Bug 153281

When running with nouveau driver, psensor shows GPU temperature 1C and fans are switching on on higher load and temperatures, when CPU is cooled down, fans are switching off correctly, see this screenshot
http://imgur.com/RCTzmnc
Once installed binary driver (v 370.23) the gpu temperature (nvidia or intel) is shown correctly, but the fans are never switched off even when the CPU/GPU temperature is dropped down to about +33C.
http://imgur.com/krxeKNU
If you need any additional info I'll be glad to provide it.

Created attachment 229991 [details]
Fans are switched off correctly on nouveau

Fans are switched off correctly on nouveau, but GPU temperature claims to be +1C

please attach the acpidump output.
please attach the output of "grep . /sys/class/thermal/*/*"

(In reply to Peter from comment #1)
> When running with nouveau driver, psensor shows GPU temperature 1C and fans
> are switching on on higher load and temperatures, when CPU is cooled down,
> fans are switching off correctly, see this screenshot
> http://imgur.com/RCTzmnc
> Once installed binary driver (v 370.23) the gpu temperature (nvidia or
> intel) is shown correctly, but the fans are never switched off even when the
> CPU/GPU temperature is dropped down to about +33C.
> http://imgur.com/krxeKNU
> If you need any additional info I'll be glad to provide it.

This is confusing, I didn't get this info from the original report.
I didn't find anyone mentioning binary driver in https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1602888

Created attachment 230051 [details]
acpidump Dell Inspiron 7559

Created attachment 230061 [details]
thermal Dell Inspiron 7559

(In reply to Zhang Rui from comment #4)
> (In reply to Peter from comment #1)
> > When running with nouveau driver, psensor shows GPU temperature 1C and fans
> > are switching on on higher load and temperatures, when CPU is cooled down,
> > fans are switching off correctly, see this screenshot
> > http://imgur.com/RCTzmnc
> > Once installed binary driver (v 370.23) the gpu temperature (nvidia or
> > intel) is shown correctly, but the fans are never switched off even when
> the
> > CPU/GPU temperature is dropped down to about +33C.
> > http://imgur.com/krxeKNU
> > If you need any additional info I'll be glad to provide it.
> 
> This is confusing, I didn't get this info from the original report.
> I didn't find anyone mentioning binary driver in
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1602888

It was not created by me on launchpad, I've created this bug here because it really bothers me this fan issue on my laptop and on launchpad seems no one cares, also there were suggested to me to create it here on other forum.

(In reply to Peter from comment #7)
> (In reply to Zhang Rui from comment #4)
> > (In reply to Peter from comment #1)
> > > When running with nouveau driver, psensor shows GPU temperature 1C and
> fans
> > > are switching on on higher load and temperatures, when CPU is cooled
> down,
> > > fans are switching off correctly, see this screenshot
> > > http://imgur.com/RCTzmnc
> > > Once installed binary driver (v 370.23) the gpu temperature (nvidia or
> > > intel) is shown correctly, but the fans are never switched off even when
> the
> > > CPU/GPU temperature is dropped down to about +33C.
> > > http://imgur.com/krxeKNU
> > > If you need any additional info I'll be glad to provide it.
> > 
> > This is confusing, I didn't get this info from the original report.
> > I didn't find anyone mentioning binary driver in
> > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1602888
> 
> It was not created by me on launchpad, I've created this bug here because it
> really bothers me this fan issue on my laptop and on launchpad seems no one
> cares, also there were suggested to me to create it here on other forum.

So are you using binary driver?
If the problem only exists with binary driver loaded, I'm sorry I can not help you on this, because we only handle problems that can be reproduced with latest vanilla kernel.

(In reply to Zhang Rui from comment #8)
> (In reply to Peter from comment #7)
> > (In reply to Zhang Rui from comment #4)
> > > (In reply to Peter from comment #1)
> > > > When running with nouveau driver, psensor shows GPU temperature 1C and
> fans
> > > > are switching on on higher load and temperatures, when CPU is cooled
> down,
> > > > fans are switching off correctly, see this screenshot
> > > > http://imgur.com/RCTzmnc
> > > > Once installed binary driver (v 370.23) the gpu temperature (nvidia or
> > > > intel) is shown correctly, but the fans are never switched off even
> when the
> > > > CPU/GPU temperature is dropped down to about +33C.
> > > > http://imgur.com/krxeKNU
> > > > If you need any additional info I'll be glad to provide it.
> > > 
> > > This is confusing, I didn't get this info from the original report.
> > > I didn't find anyone mentioning binary driver in
> > > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1602888
> > 
> > It was not created by me on launchpad, I've created this bug here because
> it
> > really bothers me this fan issue on my laptop and on launchpad seems no one
> > cares, also there were suggested to me to create it here on other forum.
> 
> So are you using binary driver?
> If the problem only exists with binary driver loaded, I'm sorry I can not
> help you on this, because we only handle problems that can be reproduced
> with latest vanilla kernel.

It is with binary driver but switched to intel only gpu using nvidia-prime with afaik kernel driver. I wouldn't care if fans work endless on nvidia gpu, but they also work endless on indel gpu too.

(In reply to Peter from comment #9)
> It is with binary driver but switched to intel only gpu using nvidia-prime
> with afaik kernel driver. I wouldn't care if fans work endless on nvidia
> gpu, but they also work endless on indel gpu too.

Good to know. But this seems to be conflict with your description below

(In reply to Peter from comment #1)
> When running with nouveau driver, psensor shows GPU temperature 1C and fans
> are switching on on higher load and temperatures, when CPU is cooled down,
> fans are switching off correctly, see this screenshot
> http://imgur.com/RCTzmnc
> Once installed binary driver (v 370.23) the gpu temperature (nvidia or
> intel) is shown correctly, but the fans are never switched off even when the
> CPU/GPU temperature is dropped down to about +33C.
> http://imgur.com/krxeKNU

My understanding of this is
When running with nouveau driver, when CPU is cooled down, fans are switching off correctly
Once installed binary driver, the fans are never switched off even when the
CPU/GPU temperature is dropped down to about +33C.

