774ac8b7eff6 ("Thermal: initialize thermal zone device correctly") causes performance drop

REGRESSION Laura Abbott 5 participants 11 messages Mar 16, 2016 – Mar 21, 2016 1 message not in the archive Thermal

Share & tools

Laura Abbott

author Tested-by Manuel Krause Tested-by szegad Tested-by prash Tested-by amish Tested-by Matthias Reviewed-by Javi Merino

2016-03-16 22:27 UTC #0 lore raw

Hi,

Fedora received a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).

Bisection showed the first bad commit was

commit 774ac8b7eff69e0786970157de2157e68b22f456

Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800

Thermal: initialize thermal zone device correctly

commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream.

After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().

Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.

Tested-by Manuel Krause <[email protected]>

Tested-by szegad <[email protected]>

Tested-by prash <[email protected]>

Tested-by amish <[email protected]>

Tested-by Matthias <[email protected]>

Reviewed-by Javi Merino <[email protected]>

Signed-off-by Zhang Rui <[email protected]>

Signed-off-by Chen Yu <[email protected]>

Signed-off-by Greg Kroah-Hartman <[email protected]>

Reverting this plus to other commits in the series (a67208e94d94 "Thermal: handle thermal zone device properly during system sleep" and 27f356149d59 "Thermal: do thermal zone update after a cooling device registered") confirmed the performance was back to normal.

Bugzilla has the full discussion but this comment from one of the reporters sums it up:

"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between 1000 and 3400 MHz. In 4.4.4, the processor would fluctuate between 400 and 700 MHz, according to /proc/cpuinfo.

Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to performance, instead of the default "powersave" forces the CPU to 2400 MHz, and improves performance greatly, but still not to the same level as in 4.4.3."

Any ideas?

Thanks,
Laura

Greg Kroah-Hartman

2016-03-16 22:46 UTC #1 lore raw

Quoted 48 lines · Laura Abbott

Hi,
Fedora received a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream. After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>
Reverting this plus to other commits in the series (a67208e94d94 "Thermal: handle thermal zone device properly during system sleep" and 27f356149d59 "Thermal: do thermal zone update after a cooling device registered") confirmed the performance was back to normal.
Bugzilla has the full discussion but this comment from one of the reporters sums it up:
"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between 1000 and 3400 MHz. In 4.4.4, the processor would fluctuate between 400 and 700 MHz, according to /proc/cpuinfo.

Quoted 6 lines

Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to performance, instead of the default "powersave" forces the CPU to 2400 MHz, and improves performance greatly, but still not to the same level as in 4.4.3."
Any ideas?

Is this same "slowdown" also seen in 4.5?

thanks,

greg k-h

Laura Abbott

author

2016-03-17 00:00 UTC #2 lore raw

Quoted 54 lines · Greg Kroah-Hartman

Quoted 53 lines · Laura Abbott
Hi,
Fedora received a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream. After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>
Reverting this plus to other commits in the series (a67208e94d94 "Thermal: handle thermal zone device properly during system sleep" and 27f356149d59 "Thermal: do thermal zone update after a cooling device registered") confirmed the performance was back to normal.
Bugzilla has the full discussion but this comment from one of the reporters sums it up:
"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between 1000 and 3400 MHz. In 4.4.4, the processor would fluctuate between 400 and 700 MHz, according to /proc/cpuinfo.
Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to performance, instead of the default "powersave" forces the CPU to 2400 MHz, and improves performance greatly, but still not to the same level as in 4.4.3."

Quoted 8 lines

Quoted 1 line
Any ideas?
Is this same "slowdown" also seen in 4.5?
thanks,
greg k-h

Yes, the same issue is seen on 4.5 according to the reporter.

