc *shdlc,
+ struct sk_buff *skb,
+ enum uframe_modifier u_frame_modifier)
+{
+ u8 w = SHDLC_MAX_WINDOW;
+ bool srej_support = SHDLC_SREJ_SUPPORT;
+ int r;
+
+ pr_debug("u_frame_modifier=%d\n", u_frame_modifier);
+
+ switch (u_frame_modifier) {
+ case U_FRAME_RSET:
+ switch (shdlc->state) {
+ case SHDLC_NEGOTIATING:
+ case SHDLC_CONNECTING:
+ /*
+ * We sent RSET, but chip wants to negociate or we
+ * got RSET before we managed to send out our.
+ */
+ if (skb->len > 0)
+ w = skb->data[0];
+
+ if (skb->len > 1)
+ srej_support = skb->data[1] & 0x01 ? true :
+ false;
+
+ if ((w <= SHDLC_MAX_WINDOW) &&
+ (SHDLC_SREJ_SUPPORT || (srej_support == false))) {
+ shdlc->w = w;
+ shdlc->srej_support = srej_support;
+ r = llc_shdlc_connect_send_ua(shdlc);
+ llc_shdlc_connect_complete(shdlc, r);
+ }
+ break;
+ case SHDLC_HALF_CONNECTED:
+ /*
+ * Chip resent RSET due to its timeout - Ignote it
+ * as we already sent UA.
+ */
+ break;
+ case SHDLC_CONNECTED:
+ /*
+ * Chip wants to reset link. This is unexpected and
+ * unsupported.
+ */
+ shdlc->hard_fault = -ECONNRESET;
+ break;
+ default:
+ break;
+ }
+ break;
+ case U_FRAME_UA:
+ if ((shdlc->state == SHDLC_CONNECTING &&
+ shdlc->connect_tries > 0) ||
+ (shdlc->state == SHDLC_NEGOTIATING)) {
+ llc_shdlc_connect_complete(shdlc, 0);
+ shdlc->state = SHDLC_CONNECTED;
+ }
+ break;
+ default:
+ break;
+ }
+
+ kfree_skb(skb);
+}
+
+static void llc_shdlc_handle_rcv_queue(struct llc_shdlc *shdlc)
+{
+ struct sk_buff *skb;
+ u8 control;
+ int nr;
+ int ns;
+ enum sframe_type s_frame_type;
+ enum uframe_modifier u_frame_modifier;
+
+ if (shdlc->rcv_q.qlen)
+ pr_debug("rcvQlen=%d\n", shdlc->rcv_q.qlen);
+
+ while ((skb = skb_dequeue(&shdlc->rcv_q)) != NULL) {
+ control = skb->data[0];
+ skb_pull(skb, 1);
+ switch (control & SHDLC_CONTROL_HEAD_MASK) {
+ case SHDLC_CONTROL_HEAD_I:
+ case SHDLC_CONTROL_HEAD_I2:
+ if (shdlc->state == SHDLC_HALF_CONNECTED)
+ shdlc->state = SHDLC_CONNECTED;
+
+ ns = (control & SHDLC_CONTROL_NS_MASK) >> 3;
+ nr = control & SHDLC_CONTROL_NR_MASK;
+ llc_shdlc_rcv_i_frame(shdlc, skb, ns, nr);
+ break;
+ case SHDLC_CONTROL_HEAD_S:
+ if (shdlc->state == SHDLC_HALF_CONNECTED)
+ shdlc->state = SHDLC_CONNECTED;
+
+ s_frame_type = (control & SHDLC_CONTROL_TYPE_MASK) >> 3;
+ nr = control & SHDLC_CONTROL_NR_MASK;
+ llc_shdlc_rcv_s_frame(shdlc, s_frame_type, nr);
+ kfree_skb(skb);
+ break;
+ case SHDLC_CONTROL_HEAD_U:
+ u_frame_modifier = control & SHDLC_CONTROL_M_MASK;
+ llc_shdlc_rcv_u_frame(shdlc, skb, u_frame_modifier);
+ break;
+ default:
+ pr_err("UNKNOWN Control=%d\n", control);