Can you please clarify?

Created attachment 230201 [details]
reproduced fans with nomdeset

I've booted with nomodeset parameter and reproduced the behavior without usage proprietary driver (I'd say that on CPU/GPU temp at about +30C should switch off the fans).

Hello,

I am the person that filed the previously mentioned launchpad.net bug report. To clarify, it does not matter what driver is in use - nouveau, nvidia, intel, the fans will not shut off once started, even when CPU temp has reached 30c  (at least this is the case with my machine)

Problem still exists in RC4

Problem still exists in RC5

Just wanted to add that I too, have this same issue as described above.  Again, proprietary or open source driver, makes no difference.  Tried many different distributions as well; they all have the same fan issue.

please attach the output of "grep . /sys/class/thermal/*/*" in both good and bad kernel.

I'm not sure what you mean by a 'good kernel'.  I'm on 4.7.2 now and I have been unable to get the fan to work correctly under any kernel up to and including 4.7.2.  I'm using Arch as well, so the kernel shouldn't have any extra patches added.

hmm, according to previous description of the problem, the fan sometimes spin while sometimes not.
So I need to check and compare the thermal status (grep . /sys/class/thermal/*/*) in both cases.

Ok. I can give you the termal before the fans spin and when they are on constantly.

Created attachment 232511 [details]
Correct fan behavior

Correct fan behavior on a freshly booted system.

Created attachment 232521 [details]
Bad fan behavior

Bad fan behavior after the fans have turned on and won't go off after about 30-45 minutes of being on.

Created attachment 232531 [details]
Fans off after boot

Output from grep . /sys/class/thermal/*/*
when the fans are off after boot

Created attachment 232541 [details]
Fans on on System load

Output from grep . /sys/class/thermal/*/* after fans are on on system load

Created attachment 232551 [details]
Fans on on cooled system

Output from grep . /sys/class/thermal/*/* when the fans are still on but the system is cooled down and the fans should be off too but they are not

(In reply to John from comment #21)
> Created attachment 232521 [details]
> Bad fan behavior
> 
> Bad fan behavior after the fans have turned on and won't go off after about
> 30-45 minutes of being on.

/sys/class/thermal/cooling_device10/cur_state:0
/sys/class/thermal/cooling_device10/max_state:1
/sys/class/thermal/cooling_device10/type:Fan
/sys/class/thermal/cooling_device11/cur_state:0
/sys/class/thermal/cooling_device11/max_state:1
/sys/class/thermal/cooling_device11/type:Fan
/sys/class/thermal/cooling_device12/cur_state:0
/sys/class/thermal/cooling_device12/max_state:1
/sys/class/thermal/cooling_device12/type:Fan
/sys/class/thermal/cooling_device8/cur_state:0
/sys/class/thermal/cooling_device8/max_state:1
/sys/class/thermal/cooling_device8/type:Fan
/sys/class/thermal/cooling_device9/cur_state:0
/sys/class/thermal/cooling_device9/max_state:1
/sys/class/thermal/cooling_device9/type:Fan

(In reply to Peter from comment #23)
> Created attachment 232541 [details]
> Fans on on System load
> 
> Output from grep . /sys/class/thermal/*/* after fans are on on system load

/sys/class/thermal/cooling_device0/cur_state:0
/sys/class/thermal/cooling_device0/max_state:3
/sys/class/thermal/cooling_device0/type:Processor
/sys/class/thermal/cooling_device10/cur_state:0
/sys/class/thermal/cooling_device10/max_state:1
/sys/class/thermal/cooling_device10/type:Fan
/sys/class/thermal/cooling_device11/cur_state:0
/sys/class/thermal/cooling_device11/max_state:1
/sys/class/thermal/cooling_device11/type:Fan
/sys/class/thermal/cooling_device12/cur_state:0
/sys/class/thermal/cooling_device12/max_state:1
/sys/class/thermal/cooling_device12/type:Fan
/sys/class/thermal/cooling_device8/cur_state:0
/sys/class/thermal/cooling_device8/max_state:1
/sys/class/thermal/cooling_device8/type:Fan
/sys/class/thermal/cooling_device9/cur_state:0
/sys/class/thermal/cooling_device9/max_state:1
/sys/class/thermal/cooling_device9/type:Fan

(In reply to Peter from comment #24)
> Created attachment 232551 [details]
> Fans on on cooled system
> 
> Output from grep . /sys/class/thermal/*/* when the fans are still on but the
> system is cooled down and the fans should be off too but they are not

/sys/class/thermal/cooling_device10/cur_state:0
/sys/class/thermal/cooling_device10/max_state:1
/sys/class/thermal/cooling_device10/type:Fan
/sys/class/thermal/cooling_device11/cur_state:0
/sys/class/thermal/cooling_device11/max_state:1
/sys/class/thermal/cooling_device11/type:Fan
/sys/class/thermal/cooling_device12/cur_state:0
/sys/class/thermal/cooling_device12/max_state:1
/sys/class/thermal/cooling_device12/type:Fan
/sys/class/thermal/cooling_device8/cur_state:0
/sys/class/thermal/cooling_device8/max_state:1
/sys/class/thermal/cooling_device8/type:Fan
/sys/class/thermal/cooling_device9/cur_state:0
/sys/class/thermal/cooling_device9/max_state:1
/sys/class/thermal/cooling_device9/type:Fan

it seems that the fan is not controlled by ACPI.

can you please check if the fan can be spinned on/off by setting cur_state to 1/0, via the thermal sysfs I/F above?

(In reply to Zhang Rui from comment #28)
> it seems that the fan is not controlled by ACPI.
> 
> can you please check if the fan can be spinned on/off by setting cur_state
> to 1/0, via the thermal sysfs I/F above?

