2009-11-30 23:47:24

by djwong

Subject: [PATCH] therm_adt746x: Record pwm invert bit at module load time

In commit 0512a9a8e277a9de2820211eef964473b714ae65, we unilaterally zero the
"pwm invert" bit in the fan behavior configuration register. On my PowerBook
G4, this results in the fans going to full speed at low temperature and
shutting off at high temperature because the pwm invert bit is supposed to be
set.

Therefore, record the pwm invert bit at driver load time, and write the bit
into the fan behavior control register. This restores correct behavior on my
PBG4 and should work around the bit being set to the wrong value after
suspend/resume (which is what the original patch was trying to fix). It also
fixes a minor omission where the pwm invert bit correction is NOT performed
when switching into automatic mode.

Signed-off-by: Darrick J. Wong <[email protected]>
---

drivers/macintosh/therm_adt746x.c | 13 +++++++++++--
1 files changed, 11 insertions(+), 2 deletions(-)


diff --git a/drivers/macintosh/therm_adt746x.c b/drivers/macintosh/therm_adt746x.c
index 556f0fe..386a797 100644
--- a/drivers/macintosh/therm_adt746x.c
+++ b/drivers/macintosh/therm_adt746x.c
@@ -79,6 +79,7 @@ struct thermostat {
u8 limits[3];
int last_speed[2];
int last_var[2];
+ int pwm_inv[2];
};

static enum {ADT7460, ADT7467} therm_type;
@@ -229,19 +230,23 @@ static void write_fan_speed(struct thermostat *th, int speed, int fan)

if (speed >= 0) {
manual = read_reg(th, MANUAL_MODE[fan]);
+ manual &= ~INVERT_MASK;
write_reg(th, MANUAL_MODE[fan],
- (manual|MANUAL_MASK) & (~INVERT_MASK));
+ manual | MANUAL_MASK | th->pwm_inv[fan]);
write_reg(th, FAN_SPD_SET[fan], speed);
} else {
/* back to automatic */
if(therm_type == ADT7460) {
manual = read_reg(th,
MANUAL_MODE[fan]) & (~MANUAL_MASK);
-
+ manual &= ~INVERT_MASK;
+ manual |= th->pwm_inv[fan];
write_reg(th,
MANUAL_MODE[fan], manual|REM_CONTROL[fan]);
} else {
manual = read_reg(th, MANUAL_MODE[fan]);
+ manual &= ~INVERT_MASK;
+ manual |= th->pwm_inv[fan];
write_reg(th, MANUAL_MODE[fan], manual&(~AUTO_MASK));
}
}
@@ -418,6 +423,10 @@ static int probe_thermostat(struct i2c_client *client,

thermostat = th;

+ /* record invert bit status because fw can corrupt it after suspend */
+ th->pwm_inv[0] = read_reg(th, MANUAL_MODE[0]) & INVERT_MASK;
+ th->pwm_inv[1] = read_reg(th, MANUAL_MODE[1]) & INVERT_MASK;
+
/* be sure to really write fan speed the first time */
th->last_speed[0] = -2;
th->last_speed[1] = -2;

2009-12-01 08:07:42

by Michel Dänzer

Subject: Re: [PATCH] therm_adt746x: Record pwm invert bit at module load time

On Mon, 2009-11-30 at 15:47 -0800, Darrick J. Wong wrote:
> In commit 0512a9a8e277a9de2820211eef964473b714ae65, we unilaterally zero the
> "pwm invert" bit in the fan behavior configuration register. On my PowerBook
> G4, this results in the fans going to full speed at low temperature and
> shutting off at high temperature because the pwm invert bit is supposed to be
> set.
>
> Therefore, record the pwm invert bit at driver load time, and write the bit
> into the fan behavior control register. This restores correct behavior on my
> PBG4 and should work around the bit being set to the wrong value after
> suspend/resume (which is what the original patch was trying to fix). It also
> fixes a minor omission where the pwm invert bit correction is NOT performed
> when switching into automatic mode.
>
> Signed-off-by: Darrick J. Wong <[email protected]>

Reviewed-by: Michel Dänzer <[email protected]>

Looks great, thanks Darrick. Also FWIW, at least loading the patched
driver works as expected on my PowerBook5,8 which has the invert bit
disabled.


> ---
>
> drivers/macintosh/therm_adt746x.c | 13 +++++++++++--
> 1 files changed, 11 insertions(+), 2 deletions(-)
>
>
> diff --git a/drivers/macintosh/therm_adt746x.c b/drivers/macintosh/therm_adt746x.c
> index 556f0fe..386a797 100644
> --- a/drivers/macintosh/therm_adt746x.c
> +++ b/drivers/macintosh/therm_adt746x.c
> @@ -79,6 +79,7 @@ struct thermostat {
> u8 limits[3];
> int last_speed[2];
> int last_var[2];
> + int pwm_inv[2];
> };
>
> static enum {ADT7460, ADT7467} therm_type;
> @@ -229,19 +230,23 @@ static void write_fan_speed(struct thermostat *th, int speed, int fan)
>
> if (speed >= 0) {
> manual = read_reg(th, MANUAL_MODE[fan]);
> + manual &= ~INVERT_MASK;
> write_reg(th, MANUAL_MODE[fan],
> - (manual|MANUAL_MASK) & (~INVERT_MASK));
> + manual | MANUAL_MASK | th->pwm_inv[fan]);
> write_reg(th, FAN_SPD_SET[fan], speed);
> } else {
> /* back to automatic */
> if(therm_type == ADT7460) {
> manual = read_reg(th,
> MANUAL_MODE[fan]) & (~MANUAL_MASK);
> -
> + manual &= ~INVERT_MASK;
> + manual |= th->pwm_inv[fan];
> write_reg(th,
> MANUAL_MODE[fan], manual|REM_CONTROL[fan]);
> } else {
> manual = read_reg(th, MANUAL_MODE[fan]);
> + manual &= ~INVERT_MASK;
> + manual |= th->pwm_inv[fan];
> write_reg(th, MANUAL_MODE[fan], manual&(~AUTO_MASK));
> }
> }
> @@ -418,6 +423,10 @@ static int probe_thermostat(struct i2c_client *client,
>
> thermostat = th;
>
> + /* record invert bit status because fw can corrupt it after suspend */
> + th->pwm_inv[0] = read_reg(th, MANUAL_MODE[0]) & INVERT_MASK;
> + th->pwm_inv[1] = read_reg(th, MANUAL_MODE[1]) & INVERT_MASK;
> +
> /* be sure to really write fan speed the first time */
> th->last_speed[0] = -2;
> th->last_speed[1] = -2;