+ kfree_skb(skb);
+ break;
+ }
+ }
+}
+
+static int llc_shdlc_w_used(int ns, int dnr)
+{
+ int unack_count;
+
+ if (dnr <= ns)
+ unack_count = ns - dnr;
+ else
+ unack_count = 8 - dnr + ns;
+
+ return unack_count;
+}
+
+/* Send frames according to algorithm at spec:10.8.1 */
+static void llc_shdlc_handle_send_queue(struct llc_shdlc *shdlc)
+{
+ struct sk_buff *skb;
+ int r;
+ unsigned long time_sent;
+
+ if (shdlc->send_q.qlen)
+ pr_debug
+ ("sendQlen=%d ns=%d dnr=%d rnr=%s w_room=%d unackQlen=%d\n",
+ shdlc->send_q.qlen, shdlc->ns, shdlc->dnr,
+ shdlc->rnr == false ? "false" : "true",
+ shdlc->w - llc_shdlc_w_used(shdlc->ns, shdlc->dnr),
+ shdlc->ack_pending_q.qlen);
+
+ while (shdlc->send_q.qlen && shdlc->ack_pending_q.qlen < shdlc->w &&
+ (shdlc->rnr == false)) {
+
+ if (shdlc->t1_active) {
+ del_timer_sync(&shdlc->t1_timer);
+ shdlc->t1_active = false;
+ pr_debug("Stopped T1(send ack)\n");
+ }
+
+ skb = skb_dequeue(&shdlc->send_q);
+
+ *(u8 *)skb_push(skb, 1) = SHDLC_CONTROL_HEAD_I | (shdlc->ns << 3) |
+ shdlc->nr;
+
+ pr_debug("Sending I-Frame %d, waiting to rcv %d\n", shdlc->ns,
+ shdlc->nr);
+ SHDLC_DUMP_SKB("shdlc frame written", skb);
+
+ r = shdlc->xmit_to_drv(shdlc->hdev, skb);
+ if (r < 0) {
+ shdlc->hard_fault = r;
+ break;
+ }
+
+ shdlc->ns = (shdlc->ns + 1) % 8;
+
+ time_sent = jiffies;
+ *(unsigned long *)skb->cb = time_sent;
+
+ skb_queue_tail(&shdlc->ack_pending_q, skb);
+
+ if (shdlc->t2_active == false) {
+ shdlc->t2_active = true;
+ mod_timer(&shdlc->t2_timer, time_sent +
+ msecs_to_jiffies(SHDLC_T2_VALUE_MS));
+ pr_debug("Started T2 (retransmit)\n");
+ }
+ }
+}
+
+static void llc_shdlc_connect_timeout(struct timer_list *t)
+{
+ struct llc_shdlc *shdlc = from_timer(shdlc, t, connect_timer);
+
+ pr_debug("\n");
+
+ schedule_work(&shdlc->sm_work);
+}
+
+static void llc_shdlc_t1_timeout(struct timer_list *t)
+{
+ struct llc_shdlc *shdlc = from_timer(shdlc, t, t1_timer);
+
+ pr_debug("SoftIRQ: need to send ack\n");
+
+ schedule_work(&shdlc->sm_work);
+}
+
+static void llc_shdlc_t2_timeout(struct timer_list *t)
+{
+ struct llc_shdlc *shdlc = from_timer(shdlc, t, t2_timer);
+
+ pr_debug("SoftIRQ: need to retransmit\n");
+
+ schedule_work(&shdlc->sm_work);
+}
+
+static void llc_shdlc_sm_work(struct work_struct *work)
+{
+ struct llc_shdlc *shdlc = container_of(work, struct llc_shdlc, sm_work);
+ int r;
+
+ pr_debug("\n");
+
+ mutex_lock(&shdlc->state_mutex);
+
+ switch (shdlc->state) {
+ case SHDLC_DISCONNECTED:
+ skb_queue_purge(&shdlc->rcv_q);
+ skb_queue_purge(&shdlc->send_q);