Setting cur_state to 1/0 does nothing for all the cooling_device sysfs I/F.

Then it seems that windows is using something else to control the fan, say, vendor specific drivers.
reassign to platform driver experts to see if this can be resolved in vendor specific driver.

Would it be possible to test Linux 4.8-rc1 or newer using the nouveau driver (no proprietary drivers or bbswitch please). There have been reports that use of the alternative power management method (ACPI power resources instead of DSM) fixes the fan issue.

Created attachment 232651 [details]
Fans behavior with nouveau

On Kernel 4.8-RC5, with nouveau driver when I force to make CPU load using:
dd if=/dev/zero of=/dev/null
I could observe a correct fans behavior for a while (fans were switching off) but then after a while the fans started to spin and now they are spinning like forever.
Also GPU temperature is always +1C.
I should mention that I'd rather to run system on Intel GPU with kernel driver.

(In reply to Peter from comment #32)
> Created attachment 232651 [details]
> Fans behavior with nouveau
> 
> On Kernel 4.8-RC5, with nouveau driver when I force to make CPU load using:
> dd if=/dev/zero of=/dev/null
> I could observe a correct fans behavior for a while (fans were switching
> off) but then after a while the fans started to spin and now they are
> spinning like forever.
> Also GPU temperature is always +1C.
> I should mention that I'd rather to run system on Intel GPU with kernel
> driver.

Same behavior I got using proprietary nvidia driver switched to Intel GPU using nvidia-prime.
So forced CPU load spins the fans, when interrupted fans off, but after a while fans are on and spinning forever CPU/GPU temperatures are at about +33C.

Please attack your dmesg (for nouveau, not the proprietary driver). If you see something going wacky after runtime resume, try the acpi_osi workaround documented for your laptop model at
https://github.com/Bumblebee-Project/Bumblebee/issues/764#issuecomment-234494238

If that solves the issue, your new problem is bug 156341.

Created attachment 232681 [details]
dmesg

dmesg on 4.8-RC5, no problems observed with kernel and nouveau so far

Problem with fans still persists on 4.8.0-040800rc6-generic #201609121119 using nouveau driver. Please note that acpi_osi=! caused that gpu temp. seems to be shown correctly in psensor whereas without it, the gpu temp is shown as 1C.

Does this bug still affect the version that comes pre-installed with ubuntu?

If not, I may have access to one of these through a friend. Is there anything I can do to help?

(In reply to pedrohjordao from comment #37)
> Does this bug still affect the version that comes pre-installed with ubuntu?
> 
> If not, I may have access to one of these through a friend. Is there
> anything I can do to help?

Personally I do not know. If there is a possibility to download Ubuntu 14.04 with drivers included drivers for Dell Inspiron 7559 I am willing to install and try it. What I know for sure about this problem is, that it lasts (in my case) since I've bought this laptop and installed Ubuntu 15.10 on it.

I also have a friend with the Dell Inspiron 7557 and managed to solve the issues with Linux 4.8.

Addition of the acpi_osi="!Windows 2015" to the kernel command line fixed lockups and suspend failures when nouveau is in use. (Or similarly, bbswitch.)

To fix the fan issue, kernel 4.8 and nouveau is required (bbswitch is not yet updated). You can grab the current stable Ubuntu kernel from
http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.8/

Created attachment 240851 [details]
Kernel 4.8 fans still on

Kernel parameter acpi_osi="!Windows 2015" did not help so the fans are still spinning, also with this parameter the GPU temp. is shown as +1C.

Have you tried using acpi_osi="..." Instead of "acpi_osi=..."?

Can you also upload a full dmesg? If nvidia or bbswitch modules are loaded, retry without them. If it matters, do a power off first instead of a normal reboot.

Created attachment 240871 [details]
dmesg output

I've used acpi_osi="!Windows 2015" - exactly as written with doubleqoutes, you can find it in dmesg output. I've found 2 weird things:
- lmsensors says that GPU temperature is still +1C (when using nouveau driver)
- once I've forced to load CPU the fans were started to spin and after a while they were switched off (it behaved sort off correctly), but after 3 attempts the fans started to spin and now they are spinning forever even when the CPU is now cooled (please see my next attachment)
So it seems like the fans are controlled somehow, but not correctly.

Created attachment 240881 [details]
semi - controlled fans

Here you can see that fans were controlled for a while somehow after the cpu was cooled after load, but then they started to spin and now are spinning forever...

Do not trust the GPU sensor when the GPU is turned off, if the hardware is off then you cannot query it for its temperature.

Do you have additional messages in your dmesg after the fan state changes?

Also, if you have specific steps to reproduce then I could try to repeat it with my friend. Preferably try to reproduce it with no GUI running, but in a terminal (to rule out other background activity).

(In reply to Peter Wu from comment #44)
> Do not trust the GPU sensor when the GPU is turned off, if the hardware is
> off then you cannot query it for its temperature.
> 
> Do you have additional messages in your dmesg after the fan state changes?
> 
> Also, if you have specific steps to reproduce then I could try to repeat it
> with my friend. Preferably try to reproduce it with no GUI running, but in a
> terminal (to rule out other background activity). 

