public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Henrik Rydberg <rydberg@euromail.se>
To: guenter.roeck@ericsson.com
Cc: Jean Delvare <khali@linux-fr.org>,
	"lm-sensors@lm-sensors.org" <lm-sensors@lm-sensors.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 04/11] hwmon: applesmc: Introduce a register lookup table (rev2)
Date: Wed, 10 Nov 2010 11:57:24 +0100	[thread overview]
Message-ID: <4CDA7A94.7070302@euromail.se> (raw)
In-Reply-To: <1289336024.22931.243.camel@groeck-laptop>


>>
>> mutex_destroy() is defined as a nop, so I guess the question is whether anything
>> could be holding the lock when entering a second init. There are no sysfs files
>> created at that point, so I would say no. The mutex could be put back with a
>> static initializer, if this is not satisfactory. The real reason to move it to
>> the smcreg struct was to force a rename of the mutex itself.
>>
> 
> Alternatively, you could move the mutex initialization to the beginning
> of applesmc_init_smcreg() and make it
> 	mutex_init(&smcreg.mutex);


Looking at this again, it seems there are two other problems as well. Firstly,
the cache memory is not freed after probe failure, my apologies. Secondly,
execution continues after a probe failure, and the initialization is retried. I
would like to push the latter problem to some other occasion, since the whole
platform logic should be rewritten for the new interface, anyways.

>>
>> With the empirical failure rate, it is extremely unlikely to get more than a
>> couple of failures in a row - information which in itself could be very useful.
> 
> You would have alternative options, though, with less noise. For
> example, something along the line of
> 
> 	for (...) {
> 		...
> 		if (!ret) {
> 			if (ms)
> 				pr_info("smcreg initialization took %d ms\n", ms);
> 			return 0;
> 		}
> 	...
> 	}
> 	pr_err("smcreg initialization failed\n");


Looks nice, have applied, but without the last line; the probe failure report
should be enough to deduce this.

>>
>> Changing the place of the mutex will ripple through all patches, so I will
>> resend from this one onwards. I suppose you have more comments on the following
>> patches?
> 
> Maybe it won't be that bad if you initialize it as I suggested above.


I tried several types of changes, and they all had some effect on later patches.
The patch below comprise the resulting changes to patch 4. Hope you like. In
addition, patch 5 and 7 needed one line of wiggling. I am resending all three.

@@ -217,7 +217,9 @@ static struct applesmc_registers {
 	unsigned int key_count;		/* number of SMC registers */
 	bool init_complete;		/* true when fully initialized */
 	struct applesmc_entry *cache;	/* cached key entries */
-} smcreg;
+} smcreg = {
+	.mutex = __MUTEX_INITIALIZER(smcreg.mutex),
+};

 static const int debug;
 static struct platform_device *pdev;
@@ -581,8 +583,6 @@ static int applesmc_init_smcreg_try(void)
 	if (s->init_complete)
 		return 0;

-	mutex_init(&s->mutex);
-
 	ret = read_register_count(&s->key_count);
 	if (ret)
 		return ret;
@@ -611,19 +611,25 @@ static int applesmc_init_smcreg(void)

 	for (ms = 0; ms < INIT_TIMEOUT_MSECS; ms += INIT_WAIT_MSECS) {
 		ret = applesmc_init_smcreg_try();
-		if (!ret)
+		if (!ret) {
+ 			if (ms)
+				pr_info("smcreg initialization took %d ms\n", ms);
 			return 0;
-		pr_warn("slow init, retrying\n");
+		}
 		msleep(INIT_WAIT_MSECS);
 	}

+ 	kfree(smcreg.cache);
+	smcreg.cache = NULL;
+
 	return ret;
 }

 static void applesmc_destroy_smcreg(void)
 {
 	kfree(smcreg.cache);
-	memset(&smcreg, 0, sizeof(smcreg));
+	smcreg.cache = NULL;
+	smcreg.init_complete = false;
 }

 /* Device model stuff */

Thanks,
Henrik

  reply	other threads:[~2010-11-10 10:57 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-11-09 15:15 [PATCH 00/11] hwmon: applesmc: Dynamic configuration rewrite (rev2) Henrik Rydberg
2010-11-09 15:15 ` [PATCH 01/11] hwmon: applesmc: Add MacBookAir3,1(3,2) support Henrik Rydberg
2010-11-09 18:56   ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 02/11] hwmon: applesmc: Relax the severity of device init failure (rev2) Henrik Rydberg
2010-11-09 18:57   ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 03/11] drivers/hwmon/applesmc.c: Use pr_fmt and pr_<level> Henrik Rydberg
2010-11-09 18:57   ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 04/11] hwmon: applesmc: Introduce a register lookup table (rev2) Henrik Rydberg
2010-11-09 19:04   ` Guenter Roeck
2010-11-09 19:32     ` Henrik Rydberg
2010-11-09 20:53       ` Guenter Roeck
2010-11-10 10:57         ` Henrik Rydberg [this message]
2010-11-10 15:42           ` Guenter Roeck
2010-11-10 17:14             ` Henrik Rydberg
2010-11-10 17:57               ` Guenter Roeck
2010-11-09 22:15       ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 05/11] hwmon: applesmc: Dynamic creation of temperature files (rev2) Henrik Rydberg
2010-11-09 15:15 ` [PATCH 06/11] hwmon: applesmc: Handle new temperature format (rev2) Henrik Rydberg
2010-11-10 18:07   ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 07/11] hwmon: applesmc: Extract all features generically (rev2) Henrik Rydberg
2010-11-09 15:15 ` [PATCH 08/11] hwmon: applesmc: Dynamic creation of fan files (rev2) Henrik Rydberg
2010-11-10 18:09   ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 09/11] hwmon: applesmc: Simplify feature sysfs handling (rev2) Henrik Rydberg
2010-11-10 18:10   ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 10/11] hwmon: applesmc: Silence driver Henrik Rydberg
2010-11-10 18:10   ` Guenter Roeck
2010-11-09 15:15 ` [PATCH 11/11] hwmon: applesmc: Update copyright information Henrik Rydberg
2010-11-10 18:10   ` Guenter Roeck
2010-11-10 18:22     ` Henrik Rydberg

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4CDA7A94.7070302@euromail.se \
    --to=rydberg@euromail.se \
    --cc=guenter.roeck@ericsson.com \
    --cc=khali@linux-fr.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lm-sensors@lm-sensors.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox