Re: [PATCH v6 0/5] Fixes for PMF and CCP drivers after S4

["Ilpo Järvinen" <ilpo.jarvinen@linux.intel.com>]

On Thu, 15 Jan 2026, Mario Limonciello (AMD) wrote:

> Lars Francke reported that the PMF driver fails to work afer S4 with:
>   ccp 0000:c3:00.2: tee: command 0x5 timed out, disabling PSP
> 
> This is because there is a TA loaded to the TEE environment that
> is lost during S4.  The TEE rings need to be reinitialized and the
> TA needs to be reloaded.
> 
> This series adds those flows to the PMF and CCP drivers.
> 
> v5->v6:
>  * Fix Tom's feedback on patch 3/5
> 
> Mario Limonciello (AMD) (4):
>   crypto: ccp - Declare PSP dead if PSP_CMD_TEE_RING_INIT fails
>   crypto: ccp - Add an S4 restore flow
>   crypto: ccp - Factor out ring destroy handling to a helper
>   crypto: ccp - Send PSP_CMD_TEE_RING_DESTROY when PSP_CMD_TEE_RING_INIT
>     fails
> 
> Shyam Sundar S K (1):
>   platform/x86/amd/pmf: Prevent TEE errors after hibernate
> 
>  drivers/crypto/ccp/psp-dev.c          | 11 +++++
>  drivers/crypto/ccp/sp-dev.c           | 12 ++++++
>  drivers/crypto/ccp/sp-dev.h           |  3 ++
>  drivers/crypto/ccp/sp-pci.c           | 16 ++++++-
>  drivers/crypto/ccp/tee-dev.c          | 56 ++++++++++++++++++------
>  drivers/crypto/ccp/tee-dev.h          |  1 +
>  drivers/platform/x86/amd/pmf/core.c   | 62 ++++++++++++++++++++++++++-
>  drivers/platform/x86/amd/pmf/pmf.h    | 10 +++++
>  drivers/platform/x86/amd/pmf/tee-if.c | 12 ++----
>  include/linux/psp.h                   |  1 +
>  10 files changed, 161 insertions(+), 23 deletions(-)

I've applied this to the review-ilpo-next branch (and reorganized the 
NPU metrics to come after this series as these seemed to have a context 
conflict).

-- 
 i.