peter@peter-Inspiron-7559 ~ $ diff dmesg-FansOff.txt dmesg-fansRun.txt 
1007a1008,1032
> [  364.851820] [drm:intel_pipe_update_end [i915]] *ERROR* Atomic update
> failure on pipe A (start=24540 end=24547) time 116536 us, min 1073, max 1079,
> scanline start 301, end 294
> [  457.718730] SGI XFS with ACLs, security attributes, realtime, no debug
> enabled
> [  457.736306] JFS: nTxBlock = 8192, nTxLock = 65536
> [  457.754707] ntfs: driver 2.1.32 [Flags: R/O MODULE].
> [  457.783206] QNX4 filesystem 0.2.3 registered.
> [  458.799716] FAT-fs (sdb3): invalid media value (0xc7)
> [  458.799724] FAT-fs (sdb3): Can't find a valid FAT filesystem
> [  458.805930] EXT4-fs (sdb3): VFS: Can't find ext4 filesystem
> [  458.809051] EXT4-fs (sdb3): VFS: Can't find ext4 filesystem
> [  458.812081] EXT4-fs (sdb3): VFS: Can't find ext4 filesystem
> [  458.825894] XFS (sdb3): Invalid superblock magic number
> [  458.831466] FAT-fs (sdb3): invalid media value (0xc7)
> [  458.831473] FAT-fs (sdb3): Can't find a valid FAT filesystem
> [  458.839230] VFS: Can't find a Minix filesystem V1 | V2 | V3 on device
> sdb3.
> [  458.846491] hfsplus: unable to find HFS+ superblock
> [  458.849395] qnx4: no qnx4 filesystem (no root dir).
> [  458.851959] ufs: You didn't specify the type of your ufs filesystem
>                
>                mount -t ufs -o
>                ufstype=sun|sunx86|44bsd|ufs2|5xbsd|old|hp|nextstep|nextstep-cd|openstep
>                ...
>                
>                >>>WARNING<<< Wrong ufstype may corrupt your filesystem,
>                default is ufstype=old
> [  458.852179] ufs: ufs_fill_super(): bad magic number
> [  458.855036] hfs: can't find a HFS filesystem on dev sdb3
> [  982.532098] [drm:intel_pipe_update_end [i915]] *ERROR* Atomic update
> failure on pipe A (start=61608 end=61615) time 116654 us, min 1073, max 1079,
> scanline start 389, end 390
> [ 1089.540709] [drm:intel_pipe_update_end [i915]] *ERROR* Atomic update
> failure on pipe A (start=68030 end=68037) time 116561 us, min 1073, max 1079,
> scanline start 150, end 145

I'll try to reproduce the behavior it on init 3

OK on init 3, I've loaded CPU with multiple dd if=/dev/zero of=/dev/null commands and after it heated up the fans were started to spin, after canceling the CPU load the fans stopped to spin. I've tried this couple ( > 5) times in row
So now the question is, why the fans are not behaving correctly when X is on?

Can you try to isolate the component? I would suspect the tool you use for reading sensors.

(In reply to Peter Wu from comment #47)
> Can you try to isolate the component? I would suspect the tool you use for
> reading sensors.

definitely not the psensor, I've removed this app and problem still occurs on gui

What I've found is: no matter nouveau or nvidia binary driver or GUI or text mode only or acpi_osi parameter in grub, if I load the cpu after fresh boot the fans are started to spin and after idling it, the fans stop, but then after a while (in my case during kernel 4.8.1 installation) the fans started to spin and are spinning forever. So something has to mess the acpi or sensor or whatever what reads temperatures to stop the fans.
After powering off and booting to kernel 4.8.1 the behavior is the same, but I've removed 4.8 to kick off the fans. Then they are spinning now again and CPU temp. is at +28C.
If I put the laptop to sleep and then awake immediately fans are remaining stopped until the next load (This is not very nice workaround to stop them when cpu is cool).

I cannot reproduce it. Friend has been using this kernel for almost a week now and did not notice fan issues.

Tested dd if=/dev/zero of=/dev/null such that all 8 cores are loaded, the fan indeed goes up with temperatures reaching over 80 degrees Celsius, but after killall dd the fan goes down quite fast. (I also saw the dell_smm sensor reporting 94 degree Celsius, but it also dropped quickly after stopping).

His laptop has BIOS 1.1.5, 01/14/2016 and is running Elementary OS (Loki) with kernel 4.8 from http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.8/. nouveau is in use with acpi_osi="!Windows 2015".

Tried to downgrade to BIOS 1.1.5 with fresh install Ubuntu 16.10 Beta 2 (kernel 4.8) with nouveau driver. Yes, after forcing to load CPU it kicks off the fans and after killall load processes the fans are off. But then after using it a while (I was optimistic too soon) the fans were kicked off by themselves at about +53C and they were running "forever" - without turning them off, whereas on Windows the fans are working correctly. I'm getting tired of this laptop and I've regret buying it so many times - and I'll still do until I get rid off it.

Can you check your dmesg for any runtime suspend/resume events from nouveau? I have noticed on my own laptop that opening the KDE PDF viewer (Okular) somehows kicks the video card briefly (maybe it does a RandR request?). This seems to briefly spin the fan up, but then it goes down quickly.

Are you able to reproduce the issue with other distributions/desktop environments (e.g. on Elementary OS (Loki)? That is what my friend uses and where he could not reproduce the issue.

I am also not seeing much improvement with the latest mainline kernel on the Inspiron 7559.  Same configuration as peter, Ubuntu 16.04,  4.8.1-040801-generic, nouveau driver.  No real improvement. Fan issue seems the same to me.  

A couple things to note with this fan issue which makes the bug hard to observe.  

1. As peter mentions, when the laptop is "cool"  (i.e it has been powered off a while) then the fans actually will shut off on their own for a brief period of time.  Maybe within the first 15 minutes of being cold booted.  However, after the laptop has been running for a while,  the fan issue rears its ugly head.  

2. I have observed that IF the ambient temp is VERY cold, then the fans will shut off on their own.  I went so far as to pop the bottom of the laptop off, and blowing cool air in the case, and the fans actually shut off.  Of course once I stopped blowing air into the laptop, they started right back up.  The laptop was -not- hot.  CPU temp around 28c.

3.  When the fans are spinning, its at the lowest speed, which is really quiet on this laptop.  In a room with a lot of ambient noise, you might not even be aware they are running.  Perhaps this is the case with your friend?

Also, I have observed the same behaviour on Arch Linux using Budgie, i3, KDE and Mate desktops personally, so I doubt is related to what desktop or window manager you are using.

