gpu: nvgpu: support binding multiple channels to a debug session

We currently bind only one channel to a debug session But some use cases might need multiple channels bound to same debug session Add this support by adding a list of channels to debug session. List structure is implemented as struct dbg_session_channel_data List node dbg_s_list_node is currently defined in struct dbg_session_gk20a. But this is inefficient when we need to add debug session to multiple channels Hence add new reference structure dbg_session_data to store dbg_session pointer and list entry For each NVGPU_DBG_GPU_IOCTL_BIND_CHANNEL call, create two reference structure dbg_session_channel_data for channel and dbg_session_data for debug session and bind them together Define API nvgpu_dbg_gpu_get_session_channel() which will get first channel in the list of debug session Use this API wherever we refer to channel bound to debug session Remove dbg_sessions define in struct gk20a since it is not being used anywhere Add new API NVGPU_DBG_GPU_IOCTL_UNBIND_CHANNEL to support unbinding of channel from debug sesssion Bug 200156699 Change-Id: I3bfa6f9cd5b90e7254a75c7e64ac893739776b7f Signed-off-by: Deepak Nibade <dnibade@nvidia.com> Reviewed-on: http://git-master/r/1120331 GVS: Gerrit_Virtual_Submit Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
2025-12-23 09:57:08 +03:00 · 2016-03-29 15:01:25 +05:30
parent c651adbeaa
commit dfac8ce704
7 changed files with 257 additions and 67 deletions
--- a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
@@ -857,6 +857,8 @@ static void gk20a_free_channel(struct channel_gk20a *ch)
 	struct vm_gk20a *ch_vm = ch->vm;
 	unsigned long timeout = gk20a_get_gr_idle_timeout(g);
 	struct dbg_session_gk20a *dbg_s;
+	struct dbg_session_data *session_data, *tmp_s;
+	struct dbg_session_channel_data *ch_data, *tmp;
 	bool was_reset;

 	gk20a_dbg_fn("");
@@ -992,14 +994,21 @@ unbind:
 	WARN_ON(ch->sync);

 	/* unlink all debug sessions */
-	mutex_lock(&ch->dbg_s_lock);
+	mutex_lock(&g->dbg_sessions_lock);

-	list_for_each_entry(dbg_s, &ch->dbg_s_list, dbg_s_list_node) {
-		dbg_s->ch = NULL;
-		list_del_init(&dbg_s->dbg_s_list_node);
+	list_for_each_entry_safe(session_data, tmp_s,
+				&ch->dbg_s_list, dbg_s_entry) {
+		dbg_s = session_data->dbg_s;
+		mutex_lock(&dbg_s->ch_list_lock);
+		list_for_each_entry_safe(ch_data, tmp,
+					&dbg_s->ch_list, ch_entry) {
+			if (ch_data->chid == ch->hw_chid)
+				dbg_unbind_single_channel_gk20a(dbg_s, ch_data);
+		}
+		mutex_unlock(&dbg_s->ch_list_lock);
 	}

-	mutex_unlock(&ch->dbg_s_lock);
+	mutex_unlock(&g->dbg_sessions_lock);

 release:
 	/* make sure we catch accesses of unopened channels in case
--- a/drivers/gpu/nvgpu/gk20a/dbg_gpu_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/dbg_gpu_gk20a.c
@@ -35,6 +35,33 @@ struct dbg_gpu_session_ops dbg_gpu_session_ops_gk20a = {
 	.exec_reg_ops = exec_regops_gk20a,
 };

+/*
+ * API to get first channel from the list of all channels
+ * bound to the debug session
+ */
+struct channel_gk20a *
+nvgpu_dbg_gpu_get_session_channel(struct dbg_session_gk20a *dbg_s)
+{
+	struct dbg_session_channel_data *ch_data;
+	struct channel_gk20a *ch;
+	struct gk20a *g = dbg_s->g;
+
+	mutex_lock(&dbg_s->ch_list_lock);
+	if (list_empty(&dbg_s->ch_list)) {
+		mutex_unlock(&dbg_s->ch_list_lock);
+		return NULL;
+	}
+
+	ch_data = list_first_entry(&dbg_s->ch_list,
+				   struct dbg_session_channel_data,
+				   ch_entry);
+	ch = g->fifo.channel + ch_data->chid;
+
+	mutex_unlock(&dbg_s->ch_list_lock);
+
+	return ch;
+}
+
 /* silly allocator - just increment session id */
 static atomic_t session_id = ATOMIC_INIT(0);
 static int generate_session_id(void)
@@ -95,8 +122,9 @@ static int gk20a_dbg_gpu_do_dev_open(struct inode *inode,
 	if (gk20a_gpu_is_virtual(dev))
 		dbg_session->is_pg_disabled = true;

