Improve pcf50633 interrupt service scheduling to enforce only servicing
when resume action is completed

Signed-off-by: Andy Green <[EMAIL PROTECTED]>
---

 drivers/i2c/chips/pcf50633.c |   67 ++++++++++++++++++++++++++++++------------
 1 files changed, 48 insertions(+), 19 deletions(-)

diff --git a/drivers/i2c/chips/pcf50633.c b/drivers/i2c/chips/pcf50633.c
index 4d92c84..33e2eaf 100644
--- a/drivers/i2c/chips/pcf50633.c
+++ b/drivers/i2c/chips/pcf50633.c
@@ -656,6 +656,26 @@ static void pcf50633_work(struct work_struct *work)
 
        mutex_lock(&pcf->working_lock);
        pcf->working = 1;
+
+       dev_info(&pcf->client.dev, "pcf50633_work called with suspended = %d\n",
+                                  pcf->have_been_suspended);
+
+       /*
+        * If we are inside suspend -> resume completion time we don't attempt
+        * service until we have fully resumed.  Although we could talk to the
+        * device as soon as I2C is up, the regs in the device which we might
+        * choose to modify as part of the service action have not been
+        * reloaded with their pre-suspend states yet.  Therefore we will
+        * defer our service if we are called like that until our resume has
+        * completed.
+        */
+
+       if (pcf->have_been_suspended && (pcf->have_been_suspended < 2)) {
+               dev_info(&pcf->client.dev, "rescheduling,  suspended = %d\n",
+                                          pcf->have_been_suspended);
+               goto reschedule;
+       }
+
        /*
         * datasheet says we have to read the five IRQ
         * status regs in one transaction
@@ -663,30 +683,25 @@ static void pcf50633_work(struct work_struct *work)
        ret = i2c_smbus_read_i2c_block_data(&pcf->client, PCF50633_REG_INT1, 5,
                                            pcfirq);
        if (ret != 5) {
-               DEBUGP("Oh crap PMU IRQ register read failed -- "
-                      "retrying later %d\n", ret);
+               dev_info(&pcf->client.dev,
+                        "Oh crap PMU IRQ register read failed -- "
+                        "retrying later %d\n", ret);
                /*
-                * this situation can happen during resume, just defer
-                * handling the interrupt until enough I2C is up we can
-                * actually talk to the PMU.  We can't just ignore this
-                * because we are on a falling edge interrupt and our
-                * PMU interrupt source does not clear until we read these
-                * interrupt source registers.
+                * it shouldn't fail, we no longer attempt to use I2C while
+                * it can be suspended.  But we don't have much option but to
+                * retry if if it ever did fail, because if we don't service
+                * the interrupt to clear it, we will never see another PMU
+                * interrupt edge.
                 */
-               if (!schedule_work(&pcf->work) && !pcf->working)
-                       dev_dbg(&pcf->client.dev, "work item may be lost\n");
-
-               /* we don't put the device here, hold it for next time */
-               mutex_unlock(&pcf->working_lock);
-               /* don't spew, delaying whatever else is happening */
-               msleep(1);
-               return;
+               goto reschedule;
        }
 
        /* hey did we just resume? */
 
        if (pcf->have_been_suspended) {
+               /* resume is officially over now then */
                pcf->have_been_suspended = 0;
+
                /*
                 * grab a copy of resume interrupt reasons
                 * from pcf50633 POV
@@ -1045,6 +1060,19 @@ static void pcf50633_work(struct work_struct *work)
        input_sync(pcf->input_dev);
        put_device(&pcf->client.dev);
        mutex_unlock(&pcf->working_lock);
+
+       return;
+
+reschedule:
+       /* don't spew, delaying whatever else is happening */
+       msleep(100);
+
+       dev_info(&pcf->client.dev, "rescheduling interrupt service\n");
+       if (!schedule_work(&pcf->work))
+               dev_err(&pcf->client.dev, "int service reschedule failed\n");
+
+       /* we don't put the device here, hold it for next time */
+       mutex_unlock(&pcf->working_lock);
 }
 
 static irqreturn_t pcf50633_irq(int irq, void *_pcf)
@@ -1052,10 +1080,11 @@ static irqreturn_t pcf50633_irq(int irq, void *_pcf)
        struct pcf50633_data *pcf = _pcf;
 
        DEBUGP("entering(irq=%u, pcf=%p): scheduling work\n", irq, _pcf);
+       dev_info(&pcf->client.dev, "pcf50633_irq scheduling work\n");
 
        get_device(&pcf->client.dev);
        if (!schedule_work(&pcf->work) && !pcf->working)
-               dev_dbg(&pcf->client.dev, "work item may be lost\n");
+               dev_err(&pcf->client.dev, "work item may be lost\n");
 
        return IRQ_HANDLED;
 }
@@ -2244,9 +2273,9 @@ static int pcf50633_suspend(struct device *dev, 
pm_message_t state)
 int pcf50633_ready(struct pcf50633_data *pcf)
 {
        if (!pcf)
-               return -EBUSY;
+               return -EACCES;
 
-       if (pcf->have_been_suspended)
+       if (pcf->have_been_suspended && (pcf->have_been_suspended < 3))
                return -EBUSY;
 
        return 0;


Reply via email to