Forgot to mention that I'm not even using the Nvidia card at the moment, just the integrated Intel card, so I don't have nouveau or nvidia Xorg drivers installed at all, just the intel driver.  I also tried downgrading to BIOS 1.1.5 and i run with the kernel parameter 'acpi_osi="!Windows 2015"' as well and have noticed the fan issue all the time.  The only thing that will clear it out for me is a complete shutdown or a suspend/resume cycle.

Downgraded BIOS to 1.1.5 and fresh install of Elementary OS Loki with 4.8.1 kernel and I have same behavior as Bob Tanner describes. Fans are managed well for a while, but then after couple minutes on idle, fans were kicked on by themselves and spinning forever. It is true that the fans are not very noisy on lowest speed, but when you watch movie in the evening or just read the web, it disturbs quite a lot, however on Windows fans are managed nicely without disturbtion.

I should also to mention that the persistent fans are draining battery much faster.

Perhaps we need someone from Dell to look at this? Dell 7559 got Ubuntu certified pre-install status. Not sure whether this problem exists in that pre-installed image, anyway it would be nice if Dell could provide all needed patches for including into upstream.
May it be that this problem is just caused by the current situation when Skylake is not getting into deeper power states?

(In reply to yourbestfriend from comment #58)
> Perhaps we need someone from Dell to look at this? Dell 7559 got Ubuntu
> certified pre-install status. Not sure whether this problem exists in that
> pre-installed image, anyway it would be nice if Dell could provide all
> needed patches for including into upstream.
> May it be that this problem is just caused by the current situation when
> Skylake is not getting into deeper power states?

Dell engineers are pretty good ignorants as I've asked for this topic quite long time ago. If anybody can escalate this to the dell engineers would be nice.

http://en.community.dell.com/support-forums/software-os/f/3525/t/19982814

I've also not had any luck with Dell support.  They seem amazingly clueless about pretty much everything. 

I've looked around but I've never found a way to actually report a bug to Dell.  Anyone know if any kind of bug reporting mechanism exists for Dell?  If such a thing exists, I'll happily open a bug report with them.

(In reply to Bob Tanner from comment #60)
> I've also not had any luck with Dell support.  They seem amazingly clueless
> about pretty much everything. 
> 
> I've looked around but I've never found a way to actually report a bug to
> Dell.  Anyone know if any kind of bug reporting mechanism exists for Dell? 
> If such a thing exists, I'll happily open a bug report with them.

Maybe just open a support ticket?  I don't think they have a public bugtracker.

Unfortunately I've tried and found their consumer tech support to be an exercise in futility.  Agents read from a script, only deal with windows, etc.

I guess there is no way to get in touch with actual Dell engineers. E.g, the ones who deal with BIOS issues.

I wonder whether this doesn't allow to go into deeper package c-states

$ dmesg | grep -I aspm
[ 0.000000] Kernel command line: initrd=\efi\nixos\f74mgfd5d4k9n154p0ffs5r0d3zazv3n-initrd-initrd.efi systemConfig=/nix/store/50044hcm6lf7fc48nlikfpbap4zjp54n-nixos-system-nixos-16.09.773.b8ede35 init=/nix/store/50044hcm6lf7fc48nlikfpbap4zjp54n-nixos-system-nixos-16.09.773.b8ede35/init loglevel=4 *pcie_aspm=force* nouveau.modeset=0
[ 0.000000] PCIe ASPM is forcibly enabled
[ 0.285133] ACPI FADT declares the system doesn't support PCIe ASPM, so disable it
[ 0.398009] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.398040] acpi PNP0A08:00: _OSC failed (AE_ERROR); *disabling ASPM*

So it seems as though we are just stuck with this problem.

Luckily fans are not that loud :-)
Regarding non-working ASPM, I've created a separate bug request: https://bugzilla.kernel.org/show_bug.cgi?id=183881
Tables as seen under Windows and under Linux are different.

Created attachment 244431 [details]
psensor reading at BIOS fan shutdown

requires opening laptop bottom cover and holding against external fan to trigger fan shutdown

* attachment 244431 [details] - for Dell 7559 i7-6700HQ

Created attachment 246771 [details]
Fans seems working correctly!

I do not know what happened, but fans are now working correctly using this setup:

- BIOS 1.2.0
- Kernel 4.8.11 or higher
- nvidia driver (doesn't matter whether nouveau or nvidia binary in my case 375.20) and I am running on Intel GPU by default using bumblebee to switch to nvidia
- grub boot argument: acpi_osi="!Windows 2015"
- For powertop - see my attached powertop - Tunables setting
I had to set tunables manually to Good and now the fans are working correctly. Problem is, after reboot the tunbales setting are gone and I have to re-set them (I've created init script to do that)

(In reply to Peter from comment #68)
> Created attachment 246771 [details]
> Fans seems working correctly!
> 
> I do not know what happened, but fans are now working correctly using this
> setup:
> 
> - BIOS 1.2.0
> - Kernel 4.8.11 or higher
> - nvidia driver (doesn't matter whether nouveau or nvidia binary in my case
> 375.20) and I am running on Intel GPU by default using bumblebee to switch
> to nvidia
> - grub boot argument: acpi_osi="!Windows 2015"
> - For powertop - see my attached powertop - Tunables setting
> I had to set tunables manually to Good and now the fans are working
> correctly. Problem is, after reboot the tunbales setting are gone and I have
> to re-set them (I've created init script to do that)

Are you sure?  I tried those settings and it still doesn't seem to be operating normally to me.