-	INIT_LIST_HEAD(&dbg_session->dbg_s_list_node);
 	init_waitqueue_head(&dbg_session->dbg_events.wait_queue);
+	INIT_LIST_HEAD(&dbg_session->ch_list);
+	mutex_init(&dbg_session->ch_list_lock);
 	dbg_session->dbg_events.events_enabled = false;
 	dbg_session->dbg_events.num_pending_events = 0;

@@ -108,18 +136,22 @@ static int gk20a_dbg_gpu_do_dev_open(struct inode *inode,
 * since it might not have an associated channel. */
 static void gk20a_dbg_session_mutex_lock(struct dbg_session_gk20a *dbg_s)
 {
-	if (dbg_s->is_profiler)
+	struct channel_gk20a *ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+
+	if (dbg_s->is_profiler || !ch)
 		mutex_lock(&dbg_s->g->dbg_sessions_lock);
 	else
-		mutex_lock(&dbg_s->ch->dbg_s_lock);
+		mutex_lock(&ch->dbg_s_lock);
 }

 static void gk20a_dbg_session_mutex_unlock(struct dbg_session_gk20a *dbg_s)
 {
-	if (dbg_s->is_profiler)
+	struct channel_gk20a *ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+
+	if (dbg_s->is_profiler || !ch)
 		mutex_unlock(&dbg_s->g->dbg_sessions_lock);
 	else
-		mutex_unlock(&dbg_s->ch->dbg_s_lock);
+		mutex_unlock(&ch->dbg_s_lock);
 }

 static void gk20a_dbg_gpu_events_enable(struct dbg_session_gk20a *dbg_s)
@@ -163,10 +195,12 @@ static int gk20a_dbg_gpu_events_ctrl(struct dbg_session_gk20a *dbg_s,
 			  struct nvgpu_dbg_gpu_events_ctrl_args *args)
 {
 	int ret = 0;
+	struct channel_gk20a *ch;

 	gk20a_dbg(gpu_dbg_fn | gpu_dbg_gpu_dbg, "dbg events ctrl cmd %d", args->cmd);

-	if (!dbg_s->ch) {
+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+	if (!ch) {
 		gk20a_err(dev_from_gk20a(dbg_s->g),
 			   "no channel bound to dbg session\n");
 		return -EINVAL;
@@ -235,6 +269,7 @@ int gk20a_prof_gpu_dev_open(struct inode *inode, struct file *filp)

 void gk20a_dbg_gpu_post_events(struct channel_gk20a *ch)
 {
+	struct dbg_session_data *session_data;
 	struct dbg_session_gk20a *dbg_s;

 	gk20a_dbg(gpu_dbg_fn | gpu_dbg_gpu_dbg, "");
@@ -242,7 +277,8 @@ void gk20a_dbg_gpu_post_events(struct channel_gk20a *ch)
 	/* guard against the session list being modified */
 	mutex_lock(&ch->dbg_s_lock);

-	list_for_each_entry(dbg_s, &ch->dbg_s_list, dbg_s_list_node) {
+	list_for_each_entry(session_data, &ch->dbg_s_list, dbg_s_entry) {
+		dbg_s = session_data->dbg_s;
 		if (dbg_s->dbg_events.events_enabled) {
 			gk20a_dbg(gpu_dbg_gpu_dbg, "posting event on session id %d",
 					dbg_s->id);
@@ -260,6 +296,7 @@ void gk20a_dbg_gpu_post_events(struct channel_gk20a *ch)

 bool gk20a_dbg_gpu_broadcast_stop_trigger(struct channel_gk20a *ch)
 {
+	struct dbg_session_data *session_data;
 	struct dbg_session_gk20a *dbg_s;
 	bool broadcast = false;

