On Mon, Apr 27, 2015 at 8:50 AM, Thibaut Verron wrote:

2015-04-27 12:44 GMT+02:00 Jan Matejka :

...

On Sat, 25 Apr 2015 10:47:58 +0200 Thibaut Verron wrote:

...

Does anyone have any clue as to how to investigate (and hopefully fix) the issue?

Your laptop is probably clogged with dust. You need to take it apart and clean it. I need to clean almost all my machines every year or two.

- -- Jan Matějka | QA Engineer for Maintenance SUSE LINUX s.r.o. | https://www.suse.com/ GPG: A33E F5BC A9F6 DAFD 2021 6FB6 3EBF D45B EEB6 CA8B

Hi,

Thank you for your answer.

Although I cannot rule this possibility out with 100% certainty, I find it unlikely that the problem would appear all of a sudden. And it would not explain the inaccurate fan speed reported by the sensors, either.

It's quite common that it happens all of a sudden. Either you notice all of a sudden, or it crosses the threshold from not-a-problem into overheating-problem all of a sudden. The inaccuracies are quite common too, I can usually see them even in the BIOS of pristine machines.

Sadly, I don't have the surgical skills required for taking apart and back together this kind of laptops.

Just get a compressed air can (make sure it's for electronics, otherwise it might have incompatible solvents), make sure your laptop is turned off, and blow some air through the cooler vent. Make sure to hold the can correctly (read the instructions) to avoid propellent to go out as well. That should move dust around enough to make things better. And if it does make things better, then you've got confirmation that dust is to blame. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 You could try your machine with the 13.1 or 13.2 live. If it doesn't heat up, it is not dust. On 2015-04-28 00:08, Thibaut Verron wrote:

...

Just get a compressed air can (make sure it's for electronics, otherwise it might have incompatible solvents), make sure your laptop is turned off, and blow some air through the cooler vent. Make sure to hold the can correctly (read the instructions) to avoid propellent to go out as well. That should move dust around enough to make things better. And if it does make things better, then you've got confirmation that dust is to blame.

I will try it, though, but I don't know when I will have time to drop by the electronics store, so it may take some time.

Around here (Spain), I can get the can at some convenience stores, that we call here Chinese stores. The cans are cheaper and contains less gas, but it suffices. The gas is typically butane or propane, so don't use it in a closed room or near fires. No solvents. It is impossible to compress enough air in a light metal can; it has to be a gas that when compressed at room temperature liquifies. And butane is just it, and cheap. Previously freon was optimal, till it was found how damaging it was to the ozone layer... so it was forbidden. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlU+xCIACgkQja8UbcUWM1wd8QEAmCzwFOPW/pkMWSK0h8R1wmPh 5VlURjMYFkmohNHdsooA/j/bOg5pShQSEj49sCa38Hqe+Bkg45m5WsRNxD8fYyno =Qtrb -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

2015-04-28 9:37 GMT+02:00 ellanios82 :

On 04/28/2015 02:20 AM, Carlos E. R. wrote:

...

And butane is just it, and cheap. Previously freon was optimal, till it was found how damaging it was to the ozone layer... so it was forbidden.

__________

- just a remark :: Electrolux swedish vacuum cleaners used to have a plug , under the air-filter , into which plug one could insert a tube to "blow" { instead of suck}

Don't worry, I know where I can get a can of compressed air, I will just need some time. ;) I found some more information on the issue, while randomly googling. It turns out that not all archives on my past issue are on a now-deleted distribution forum, I also asked a question on superuser : http://superuser.com/q/446610/119172 So apparently I remembered things wrong, the fan sensor has always been misreporting low speeds as 4090-ish rpm. It also appears that the problem I had 3 years ago was really close to the one I have today, except that back then, pwm1_enable was working as expected (and so I could turn the fan to max speed at start, so that I could investigate the problem without the computer shutting down). I should have kept a trace of what I did to fix the issue back then... All that I remember was that it involved booting with different acpi_osi settings, maybe tweaking some values in the /sys/class/hwmon folders, maybe also loading and unloading some modules (there were several competing drivers for eeepc acpi)... Ok, no panic, easy things first, first the dust. Thibaut -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

2015-04-28 13:46 GMT+02:00 Thibaut Verron :

2015-04-28 10:07 GMT+02:00 Thibaut Verron :

...

2015-04-28 9:37 GMT+02:00 ellanios82 :

...

On 04/28/2015 02:20 AM, Carlos E. R. wrote:

...

And butane is just it, and cheap. Previously freon was optimal, till it was found how damaging it was to the ozone layer... so it was forbidden.