Thanks,
Laura

Greg Kroah-Hartman

2016-03-17 00:11 UTC #3 lore raw

Quoted 57 lines · Laura Abbott

Quoted 56 lines · Greg Kroah-Hartman
Quoted 54 lines · Laura Abbott
Hi,
Fedora received a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream. After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>
Reverting this plus to other commits in the series (a67208e94d94 "Thermal: handle thermal zone device properly during system sleep" and 27f356149d59 "Thermal: do thermal zone update after a cooling device registered") confirmed the performance was back to normal.
Bugzilla has the full discussion but this comment from one of the reporters sums it up:
"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between 1000 and 3400 MHz. In 4.4.4, the processor would fluctuate between 400 and 700 MHz, according to /proc/cpuinfo.
Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to performance, instead of the default "powersave" forces the CPU to 2400 MHz, and improves performance greatly, but still not to the same level as in 4.4.3."
Any ideas?

Quoted 8 lines

Quoted 6 lines
Is this same "slowdown" also seen in 4.5?
thanks,
greg k-h
Yes, the same issue is seen on 4.5 according to the reporter.

Great, we are "bug compatible" :)

Can you please work to get this resolved in Linus's tree and then we can backport the needed changes into the 4.4-stable release.

Zhang, any ideas here?

thanks,

greg k-h

Zhang, Rui

reviewer

2016-03-17 00:17 UTC #7 lore raw

In reply to Laura Abbott

Can you send me the output of "grep . /sys/class/thermal/*/*" both w/ and w/o the broken patch series?

Thanks,
rui

Quoted 76 lines

-----Original Message-----
From: Laura Abbott [mailto:[email protected]]
Sent: Thursday, March 17, 2016 6:28 AM
To: Zhang, Rui <[email protected]>; Javi Merino <[email protected]>; Chen, Yu C <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>; Manuel Krause <[email protected]>; szegad <[email protected]>; prash <[email protected]>; amish <[email protected]>; Matthias <[email protected]>; [email protected]; Linux Kernel Mailing List <[email protected]>
Subject: [REGRESSION] 774ac8b7eff6 ("Thermal: initialize thermal zone device correctly") causes performance drop
Importance: High
Hi,
Fedora received a bug report
(https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly
commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream.
After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>
Reverting this plus to other commits in the series (a67208e94d94 "Thermal: handle thermal zone device properly during system sleep" and 27f356149d59 "Thermal: do thermal zone update after a cooling device registered") confirmed the performance was back to normal.
Bugzilla has the full discussion but this comment from one of the reporters sums it up:
"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between 1000 and 3400 MHz. In 4.4.4, the processor would fluctuate between 400 and 700 MHz, according to /proc/cpuinfo.
Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to performance, instead of the default "powersave" forces the CPU to 2400 MHz, and improves performance greatly, but still not to the same level as in 4.4.3."
Any ideas?
Thanks,
Laura

Pandruvada, Srinivas

2016-03-17 00:20 UTC #4 lore raw

In reply to Laura Abbott

Quoted 66 lines · Laura Abbott

Quoted 65 lines · Greg Kroah-Hartman
Quoted 63 lines · Laura Abbott
Hi,
Fedora received a bug report (https://bugzilla.redhat.com/show_bu
g.cgi?id=1317190)
of a major performance drop on various bench marks and general system
sluggishness with the 4.4.4 kernel update. The benchmarks were showing
a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
     Thermal: initialize thermal zone device correctly commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream. After thermal zone device registered, as we have not read any
     temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
     Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]
g>
Reverting this plus to other commits in the series (a67208e94d94 "Thermal: handle thermal zone device properly during system sleep"
and 27f356149d59 "Thermal: do thermal zone update after a cooling device registered") confirmed the performance was back to normal.
Bugzilla has the full discussion but this comment from one of the reporters sums it up:
"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between
1000 and 3400 MHz.  In 4.4.4, the processor would fluctuate between
400 and 700 MHz, according to /proc/cpuinfo.
Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to
performance, instead of the default "powersave" forces the CPU to 2400 MHz, and improves performance greatly, but still not to the same level as in 4.4.3."
Any ideas?

Quoted 8 lines

Quoted 6 lines
Is this same "slowdown" also seen in 4.5?
thanks,
greg k-h
Yes, the same issue is seen on 4.5 according to the reporter.