(In reply to John from comment #69)
> (In reply to Peter from comment #68)
> > Created attachment 246771 [details]
> > Fans seems working correctly!
> > 
> > I do not know what happened, but fans are now working correctly using this
> > setup:
> > 
> > - BIOS 1.2.0
> > - Kernel 4.8.11 or higher
> > - nvidia driver (doesn't matter whether nouveau or nvidia binary in my case
> > 375.20) and I am running on Intel GPU by default using bumblebee to switch
> > to nvidia
> > - grub boot argument: acpi_osi="!Windows 2015"
> > - For powertop - see my attached powertop - Tunables setting
> > I had to set tunables manually to Good and now the fans are working
> > correctly. Problem is, after reboot the tunbales setting are gone and I
> have
> > to re-set them (I've created init script to do that)
> 
> Are you sure?  I tried those settings and it still doesn't seem to be
> operating normally to me.

Did you set Tunables to good (according my attached setting) in powertop? I've figured out that this is the crucial thing to do.

(In reply to Peter from comment #70)
> (In reply to John from comment #69)
> > (In reply to Peter from comment #68)
> > > Created attachment 246771 [details]
> > > Fans seems working correctly!
> > > 
> > > I do not know what happened, but fans are now working correctly using
> this
> > > setup:
> > > 
> > > - BIOS 1.2.0
> > > - Kernel 4.8.11 or higher
> > > - nvidia driver (doesn't matter whether nouveau or nvidia binary in my
> case
> > > 375.20) and I am running on Intel GPU by default using bumblebee to
> switch
> > > to nvidia
> > > - grub boot argument: acpi_osi="!Windows 2015"
> > > - For powertop - see my attached powertop - Tunables setting
> > > I had to set tunables manually to Good and now the fans are working
> > > correctly. Problem is, after reboot the tunbales setting are gone and I
> have
> > > to re-set them (I've created init script to do that)
> > 
> > Are you sure?  I tried those settings and it still doesn't seem to be
> > operating normally to me.
> 
> Did you set Tunables to good (according my attached setting) in powertop?
> I've figured out that this is the crucial thing to do.

I don't use Powertop, I use TLP.  TLP sets all that stuff automatically.  As a matter of fact, if I go into Powertop, it shows everything is optimal.

What about if you are on AC?  Do the fans run properly there as well?

(In reply to John from comment #71)
> (In reply to Peter from comment #70)
> > (In reply to John from comment #69)
> > > (In reply to Peter from comment #68)
> > > > Created attachment 246771 [details]
> > > > Fans seems working correctly!
> > > > 
> > > > I do not know what happened, but fans are now working correctly using
> this
> > > > setup:
> > > > 
> > > > - BIOS 1.2.0
> > > > - Kernel 4.8.11 or higher
> > > > - nvidia driver (doesn't matter whether nouveau or nvidia binary in my
> case
> > > > 375.20) and I am running on Intel GPU by default using bumblebee to
> switch
> > > > to nvidia
> > > > - grub boot argument: acpi_osi="!Windows 2015"
> > > > - For powertop - see my attached powertop - Tunables setting
> > > > I had to set tunables manually to Good and now the fans are working
> > > > correctly. Problem is, after reboot the tunbales setting are gone and I
> have
> > > > to re-set them (I've created init script to do that)
> > > 
> > > Are you sure?  I tried those settings and it still doesn't seem to be
> > > operating normally to me.
> > 
> > Did you set Tunables to good (according my attached setting) in powertop?
> > I've figured out that this is the crucial thing to do.
> 
> I don't use Powertop, I use TLP.  TLP sets all that stuff automatically.  As
> a matter of fact, if I go into Powertop, it shows everything is optimal.
> 
> What about if you are on AC?  Do the fans run properly there as well?

I'm running on AC and the fans running properly. Afaik TLP is for Lenovo Thinkpad laptops, I've opened this bug for Dell Inspiron 7559 model.

No, TLP is not just for Lenovo laptops, it can be used on any laptop.  It has battery charge threshold components for Thinkpads, but those aren't required.  I'm using it with my Dell 7559 and it works fine.

Anyway, I turned off TLP and just went straight to Powertop and my fans are still just going crazy.

Created attachment 246921 [details]
Fans seems working properly!

@John: You have to wait a bit to system cool down and then the fans should turn off. As you can see on my screenshot fans got turned off after system is cooled down. I am monitoring the system for more than 5 days and I have to say that the CPU temperatures are still worse than on Windows (Idle: +35C Windows vs +45C on Linux) but at last the fans are shutting off properly. I think turning off Tunables is just a workaround and we have to wait and hope that someone makes proper solution in the kernel/drivers.

(In reply to Peter from comment #74)
> Created attachment 246921 [details]
> Fans seems working properly!
> 
> @John: You have to wait a bit to system cool down and then the fans should
> turn off. As you can see on my screenshot fans got turned off after system
> is cooled down. I am monitoring the system for more than 5 days and I have
> to say that the CPU temperatures are still worse than on Windows (Idle: +35C
> Windows vs +45C on Linux) but at last the fans are shutting off properly. I
> think turning off Tunables is just a workaround and we have to wait and hope
> that someone makes proper solution in the kernel/drivers.

Ok, thanks for info.

So finally it seems that fans are turning off only in case of leaving system idle for about 20+ minutes when the CPU temp drops down to ~30C, that means problem with fans still exist, however it is an improvement. Anyway I have to say that with 4.9 kernel, CPU temperatures are comparable to Windows when the fans are off.

@Peter : just to be clear, is this observation based on the kernel update alone, or will it require the powertop / grub parameter discussed above?

(In reply to lpturoyo from comment #77)
> @Peter : just to be clear, is this observation based on the kernel update
> alone, or will it require the powertop / grub parameter discussed above?

powertop / grub parameter is needed Also Now I've noticed that I was running on RC8 kernel (for some reason after installation 4.9 final I still had RC8 as the default one), on 4.9 final the fans are now spinning more than 40 mins. I'll try to reboot back to RC8 and keep you informed if it helped to turn them off

yup. On 4.9.0-040900rc8-generic I've got fans off after few mins with mentioned grub/powertop setup

(In reply to Peter from comment #79)
> yup. On 4.9.0-040900rc8-generic I've got fans off after few mins with
> mentioned grub/powertop setup

Thanks for checking.  I was partially hoping that the kernel update made the resolution by its own.

