iTCO_wdt on Intel NUC

iTCO_wdt on Intel NUC

[Valentin Vidić]

Hello,

I'm having trouble getting iTCO_wdt to work on NUC8i5BEH with kernel 4.19.37-5+deb10u1:

[    9.346099] iTCO_vendor_support: vendor-support=0
[    9.377211] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11
[    9.377397] iTCO_wdt iTCO_wdt: can't request region for resource [mem 0x00c5fffc-0x00c5ffff]
[    9.377407] iTCO_wdt: probe of iTCO_wdt failed with error -16

Is there a way to check if the TCO hardware is missing in this machine
or something else needs to be updated to get it working?

Valentin

Re: iTCO_wdt on Intel NUC

[Guenter Roeck]

On 7/21/19 1:30 PM, Valentin Vidić wrote:
> Hello,
> 
> I'm having trouble getting iTCO_wdt to work on NUC8i5BEH with kernel 4.19.37-5+deb10u1:
> 
> [    9.346099] iTCO_vendor_support: vendor-support=0
> [    9.377211] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11
> [    9.377397] iTCO_wdt iTCO_wdt: can't request region for resource [mem 0x00c5fffc-0x00c5ffff]
> [    9.377407] iTCO_wdt: probe of iTCO_wdt failed with error -16
> 
> Is there a way to check if the TCO hardware is missing in this machine
> or something else needs to be updated to get it working?
> 

Most likely the memory region is reserved/used by something else. /proc/iomem
might show you.

Guenter

Re: iTCO_wdt on Intel NUC

[Valentin Vidić]

On Sun, Jul 21, 2019 at 05:05:17PM -0700, Guenter Roeck wrote:
> Most likely the memory region is reserved/used by something else. /proc/iomem
> might show you.

The output is a bit strange since it mentions iTCO_wdt for that address range:

[    9.377397] iTCO_wdt iTCO_wdt: can't request region for resource [mem 0x00c5fffc-0x00c5ffff]

00000000-00000fff : Reserved
00001000-0009efff : System RAM
0009f000-000fffff : Reserved
  000a0000-000bffff : PCI Bus 0000:00
  000e0000-000e3fff : PCI Bus 0000:00
  000e4000-000e7fff : PCI Bus 0000:00
  000e8000-000ebfff : PCI Bus 0000:00
  000ec000-000effff : PCI Bus 0000:00
  000f0000-000fffff : System ROM
00100000-6f8d8017 : System RAM
  00c5fffc-00c5ffff : iTCO_wdt
6f8d8018-6f8e8e57 : System RAM
6f8e8e58-79d52fff : System RAM
79d53000-7a1bdfff : Reserved
7a1be000-7a23afff : ACPI Tables
7a23b000-7a2ebfff : ACPI Non-volatile Storage
7a2ec000-7ac0dfff : Reserved
7ac0e000-7ac0efff : System RAM
7ac0f000-7fffffff : Reserved
  7c000000-7fffffff : Graphics Stolen Memory

-- 
Valentin

Re: iTCO_wdt on Intel NUC

[Guenter Roeck]

On Mon, Jul 22, 2019 at 06:55:57AM +0200, Valentin Vidić wrote:
> On Sun, Jul 21, 2019 at 05:05:17PM -0700, Guenter Roeck wrote:
> > Most likely the memory region is reserved/used by something else. /proc/iomem
> > might show you.
> 
> The output is a bit strange since it mentions iTCO_wdt for that address range:
> 
> [    9.377397] iTCO_wdt iTCO_wdt: can't request region for resource [mem 0x00c5fffc-0x00c5ffff]
> 
> 00000000-00000fff : Reserved
> 00001000-0009efff : System RAM
> 0009f000-000fffff : Reserved
>   000a0000-000bffff : PCI Bus 0000:00
>   000e0000-000e3fff : PCI Bus 0000:00
>   000e4000-000e7fff : PCI Bus 0000:00
>   000e8000-000ebfff : PCI Bus 0000:00
>   000ec000-000effff : PCI Bus 0000:00
>   000f0000-000fffff : System ROM
> 00100000-6f8d8017 : System RAM
>   00c5fffc-00c5ffff : iTCO_wdt
> 6f8d8018-6f8e8e57 : System RAM
> 6f8e8e58-79d52fff : System RAM
> 79d53000-7a1bdfff : Reserved
> 7a1be000-7a23afff : ACPI Tables
> 7a23b000-7a2ebfff : ACPI Non-volatile Storage
> 7a2ec000-7ac0dfff : Reserved
> 7ac0e000-7ac0efff : System RAM
> 7ac0f000-7fffffff : Reserved
>   7c000000-7fffffff : Graphics Stolen Memory
> 

Weird. Is iTCO_wdt instantiated twice ?

