Re: intel_soc_dts_thermal binding issues

[Pandruvada, Srinivas]

On Fri, 2015-03-06 at 08:51 -0500, Josh Boyer wrote:
> On Thu, Mar 5, 2015 at 8:37 PM, Zhang, Rui <rui.zhang@xxxxxxxxx> wrote:
> > Please attach the acpidump output.
> 
> Attached for my machine.
> 
There is no INT3401 in the acpi dump. Also no one registered IRQ 86.
So the request_irq failing.
It is not a problem, this platform can't support.
Do we want to silently bail out?

Thanks,
Srinivas


> josh
> 
> >
> > Thanks,
> > rui
> >
> > -----Original Message-----
> > From: linux-pm-owner@xxxxxxxxxxxxxxx [mailto:linux-pm-owner@xxxxxxxxxxxxxxx] On Behalf Of Josh Boyer
> > Sent: Thursday, March 5, 2015 10:30 PM
> > To: Zhang, Rui
> > Cc: Eduardo Valentin; Linux PM list; Linux-Kernel@Vger. Kernel. Org
> > Subject: Re: intel_soc_dts_thermal binding issues
> > Importance: High
> >
> > On Wed, Mar 4, 2015 at 8:55 PM, Zhang, Rui <rui.zhang@xxxxxxxxx> wrote:
> >> What kernel are you using?
> >> I think it should be fixed by commit 014d9d5d0cc1da79bbe48fbc5e1068c5616238d2, which is merged in 3.19-rc5.
> >
> > The reporter with the ASUS machine was running 3.18.7.  However, on my NUC machine I still see this with 4.0-rc2 and the latest Linus tree:
> >
> > [jwboyer@nuc-celeron ~]$ uname -a
> > Linux nuc-celeron 4.0.0-0.rc2.git1.1.fc23.x86_64 #1 SMP Thu Mar 5
> > 08:37:23 EST 2015 x86_64 x86_64 x86_64 GNU/Linux [jwboyer@nuc-celeron ~]$ dmesg | grep request_threaded
> > [   11.187239] intel_soc_dts_thermal: request_threaded_irq ret -22
> > [   11.261738] intel_soc_dts_thermal: request_threaded_irq ret -22
> > [jwboyer@nuc-celeron ~]$ cat /proc/cpuinfo | grep "model name"
> > model name    : Intel(R) Celeron(R) CPU  N2820  @ 2.13GHz
> > model name    : Intel(R) Celeron(R) CPU  N2820  @ 2.13GHz
> > [jwboyer@nuc-celeron ~]$
> >
> > The 4.0.0-rc2.git1.1 above corresponds to Linux v4.0-rc2-150-g6587457b4b3d in mainline.  We aren't carrying any patches that would impact this.
> >
> > The commit you pointed to seems to imply that either INT340X_THERMAL or INTEL_SOC_DTS_THERMAL could be set, but not both.  In Fedora, we have them both enabled as modules:
> >
> > [jwboyer@nuc-celeron ~]$ grep _THERMAL=m
> > /boot/config-4.0.0-0.rc2.git1.1.fc23.x86_64
> > CONFIG_X86_PKG_TEMP_THERMAL=m
> > CONFIG_INTEL_SOC_DTS_THERMAL=m
> > CONFIG_INT340X_THERMAL=m
> > [jwboyer@nuc-celeron ~]$
> >
> > Maybe the commit should be checking for INTEL_SOC_DTS_THERMAL unconditionally, and not as an #elfi?
> >
> > josh
> >
> >>
> >> Thanks,
> >> rui
> >>
> >> -----Original Message-----
> >> From: jwboyer@xxxxxxxxx [mailto:jwboyer@xxxxxxxxx] On Behalf Of Josh
> >> Boyer
> >> Sent: Wednesday, March 4, 2015 10:05 PM
> >> To: Zhang, Rui; Eduardo Valentin
> >> Cc: Linux PM list; Linux-Kernel@Vger. Kernel. Org
> >> Subject: intel_soc_dts_thermal binding issues
> >> Importance: High
> >>
> >> Hi All,
> >>
> >> We've seen a number of reports where people see:
> >>
> >> [   20.499427] intel_soc_dts_thermal: request_threaded_irq ret -22
> >> [   20.519160] intel_soc_dts_thermal: request_threaded_irq ret -22
> >> [   20.532660] intel_soc_dts_thermal: request_threaded_irq ret -22
> >>
> >> in their boot logs.  They view this as a kernel bug and report is as such.  The above log came from an ASUS x200ma machine, which has a quad core Baytrail-M Pentium CPU (N3530).
> >>
> >> I can't tell if the issue is because the driver is trying to bind to all 4 cores and only CPU0 works, or what other issue would cause this.
> >> I personally have a Celeron based NUC with a N2820 that gets the error above for both cores and the driver doesn't load at all.
> >>
> >> Is there something else the driver should be doing to bind to the various CPUs?  If there's nothing else it can key off of, can we drop the pr_err so it stops showing up in dmesg?
> >>
> >> josh
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-pm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at  http://vger.kernel.org/majordomo-info.html