Possible issue with tpm_crb runtime PM

Possible issue with tpm_crb runtime PM

[Jarkko Sakkinen]

Hi

I get this (probably every time) when I do rmmod:

[ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the TPM

That is probably caused by failing TPM2_Shutdown. My guess is that
runtime PM patches cause this.

I'll verify that the problem does not occur with the next branch,
which does not include runtime PM patches.

/Jarkko

------------------------------------------------------------------------------

Re: Possible issue with tpm_crb runtime PM

[Winkler, Tomas]

> 
> Hi
> 
> I get this (probably every time) when I do rmmod:
> 
> [ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the TPM
> 
> That is probably caused by failing TPM2_Shutdown. My guess is that runtime
> PM patches cause this.
> 
> I'll verify that the problem does not occur with the next branch, which does
> not include runtime PM patches.

Can you produce the full log?
Tomas


------------------------------------------------------------------------------

Re: Possible issue with tpm_crb runtime PM

[Jarkko Sakkinen]

On Tue, Sep 20, 2016 at 01:30:58PM +0000, Winkler, Tomas wrote:
> 
> > 
> > Hi
> > 
> > I get this (probably every time) when I do rmmod:
> > 
> > [ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the TPM
> > 
> > That is probably caused by failing TPM2_Shutdown. My guess is that runtime
> > PM patches cause this.
> > 
> > I'll verify that the problem does not occur with the next branch, which does
> > not include runtime PM patches.
> 
> Can you produce the full log?

I just compiled kernel from the 'next' branch and it doesn't have this
issue. If you want, I can recompile the kernel but that's the only error
message that is emitted and it is obviously emitted by tpm2_shutdown().
This is a Skylake m7-6Y75 machine that I use for ad-hoc testing (runs
Ubuntu 16.04 LTS, usually I test kernel with buildroot initramfs) of TPM
and SGX.

> Tomas

/Jarkko

------------------------------------------------------------------------------

Re: Possible issue with tpm_crb runtime PM

[Winkler, Tomas]

> On Tue, Sep 20, 2016 at 01:30:58PM +0000, Winkler, Tomas wrote:
> >
> > >
> > > Hi
> > >
> > > I get this (probably every time) when I do rmmod:
> > >
> > > [ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the TPM
> > >
> > > That is probably caused by failing TPM2_Shutdown. My guess is that
> > > runtime PM patches cause this.
> > >
> > > I'll verify that the problem does not occur with the next branch,
> > > which does not include runtime PM patches.
> >
> > Can you produce the full log?
> 
> I just compiled kernel from the 'next' branch and it doesn't have this issue. If
> you want, I can recompile the kernel but that's the only error message that is
> emitted and it is obviously emitted by tpm2_shutdown().
> This is a Skylake m7-6Y75 machine that I use for ad-hoc testing (runs Ubuntu
> 16.04 LTS, usually I test kernel with buildroot initramfs) of TPM and SGX.

If you can just revert the last patch, so the device  will be always up, see if  this helps? 
Thanks
Tomas


------------------------------------------------------------------------------

Re: Possible issue with tpm_crb runtime PM

[Jarkko Sakkinen]

On Tue, Sep 20, 2016 at 07:26:08PM +0000, Winkler, Tomas wrote:
> > On Tue, Sep 20, 2016 at 01:30:58PM +0000, Winkler, Tomas wrote:
> > >
> > > >
> > > > Hi
> > > >
> > > > I get this (probably every time) when I do rmmod:
> > > >
> > > > [ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the TPM
> > > >
> > > > That is probably caused by failing TPM2_Shutdown. My guess is that
> > > > runtime PM patches cause this.
> > > >
> > > > I'll verify that the problem does not occur with the next branch,
> > > > which does not include runtime PM patches.
> > >
> > > Can you produce the full log?
> > 
> > I just compiled kernel from the 'next' branch and it doesn't have this issue. If
> > you want, I can recompile the kernel but that's the only error message that is
> > emitted and it is obviously emitted by tpm2_shutdown().
> > This is a Skylake m7-6Y75 machine that I use for ad-hoc testing (runs Ubuntu
> > 16.04 LTS, usually I test kernel with buildroot initramfs) of TPM and SGX.
> 
> If you can just revert the last patch, so the device  will be always up, see if  this helps? 

This error does not occur with

0c22db435bf7 ("tpm/tpm_crb: open code the crb_init into acpi_add")

/Jarkko

------------------------------------------------------------------------------

Re: Possible issue with tpm_crb runtime PM

[Winkler, Tomas]