__________

- just a remark :: Electrolux swedish vacuum cleaners used to have a plug , under the air-filter , into which plug one could insert a tube to "blow" { instead of suck}

Don't worry, I know where I can get a can of compressed air, I will just need some time. ;)

I found some more information on the issue, while randomly googling. It turns out that not all archives on my past issue are on a now-deleted distribution forum, I also asked a question on superuser : http://superuser.com/q/446610/119172

So apparently I remembered things wrong, the fan sensor has always been misreporting low speeds as 4090-ish rpm. It also appears that the problem I had 3 years ago was really close to the one I have today, except that back then, pwm1_enable was working as expected (and so I could turn the fan to max speed at start, so that I could investigate the problem without the computer shutting down).

I should have kept a trace of what I did to fix the issue back then... All that I remember was that it involved booting with different acpi_osi settings, maybe tweaking some values in the /sys/class/hwmon folders, maybe also loading and unloading some modules (there were several competing drivers for eeepc acpi)...

Ok, no panic, easy things first, first the dust.

Thibaut

Just one more thing I noticed : the last update before the system stopped working properly included an update to grub2, which is (to the best of my knowledge) one of the packages with the power to break this kind of features.

I couldn't manage to revert to 2.02~beta2.35.1, so instead I reinstalled grub2 from stable (2.02~beta2.20.1.5). It didn't help. But since I have already seen non-trivially reversible changes of behavior of this bios, I must ask : was there any change in the package that could affect acpi features? For reference, the update to grub2 was in snapshot 20150421.

Thibaut

So. I bought a can of compressed air, I blowed thoroughly in the input vent, it definitely made some dust go out the output vent (= between the keyboard keys), but the cpu still overheats even when idle. The fan appears to be able to spin at reasonable speeds, since it reports 3900 RPM at some times. Based on previous experience, if the fan sensor is as accurate as before, this report should be taken as exact. However, this fan speed is only reported for a very short amount of time, as if the cooling policy specified to turn the fan to full speed only between 60 and 62°. Thibaut -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

2015-04-29 14:10 GMT+02:00 Jan Matejka :

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512

On Wed, 29 Apr 2015 13:27:55 +0200 Thibaut Verron wrote:

...

So. I bought a can of compressed air, I blowed thoroughly in the input vent, it definitely made some dust go out the output vent (= between the keyboard keys), but the cpu still overheats even when idle.

The fan appears to be able to spin at reasonable speeds, since it reports 3900 RPM at some times. Based on previous experience, if the fan sensor is as accurate as before, this report should be taken as exact. However, this fan speed is only reported for a very short amount of time, as if the cooling policy specified to turn the fan to full speed only between 60 and 62°.

Question is not so much if it doesn't overheat anymore but if it takes longer. I'm not using the compressed air as I find it's pretty much impossible to really clean the parts properly with it. It just doesn't generate enough force. Especially in spaces that are enclosed a lot.

It's just easier to take it apart. You can find service manuals that guide your step by step through the disassembly on the notebook's manufacturer's web pages.

There is no sign that it overheats any slower. To be 100% sure, I should have measured the time before and after "cleaning", but my rough estimations will have to do. As for taking the netbook apart, this is a step I will not take. The compressed can thing was easy to test, and I should have had one of these at hand for a long time anyway. Taking the computer apart will take an afternoon, and I have no certainty that I will be able to recreate a thermal contact as good as the current, or even that I will be able to reassemble the netbook at all. The computer is not entirely functional with this thermal issue, but it is still way more useful than a brick. I know how this whole story looks like, but you have to believe me, I had this kind of issues twice already, and each time it was caused by a software issue (combination of drivers, grub options and bios black magic). Here the problem appeared with a system update, and it has at least one purely software symptom (the fact that I cannot get manual control of the fan, I am not even able to manually stop it). Could we please try to investigate this? Thibaut -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2015-04-29 14:48, Thibaut Verron wrote:

I know how this whole story looks like, but you have to believe me, I had this kind of issues twice already, and each time it was caused by a software issue (combination of drivers, grub options and bios black magic). Here the problem appeared with a system update, and it has at least one purely software symptom (the fact that I cannot get manual control of the fan, I am not even able to manually stop it). Could we please try to investigate this?