--
Earthling Michel Dänzer | http://www.vmware.com
Libre software enthusiast | Debian, X and DRI developer

2009-12-04 02:21:14

by Benjamin Herrenschmidt

Subject: Re: [PATCH] therm_adt746x: Record pwm invert bit at module load time

On Mon, 2009-11-30 at 15:47 -0800, Darrick J. Wong wrote:
> In commit 0512a9a8e277a9de2820211eef964473b714ae65, we unilaterally zero the
> "pwm invert" bit in the fan behavior configuration register. On my PowerBook
> G4, this results in the fans going to full speed at low temperature and
> shutting off at high temperature because the pwm invert bit is supposed to be
> set.
>
> Therefore, record the pwm invert bit at driver load time, and write the bit
> into the fan behavior control register. This restores correct behavior on my
> PBG4 and should work around the bit being set to the wrong value after
> suspend/resume (which is what the original patch was trying to fix). It also
> fixes a minor omission where the pwm invert bit correction is NOT performed
> when switching into automatic mode.

Thanks, I'll apply.

Cheers,
Ben.

> Signed-off-by: Darrick J. Wong <[email protected]>
> ---
>
> drivers/macintosh/therm_adt746x.c | 13 +++++++++++--
> 1 files changed, 11 insertions(+), 2 deletions(-)
>
>
> diff --git a/drivers/macintosh/therm_adt746x.c b/drivers/macintosh/therm_adt746x.c
> index 556f0fe..386a797 100644
> --- a/drivers/macintosh/therm_adt746x.c
> +++ b/drivers/macintosh/therm_adt746x.c
> @@ -79,6 +79,7 @@ struct thermostat {
> u8 limits[3];
> int last_speed[2];
> int last_var[2];
> + int pwm_inv[2];
> };
>
> static enum {ADT7460, ADT7467} therm_type;
> @@ -229,19 +230,23 @@ static void write_fan_speed(struct thermostat *th, int speed, int fan)
>
> if (speed >= 0) {
> manual = read_reg(th, MANUAL_MODE[fan]);
> + manual &= ~INVERT_MASK;
> write_reg(th, MANUAL_MODE[fan],
> - (manual|MANUAL_MASK) & (~INVERT_MASK));
> + manual | MANUAL_MASK | th->pwm_inv[fan]);
> write_reg(th, FAN_SPD_SET[fan], speed);
> } else {
> /* back to automatic */
> if(therm_type == ADT7460) {
> manual = read_reg(th,
> MANUAL_MODE[fan]) & (~MANUAL_MASK);
> -
> + manual &= ~INVERT_MASK;
> + manual |= th->pwm_inv[fan];
> write_reg(th,
> MANUAL_MODE[fan], manual|REM_CONTROL[fan]);
> } else {
> manual = read_reg(th, MANUAL_MODE[fan]);
> + manual &= ~INVERT_MASK;
> + manual |= th->pwm_inv[fan];
> write_reg(th, MANUAL_MODE[fan], manual&(~AUTO_MASK));
> }
> }
> @@ -418,6 +423,10 @@ static int probe_thermostat(struct i2c_client *client,
>
> thermostat = th;
>
> + /* record invert bit status because fw can corrupt it after suspend */
> + th->pwm_inv[0] = read_reg(th, MANUAL_MODE[0]) & INVERT_MASK;
> + th->pwm_inv[1] = read_reg(th, MANUAL_MODE[1]) & INVERT_MASK;
> +
> /* be sure to really write fan speed the first time */
> th->last_speed[0] = -2;
> th->last_speed[1] = -2;

2009-12-04 08:44:04

by Jean Delvare

Subject: Re: [lm-sensors] [PATCH] therm_adt746x: Record pwm invert bit at module load time

On Fri, 04 Dec 2009 13:20:11 +1100, Benjamin Herrenschmidt wrote:
> On Mon, 2009-11-30 at 15:47 -0800, Darrick J. Wong wrote:
> > In commit 0512a9a8e277a9de2820211eef964473b714ae65, we unilaterally zero the
> > "pwm invert" bit in the fan behavior configuration register. On my PowerBook
> > G4, this results in the fans going to full speed at low temperature and
> > shutting off at high temperature because the pwm invert bit is supposed to be
> > set.
> >
> > Therefore, record the pwm invert bit at driver load time, and write the bit
> > into the fan behavior control register. This restores correct behavior on my
> > PBG4 and should work around the bit being set to the wrong value after
> > suspend/resume (which is what the original patch was trying to fix). It also
> > fixes a minor omission where the pwm invert bit correction is NOT performed
> > when switching into automatic mode.
>
> Thanks, I'll apply.

This is a pretty serious bug, please make sure to send the fix to the
stable team for kernels 2.6.31 and 2.6.32.

--
Jean Delvare