> 
> On Tue, Sep 20, 2016 at 07:26:08PM +0000, Winkler, Tomas wrote:
> > > On Tue, Sep 20, 2016 at 01:30:58PM +0000, Winkler, Tomas wrote:
> > > >
> > > > >
> > > > > Hi
> > > > >
> > > > > I get this (probably every time) when I do rmmod:
> > > > >
> > > > > [ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the
> > > > > TPM
> > > > >
> > > > > That is probably caused by failing TPM2_Shutdown. My guess is
> > > > > that runtime PM patches cause this.
> > > > >
> > > > > I'll verify that the problem does not occur with the next
> > > > > branch, which does not include runtime PM patches.
> > > >
> > > > Can you produce the full log?
> > >
> > > I just compiled kernel from the 'next' branch and it doesn't have
> > > this issue. If you want, I can recompile the kernel but that's the
> > > only error message that is emitted and it is obviously emitted by
> tpm2_shutdown().
> > > This is a Skylake m7-6Y75 machine that I use for ad-hoc testing
> > > (runs Ubuntu
> > > 16.04 LTS, usually I test kernel with buildroot initramfs) of TPM and SGX.
> >
> > If you can just revert the last patch, so the device  will be always up, see if
> this helps?
> 
> This error does not occur with
> 
> 0c22db435bf7 ("tpm/tpm_crb: open code the crb_init into acpi_add")
>
Okay,  can you apply the my original patch with pm_rurntime_put_sync ? 
Thanks


------------------------------------------------------------------------------

Re: Possible issue with tpm_crb runtime PM

[Jarkko Sakkinen]

On Sun, Sep 25, 2016 at 12:38:56PM +0000, Winkler, Tomas wrote:
> > 
> > On Tue, Sep 20, 2016 at 07:26:08PM +0000, Winkler, Tomas wrote:
> > > > On Tue, Sep 20, 2016 at 01:30:58PM +0000, Winkler, Tomas wrote:
> > > > >
> > > > > >
> > > > > > Hi
> > > > > >
> > > > > > I get this (probably every time) when I do rmmod:
> > > > > >
> > > > > > [ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the
> > > > > > TPM
> > > > > >
> > > > > > That is probably caused by failing TPM2_Shutdown. My guess is
> > > > > > that runtime PM patches cause this.
> > > > > >
> > > > > > I'll verify that the problem does not occur with the next
> > > > > > branch, which does not include runtime PM patches.
> > > > >
> > > > > Can you produce the full log?
> > > >
> > > > I just compiled kernel from the 'next' branch and it doesn't have
> > > > this issue. If you want, I can recompile the kernel but that's the
> > > > only error message that is emitted and it is obviously emitted by
> > tpm2_shutdown().
> > > > This is a Skylake m7-6Y75 machine that I use for ad-hoc testing
> > > > (runs Ubuntu
> > > > 16.04 LTS, usually I test kernel with buildroot initramfs) of TPM and SGX.
> > >
> > > If you can just revert the last patch, so the device  will be always up, see if
> > this helps?
> > 
> > This error does not occur with
> > 
> > 0c22db435bf7 ("tpm/tpm_crb: open code the crb_init into acpi_add")
> >
> Okay, can you apply the my original patch with pm_rurntime_put_sync ? 
> Thanks

The same error still happens with pm_runtime_put_sync().

/Jarkko

------------------------------------------------------------------------------

Re: Possible issue with tpm_crb runtime PM

[Jarkko Sakkinen]

On Tue, Sep 27, 2016 at 01:14:27AM +0300, Jarkko Sakkinen wrote:
> On Sun, Sep 25, 2016 at 12:38:56PM +0000, Winkler, Tomas wrote:
> > > 
> > > On Tue, Sep 20, 2016 at 07:26:08PM +0000, Winkler, Tomas wrote:
> > > > > On Tue, Sep 20, 2016 at 01:30:58PM +0000, Winkler, Tomas wrote:
> > > > > >
> > > > > > >
> > > > > > > Hi
> > > > > > >
> > > > > > > I get this (probably every time) when I do rmmod:
> > > > > > >
> > > > > > > [ 4372.725616] tpm tpm0: A TPM error (325) occurred stopping the
> > > > > > > TPM
> > > > > > >
> > > > > > > That is probably caused by failing TPM2_Shutdown. My guess is
> > > > > > > that runtime PM patches cause this.
> > > > > > >
> > > > > > > I'll verify that the problem does not occur with the next
> > > > > > > branch, which does not include runtime PM patches.
> > > > > >
> > > > > > Can you produce the full log?
> > > > >
> > > > > I just compiled kernel from the 'next' branch and it doesn't have
> > > > > this issue. If you want, I can recompile the kernel but that's the
> > > > > only error message that is emitted and it is obviously emitted by
> > > tpm2_shutdown().
> > > > > This is a Skylake m7-6Y75 machine that I use for ad-hoc testing
> > > > > (runs Ubuntu
> > > > > 16.04 LTS, usually I test kernel with buildroot initramfs) of TPM and SGX.
> > > >
> > > > If you can just revert the last patch, so the device  will be always up, see if
> > > this helps?
> > > 
> > > This error does not occur with
> > > 
> > > 0c22db435bf7 ("tpm/tpm_crb: open code the crb_init into acpi_add")
> > >
> > Okay, can you apply the my original patch with pm_rurntime_put_sync ? 
> > Thanks
> 
> The same error still happens with pm_runtime_put_sync().

Now that I replied the patches the problem does not occur anymore.
So it should be ready for 4.10. I guess the problem was really this.
Probably put callback was delayed. If we change this behaviour it's
better to do as a separate commit with its own test cycle.

/Jarkko

------------------------------------------------------------------------------