@@ -268,7 +305,8 @@ bool gk20a_dbg_gpu_broadcast_stop_trigger(struct channel_gk20a *ch)
 	/* guard against the session list being modified */
 	mutex_lock(&ch->dbg_s_lock);

-	list_for_each_entry(dbg_s, &ch->dbg_s_list, dbg_s_list_node) {
+	list_for_each_entry(session_data, &ch->dbg_s_list, dbg_s_entry) {
+		dbg_s = session_data->dbg_s;
 		if (dbg_s->broadcast_stop_trigger) {
 			gk20a_dbg(gpu_dbg_gpu_dbg | gpu_dbg_fn | gpu_dbg_intr,
 					"stop trigger broadcast enabled");
@@ -284,6 +322,7 @@ bool gk20a_dbg_gpu_broadcast_stop_trigger(struct channel_gk20a *ch)

 int gk20a_dbg_gpu_clear_broadcast_stop_trigger(struct channel_gk20a *ch)
 {
+	struct dbg_session_data *session_data;
 	struct dbg_session_gk20a *dbg_s;

 	gk20a_dbg(gpu_dbg_fn | gpu_dbg_gpu_dbg | gpu_dbg_intr, "");
@@ -291,7 +330,8 @@ int gk20a_dbg_gpu_clear_broadcast_stop_trigger(struct channel_gk20a *ch)
 	/* guard against the session list being modified */
 	mutex_lock(&ch->dbg_s_lock);

-	list_for_each_entry(dbg_s, &ch->dbg_s_list, dbg_s_list_node) {
+	list_for_each_entry(session_data, &ch->dbg_s_list, dbg_s_entry) {
+		dbg_s = session_data->dbg_s;
 		if (dbg_s->broadcast_stop_trigger) {
 			gk20a_dbg(gpu_dbg_gpu_dbg | gpu_dbg_fn | gpu_dbg_intr,
 					"stop trigger broadcast disabled");
@@ -347,36 +387,87 @@ static int nvgpu_dbg_timeout_enable(struct dbg_session_gk20a *dbg_s,
 	return err;
 }

-static int dbg_unbind_channel_gk20a(struct dbg_session_gk20a *dbg_s)
+int dbg_unbind_single_channel_gk20a(struct dbg_session_gk20a *dbg_s,
+			struct dbg_session_channel_data *ch_data)
 {
-	struct channel_gk20a *ch_gk20a = dbg_s->ch;
 	struct gk20a *g = dbg_s->g;
+	int chid;
+	struct channel_gk20a *ch;
+	struct dbg_session_data *session_data;

 	gk20a_dbg(gpu_dbg_fn | gpu_dbg_gpu_dbg, "");

-	/* wasn't bound to start with ? */
-	if (!ch_gk20a) {
-		gk20a_dbg(gpu_dbg_gpu_dbg | gpu_dbg_fn, "not bound already?");
-		return -ENODEV;
-	}
+	chid = ch_data->chid;
+	ch = g->fifo.channel + chid;
+
+	list_del_init(&ch_data->ch_entry);
+
+	session_data = ch_data->session_data;
+	list_del_init(&session_data->dbg_s_entry);
+	kfree(session_data);
+
+	fput(ch_data->ch_f);
+	kfree(ch_data);
+
+	return 0;
+}
+
+static int dbg_unbind_all_channels_gk20a(struct dbg_session_gk20a *dbg_s)
+{
+	struct dbg_session_channel_data *ch_data, *tmp;
+	struct gk20a *g = dbg_s->g;

 	mutex_lock(&g->dbg_sessions_lock);
-	mutex_lock(&ch_gk20a->dbg_s_lock);
-
-	--g->dbg_sessions;
-
-	dbg_s->ch = NULL;
-	fput(dbg_s->ch_f);
-	dbg_s->ch_f = NULL;
-
-	list_del_init(&dbg_s->dbg_s_list_node);
-
-	mutex_unlock(&ch_gk20a->dbg_s_lock);
+	mutex_lock(&dbg_s->ch_list_lock);
+	list_for_each_entry_safe(ch_data, tmp, &dbg_s->ch_list, ch_entry)
+		dbg_unbind_single_channel_gk20a(dbg_s, ch_data);
+	mutex_unlock(&dbg_s->ch_list_lock);
 	mutex_unlock(&g->dbg_sessions_lock);