+ skb_queue_purge(&shdlc->ack_pending_q);
+ break;
+ case SHDLC_CONNECTING:
+ if (shdlc->hard_fault) {
+ llc_shdlc_connect_complete(shdlc, shdlc->hard_fault);
+ break;
+ }
+
+ if (shdlc->connect_tries++ < 5)
+ r = llc_shdlc_connect_initiate(shdlc);
+ else
+ r = -ETIME;
+ if (r < 0) {
+ llc_shdlc_connect_complete(shdlc, r);
+ } else {
+ mod_timer(&shdlc->connect_timer, jiffies +
+ msecs_to_jiffies(SHDLC_CONNECT_VALUE_MS));
+
+ shdlc->state = SHDLC_NEGOTIATING;
+ }
+ break;
+ case SHDLC_NEGOTIATING:
+ if (timer_pending(&shdlc->connect_timer) == 0) {
+ shdlc->state = SHDLC_CONNECTING;
+ schedule_work(&shdlc->sm_work);
+ }
+
+ llc_shdlc_handle_rcv_queue(shdlc);
+
+ if (shdlc->hard_fault) {
+ llc_shdlc_connect_complete(shdlc, shdlc->hard_fault);
+ break;
+ }
+ break;
+ case SHDLC_HALF_CONNECTED:
+ case SHDLC_CONNECTED:
+ llc_shdlc_handle_rcv_queue(shdlc);
+ llc_shdlc_handle_send_queue(shdlc);
+
+ if (shdlc->t1_active && timer_pending(&shdlc->t1_timer) == 0) {
+ pr_debug
+ ("Handle T1(send ack) elapsed (T1 now inactive)\n");
+
+ shdlc->t1_active = false;
+ r = llc_shdlc_send_s_frame(shdlc, S_FRAME_RR,
+ shdlc->nr);
+ if (r < 0)
+ shdlc->hard_fault = r;
+ }
+
+ if (shdlc->t2_active && timer_pending(&shdlc->t2_timer) == 0) {
+ pr_debug
+ ("Handle T2(retransmit) elapsed (T2 inactive)\n");
+
+ shdlc->t2_active = false;
+
+ llc_shdlc_requeue_ack_pending(shdlc);
+ llc_shdlc_handle_send_queue(shdlc);
+ }
+
+ if (shdlc->hard_fault)
+ shdlc->llc_failure(shdlc->hdev, shdlc->hard_fault);
+ break;
+ default:
+ break;
+ }
+ mutex_unlock(&shdlc->state_mutex);
+}
+
+/*
+ * Called from syscall context to establish shdlc link. Sleeps until
+ * link is ready or failure.
+ */
+static int llc_shdlc_connect(struct llc_shdlc *shdlc)
+{
+ DECLARE_WAIT_QUEUE_HEAD_ONSTACK(connect_wq);
+
+ pr_debug("\n");
+
+ mutex_lock(&shdlc->state_mutex);
+
+ shdlc->state = SHDLC_CONNECTING;
+ shdlc->connect_wq = &connect_wq;
+ shdlc->connect_tries = 0;
+ shdlc->connect_result = 1;
+
+ mutex_unlock(&shdlc->state_mutex);
+
+ schedule_work(&shdlc->sm_work);
+
+ wait_event(connect_wq, shdlc->connect_result != 1);
+
+ return shdlc->connect_result;
+}
+
+static void llc_shdlc_disconnect(struct llc_shdlc *shdlc)
+{
+ pr_debug("\n");
+
+ mutex_lock(&shdlc->state_mutex);
+
+ shdlc->state = SHDLC_DISCONNECTED;
+
+ mutex_unlock(&shdlc->state_mutex);
+
+ schedule_work(&shdlc->sm_work);
+}
+
+/*
+ * Receive an incoming shdlc frame. Frame has already been crc-validated.
+ * skb contains only LLC header and payload.
+ * If skb == NULL, it is a notification that the link below is dead.
