On Sun, Dec 28, 2014 at 12:24:47AM -0800, Guenter Roeck wrote:

On Sat, Dec 27, 2014 at 11:35:16PM +0100, Pavel Machek wrote:

quoted

On Sat 2014-12-27 20:58:25, Pavel Machek wrote:

quoted

On Fri 2014-12-26 13:34:53, Sebastian Reichel wrote:

quoted

OMAP34xx and OMAP36xx processors contain a register in the syscon area,
which can be used to determine the SoCs temperature. This patch provides
a DT based driver for the temperature sensor based on an older driver
written by Peter De Schrijver for the Nokia N900 and N9.

Signed-off-by: Sebastian Reichel <sre@kernel.org>
---
 drivers/hwmon/Kconfig      |   8 ++
 drivers/hwmon/Makefile     |   1 +
 drivers/hwmon/omap3-temp.c | 307 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 316 insertions(+)
 create mode 100644 drivers/hwmon/omap3-temp.c

When it hangs, it loops here:

do {
regmap_read(data->syscon, SYSCON_TEMP_REG,&temp_sensor_reg);
if ((temp_sensor_reg & eocz_mask) == level)
   return true;
printk("=");
}
while (ktime_us_delta(expire, ktime_get()) > 0);

And this fixes the hang, and makes level handling more readable.

Fix the timeout code, now it actually works. Driver still fails after
a while.

Signed-off-by: Pavel Machek <redacted>

diff --git a/drivers/hwmon/omap3-temp.c b/drivers/hwmon/omap3-temp.c
index 8a69604..1b8c768 100644
--- a/drivers/hwmon/omap3-temp.c
+++ b/drivers/hwmon/omap3-temp.c

@@ -130,9 +130,7 @@ static inline bool wait_for_eocz(struct omap3_temp_data *data,
 	ktime_t timeout, expire;
 	u32 temp_sensor_reg, eocz_mask;
 
 	eocz_mask = BIT(data->hwdata->eocz_bit);
-	level &= 1;
-	level *= eocz_mask;
 
 	expire = ktime_add_ns(ktime_get(), max_delay);
 	timeout = ktime_set(0, min_delay);

@@ -140,9 +141,9 @@ static inline bool wait_for_eocz(struct omap3_temp_data *data,
 	schedule_hrtimeout(&timeout, HRTIMER_MODE_REL);
 	do {
 		regmap_read(data->syscon, SYSCON_TEMP_REG, &temp_sensor_reg);
-		if ((temp_sensor_reg & eocz_mask) == level)
+		if (!!(temp_sensor_reg & eocz_mask) == level)
 			return true;
-	} while (ktime_us_delta(expire, ktime_get()) > 0);
+	} while (ktime_after(expire, ktime_get()));
 

Does this have to be a hard loop, without sleep ? I am a bit concerned that it
may loop for more than a ms.

Other than that, I assume we'll see an updated version with the coding style
issues and hang-up problems fixed.

I will send a v2 patchset with Pavels change requests. It may take a
few days though, since I'm currently at a conference (31C3).

-- Sebastian
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20141228/f8bac595/attachment.sig>