 	return 0;
 }

+static int dbg_unbind_channel_gk20a(struct dbg_session_gk20a *dbg_s,
+			  struct nvgpu_dbg_gpu_unbind_channel_args *args)
+{
+	struct dbg_session_channel_data *ch_data;
+	struct gk20a *g = dbg_s->g;
+	bool channel_found = false;
+	struct channel_gk20a *ch;
+	int err;
+
+	gk20a_dbg(gpu_dbg_fn|gpu_dbg_gpu_dbg, "%s fd=%d",
+		   dev_name(dbg_s->dev), args->channel_fd);
+
+	ch = gk20a_get_channel_from_file(args->channel_fd);
+	if (!ch) {
+		gk20a_dbg_fn("no channel found for fd");
+		return -EINVAL;
+	}
+
+	mutex_lock(&dbg_s->ch_list_lock);
+	list_for_each_entry(ch_data, &dbg_s->ch_list, ch_entry) {
+		if (ch->hw_chid == ch_data->chid) {
+			channel_found = true;
+			break;
+		}
+	}
+	mutex_unlock(&dbg_s->ch_list_lock);
+
+	if (!channel_found) {
+		gk20a_dbg_fn("channel not bounded, fd=%d\n", args->channel_fd);
+		return -EINVAL;
+	}
+
+	mutex_lock(&g->dbg_sessions_lock);
+	mutex_lock(&dbg_s->ch_list_lock);
+	err = dbg_unbind_single_channel_gk20a(dbg_s, ch_data);
+	mutex_unlock(&dbg_s->ch_list_lock);
+	mutex_unlock(&g->dbg_sessions_lock);
+
+	return err;
+}
+
 int gk20a_dbg_gpu_dev_release(struct inode *inode, struct file *filp)
 {
 	struct dbg_session_gk20a *dbg_s = filp->private_data;
@@ -384,9 +475,8 @@ int gk20a_dbg_gpu_dev_release(struct inode *inode, struct file *filp)

 	gk20a_dbg(gpu_dbg_gpu_dbg | gpu_dbg_fn, "%s", dev_name(dbg_s->dev));

-	/* unbind if it was bound */
-	if (dbg_s->ch)
-		dbg_unbind_channel_gk20a(dbg_s);
+	/* unbind channels */
+	dbg_unbind_all_channels_gk20a(dbg_s);

 	/* Powergate/Timeout enable is called here as possibility of dbg_session
 	 * which called powergate/timeout disable ioctl, to be killed without
@@ -405,14 +495,28 @@ static int dbg_bind_channel_gk20a(struct dbg_session_gk20a *dbg_s,
 			  struct nvgpu_dbg_gpu_bind_channel_args *args)
 {
 	struct file *f;
-	struct gk20a *g;
+	struct gk20a *g = dbg_s->g;
 	struct channel_gk20a *ch;
+	struct dbg_session_channel_data *ch_data, *tmp;
+	struct dbg_session_data *session_data;

 	gk20a_dbg(gpu_dbg_fn|gpu_dbg_gpu_dbg, "%s fd=%d",
 		   dev_name(dbg_s->dev), args->channel_fd);

-	if (args->channel_fd == ~0)
-		return dbg_unbind_channel_gk20a(dbg_s);
+	if (args->channel_fd == ~0) {
+		ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+		if (!ch)
+			return -EINVAL;
+
+		mutex_lock(&dbg_s->ch_list_lock);
+		list_for_each_entry_safe(ch_data, tmp,
+					 &dbg_s->ch_list, ch_entry) {
+			if (ch_data->chid == ch->hw_chid)
+				dbg_unbind_single_channel_gk20a(dbg_s, ch_data);
+		}
+		mutex_unlock(&dbg_s->ch_list_lock);
+		return 0;
+	}