+ */
+static void llc_shdlc_recv_frame(struct llc_shdlc *shdlc, struct sk_buff *skb)
+{
+ if (skb == NULL) {
+ pr_err("NULL Frame -> link is dead\n");
+ shdlc->hard_fault = -EREMOTEIO;
+ } else {
+ SHDLC_DUMP_SKB("incoming frame", skb);
+ skb_queue_tail(&shdlc->rcv_q, skb);
+ }
+
+ schedule_work(&shdlc->sm_work);
+}
+
+static void *llc_shdlc_init(struct nfc_hci_dev *hdev, xmit_to_drv_t xmit_to_drv,
+ rcv_to_hci_t rcv_to_hci, int tx_headroom,
+ int tx_tailroom, int *rx_headroom, int *rx_tailroom,
+ llc_failure_t llc_failure)
+{
+ struct llc_shdlc *shdlc;
+
+ *rx_headroom = SHDLC_LLC_HEAD_ROOM;
+ *rx_tailroom = 0;
+
+ shdlc = kzalloc(sizeof(struct llc_shdlc), GFP_KERNEL);
+ if (shdlc == NULL)
+ return NULL;
+
+ mutex_init(&shdlc->state_mutex);
+ shdlc->state = SHDLC_DISCONNECTED;
+
+ timer_setup(&shdlc->connect_timer, llc_shdlc_connect_timeout, 0);
+ timer_setup(&shdlc->t1_timer, llc_shdlc_t1_timeout, 0);
+ timer_setup(&shdlc->t2_timer, llc_shdlc_t2_timeout, 0);
+
+ shdlc->w = SHDLC_MAX_WINDOW;
+ shdlc->srej_support = SHDLC_SREJ_SUPPORT;
+
+ skb_queue_head_init(&shdlc->rcv_q);
+ skb_queue_head_init(&shdlc->send_q);
+ skb_queue_head_init(&shdlc->ack_pending_q);
+
+ INIT_WORK(&shdlc->sm_work, llc_shdlc_sm_work);
+
+ shdlc->hdev = hdev;
+ shdlc->xmit_to_drv = xmit_to_drv;
+ shdlc->rcv_to_hci = rcv_to_hci;
+ shdlc->tx_headroom = tx_headroom;
+ shdlc->tx_tailroom = tx_tailroom;
+ shdlc->llc_failure = llc_failure;
+
+ return shdlc;
+}
+
+static void llc_shdlc_deinit(struct nfc_llc *llc)
+{
+ struct llc_shdlc *shdlc = nfc_llc_get_data(llc);
+
+ skb_queue_purge(&shdlc->rcv_q);
+ skb_queue_purge(&shdlc->send_q);
+ skb_queue_purge(&shdlc->ack_pending_q);
+
+ kfree(shdlc);
+}
+
+static int llc_shdlc_start(struct nfc_llc *llc)
+{
+ struct llc_shdlc *shdlc = nfc_llc_get_data(llc);
+
+ return llc_shdlc_connect(shdlc);
+}
+
+static int llc_shdlc_stop(struct nfc_llc *llc)
+{
+ struct llc_shdlc *shdlc = nfc_llc_get_data(llc);
+
+ llc_shdlc_disconnect(shdlc);
+
+ return 0;
+}
+
+static void llc_shdlc_rcv_from_drv(struct nfc_llc *llc, struct sk_buff *skb)
+{
+ struct llc_shdlc *shdlc = nfc_llc_get_data(llc);
+
+ llc_shdlc_recv_frame(shdlc, skb);
+}
+
+static int llc_shdlc_xmit_from_hci(struct nfc_llc *llc, struct sk_buff *skb)
+{
+ struct llc_shdlc *shdlc = nfc_llc_get_data(llc);
+
+ skb_queue_tail(&shdlc->send_q, skb);
+
+ schedule_work(&shdlc->sm_work);
+
+ return 0;
+}
+
+static struct nfc_llc_ops llc_shdlc_ops = {
+ .init = llc_shdlc_init,
+ .deinit = llc_shdlc_deinit,
+ .start = llc_shdlc_start,
+ .stop = llc_shdlc_stop,
+ .rcv_from_drv = llc_shdlc_rcv_from_drv,
+ .xmit_from_hci = llc_shdlc_xmit_from_hci,
+};
+
+int nfc_llc_shdlc_register(void)
+{
+ return nfc_llc_register(LLC_SHDLC_NAME, &llc_shdlc_ops);
+}