Well, I suggested you try running the live of 13.1 or 13.2 to find out. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlVA6yYACgkQja8UbcUWM1xHZQD+NIrY8aW/8nm073inY/9mhiqm oJFUWxdu2QkNn1ptHLcA/1lyoHUPhmdkWgFRAmQbrMZyh8xPT0o/LUuK0iSQwVvy =aCTy -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Am 29.04.2015 um 17:46 schrieb Claudio Freire:

If I were you, I'd report that with your vendor. Seems like a hardware or firmware issue that they should be able or at least try to resolve. Even if it's just dust that the compressed air didn't unclog, they'd be able to disassemble and properly clean it.

This is not useful. No fscking way is ASUS going to fix their BIOS for Linux users. The machine was working before. After a software update (I'd guess it is a kernel problem), it is not working anymore. The heatsink has been cleaned (at least partially), and it did not help at all. Can we now please start investigating the software issue and stop blaming Thibaut? Thibaut: the 13.2-livemedium test would really be useful, just to convince the infidels that it *really* is a software issue :-) I believe you, I have seen such things before. And no, claiming "but the BIOS is broken" will not help. I don't think ASUS is going to fix their crap. -- Stefan Seyfried "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Richard Feynman -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Can we now please start investigating the software issue and stop blaming Thibaut?

No offense taken, it was really the first step to take. But I was somewhat afraid that past the mandatory "have you turned it off and on again" (oops, sorry, "please take out the dust") the thread would die a quick death. Thank you very much for your support.

Thibaut: the 13.2-livemedium test would really be useful, just to convince the infidels that it *really* is a software issue :-)

I am downloading it right now, for some reason I missed it in the first mail from Carlos, sorry.

I forgot: Thibaut, once you confirmed that it works with 13.2 but fails with factory/tumbleweed, proceed directly to the opensuse-kernel list, because that's the place where the people are that are more likely to actually help you.

I like your confidence. :) But thanks for the pointers! Thibaut -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Thu, 2015-04-30 at 00:47 +0200, Carlos E. R. wrote:

On 2015-04-29 22:01, Stefan Seyfried wrote:

...

This is not useful.

No fscking way is ASUS going to fix their BIOS for Linux users.

I understand he said that Windows had the same problem.

That's what I got too, problem is global, but the magnitude changed. -Mike -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Thu, Apr 30, 2015 at 1:32 AM, Claudio Freire wrote:

On Wed, Apr 29, 2015 at 7:47 PM, Carlos E. R. wrote:

...

On 2015-04-29 22:01, Stefan Seyfried wrote:

...

This is not useful.

No fscking way is ASUS going to fix their BIOS for Linux users.

I understand he said that Windows had the same problem.

That's why I think it unlikely treating it as a software bug will help.

Even if it is.

There's no way to test fixes, because supposedly working kernels (windows) still exhibit the issue.

But sure, testing a live media can't hurt. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2015-04-30 10:43, Thibaut Verron wrote:

Just an update on the tests you asked me to run: until now I could not get the live USB to boot (it hangs at some point), and I can't let it heat idle to see what happens, because it reboots after a while. The fan does sound a bit louder, but this is only my inaccurate ear measuring. I had to go to work, I will try troubleshooting why the boot fails this evening, and if I still cannot boot, I'll try to start the live USB with a hot CPU, to see if the fan goes berserk as it should.

I place my laptop on a... "thing" with a largish fan inside, which can plug to one of the USB sockets of the laptop, or to a mains-usb converter (I do the later). It keeps the base of the laptop cool. The internal fan cools well only the CPU. Ram and HD can get quite hot. If I can't do this, I place the laptop on 1 centimeter pegs, not just on a table, so that air can circulate. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlVCJmUACgkQja8UbcUWM1wTRwD+PTjFgOiJL34d3xn66BAmi2cI bWPtgjRDosfZImyEymwA/3Id/lqlXriV5JWsZVHkpu7KnK5GP/J8q+Seyixwhevs =fOHz -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

2015-04-30 10:43 GMT+02:00 Thibaut Verron :

...

On Thu, Apr 30, 2015 at 1:32 AM, Claudio Freire wrote:

...

On Wed, Apr 29, 2015 at 7:47 PM, Carlos E. R. wrote:

...

On 2015-04-29 22:01, Stefan Seyfried wrote:

...

This is not useful.

No fscking way is ASUS going to fix their BIOS for Linux users.

I understand he said that Windows had the same problem.

That's why I think it unlikely treating it as a software bug will help.

Even if it is.

There's no way to test fixes, because supposedly working kernels (windows) still exhibit the issue.