 	/* even though get_file_channel is doing this it releases it as well */
 	/* by holding it here we'll keep it from disappearing while the
@@ -428,20 +532,40 @@ static int dbg_bind_channel_gk20a(struct dbg_session_gk20a *dbg_s,
 		return -EINVAL;
 	}

-	g = dbg_s->g;
 	gk20a_dbg_fn("%s hwchid=%d", dev_name(dbg_s->dev), ch->hw_chid);

 	mutex_lock(&g->dbg_sessions_lock);
 	mutex_lock(&ch->dbg_s_lock);

-	dbg_s->ch_f = f;
-	dbg_s->ch = ch;
-	list_add(&dbg_s->dbg_s_list_node, &dbg_s->ch->dbg_s_list);
+	ch_data = kzalloc(sizeof(*ch_data), GFP_KERNEL);
+	if (!ch_data) {
+		fput(f);
+		return -ENOMEM;
+	}
+	ch_data->ch_f = f;
+	ch_data->channel_fd = args->channel_fd;
+	ch_data->chid = ch->hw_chid;
+	INIT_LIST_HEAD(&ch_data->ch_entry);

-	g->dbg_sessions++;
+	session_data = kzalloc(sizeof(*session_data), GFP_KERNEL);
+	if (!session_data) {
+		kfree(ch_data);
+		fput(f);
+		return -ENOMEM;
+	}
+	session_data->dbg_s = dbg_s;
+	INIT_LIST_HEAD(&session_data->dbg_s_entry);
+	ch_data->session_data = session_data;
+
+	list_add(&session_data->dbg_s_entry, &ch->dbg_s_list);
+
+	mutex_lock(&dbg_s->ch_list_lock);
+	list_add_tail(&ch_data->ch_entry, &dbg_s->ch_list);
+	mutex_unlock(&dbg_s->ch_list_lock);

 	mutex_unlock(&ch->dbg_s_lock);
 	mutex_unlock(&g->dbg_sessions_lock);
+
 	return 0;
 }

@@ -470,8 +594,12 @@ static int gk20a_perfbuf_unmap(struct dbg_session_gk20a *dbg_s,
 static int gk20a_dbg_pc_sampling(struct dbg_session_gk20a *dbg_s,
 			  struct nvgpu_dbg_gpu_pc_sampling_args *args)
 {
-	struct channel_gk20a *ch = dbg_s->ch;
-	struct gk20a *g = ch->g;
+	struct channel_gk20a *ch;
+	struct gk20a *g = dbg_s->g;
+
+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+	if (!ch)
+		return -EINVAL;

 	gk20a_dbg_fn("");

@@ -571,9 +699,13 @@ static int nvgpu_dbg_gpu_ioctl_clear_single_sm_error_state(
 	struct gk20a *g = get_gk20a(dbg_s->dev);
 	struct gr_gk20a *gr = &g->gr;
 	u32 sm_id;
-	struct channel_gk20a *ch = dbg_s->ch;
+	struct channel_gk20a *ch;
 	int err = 0;

+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+	if (!ch)
+		return -EINVAL;
+
 	sm_id = args->sm_id;

 	if (sm_id >= gr->no_of_sm)
@@ -598,10 +730,14 @@ static int nvgpu_dbg_gpu_ioctl_write_single_sm_error_state(
 	struct gk20a *g = get_gk20a(dbg_s->dev);
 	struct gr_gk20a *gr = &g->gr;
 	u32 sm_id;
-	struct channel_gk20a *ch = dbg_s->ch;
+	struct channel_gk20a *ch;
 	struct nvgpu_dbg_gpu_sm_error_state_record *sm_error_state;
 	int err = 0;

+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+	if (!ch)
+		return -EINVAL;
+
 	sm_id = args->sm_id;
 	if (sm_id >= gr->no_of_sm)
 		return -EINVAL;
@@ -756,6 +892,11 @@ long gk20a_dbg_gpu_dev_ioctl(struct file *filp, unsigned int cmd,
 		  (struct nvgpu_dbg_gpu_write_single_sm_error_state_args *)buf);
 		break;

