[PATCH 03/13] ANDROID: binder: add support for RT prio inheritance.

[Martijn Coenen]

Adds support for SCHED_BATCH/SCHED_FIFO/SCHED_RR
priority inheritance.

Signed-off-by: Martijn Coenen <maco@xxxxxxxxxxx>
---
 drivers/android/binder.c | 164 ++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 135 insertions(+), 29 deletions(-)

diff --git a/drivers/android/binder.c b/drivers/android/binder.c
index 2f27bce31baf..bb61116d1c0c 100644
--- a/drivers/android/binder.c
+++ b/drivers/android/binder.c
@@ -77,6 +77,7 @@
 #endif
 
 #include <uapi/linux/android/binder.h>
+#include <uapi/linux/sched/types.h>
 #include "binder_alloc.h"
 #include "binder_trace.h"
 
@@ -463,6 +464,22 @@ enum binder_deferred_state {
 	BINDER_DEFERRED_RELEASE      = 0x04,
 };
 
+/**
+ * struct binder_priority - scheduler policy and priority
+ * @sched_policy            scheduler policy
+ * @prio                    [100..139] for SCHED_NORMAL, [0..99] for FIFO/RT
+ *
+ * The binder driver supports inheriting the following scheduler policies:
+ * SCHED_NORMAL
+ * SCHED_BATCH
+ * SCHED_FIFO
+ * SCHED_RR
+ */
+struct binder_priority {
+	unsigned int sched_policy;
+	int prio;
+};
+
 /**
  * struct binder_proc - binder process bookkeeping
  * @proc_node:            element for binder_procs list
@@ -542,7 +559,7 @@ struct binder_proc {
 	int requested_threads;
 	int requested_threads_started;
 	int tmp_ref;
-	long default_priority;
+	struct binder_priority default_priority;
 	struct dentry *debugfs_entry;
 	struct binder_alloc alloc;
 	struct binder_context *context;
@@ -624,8 +641,8 @@ struct binder_transaction {
 	struct binder_buffer *buffer;
 	unsigned int	code;
 	unsigned int	flags;
-	long	priority;
-	long	saved_priority;
+	struct binder_priority	priority;
+	struct binder_priority	saved_priority;
 	kuid_t	sender_euid;
 	/**
 	 * @lock:  protects @from, @to_proc, and @to_thread
@@ -1051,22 +1068,93 @@ static void binder_wakeup_proc_ilocked(struct binder_proc *proc)
 	binder_wakeup_thread_ilocked(proc, thread, /* sync = */false);
 }
 
-static void binder_set_nice(long nice)
+static bool is_rt_policy(int policy)
+{
+	return policy == SCHED_FIFO || policy == SCHED_RR;
+}
+
+static bool is_fair_policy(int policy)
+{
+	return policy == SCHED_NORMAL || policy == SCHED_BATCH;
+}
+
+static bool binder_supported_policy(int policy)
+{
+	return is_fair_policy(policy) || is_rt_policy(policy);
+}
+
+static int to_userspace_prio(int policy, int kernel_priority)
+{
+	if (is_fair_policy(policy))
+		return PRIO_TO_NICE(kernel_priority);
+	else
+		return MAX_USER_RT_PRIO - 1 - kernel_priority;
+}
+
+static int to_kernel_prio(int policy, int user_priority)
+{
+	if (is_fair_policy(policy))
+		return NICE_TO_PRIO(user_priority);
+	else
+		return MAX_USER_RT_PRIO - 1 - user_priority;
+}
+
+static void binder_set_priority(struct task_struct *task,
+				struct binder_priority desired)
 {
-	long min_nice;
+	int priority; /* user-space prio value */
+	bool has_cap_nice;
+	unsigned int policy = desired.sched_policy;
 
-	if (can_nice(current, nice)) {
-		set_user_nice(current, nice);
+	if (task->policy == policy && task->normal_prio == desired.prio)
 		return;
+
+	has_cap_nice = has_capability_noaudit(task, CAP_SYS_NICE);
+
+	priority = to_userspace_prio(policy, desired.prio);
+
+	if (is_rt_policy(policy) && !has_cap_nice) {
+		long max_rtprio = task_rlimit(task, RLIMIT_RTPRIO);
+
+		if (max_rtprio == 0) {
+			policy = SCHED_NORMAL;
+			priority = MIN_NICE;
+		} else if (priority > max_rtprio) {
+			priority = max_rtprio;
+		}
 	}
-	min_nice = rlimit_to_nice(rlimit(RLIMIT_NICE));
-	binder_debug(BINDER_DEBUG_PRIORITY_CAP,
-		     "%d: nice value %ld not allowed use %ld instead\n",
-		      current->pid, nice, min_nice);
-	set_user_nice(current, min_nice);
-	if (min_nice <= MAX_NICE)
-		return;
-	binder_user_error("%d RLIMIT_NICE not set\n", current->pid);
+
+	if (is_fair_policy(policy) && !has_cap_nice) {
+		long min_nice = rlimit_to_nice(task_rlimit(task, RLIMIT_NICE));
+
+		if (min_nice > MAX_NICE) {
+			binder_user_error("%d RLIMIT_NICE not set\n",
+					  task->pid);
+			return;
+		} else if (priority < min_nice) {
+			priority = min_nice;
+		}
+	}
+
+	if (policy != desired.sched_policy ||
+	    to_kernel_prio(policy, priority) != desired.prio)
+		binder_debug(BINDER_DEBUG_PRIORITY_CAP,
+			     "%d: priority %d not allowed, using %d instead\n",
+			      task->pid, desired.prio,
+			      to_kernel_prio(policy, priority));
+
+	/* Set the actual priority */
+	if (task->policy != policy || is_rt_policy(policy)) {
+		struct sched_param params;
+
+		params.sched_priority = is_rt_policy(policy) ? priority : 0;
+
+		sched_setscheduler_nocheck(task,
+					   policy | SCHED_RESET_ON_FORK,
+					   &params);
+	}
+	if (is_fair_policy(policy))
+		set_user_nice(task, priority);
 }
 