But sure, testing a live media can't hurt. Just an update on the tests you asked me to run: until now I could not get the live USB to boot (it hangs at some point), and I can't let it heat idle to see what happens, because it reboots after a while. The fan does sound a bit louder, but this is only my inaccurate ear measuring. I had to go to work, I will try troubleshooting why the boot fails this evening, and if I still cannot boot, I'll try to start

2015-04-30 6:35 GMT+02:00 Claudio Freire : the live USB with a hot CPU, to see if the fan goes berserk as it should.

Good news, I think! The live USB did not seem to work: it still could not get it to boot, and even if I start it with a hot (>80°C) CPU, the fan doesn't seem to step up. But. I landed on a list of kernel flags, and I tried a couple of them out... just to see? Anyway, it turns out that if I boot my tumbleweed system with "thermal.off=1", I lose the main temperature sensor, and I *can* force the fan to full speed with pwm1_enable=1. The fan speed reading is not any useful, but I can hear that the fan is notably faster. And the remaining temperature readings are stabilized. So, the fan is working at least. After seeing all that, I rebooted (with the default boot flags this time), pwm1_enable was still set to 1 (before that, it was always 2 after a reboot). Again, I could control the fan (stop it or set it to full speed), and this time the fan speed readings seem to be accurate (at least for high speeds). I have now set pwm1_enable to 2 (automatic), I will see how the temperature is in 30-60 minutes. Thibaut -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Ok, thanks for the help guys! Thibaut 2015-04-30 22:52 GMT+02:00 Claudio Freire :

On Thu, Apr 30, 2015 at 5:23 PM, Thibaut Verron wrote:

...

2015-04-30 10:43 GMT+02:00 Thibaut Verron :

...

...

On Thu, Apr 30, 2015 at 1:32 AM, Claudio Freire wrote:

...

On Wed, Apr 29, 2015 at 7:47 PM, Carlos E. R. wrote:

...

On 2015-04-29 22:01, Stefan Seyfried wrote:

> This is not useful. > > No fscking way is ASUS going to fix their BIOS for Linux users.

I understand he said that Windows had the same problem.

That's why I think it unlikely treating it as a software bug will help.

Even if it is.

There's no way to test fixes, because supposedly working kernels (windows) still exhibit the issue.

But sure, testing a live media can't hurt. Just an update on the tests you asked me to run: until now I could not get the live USB to boot (it hangs at some point), and I can't let it heat idle to see what happens, because it reboots after a while. The fan does sound a bit louder, but this is only my inaccurate ear measuring. I had to go to work, I will try troubleshooting why the boot fails this evening, and if I still cannot boot, I'll try to start

2015-04-30 6:35 GMT+02:00 Claudio Freire : the live USB with a hot CPU, to see if the fan goes berserk as it should.

Good news, I think! The live USB did not seem to work: it still could not get it to boot, and even if I start it with a hot (>80°C) CPU, the fan doesn't seem to step up.

But. I landed on a list of kernel flags, and I tried a couple of them out... just to see? Anyway, it turns out that if I boot my tumbleweed system with "thermal.off=1", I lose the main temperature sensor, and I *can* force the fan to full speed with pwm1_enable=1. The fan speed reading is not any useful, but I can hear that the fan is notably faster. And the remaining temperature readings are stabilized.

So, the fan is working at least.

After seeing all that, I rebooted (with the default boot flags this time), pwm1_enable was still set to 1 (before that, it was always 2 after a reboot). Again, I could control the fan (stop it or set it to full speed), and this time the fan speed readings seem to be accurate (at least for high speeds). I have now set pwm1_enable to 2 (automatic), I will see how the temperature is in 30-60 minutes.

Well, you've proven it's a kernel thing. Or a kernel-BIOS interaction.

So I'd suggest trying to get help from the kernel ML mentioned upthread, where the wise guys are. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

2015-04-28 9:19 GMT+02:00 Mike Galbraith :

On Sat, 2015-04-25 at 10:47 +0200, Thibaut Verron wrote:

...

Booting on a previous kernel did not help. If I boot on a different distribution or on windows, the OS does not see the fan speed sensor, and the computer overheats slowly all the same. But as I said, I already had similar problems with this computer (the fan was not spinning at all, then), and I remember that it showed across distributions in the same way. If I remember, I concluded back then that something had set a bios flag that prevented the bios from handling the fan properly for other OSes.

Hm. On top of what others have said about crud maybe plugging things up, this paragraph suggests to me that your laptop may have crap for bios, that you should perhaps look for an update.

-Mike

Indeed that's something I hadn't thought of. But I updated that bios the first time I had an issue with the fans (back then, it didn't help, or at least not directly), and there has been no new version since then. Thibaut -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org