+	case NVGPU_DBG_GPU_IOCTL_UNBIND_CHANNEL:
+		err = dbg_unbind_channel_gk20a(dbg_s,
+			     (struct nvgpu_dbg_gpu_unbind_channel_args *)buf);
+		break;
+
 	default:
 		gk20a_err(dev_from_gk20a(g),
 			   "unrecognized dbg gpu ioctl cmd: 0x%x",
@@ -805,6 +946,7 @@ static int nvgpu_ioctl_channel_reg_ops(struct dbg_session_gk20a *dbg_s,
 	struct device *dev = dbg_s->dev;
 	struct gk20a *g = get_gk20a(dbg_s->dev);
 	struct nvgpu_dbg_gpu_reg_op *ops;
+	struct channel_gk20a *ch;
 	u64 ops_size = sizeof(ops[0]) * args->num_ops;

 	gk20a_dbg_fn("%d ops, total size %llu", args->num_ops, ops_size);
@@ -814,7 +956,8 @@ static int nvgpu_ioctl_channel_reg_ops(struct dbg_session_gk20a *dbg_s,
 		return -EINVAL;
 	}

-	if (!dbg_s->is_profiler && !dbg_s->ch) {
+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+	if (!dbg_s->is_profiler && !ch) {
 		gk20a_err(dev, "bind a channel before regops for a debugging session");
 		return -EINVAL;
 	}
@@ -1016,8 +1159,7 @@ static int nvgpu_dbg_gpu_ioctl_smpc_ctxsw_mode(struct dbg_session_gk20a *dbg_s,
 	/* Take the global lock, since we'll be doing global regops */
 	mutex_lock(&g->dbg_sessions_lock);

-	ch_gk20a = dbg_s->ch;
-
+	ch_gk20a = nvgpu_dbg_gpu_get_session_channel(dbg_s);
 	if (!ch_gk20a) {
 		gk20a_err(dev_from_gk20a(g),
 			  "no bound channel for smpc ctxsw mode update\n");
@@ -1052,8 +1194,7 @@ static int nvgpu_dbg_gpu_ioctl_hwpm_ctxsw_mode(struct dbg_session_gk20a *dbg_s,
 	/* Take the global lock, since we'll be doing global regops */
 	mutex_lock(&g->dbg_sessions_lock);

-	ch_gk20a = dbg_s->ch;
-
+	ch_gk20a = nvgpu_dbg_gpu_get_session_channel(dbg_s);
 	if (!ch_gk20a) {
 		gk20a_err(dev_from_gk20a(g),
 			  "no bound channel for pm ctxsw mode update\n");
@@ -1080,12 +1221,16 @@ static int nvgpu_dbg_gpu_ioctl_suspend_resume_sm(
 		struct nvgpu_dbg_gpu_suspend_resume_all_sms_args *args)
 {
 	struct gk20a *g = get_gk20a(dbg_s->dev);
-	struct channel_gk20a *ch = dbg_s->ch;
+	struct channel_gk20a *ch;
 	bool ch_is_curr_ctx;
 	int err = 0, action = args->mode;

 	gk20a_dbg(gpu_dbg_fn | gpu_dbg_gpu_dbg, "action: %d", args->mode);

+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
+	if (!ch)
+		return -EINVAL;
+
 	mutex_lock(&g->dbg_sessions_lock);

 	/* Suspend GPU context switching */
--- a/drivers/gpu/nvgpu/gk20a/dbg_gpu_gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/dbg_gpu_gk20a.h
@@ -31,6 +31,9 @@ int gk20a_prof_gpu_dev_open(struct inode *inode, struct file *filp);
 /* used by the interrupt handler to post events */
 void gk20a_dbg_gpu_post_events(struct channel_gk20a *fault_ch);

+struct channel_gk20a *
+nvgpu_dbg_gpu_get_session_channel(struct dbg_session_gk20a *dbg_s);
+
 struct dbg_gpu_session_ops {
 	int (*exec_reg_ops)(struct dbg_session_gk20a *dbg_s,
 			    struct nvgpu_dbg_gpu_reg_op *ops,
@@ -68,22 +71,37 @@ struct dbg_session_gk20a {
 	struct device             *dev;
 	struct gk20a              *g;