 static struct binder_node *binder_get_node_ilocked(struct binder_proc *proc,
@@ -1150,7 +1238,8 @@ static struct binder_node *binder_init_node_ilocked(
 	node->ptr = ptr;
 	node->cookie = cookie;
 	node->work.type = BINDER_WORK_NODE;
-	node->min_priority = flags & FLAT_BINDER_FLAG_PRIORITY_MASK;
+	node->min_priority = NICE_TO_PRIO(
+			flags & FLAT_BINDER_FLAG_PRIORITY_MASK);
 	node->accept_fds = !!(flags & FLAT_BINDER_FLAG_ACCEPTS_FDS);
 	spin_lock_init(&node->lock);
 	INIT_LIST_HEAD(&node->work.entry);
@@ -2646,7 +2735,7 @@ static void binder_transaction(struct binder_proc *proc,
 		}
 		thread->transaction_stack = in_reply_to->to_parent;
 		binder_inner_proc_unlock(proc);
-		binder_set_nice(in_reply_to->saved_priority);
+		binder_set_priority(current, in_reply_to->saved_priority);
 		target_thread = binder_get_txn_from_and_acq_inner(in_reply_to);
 		if (target_thread == NULL) {
 			return_error = BR_DEAD_REPLY;
@@ -2819,7 +2908,15 @@ static void binder_transaction(struct binder_proc *proc,
 	t->to_thread = target_thread;
 	t->code = tr->code;
 	t->flags = tr->flags;
-	t->priority = task_nice(current);
+	if (!(t->flags & TF_ONE_WAY) &&
+	    binder_supported_policy(current->policy)) {
+		/* Inherit supported policies for synchronous transactions */
+		t->priority.sched_policy = current->policy;
+		t->priority.prio = current->normal_prio;
+	} else {
+		/* Otherwise, fall back to the default priority */
+		t->priority = target_proc->default_priority;
+	}
 
 	trace_binder_transaction(reply, t, target_node);
 
@@ -3735,7 +3832,7 @@ static int binder_thread_read(struct binder_proc *proc,
 			wait_event_interruptible(binder_user_error_wait,
 						 binder_stop_on_user_error < 2);
 		}
-		binder_set_nice(proc->default_priority);
+		binder_set_priority(current, proc->default_priority);
 	}
 
 	if (non_block) {
@@ -3947,16 +4044,17 @@ static int binder_thread_read(struct binder_proc *proc,
 		BUG_ON(t->buffer == NULL);
 		if (t->buffer->target_node) {
 			struct binder_node *target_node = t->buffer->target_node;
+			struct binder_priority prio = t->priority;
 
 			tr.target.ptr = target_node->ptr;
 			tr.cookie =  target_node->cookie;
-			t->saved_priority = task_nice(current);
-			if (t->priority < target_node->min_priority &&
-			    !(t->flags & TF_ONE_WAY))
-				binder_set_nice(t->priority);
-			else if (!(t->flags & TF_ONE_WAY) ||
-				 t->saved_priority > target_node->min_priority)
-				binder_set_nice(target_node->min_priority);
+			t->saved_priority.sched_policy = current->policy;
+			t->saved_priority.prio = current->normal_prio;
+			if (target_node->min_priority < t->priority.prio) {
+				prio.sched_policy = SCHED_NORMAL;
+				prio.prio = target_node->min_priority;
+			}
+			binder_set_priority(current, prio);
 			cmd = BR_TRANSACTION;
 		} else {
 			tr.target.ptr = 0;
@@ -4562,7 +4660,14 @@ static int binder_open(struct inode *nodp, struct file *filp)
 	get_task_struct(current->group_leader);
 	proc->tsk = current->group_leader;
 	INIT_LIST_HEAD(&proc->todo);
-	proc->default_priority = task_nice(current);
+	if (binder_supported_policy(current->policy)) {
+		proc->default_priority.sched_policy = current->policy;
+		proc->default_priority.prio = current->normal_prio;
+	} else {
+		proc->default_priority.sched_policy = SCHED_NORMAL;
+		proc->default_priority.prio = NICE_TO_PRIO(0);
+	}
+
 	binder_dev = container_of(filp->private_data, struct binder_device,
 				  miscdev);
 	proc->context = &binder_dev->context;
@@ -4855,13 +4960,14 @@ static void print_binder_transaction_ilocked(struct seq_file *m,
 	spin_lock(&t->lock);
 	to_proc = t->to_proc;
 	seq_printf(m,
-		   "%s %d: %p from %d:%d to %d:%d code %x flags %x pri %ld r%d",
+		   "%s %d: %p from %d:%d to %d:%d code %x flags %x pri %d:%d r%d",
 		   prefix, t->debug_id, t,
 		   t->from ? t->from->proc->pid : 0,
 		   t->from ? t->from->pid : 0,
 		   to_proc ? to_proc->pid : 0,
 		   t->to_thread ? t->to_thread->pid : 0,
-		   t->code, t->flags, t->priority, t->need_reply);
+		   t->code, t->flags, t->priority.sched_policy,
+		   t->priority.prio, t->need_reply);
 	spin_unlock(&t->lock);
 
 	if (proc != to_proc) {
-- 
2.14.1.480.gb18f417b89-goog