What does it show here when performance drops?
grep . /sys/devices/system/cpu/intel_pstate/*

Is the problem still occurs if you set /sys/class/thermal/thermal_zone*/mode to "disabled"

Thanks,
Srinivas

Quoted 8 lines

Thanks,
Laura
--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Laura Abbott

author

2016-03-18 22:28 UTC #5 lore raw

(bringing this back to the main thread)

Quoted 75 lines · Srinivas

Quoted 66 lines · Laura Abbott
Quoted 65 lines · Greg Kroah-Hartman
Quoted 63 lines · Laura Abbott
Hi,
Fedora received a bug report (https://bugzilla.redhat.com/show_bu
g.cgi?id=1317190)
of a major performance drop on various bench marks and general system
sluggishness with the 4.4.4 kernel update. The benchmarks were showing
a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream. After thermal zone device registered, as we have not read
any
temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]
g>
Reverting this plus to other commits in the series (a67208e94d94 "Thermal: handle thermal zone device properly during system sleep"
and 27f356149d59 "Thermal: do thermal zone update after a cooling device registered") confirmed the performance was back to normal.
Bugzilla has the full discussion but this comment from one of the reporters sums it up:
"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between
1000 and 3400 MHz. In 4.4.4, the processor would fluctuate between
400 and 700 MHz, according to /proc/cpuinfo.
Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to
performance, instead of the default "powersave" forces the CPU to 2400 MHz, and improves performance greatly, but still not to the same level as in 4.4.3."
Any ideas?
Quoted 8 lines
Quoted 6 lines
Is this same "slowdown" also seen in 4.5?
thanks,
greg k-h
Yes, the same issue is seen on 4.5 according to the reporter.

Quoted 9 lines

What does it show here when performance drops?
grep . /sys/devices/system/cpu/intel_pstate/*
Is the problem still occurs if you set /sys/class/thermal/thermal_zone*/mode to "disabled"
Thanks,
Srinivas

A separate thread was started which gave this insight:

"I think
the problem is your device has a passive trip temp of 0 /sys/class/thermal/thermal_zone0/trip_point_2_temp:0 /sys/class/thermal/thermal_zone0/trip_point_2_type:passive

Which triggers a false throttle = true. I think we should this trip as invalid in the case of
if (tz->temperature >= trip_temp) {} check
in thermal_zone_trip_update()."

So would something like the following work?

drivers/thermal/step_wise.c +1 -1

index ea9366a..1228797 100644
@@ -143,7 +143,7 @@ static void thermal_zone_trip_update(struct thermal_zone_device *tz, int trip)
 
        trend = get_tz_trend(tz, trip);
 
-       if (tz->temperature >= trip_temp) {
+       if (trip_type != THERMAL_TRIPS_NONE && tz->temperature >= trip_temp) {
                throttle = true;
                trace_thermal_zone_trip(tz, trip, trip_type);
        }

(completely untested, no idea if I'm even close)

