Unfortunelly (my fault) the kernel thread that now handles IR processing
has classical races in regard to wakeup and stop.
This patch hopefully closes them all.
Tested with module reload running in a loop, while receiver is blasted
with IR data for 10 minutes.

Signed-off-by: Maxim Levitsky <maximlevit...@gmail.com>
Acked-by: Jarod Wilson <ja...@redhat.com>
---
 drivers/media/IR/ir-core-priv.h |    2 ++
 drivers/media/IR/ir-raw-event.c |   34 +++++++++++++++++++++++++---------
 2 files changed, 27 insertions(+), 9 deletions(-)

diff --git a/drivers/media/IR/ir-core-priv.h b/drivers/media/IR/ir-core-priv.h
index a85a8c7..761e7f4 100644
--- a/drivers/media/IR/ir-core-priv.h
+++ b/drivers/media/IR/ir-core-priv.h
@@ -17,6 +17,7 @@
 #define _IR_RAW_EVENT
 
 #include <linux/slab.h>
+#include <linux/spinlock.h>
 #include <media/ir-core.h>
 
 struct ir_raw_handler {
@@ -33,6 +34,7 @@ struct ir_raw_handler {
 struct ir_raw_event_ctrl {
        struct list_head                list;           /* to keep track of raw 
clients */
        struct task_struct              *thread;
+       spinlock_t                      lock;
        struct kfifo                    kfifo;          /* fifo for the 
pulse/space durations */
        ktime_t                         last_event;     /* when last event 
occurred */
        enum raw_event_type             last_type;      /* last event type */
diff --git a/drivers/media/IR/ir-raw-event.c b/drivers/media/IR/ir-raw-event.c
index 8e0e1b1..119b567 100644
--- a/drivers/media/IR/ir-raw-event.c
+++ b/drivers/media/IR/ir-raw-event.c
@@ -39,22 +39,34 @@ static int ir_raw_event_thread(void *data)
        struct ir_raw_event ev;
        struct ir_raw_handler *handler;
        struct ir_raw_event_ctrl *raw = (struct ir_raw_event_ctrl *)data;
+       int retval;
 
        while (!kthread_should_stop()) {
-               try_to_freeze();
 
-               mutex_lock(&ir_raw_handler_lock);
+               spin_lock_irq(&raw->lock);
+               retval = kfifo_out(&raw->kfifo, &ev, sizeof(ev));
+
+               if (!retval) {
+                       set_current_state(TASK_INTERRUPTIBLE);
 
-               while (kfifo_out(&raw->kfifo, &ev, sizeof(ev)) == sizeof(ev)) {
-                       list_for_each_entry(handler, &ir_raw_handler_list, list)
-                               handler->decode(raw->input_dev, ev);
-                       raw->prev_ev = ev;
+                       if (kthread_should_stop())
+                               set_current_state(TASK_RUNNING);
+
+                       spin_unlock_irq(&raw->lock);
+                       schedule();
+                       continue;
                }
 
-               mutex_unlock(&ir_raw_handler_lock);
+               spin_unlock_irq(&raw->lock);
 
-               set_current_state(TASK_INTERRUPTIBLE);
-               schedule();
+
+               BUG_ON(retval != sizeof(ev));
+
+               mutex_lock(&ir_raw_handler_lock);
+               list_for_each_entry(handler, &ir_raw_handler_list, list)
+                       handler->decode(raw->input_dev, ev);
+               raw->prev_ev = ev;
+               mutex_unlock(&ir_raw_handler_lock);
        }
 
        return 0;
@@ -232,11 +244,14 @@ EXPORT_SYMBOL_GPL(ir_raw_event_set_idle);
 void ir_raw_event_handle(struct input_dev *input_dev)
 {
        struct ir_input_dev *ir = input_get_drvdata(input_dev);
+       unsigned long flags;
 
        if (!ir->raw)
                return;
 
+       spin_lock_irqsave(&ir->raw->lock, flags);
        wake_up_process(ir->raw->thread);
+       spin_unlock_irqrestore(&ir->raw->lock, flags);
 }
 EXPORT_SYMBOL_GPL(ir_raw_event_handle);
 
@@ -275,6 +290,7 @@ int ir_raw_event_register(struct input_dev *input_dev)
                return rc;
        }
 
+       spin_lock_init(&ir->raw->lock);
        ir->raw->thread = kthread_run(ir_raw_event_thread, ir->raw,
                        "rc%u",  (unsigned int)ir->devno);
 
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to