-	/* bound channel, if any */
-	struct file          *ch_f;
-	struct channel_gk20a *ch;
+	/* list of bound channels, if any */
+	struct list_head ch_list;
+	struct mutex ch_list_lock;

 	/* session operations */
 	struct dbg_gpu_session_ops *ops;

 	/* event support */
 	struct dbg_gpu_session_events dbg_events;
-	struct list_head dbg_s_list_node;

 	bool broadcast_stop_trigger;
 };

+struct dbg_session_data {
+	struct dbg_session_gk20a *dbg_s;
+	struct list_head dbg_s_entry;
+};
+
+struct dbg_session_channel_data {
+	struct file          *ch_f;
+	int channel_fd;
+	int chid;
+	struct list_head ch_entry;
+	struct dbg_session_data *session_data;
+};
+
 extern struct dbg_gpu_session_ops dbg_gpu_session_ops_gk20a;

+int dbg_unbind_single_channel_gk20a(struct dbg_session_gk20a *dbg_s,
+			struct dbg_session_channel_data *ch_data);
+
 bool gk20a_dbg_gpu_broadcast_stop_trigger(struct channel_gk20a *ch);
 int gk20a_dbg_gpu_clear_broadcast_stop_trigger(struct channel_gk20a *ch);

--- a/drivers/gpu/nvgpu/gk20a/gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/gk20a.h
@@ -647,7 +647,6 @@ struct gk20a {
 	/* held while manipulating # of debug/profiler sessions present */
 	/* also prevents debug sessions from attaching until released */
 	struct mutex dbg_sessions_lock;
-	int dbg_sessions; /* number attached */
 	int dbg_powergating_disabled_refcount; /*refcount for pg disable */
 	int dbg_timeout_disabled_refcount; /*refcount for timeout disable */

--- a/drivers/gpu/nvgpu/gk20a/regops_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/regops_gk20a.c
@@ -390,7 +390,7 @@ int exec_regops_gk20a(struct dbg_session_gk20a *dbg_s,

 	gk20a_dbg(gpu_dbg_fn | gpu_dbg_gpu_dbg, "");

-	ch = dbg_s->ch;
+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);

 	/* For vgpu, the regops routines need to be handled in the
 	 * context of the server and support for that does not exist.
@@ -559,6 +559,9 @@ static bool check_whitelists(struct dbg_session_gk20a *dbg_s,
 {
 	struct gk20a *g = dbg_s->g;
 	bool valid = false;
+	struct channel_gk20a *ch;
+
+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);

 	if (op->type == REGOP(TYPE_GLOBAL)) {
 		/* search global list */
@@ -570,7 +573,7 @@ static bool check_whitelists(struct dbg_session_gk20a *dbg_s,
 			regop_bsearch_range_cmp);

 		/* if debug session and channel is bound search context list */
-		if ((!valid) && (!dbg_s->is_profiler && dbg_s->ch)) {
+		if ((!valid) && (!dbg_s->is_profiler && ch)) {
 			/* binary search context list */
 			valid = g->ops.regops.get_context_whitelist_ranges &&
 				!!bsearch(&offset,
@@ -581,7 +584,7 @@ static bool check_whitelists(struct dbg_session_gk20a *dbg_s,
 		}

 		/* if debug session and channel is bound search runcontrol list */
-		if ((!valid) && (!dbg_s->is_profiler && dbg_s->ch)) {
+		if ((!valid) && (!dbg_s->is_profiler && ch)) {
 			valid = g->ops.regops.get_runcontrol_whitelist &&
 				linear_search(offset,
 				g->ops.regops.get_runcontrol_whitelist(),
@@ -589,7 +592,7 @@ static bool check_whitelists(struct dbg_session_gk20a *dbg_s,
 		}
 	} else if (op->type == REGOP(TYPE_GR_CTX)) {
 		/* it's a context-relative op */
-		if (!dbg_s->ch) {
+		if (!ch) {
 			gk20a_err(dbg_s->dev, "can't perform ctx regop unless bound");
 			op->status = REGOP(STATUS_UNSUPPORTED_OP);
 			return valid;
@@ -604,7 +607,7 @@ static bool check_whitelists(struct dbg_session_gk20a *dbg_s,
 			regop_bsearch_range_cmp);