Thanks,
Laura

Pandruvada, Srinivas

2016-03-18 23:36 UTC #6 lore raw

Quoted 127 lines · Laura Abbott

(bringing this back to the main thread)
Quoted 92 lines · Srinivas
Quoted 83 lines · Laura Abbott
Quoted 82 lines · Greg Kroah-Hartman
Quoted 80 lines · Laura Abbott
Hi,
Fedora received a bug report (https://bugzilla.redhat.com/sho
w_bu
g.cgi?id=1317190)
of a major performance drop on various bench marks and general
system
sluggishness with the 4.4.4 kernel update. The benchmarks were
showing
a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
      Thermal: initialize thermal zone device correctly commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream.
      After thermal zone device registered, as we have not read
any
      temperature before, thus tz->temperature should not be 0,
      which actually means 0C, and thermal trend is not available.
      In this case, we need specially handling for the first thermal_zone_device_update().
      Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor
      is the only one that uses trends, so it's the only thermal
      governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <gregkh@linuxfoundati
on.or
g>
Reverting this plus to other commits in the series (a67208e94d94
"Thermal: handle thermal zone device properly during system sleep"
and 27f356149d59 "Thermal: do thermal zone update after a cooling
device registered") confirmed the performance was back to normal.
Bugzilla has the full discussion but this comment from one of the
reporters sums it up:
"In 4.4.3 and prior, my 2.40 MHz processor would fluctuate between
1000 and 3400 MHz.  In 4.4.4, the processor would fluctuate between
400 and 700 MHz, according to /proc/cpuinfo.
Setting /sys/devices/system/cpu/cpufreq/policy0/scaling_governor to
performance, instead of the default "powersave" forces the CPU to
2400 MHz, and improves performance greatly, but still not to the
same level as in 4.4.3."
Any ideas?
Quoted 8 lines
Quoted 6 lines
Is this same "slowdown" also seen in 4.5?
thanks,
greg k-h
Yes, the same issue is seen on 4.5 according to the reporter.
Quoted 9 lines
What does it show here when performance drops?
grep . /sys/devices/system/cpu/intel_pstate/*
Is the problem still occurs if you set /sys/class/thermal/thermal_zone*/mode to "disabled"
Thanks,
Srinivas
A separate thread was started which gave this insight:
"I think
the problem is your device has a passive trip temp of 0 /sys/class/thermal/thermal_zone0/trip_point_2_temp:0 /sys/class/thermal/thermal_zone0/trip_point_2_type:passive
Which triggers a false throttle = true. I think we should this trip as
invalid in the case of
if (tz->temperature >= trip_temp) {} check
in thermal_zone_trip_update()."
So would something like the following work?
drivers/thermal/step_wise.c +0 -0
index ea9366a..1228797 100644
@@ -143,7 +143,7 @@ static void thermal_zone_trip_update(struct
 thermal_zone_device *tz, int trip)

Quoted 6 lines

         trend = get_tz_trend(tz, trip);

-       if (tz->temperature >= trip_temp) {
+       if (trip_type != THERMAL_TRIPS_NONE && tz->temperature >= trip_temp) {

	if (trip_temp && tz->temperature >= trip_temp) {

throttle = true;

Quoted 3 lines

trace_thermal_zone_trip(tz, trip, trip_type); }

I think Rui is working on some change.

Thanks,
Srinivas

Quoted 4 lines

(completely untested, no idea if I'm even close)
Thanks,
Laura

Zhang Rui

reviewer Tested-by Manuel Krause Tested-by szegad Tested-by prash Tested-by amish Tested-by Matthias Reviewed-by Javi Merino

2016-03-19 04:43 UTC #8 lore raw

In reply to Laura Abbott

Hi, all,

Quoted 28 lines · Laura Abbott

Hi,
Fedora received a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly
commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream.
After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.

Quoted 9 lines

Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>

According to https://bugzilla.redhat.com/show_bug.cgi?id=1317190#c30
and https://bugzilla.kernel.org/show_bug.cgi?id=114551,
the root cause of the problem is that
1. BIOS reports bogus passive trip point value, which is 0 degree Celsius
2. Without the thermal patches, thermal core does nothing upon driver probe, thus processor cooling states are not changed.
3. With the thermal patches applied, thermal core checks the new registered thermal zone driver when it is probed, and set the cooling devices to proper state, for example, fan device could be spinning during boot, even if the system is cool.
4. Thus, on these Lenovo laptops, processor cooling devices are throttled because the current temperature (around 50C) is higher than the passive trip point (0C), and the thermal zone is considered as overheating.

In order to workaround this bogus BIOS, we should disable those invalid trip points by checking the trip point value, like the patch below, Laura and Vlad,
can you please try the patch below and confirm if it fixes the problem for you?

Quoted 1 line

From 81ad4276b505e987dd8ebbdf63605f92cd172b52 Mon Sep 17 00:00:00 2001

From: Zhang Rui <[email protected]>
Date: Fri, 18 Mar 2016 10:03:24 +0800
Subject: [PATCH] Thermal: Ignore invalid trip points

In some cases, platform thermal driver may report invalid trip points, thermal core should not take any action for these trip points.

