mbox series

[RFC,v2,0/3] powerpc/pseries: add support for local secure storage called Platform KeyStore(PKS)

Message ID 20220622215648.96723-1-nayna@linux.ibm.com (mailing list archive)
Headers show
Series powerpc/pseries: add support for local secure storage called Platform KeyStore(PKS) | expand


Nayna Jain June 22, 2022, 9:56 p.m. UTC
PowerVM provides an isolated Platform KeyStore(PKS)[1] storage allocation
for each partition(LPAR) with individually managed access controls to store
sensitive information securely. Linux Kernel can access this storage by
interfacing with hypervisor using a new set of hypervisor calls. 

PowerVM guest secure boot feature intend to use Platform KeyStore for
the purpose of storing public keys. Secure boot requires public keys to
be able to verify the grub and boot kernel. To allow authenticated
 manipulation of keys, it supports variables to store key authorities
- PK/KEK. Other variables are used to store code signing keys - db/grubdb.
It also supports denied list to disallow booting even if signed with
valid key. This is done via denied list database - dbx or sbat. These
variables would be stored in PKS, and are managed and controlled by

The purpose of this patchset is to add userspace interface to manage
these variables.

For v1[2] version, we received following feedback
"Ok, this is like the 3rd or 4th different platform-specific proposal for
this type of functionality.  I think we need to give up on
platform-specific user/kernel apis on this (random sysfs/securityfs
files scattered around the tree), and come up with a standard place for
all of this."

Currently, OpenPOWER exposes variables via sysfs, while EFI platforms
have used sysfs and then moved to their own efivarfs filesystem.
Recently, coco feature is using securityfs to expose their
secrets. All of these environments are different both syntactically and

securityfs is meant for linux security subsystems to expose policies/logs
or any other information, and do not interact with firmware for managing
these variables. However, there are various firmware security
features which expose their variables for user management via kernel as
discussed above. There is currently no single place to expose these
variables. Different platforms use sysfs/platform specific
filesystem(efivarfs)/securityfs interface as find appropriate. This has
resulted in interfaces scattered around the tree.

This resulted in demand of a need for a common single place for new
platform interfaces to expose their variables for firmware security
features. This would simplify the interface for users of these platforms.
This patchset proposes firmware security filesystem(fwsecurityfs). Any
platform can expose the variables which are required by firmware security
features via this interface. Going forward, this would give a common place
for exposing variables managed by firmware while still allowing platforms
to implement their own underlying semantics.

This design consists of two parts:
1. firmware security filesystem(fwsecurityfs) that provides platforms with
APIs to create their own underlying directory and file structure. It is
recommended to establish a well known mount point:
i.e. /sys/firmware/security/

2. platform specific implementation for these variables which implements
underlying semantics. Platforms can expose their variables as files
allowing read/write/add/delete operations by defining their own inode and
file operations.

This patchset defines:
1. pseries driver to access LPAR Platform Key Store(PLPKS)
2. firmware security filesystem named fwsecurityfs
3. Interface to expose secure variables stored in LPAR PKS via fwsecurityfs

[1] https://community.ibm.com/community/user/power/blogs/chris-engel1/2020/11/20/powervm-introduces-the-platform-keystore
[2] https://lore.kernel.org/linuxppc-dev/20220122005637.28199-1-nayna@linux.ibm.com/



* Defined unified interface(firmware security filesystem) for all platforms
to expose their variables used for security features. 
* Expose secvars using firmware security fileystem.
* Renamed PKS driver to PLPKS to avoid naming conflict as mentioned by
Dave Hanson.

Nayna Jain (3):
  powerpc/pseries: define driver for Platform KeyStore
  fs: define a firmware security filesystem named fwsecurityfs
  powerpc/pseries: expose authenticated variables stored in LPAR PKS

 arch/powerpc/include/asm/hvcall.h             |  12 +-
 arch/powerpc/include/asm/plpks.h              |  92 ++++
 arch/powerpc/platforms/pseries/Kconfig        |  27 +
 arch/powerpc/platforms/pseries/Makefile       |   2 +
 arch/powerpc/platforms/pseries/plpks/Makefile |   9 +
 .../pseries/plpks/fwsecurityfs_arch.c         |  16 +
 .../platforms/pseries/plpks/internal.h        |  18 +
 arch/powerpc/platforms/pseries/plpks/plpks.c  | 517 ++++++++++++++++++
 .../powerpc/platforms/pseries/plpks/secvars.c | 239 ++++++++
 fs/Kconfig                                    |   1 +
 fs/Makefile                                   |   1 +
 fs/fwsecurityfs/Kconfig                       |  14 +
 fs/fwsecurityfs/Makefile                      |  10 +
 fs/fwsecurityfs/inode.c                       | 159 ++++++
 fs/fwsecurityfs/internal.h                    |  13 +
 fs/fwsecurityfs/super.c                       | 154 ++++++
 include/linux/fwsecurityfs.h                  |  33 ++
 include/uapi/linux/magic.h                    |   1 +
 18 files changed, 1317 insertions(+), 1 deletion(-)
 create mode 100644 arch/powerpc/include/asm/plpks.h
 create mode 100644 arch/powerpc/platforms/pseries/plpks/Makefile
 create mode 100644 arch/powerpc/platforms/pseries/plpks/fwsecurityfs_arch.c
 create mode 100644 arch/powerpc/platforms/pseries/plpks/internal.h
 create mode 100644 arch/powerpc/platforms/pseries/plpks/plpks.c
 create mode 100644 arch/powerpc/platforms/pseries/plpks/secvars.c
 create mode 100644 fs/fwsecurityfs/Kconfig
 create mode 100644 fs/fwsecurityfs/Makefile
 create mode 100644 fs/fwsecurityfs/inode.c
 create mode 100644 fs/fwsecurityfs/internal.h
 create mode 100644 fs/fwsecurityfs/super.c
 create mode 100644 include/linux/fwsecurityfs.h


Dave Hansen June 27, 2022, 9:10 p.m. UTC | #1
On 6/22/22 14:56, Nayna Jain wrote:
> * Renamed PKS driver to PLPKS to avoid naming conflict as mentioned by
> Dave Hanson.

Thank you for doing this!  The new naming looks much less likely to
cause confusion.