 		/* if debug session and channel is bound search runcontrol list */
-		if ((!valid) && (!dbg_s->is_profiler && dbg_s->ch)) {
+		if ((!valid) && (!dbg_s->is_profiler && ch)) {
 			valid = g->ops.regops.get_runcontrol_whitelist &&
 				linear_search(offset,
 				g->ops.regops.get_runcontrol_whitelist(),
--- a/drivers/gpu/nvgpu/vgpu/dbg_vgpu.c
+++ b/drivers/gpu/nvgpu/vgpu/dbg_vgpu.c
@@ -26,7 +26,7 @@ static int vgpu_exec_regops(struct dbg_session_gk20a *dbg_s,
 		      struct nvgpu_dbg_gpu_reg_op *ops,
 		      u64 num_ops)
 {
-	struct channel_gk20a *ch = dbg_s->ch;
+	struct channel_gk20a *ch;
 	struct gk20a_platform *platform = gk20a_get_platform(dbg_s->g->dev);
 	struct tegra_vgpu_cmd_msg msg;
 	struct tegra_vgpu_reg_ops_params *p = &msg.params.reg_ops;
@@ -55,6 +55,7 @@ static int vgpu_exec_regops(struct dbg_session_gk20a *dbg_s,

 	msg.cmd = TEGRA_VGPU_CMD_REG_OPS;
 	msg.handle = platform->virt_handle;
+	ch = nvgpu_dbg_gpu_get_session_channel(dbg_s);
 	p->handle = ch ? ch->virt_ctx : 0;
 	p->num_ops = num_ops;
 	p->is_profiler = dbg_s->is_profiler;
--- a/include/uapi/linux/nvgpu.h
+++ b/include/uapi/linux/nvgpu.h
@@ -496,8 +496,7 @@ struct nvgpu_gpu_get_gpu_time_args {
 * Binding/attaching a debugger session to an nvgpu channel
 *
 * The 'channel_fd' given here is the fd used to allocate the
- * gpu channel context.  To detach/unbind the debugger session
- * use a channel_fd of -1.
+ * gpu channel context.
 *
 */
 struct nvgpu_dbg_gpu_bind_channel_args {
@@ -510,6 +509,8 @@ struct nvgpu_dbg_gpu_bind_channel_args {

 /*
 * Register operations
+ * All operations are targeted towards first channel
+ * attached to debug session
 */
 /* valid op values */
 #define NVGPU_DBG_GPU_REG_OP_READ_32                             (0x00000000)
@@ -722,9 +723,23 @@ struct nvgpu_dbg_gpu_write_single_sm_error_state_args {
 #define NVGPU_DBG_GPU_IOCTL_WRITE_SINGLE_SM_ERROR_STATE			\
 	_IOW(NVGPU_DBG_GPU_IOCTL_MAGIC, 16, struct nvgpu_dbg_gpu_write_single_sm_error_state_args)

+/*
+ * Unbinding/detaching a debugger session from a nvgpu channel
+ *
+ * The 'channel_fd' given here is the fd used to allocate the
+ * gpu channel context.
+ */
+struct nvgpu_dbg_gpu_unbind_channel_args {
+	__u32 channel_fd; /* in */
+	__u32 _pad0[1];
+};
+
+#define NVGPU_DBG_GPU_IOCTL_UNBIND_CHANNEL				\
+	_IOW(NVGPU_DBG_GPU_IOCTL_MAGIC, 17, struct nvgpu_dbg_gpu_unbind_channel_args)
+

 #define NVGPU_DBG_GPU_IOCTL_LAST		\
-	_IOC_NR(NVGPU_DBG_GPU_IOCTL_WRITE_SINGLE_SM_ERROR_STATE)
+	_IOC_NR(NVGPU_DBG_GPU_IOCTL_UNBIND_CHANNEL)

 #define NVGPU_DBG_GPU_IOCTL_MAX_ARG_SIZE		\
 	sizeof(struct nvgpu_dbg_gpu_perfbuf_map_args)