This fixed a regression that bogus trip point starts to screw up thermal control on some Lenovo laptops, after
commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461

Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800

Thermal: initialize thermal zone device correctly

Tested-by Manuel Krause <[email protected]>

Tested-by szegad <[email protected]>

Tested-by prash <[email protected]>

Tested-by amish <[email protected]>

Tested-by Matthias <[email protected]>

Reviewed-by Javi Merino <[email protected]>

Signed-off-by Zhang Rui <[email protected]>

Signed-off-by Chen Yu <[email protected]>

CC <[email protected]> #3.18+

Link https://bugzilla.redhat.com/show_bug.cgi?id=1317190

Link https://bugzilla.kernel.org/show_bug.cgi?id=114551

Signed-off-by Zhang Rui <[email protected]>

---
 drivers/thermal/thermal_core.c | 13 ++++++++++++-
 include/linux/thermal.h        |  2 ++
 2 files changed, 14 insertions(+), 1 deletion(-)

drivers/thermal/thermal_core.c +12 -1

index a0a8fd1..d4b5465 100644
@@ -454,6 +454,10 @@ static void handle_thermal_trip(struct thermal_zone_device *tz, int trip)
 {
 	enum thermal_trip_type type;
 
+	/* Ignore disabled trip points */
+	if (test_bit(trip, &tz->trips_disabled))
+		return;
+
 	tz->ops->get_trip_type(tz, trip, &type);
 
 	if (type == THERMAL_TRIP_CRITICAL || type == THERMAL_TRIP_HOT)

@@ -1800,6 +1804,7 @@ struct thermal_zone_device *thermal_zone_device_register(const char *type,
 {
 	struct thermal_zone_device *tz;
 	enum thermal_trip_type trip_type;
+	int trip_temp;
 	int result;
 	int count;
 	int passive = 0;

@@ -1871,9 +1876,15 @@ struct thermal_zone_device *thermal_zone_device_register(const char *type,
 		goto unregister;
 
 	for (count = 0; count < trips; count++) {
-		tz->ops->get_trip_type(tz, count, &trip_type);
+		if (tz->ops->get_trip_type(tz, count, &trip_type))
+			set_bit(count, &tz->trips_disabled);
 		if (trip_type == THERMAL_TRIP_PASSIVE)
 			passive = 1;
+		if (tz->ops->get_trip_temp(tz, count, &trip_temp))
+			set_bit(count, &tz->trips_disabled);
+		/* Check for bogus trip points */
+		if (trip_temp == 0)
+			set_bit(count, &tz->trips_disabled);
 	}
 
 	if (!passive) {

include/linux/thermal.h +2 -0

index 9c48199..a55d052 100644
@@ -156,6 +156,7 @@ struct thermal_attr {
  * @trip_hyst_attrs:	attributes for trip points for sysfs: trip hysteresis
  * @devdata:	private pointer for device private data
  * @trips:	number of trip points the thermal zone supports
+ * @trips_disabled;	bitmap for disabled trips
  * @passive_delay:	number of milliseconds to wait between polls when
  *			performing passive cooling.
  * @polling_delay:	number of milliseconds to wait between polls when

@@ -191,6 +192,7 @@ struct thermal_zone_device {
 	struct thermal_attr *trip_hyst_attrs;
 	void *devdata;
 	int trips;
+	unsigned long trips_disabled;	/* bitmap for disabled trips */
 	int passive_delay;
 	int polling_delay;
 	int temperature;

Signature · 1 line

1.9.1

Vladyslav Frolov

2016-03-21 11:15 UTC #10 lore raw

In reply to a message not in the archive

Hi Rui,

I'm sorry for the delay; Reading the e-mail the first time I have missed the part about the patch, so I'm building the kernel now.

By the way, the patch reported a slight offset missmatch:

patching file drivers/thermal/thermal_core.c
Hunk #2 succeeded at 1800 (offset -4 lines).
Hunk #3 succeeded at 1872 (offset -4 lines).

Is this fine?