Same here, powertop and grub parameter are needed for the improved fan behavior.  

Still not quite right, but there is an improvement for sure.

It would be good to compare it with Dell Inspiron 7566. Fans behavior in 7559 is still not right, according this guy, model 7566 (which is pretty similar) works fine with the fans
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1602888/comments/81

Just wanted to check in and see if any more headway was made with this bug. 

I'm running the latest stable kernel with ArchLinux (4.9.9-1-ARCH
) and so far, fans are still a problem with this laptop.  Even with the "!Windows 2015"  and "powertop tunables set to good" workarounds, it's still kind of flakey. Although, these do appear to help somewhat.

I notice in powertop, CPU package still never gets below C3.  And it will only reach C3 with most tunables set to good in powertop.  

Battery life is horrible on this laptop as well.  Perhaps the CPU package not reaching deep states is the cause of all the problems?

(In reply to Bob Tanner from comment #83)
> Just wanted to check in and see if any more headway was made with this bug. 
> 
> I'm running the latest stable kernel with ArchLinux (4.9.9-1-ARCH
> ) and so far, fans are still a problem with this laptop.  Even with the
> "!Windows 2015"  and "powertop tunables set to good" workarounds, it's still
> kind of flakey. Although, these do appear to help somewhat.
> 
> I notice in powertop, CPU package still never gets below C3.  And it will
> only reach C3 with most tunables set to good in powertop.  
> 
> Battery life is horrible on this laptop as well.  Perhaps the CPU package
> not reaching deep states is the cause of all the problems?

Unfortunately there seems to be no improvement, I can confirm the problem still persists even with 4.10.0-041000. I'm giving it up anyway and not in the mood with this model. Going to sell it and buy different one, but I'll think twice (understand: spend some more time with google) if it is Dell next time.

Just wanted to report that I no longer have this issue since I installed the Solus distribution: https://solus-project.com/

I was also affected by this much nastier bug, which also seems to have been solved with Solus:  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1530405 

I had tried the latest Fedora 25 just before Solus, but couldn't even complete the installation wizard: it hanged with "soft lockup - CPU#0 stuck for 22s!".

I've been running the Solus distribution for some days now and didn't get fans or NMI watchdog issues anymore. Also no boot options have been required.

My machine is an Asus K550V laptop, Intel Core i7-6700HQ with NVidia Geforce GTX 950M.

Hope this helps.

(In reply to Roberto Longobardi from comment #85)
> Just wanted to report that I no longer have this issue since I installed the
> Solus distribution: https://solus-project.com/
> 
> I was also affected by this much nastier bug, which also seems to have been
> solved with Solus: 
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1530405 
> 
> I had tried the latest Fedora 25 just before Solus, but couldn't even
> complete the installation wizard: it hanged with "soft lockup - CPU#0 stuck
> for 22s!".
> 
> I've been running the Solus distribution for some days now and didn't get
> fans or NMI watchdog issues anymore. Also no boot options have been required.
> 
> My machine is an Asus K550V laptop, Intel Core i7-6700HQ with NVidia Geforce
> GTX 950M.
> 
> Hope this helps.


Using the latest version of Solus does NOT help for the 7559.  I have been using Solus since day 1 and it's gotten slightly better, but it still much worse than Windows.

Also using Solus with Kernel 4.12 and the parameters acpi_osi="!Windows 2015" which is the only thing that helps.  Even so, activating ANY power management such as powertop causes the system to hard lock.

(In reply to Roberto Longobardi from comment #85)
> Just wanted to report that I no longer have this issue since I installed the
> Solus distribution: https://solus-project.com/
> 
> I was also affected by this much nastier bug, which also seems to have been
> solved with Solus: 
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1530405 
> 
> I had tried the latest Fedora 25 just before Solus, but couldn't even
> complete the installation wizard: it hanged with "soft lockup - CPU#0 stuck
> for 22s!".
> 
> I've been running the Solus distribution for some days now and didn't get
> fans or NMI watchdog issues anymore. Also no boot options have been required.
> 
> My machine is an Asus K550V laptop, Intel Core i7-6700HQ with NVidia Geforce
> GTX 950M.
> 
> Hope this helps.


This bug report is not about an Asus laptop.   It's specifically about the Dell Inspiron 7559. So your experience on Solus is probably not relevant to this issue.

Latest BIOS update seems to have changed nothing on this issue.

Is there any progress with this bug?
I am having this bug on my Dell Inspiron 7559 with latest BIOS 1.2.2 on kernel  4.10.10-041010

Is there a chance that anybody is/will be working on this or this bug thread is dead too? 
I've seen a lot's of ACPI changes on the 4.12RC1 so I installed it but the problem still persists with  4.12.0-041200rc1. Am I the only one who is affected by this problem?

Dell recently released a new BIOS (1.2.3, June 2017): does this fix the issue?

(In reply to fm_bino from comment #92)
> Dell recently released a new BIOS (1.2.3, June 2017): does this fix the
> issue?

No. Also no improvement with kernel 4.13 RC1.

Just FYI: Newest BIOS 1.2.4 released July 27, no change (improvement) with fans kernel 4.13 RC2.

All Inspiron 7559 users suffer from this bug.   I think it is related to the CPU package never getting below C3 power state.  E.g if you view in powertop it never goes below C3.

There are several users subscribed to this bug report and also the one on Ubuntu Launchpad.  I wonder if there is a way we can pester Dell to look into this?

I have two notebooks from Dell that will not spend any significant amount of time in C7 states with newer 4.x kernels. As a consequence, the fans will run almost constantly and temperatures stay high.