I have no idea how to proceed, sorry.

Guenter

Re: iTCO_wdt on Intel NUC

[Valentin Vidić]

On Mon, Jul 22, 2019 at 08:45:50AM -0700, Guenter Roeck wrote:
> Weird. Is iTCO_wdt instantiated twice ?

I don't think so. I blacklisted iTCO_wdt module and rebooted but
the iTCO_wdt line in /proc/iomem still appears although the module
was not loaded.

I suspect that it might have to do with one of these modules?

drivers/platform/x86/intel_pmc_ipc.c:#define TCO_DEVICE_NAME                    "iTCO_wdt"
drivers/mfd/lpc_ich.c:  .name = "iTCO_wdt",
drivers/i2c/busses/i2c-i801.c:  pdev = platform_device_register_resndata(&pci_dev->dev, "iTCO_wdt", -1,

-- 
Valentin

Re: iTCO_wdt on Intel NUC

[Guenter Roeck]

On Mon, Jul 22, 2019 at 06:10:26PM +0200, Valentin Vidić wrote:
> On Mon, Jul 22, 2019 at 08:45:50AM -0700, Guenter Roeck wrote:
> > Weird. Is iTCO_wdt instantiated twice ?
> 
> I don't think so. I blacklisted iTCO_wdt module and rebooted but
> the iTCO_wdt line in /proc/iomem still appears although the module
> was not loaded.
> 
> I suspect that it might have to do with one of these modules?
> 
> drivers/platform/x86/intel_pmc_ipc.c:#define TCO_DEVICE_NAME                    "iTCO_wdt"
> drivers/mfd/lpc_ich.c:  .name = "iTCO_wdt",
> drivers/i2c/busses/i2c-i801.c:  pdev = platform_device_register_resndata(&pci_dev->dev, "iTCO_wdt", -1,
> 

Correct.

I suspect that the driver may be instantiated from two locations
on your system. On top of that, there is also wdat_wdt, which
instantiates a watchdog device through ACPI. Overall there are lots
of areas where things can go wrong with this driver.

Guenter

Re: iTCO_wdt on Intel NUC

[Valentin Vidić]

On Mon, Jul 22, 2019 at 09:21:52AM -0700, Guenter Roeck wrote:
> I suspect that the driver may be instantiated from two locations
> on your system. On top of that, there is also wdat_wdt, which
> instantiates a watchdog device through ACPI. Overall there are lots
> of areas where things can go wrong with this driver.

Yes, it seems to be related to i2c_i801 because that tries to
load iTCO_wdt:

# modprobe i2c_i801
[40450.070587] i801_smbus 0000:00:1f.4: SPD Write Disable is set
[40450.070652] i801_smbus 0000:00:1f.4: SMBus using PCI interrupt
[40450.072919] iTCO_vendor_support: vendor-support=0
[40450.073485] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11
[40450.073546] iTCO_wdt iTCO_wdt: can't request region for resource [mem 0x00c5fffc-0x00c5ffff]
[40450.073578] iTCO_wdt: probe of iTCO_wdt failed with error -16

-- 
Valentin

Re: iTCO_wdt on Intel NUC

[Guenter Roeck]

On Mon, Jul 22, 2019 at 07:41:29PM +0200, Valentin Vidić wrote:
> On Mon, Jul 22, 2019 at 09:21:52AM -0700, Guenter Roeck wrote:
> > I suspect that the driver may be instantiated from two locations
> > on your system. On top of that, there is also wdat_wdt, which
> > instantiates a watchdog device through ACPI. Overall there are lots
> > of areas where things can go wrong with this driver.
> 
> Yes, it seems to be related to i2c_i801 because that tries to
> load iTCO_wdt:
> 
> # modprobe i2c_i801
> [40450.070587] i801_smbus 0000:00:1f.4: SPD Write Disable is set
> [40450.070652] i801_smbus 0000:00:1f.4: SMBus using PCI interrupt
> [40450.072919] iTCO_vendor_support: vendor-support=0
> [40450.073485] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11
> [40450.073546] iTCO_wdt iTCO_wdt: can't request region for resource [mem 0x00c5fffc-0x00c5ffff]
> [40450.073578] iTCO_wdt: probe of iTCO_wdt failed with error -16
> 

Is this the first mention of iTCO_wdt in the log ?
The easiest explanation that this point would be that lpc_ich instantiates
iTCO_wdt as well.