Quoted 198 lines · Vladyslav Frolov

Hi Rui,
I'm sorry for the delay; Reading the e-mail the first time I have missed the part about the patch, so I'm building the kernel now.
By the way, the patch reported a slight offset missmatch:
patching file drivers/thermal/thermal_core.c
Hunk #2 succeeded at 1800 (offset -4 lines).
Hunk #3 succeeded at 1872 (offset -4 lines).
Is this fine?
Quoted 181 lines · Zhang Rui
Hi, all,
Quoted 29 lines · Laura Abbott
Hi,
Fedora received a bug report
(https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly
commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream.
After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Quoted 9 lines
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>
According to https://bugzilla.redhat.com/show_bug.cgi?id=1317190#c30
and https://bugzilla.kernel.org/show_bug.cgi?id=114551,
the root cause of the problem is that
1. BIOS reports bogus passive trip point value, which is 0 degree Celsius
2. Without the thermal patches, thermal core does nothing upon driver probe, thus processor cooling states are not changed.
3. With the thermal patches applied, thermal core checks the new registered thermal zone driver when it is probed, and set the cooling devices to proper state, for example, fan device could be spinning during boot, even if the system is cool.
4. Thus, on these Lenovo laptops, processor cooling devices are throttled because the current temperature (around 50C) is higher than the passive trip point (0C), and the thermal zone is considered as overheating.
In order to workaround this bogus BIOS, we should disable those invalid trip points by checking the trip point value, like the patch below, Laura and Vlad,
can you please try the patch below and confirm if it fixes the problem for you?
From 81ad4276b505e987dd8ebbdf63605f92cd172b52 Mon Sep 17 00:00:00 2001
From: Zhang Rui <[email protected]>
Date: Fri, 18 Mar 2016 10:03:24 +0800
Subject: [PATCH] Thermal: Ignore invalid trip points
In some cases, platform thermal driver may report invalid trip points, thermal core should not take any action for these trip points.
This fixed a regression that bogus trip point starts to screw up thermal control on some Lenovo laptops, after
commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly
After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
CC <[email protected]> #3.18+
Link https://bugzilla.redhat.com/show_bug.cgi?id=1317190
Link https://bugzilla.kernel.org/show_bug.cgi?id=114551
Signed-off-by Zhang Rui <[email protected]>
---
 drivers/thermal/thermal_core.c | 13 ++++++++++++-
 include/linux/thermal.h        |  2 ++
 2 files changed, 14 insertions(+), 1 deletion(-)
drivers/thermal/thermal_core.c +4 -0
index a0a8fd1..d4b5465 100644
@@ -454,6 +454,10 @@ static void handle_thermal_trip(struct
 thermal_zone_device *tz, int trip)
 {
        enum thermal_trip_type type;
 
+       /* Ignore disabled trip points */
+       if (test_bit(trip, &tz->trips_disabled))
+               return;
+
        tz->ops->get_trip_type(tz, trip, &type);
 
        if (type == THERMAL_TRIP_CRITICAL || type == THERMAL_TRIP_HOT)
drivers/thermal/thermal_core.c +1 -0
@@ -1800,6 +1804,7 @@ struct thermal_zone_device
 *thermal_zone_device_register(const char *type,
 {
        struct thermal_zone_device *tz;
        enum thermal_trip_type trip_type;
+       int trip_temp;
        int result;
        int count;
        int passive = 0;
drivers/thermal/thermal_core.c +7 -1
@@ -1871,9 +1876,15 @@ struct thermal_zone_device
 *thermal_zone_device_register(const char *type,
                goto unregister;
 
        for (count = 0; count < trips; count++) {
-               tz->ops->get_trip_type(tz, count, &trip_type);
+               if (tz->ops->get_trip_type(tz, count, &trip_type))
+                       set_bit(count, &tz->trips_disabled);
                if (trip_type == THERMAL_TRIP_PASSIVE)
                        passive = 1;
+               if (tz->ops->get_trip_temp(tz, count, &trip_temp))
+                       set_bit(count, &tz->trips_disabled);
+               /* Check for bogus trip points */
+               if (trip_temp == 0)
+                       set_bit(count, &tz->trips_disabled);
        }
 