KERNELS

Affected: 4.13.5

Unaffected: 4.4.92 LTS

If I remember correctly, kernels up to 4.9.x were unaffected, 4.10.x and newer have been affected.

C-STATES

I am using i7z to monitor C-states. With affected kernels, the CPU stays at about 3200–3350 MHz at all times. C7 percentages are rarely above 40%. Temperatures stay around 60 °C. Fans run at maximum speed.

With unaffected kernels, the CPU stays at around 1250 MHz, about 97.5% in C7 and below 40 °C.

HARDWARE

* Dell Latitude E6530 with an i7-3520M (Intel HD Graphics 4000 as the sole GPU)
* Dell Inspiron 15R SE with an i7-3632QM

Hello,
I also have this problem, but when the system is waiting and awakening the system.
I don't know how to solve it,because noise makes me unable to work.

/bin/grep . /sys/class/thermal/*/* 2>&1

- good:
https://pastebin.com/raw/W0w9TQdG
https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=linux-lts49
-- config used:
https://pastebin.com/raw/YbG4jcUE

- bad:
https://pastebin.com/raw/9Xp626t9
https://git.archlinux.org/svntogit/packages.git/tree/trunk?h=packages/linux

cpu c2d t7500
gpu rv630
laptop hp compaq

this has been an issue since linux ~ 4.12 [or .13]

with the 'bad' kernel the fan noise issue may go away if
the pc is powered off for 15 mins [not just suspend]
on power on the fan may spin down normally again

4.9 does not have this noise / fan speed step issue

(In reply to x3ask from comment #98)
> 4.9 does not have this noise / fan speed step issue

i spoke too soon, 4.9.161-1-lts49 also has this issue but not alwayls

(In reply to x3ask from comment #99)
sorry for the spam, just to clarify
4.9 works normal as expected [usually]
but if the fan's rev gets 'stuck' a
quick suspend to ram fixes it

I can confirm that kernel 4.9.11 does not have this issue at all
and that kernels 4.9.78 and later do - including 4.9.161

narrowed it down to

last good kernel - 4.9.70

https://git.archlinux.org/svntogit/packages.git/tree/trunk?h=packages/linux-lts&id=dcea44f5b32a97ab6be3eb23e69ff8021328d2ce

first bad kernel - 4.9.71

https://git.archlinux.org/svntogit/packages.git/tree/trunk?h=packages/linux-lts&id=ff00e51b6d26e7eb164ff45e53ae1f19d994e838

tested

linux-4.9.11-1-x86_64.pkg.tar.xz
linux-lts-4.9.60-1-x86_64.pkg.tar.xz
linux-lts-4.9.70-1-x86_64.pkg.tar.xz
linux-lts-4.9.71-1-x86_64.pkg.tar.xz
linux-lts-4.9.72-1-x86_64.pkg.tar.xz
linux-lts-4.9.73-1-x86_64.pkg.tar.xz
linux-lts-4.9.74-1-x86_64.pkg.tar.xz
linux-lts-4.9.75-1-x86_64.pkg.tar.xz
linux-lts-4.9.78-1-x86_64.pkg.tar.xz

A little and maybe a final update (not a dev).

I noticed newer kernels have a lower threshold when the fan is sped up.
In my case this is temp5 (see below).

If temp5 is +45 temp4 goes up from +30 to +50
Previously when temp5 was at +45 temp4 did not step up to +50
(temp4 and temp5 are not real temperature measurements)

temp4 +50 is when the fan is laud (whine), makes pc unusable
temp4 +30 is when the fan is quiet and still spinning
temp4 0 is when the fan does not spin at all
temp5's "temperature" steps are +35 +40 +45
(there may be more but those are of interest)

new (laud)
----------
temp4: +50
temp5: +45

old (quiet)
-----------
temp4: +30
temp5: +45

acpitz-virtual-0
Adapter: Virtual device
temp1:        +39.0°C  (crit = +105.0°C)
temp2:        +32.0°C  (crit = +105.0°C)
temp3:        +27.8°C  (crit = +105.0°C)
temp4:        +30.0°C  (crit = +110.0°C)
temp5:        +40.0°C  (crit = +256.0°C)
temp6:        +40.0°C  (crit = +105.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +36.0°C  (high = +100.0°C, crit = +100.0°C)
Core 1:       +37.0°C  (high = +100.0°C, crit = +100.0°C)

I did that silly trick (power off for 15+ minutes),
powered back on and kernel 5.3.1-arch1-1-ARCH does not make the fan laud.

acpitz-acpi-0
Adapter: ACPI interface
temp1:        +41.0°C  (crit = +105.0°C)
temp2:        +35.0°C  (crit = +105.0°C)
temp3:        +26.0°C  (crit = +105.0°C)
temp4:        +30.0°C  (crit = +110.0°C)
temp5:        +45.0°C  (crit = +256.0°C)
temp6:        +39.0°C  (crit = +105.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +42.0°C  (high = +100.0°C, crit = +100.0°C)
Core 1:       +42.0°C  (high = +100.0°C, crit = +100.0°C)

Created attachment 286297 [details]
psensor graph (top output is on backgroud)

The issue still presents at least for my ASUS G752VL. 
dmidecode, acpidump and actual psensor traffic are attached.
Affected kernel version is:
5.3.0-25-generic #27-Ubuntu SMP Wed Dec 4 13:36:32 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux

Created attachment 286299 [details]
dmidecode

Created attachment 286301 [details]
acpidump

it's possible that the cause of this is the gpu

Does anyone tested a bug against Language change. It is 100% reproducible in my case:

1) Turn on notebook with normal fan speed
2) Switch to RU language and start pressing buttons under it in some editor.
3) Fan goes maximal speed without any chance to stop it.

For a long time, I tried to avoid inputting in RU and my fans are OK.


see also 

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1775717

Can someone explain to me how inputting in non-EN language could influence on the speed of fans? Are these dedicated drivers without any chance of memory overlap?