Guenter

Re: iTCO_wdt on Intel NUC

[Valentin Vidić]

On Mon, Jul 22, 2019 at 10:52:03AM -0700, Guenter Roeck wrote:
> Is this the first mention of iTCO_wdt in the log ?
> The easiest explanation that this point would be that lpc_ich instantiates
> iTCO_wdt as well.

I have removed lpc_ich and the other relevant modules and than try to
load i2c_i801. It seems to me that perhaps i2c_i801 passes the wrong
parameters to iTCO_wdt or something like that?

-- 
Valentin

Re: iTCO_wdt on Intel NUC

[Guenter Roeck]

On Mon, Jul 22, 2019 at 07:56:58PM +0200, Valentin Vidić wrote:
> On Mon, Jul 22, 2019 at 10:52:03AM -0700, Guenter Roeck wrote:
> > Is this the first mention of iTCO_wdt in the log ?
> > The easiest explanation that this point would be that lpc_ich instantiates
> > iTCO_wdt as well.
> 
> I have removed lpc_ich and the other relevant modules and than try to
> load i2c_i801. It seems to me that perhaps i2c_i801 passes the wrong
> parameters to iTCO_wdt or something like that?
> 

No idea, sorry. I don't have the system, and I have no means to debug it.
The authors of the code in i2c_i801 might be able to help.

Guenter

Re: iTCO_wdt on Intel NUC

[Valentin Vidić]

On Mon, Jul 22, 2019 at 11:03:10AM -0700, Guenter Roeck wrote:
> No idea, sorry. I don't have the system, and I have no means to debug it.
> The authors of the code in i2c_i801 might be able to help.

Sure, I will try to contact the authors and linux-i2c list.

Thanks for the help,
Valentin

Re: iTCO_wdt on Intel NUC

[Guenter Roeck]

On 7/21/19 5:05 PM, Guenter Roeck wrote:
> On 7/21/19 1:30 PM, Valentin Vidić wrote:
>> Hello,
>>
>> I'm having trouble getting iTCO_wdt to work on NUC8i5BEH with kernel 4.19.37-5+deb10u1:
>>
>> [    9.346099] iTCO_vendor_support: vendor-support=0
>> [    9.377211] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11
>> [    9.377397] iTCO_wdt iTCO_wdt: can't request region for resource [mem 0x00c5fffc-0x00c5ffff]
>> [    9.377407] iTCO_wdt: probe of iTCO_wdt failed with error -16
>>
>> Is there a way to check if the TCO hardware is missing in this machine
>> or something else needs to be updated to get it working?
>>
> 
> Most likely the memory region is reserved/used by something else. /proc/iomem
> might show you.
> 

Here is a possible explanation for your problem.

https://lore.kernel.org/patchwork/patch/770990/

It seems like the memory resource is indeed reserved by the intel_pmc_ipc driver.

Guenter

Re: iTCO_wdt on Intel NUC

[Valentin Vidić]

On Wed, Jul 24, 2019 at 02:43:58AM -0700, Guenter Roeck wrote:
> Here is a possible explanation for your problem.
> 
> https://lore.kernel.org/patchwork/patch/770990/
> 
> It seems like the memory resource is indeed reserved by the intel_pmc_ipc
> driver.

Thanks, I will check if I have that one loaded. Here is the
discussion on linux-i2c, they say it might be a BIOS problem:

  https://www.spinics.net/lists/linux-i2c/msg41504.html

-- 
Valentin

Re: iTCO_wdt on Intel NUC

[Guenter Roeck]

On 7/24/19 3:00 AM, Valentin Vidić wrote:
> On Wed, Jul 24, 2019 at 02:43:58AM -0700, Guenter Roeck wrote:
>> Here is a possible explanation for your problem.
>>
>> https://lore.kernel.org/patchwork/patch/770990/
>>
>> It seems like the memory resource is indeed reserved by the intel_pmc_ipc
>> driver.
> 
> Thanks, I will check if I have that one loaded. Here is the
> discussion on linux-i2c, they say it might be a BIOS problem:
> 
>    https://www.spinics.net/lists/linux-i2c/msg41504.html
> 
Yes, Jean's explanation does make a lot of sense.

Guenter