        if (!passive) {
include/linux/thermal.h +0 -0
index 9c48199..a55d052 100644
@@ -156,6 +156,7 @@ struct thermal_attr {
  * @trip_hyst_attrs:   attributes for trip points for sysfs: trip
hysteresis
* @devdata: private pointer for device private data
* @trips: number of trip points the thermal zone supports
+ * @trips_disabled;    bitmap for disabled trips
  * @passive_delay:     number of milliseconds to wait between polls when
  *                     performing passive cooling.
  * @polling_delay:     number of milliseconds to wait between polls when
include/linux/thermal.h +1 -1
@@ -191,6 +192,7 @@ struct thermal_zone_device {
        struct thermal_attr *trip_hyst_attrs;
        void *devdata;
        int trips;
+       unsigned long trips_disabled;   /* bitmap for disabled trips */
        int passive_delay;
        int polling_delay;
        int temperature;
--
1.9.1

Vladyslav Frolov

2016-03-21 12:14 UTC #11 lore raw

Hi Rui,

The patch seems to work fine! I have tried both rmmod & modprobe thermal and suspend & resume. Great!

Quoted 213 lines · Vladyslav Frolov

Hi Rui,
I'm sorry for the delay; Reading the e-mail the first time I have missed the part about the patch, so I'm building the kernel now.
By the way, the patch reported a slight offset missmatch:
patching file drivers/thermal/thermal_core.c
Hunk #2 succeeded at 1800 (offset -4 lines).
Hunk #3 succeeded at 1872 (offset -4 lines).
Is this fine?
Quoted 198 lines · Vladyslav Frolov
Hi Rui,
I'm sorry for the delay; Reading the e-mail the first time I have missed the part about the patch, so I'm building the kernel now.
By the way, the patch reported a slight offset missmatch:
patching file drivers/thermal/thermal_core.c
Hunk #2 succeeded at 1800 (offset -4 lines).
Hunk #3 succeeded at 1872 (offset -4 lines).
Is this fine?
Quoted 181 lines · Zhang Rui
Hi, all,
Quoted 29 lines · Laura Abbott
Hi,
Fedora received a bug report
(https://bugzilla.redhat.com/show_bug.cgi?id=1317190)
of a major performance drop on various bench marks and general system sluggishness with the 4.4.4 kernel update. The benchmarks were showing a reduction to about 18% performance (not minor).
Bisection showed the first bad commit was
commit 774ac8b7eff69e0786970157de2157e68b22f456
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly
commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461 upstream.
After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Quoted 9 lines
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
Signed-off-by Greg Kroah-Hartman <[email protected]>
According to https://bugzilla.redhat.com/show_bug.cgi?id=1317190#c30
and https://bugzilla.kernel.org/show_bug.cgi?id=114551,
the root cause of the problem is that
1. BIOS reports bogus passive trip point value, which is 0 degree Celsius
2. Without the thermal patches, thermal core does nothing upon driver probe, thus processor cooling states are not changed.
3. With the thermal patches applied, thermal core checks the new registered thermal zone driver when it is probed, and set the cooling devices to proper state, for example, fan device could be spinning during boot, even if the system is cool.
4. Thus, on these Lenovo laptops, processor cooling devices are throttled because the current temperature (around 50C) is higher than the passive trip point (0C), and the thermal zone is considered as overheating.
In order to workaround this bogus BIOS, we should disable those invalid trip points by checking the trip point value, like the patch below, Laura and Vlad,
can you please try the patch below and confirm if it fixes the problem for you?
From 81ad4276b505e987dd8ebbdf63605f92cd172b52 Mon Sep 17 00:00:00 2001
From: Zhang Rui <[email protected]>
Date: Fri, 18 Mar 2016 10:03:24 +0800
Subject: [PATCH] Thermal: Ignore invalid trip points
In some cases, platform thermal driver may report invalid trip points, thermal core should not take any action for these trip points.
This fixed a regression that bogus trip point starts to screw up thermal control on some Lenovo laptops, after
commit bb431ba26c5cd0a17c941ca6c3a195a3a6d5d461
Author: Zhang Rui <[email protected]>
Date:   Fri Oct 30 16:31:47 2015 +0800
Thermal: initialize thermal zone device correctly
After thermal zone device registered, as we have not read any temperature before, thus tz->temperature should not be 0, which actually means 0C, and thermal trend is not available. In this case, we need specially handling for the first thermal_zone_device_update().
Both thermal core framework and step_wise governor is enhanced to handle this. And since the step_wise governor is the only one that uses trends, so it's the only thermal governor that needs to be updated.
Tested-by Manuel Krause <[email protected]>
Tested-by szegad <[email protected]>
Tested-by prash <[email protected]>
Tested-by amish <[email protected]>
Tested-by Matthias <[email protected]>
Reviewed-by Javi Merino <[email protected]>
Signed-off-by Zhang Rui <[email protected]>
Signed-off-by Chen Yu <[email protected]>
CC <[email protected]> #3.18+
Link https://bugzilla.redhat.com/show_bug.cgi?id=1317190
Link https://bugzilla.kernel.org/show_bug.cgi?id=114551
Signed-off-by Zhang Rui <[email protected]>
---
 drivers/thermal/thermal_core.c | 13 ++++++++++++-
 include/linux/thermal.h        |  2 ++
 2 files changed, 14 insertions(+), 1 deletion(-)
drivers/thermal/thermal_core.c +4 -0
index a0a8fd1..d4b5465 100644
@@ -454,6 +454,10 @@ static void handle_thermal_trip(struct
 thermal_zone_device *tz, int trip)
 {
        enum thermal_trip_type type;
 
+       /* Ignore disabled trip points */
+       if (test_bit(trip, &tz->trips_disabled))
+               return;
+
        tz->ops->get_trip_type(tz, trip, &type);
 
        if (type == THERMAL_TRIP_CRITICAL || type == THERMAL_TRIP_HOT)
drivers/thermal/thermal_core.c +1 -0
@@ -1800,6 +1804,7 @@ struct thermal_zone_device
 *thermal_zone_device_register(const char *type,
 {
        struct thermal_zone_device *tz;
        enum thermal_trip_type trip_type;
+       int trip_temp;
        int result;
        int count;
        int passive = 0;
drivers/thermal/thermal_core.c +7 -1
@@ -1871,9 +1876,15 @@ struct thermal_zone_device
 *thermal_zone_device_register(const char *type,
                goto unregister;
 
        for (count = 0; count < trips; count++) {
-               tz->ops->get_trip_type(tz, count, &trip_type);
+               if (tz->ops->get_trip_type(tz, count, &trip_type))
+                       set_bit(count, &tz->trips_disabled);
                if (trip_type == THERMAL_TRIP_PASSIVE)
                        passive = 1;
+               if (tz->ops->get_trip_temp(tz, count, &trip_temp))
+                       set_bit(count, &tz->trips_disabled);
+               /* Check for bogus trip points */
+               if (trip_temp == 0)
+                       set_bit(count, &tz->trips_disabled);
        }
 
        if (!passive) {
include/linux/thermal.h +0 -0
index 9c48199..a55d052 100644
@@ -156,6 +156,7 @@ struct thermal_attr {
  * @trip_hyst_attrs:   attributes for trip points for sysfs: trip
hysteresis
* @devdata: private pointer for device private data
* @trips: number of trip points the thermal zone supports
+ * @trips_disabled;    bitmap for disabled trips
  * @passive_delay:     number of milliseconds to wait between polls when
  *                     performing passive cooling.
  * @polling_delay:     number of milliseconds to wait between polls when
include/linux/thermal.h +1 -1
@@ -191,6 +192,7 @@ struct thermal_zone_device {
        struct thermal_attr *trip_hyst_attrs;
        void *devdata;
        int trips;
+       unsigned long trips_disabled;   /* bitmap for disabled trips */
        int passive_delay;
        int polling_delay;
        int temperature;
--
1.9.1