Thu, Oct 22, 2026 at 12:30:47PM +0200, aneesh.kumar@kernel.org wrote:
Jiri Pirko [off-list ref] writes:
quoted
From: Jiri Pirko <redacted>
Linux has several attestation interfaces tied to individual technologies.
That makes evidence retrieval, measurement registers, device state and
kernel consumers provider-specific.
This series introduces a common attestation subsystem. Its goal is to
support platform, TEE and device providers through one model, including
TPM2, TDX, SEV-SNP, Arm CCA, FIRME, Caliptra and PCI/TSM-backed devices.
This first series implements the common subsystem and a TDX provider.
Existing native TDX tsm_mr and tsm_report interfaces remain independent.
Attestation devices use CTLV, a schema-driven character-device TLV UAPI.
Compared with sysfs or configfs, CTLV executes one validated request
atomically instead of assembling a transaction through several files.
Compared with netlink, an opened device fd directly identifies authority
and lifetime, can be delegated normally, and owns its event queue. CTLV
also provides:
- generated kernel metadata, UAPI constants and documentation;
- strict type, bounds, nesting and required-field validation;
- runtime operation and schema introspection;
- family inheritance and extension points;
- fragmented queries with generation checks;
- per-fd event subscription, overflow reporting and removal notification;
- large input and output blobs.
The attestation family exposes capabilities, evidence, measurement-register
topology and operations, an append-only MR journal and journal
notifications. This first ABI contains the paths exercised by TDX. Extended
evidence classes, dynamic registers, record payloads, verdict recording and
device-security transitions follow with providers that need them and are out
of scope of this initial patchset. The broader direction is illustrated here:
https://github.com/jpirko/linux_mlxsw/commits/wip_attestation_all_1
That branch also contains additional providers (test and sim),
attestation core extensions, libctlv and ctlvctl. With the tools
and kernel specifications, an attestation device can be used as follows:
# Discover CTLV devices and their family chains.
ctlvctl --spec-dir Documentation/ctlv/specs list
# Inspect the root-family lineage and attestation capabilities.
ctlvctl --spec-dir Documentation/ctlv/specs query ctlv0 families-dump
ctlvctl --spec-dir Documentation/ctlv/specs query ctlv0 device-get
# List the measurement registers of the attestation device.
ctlvctl --spec-dir Documentation/ctlv/specs query ctlv0 mr-dump
# Read RTMR0 (MR 1).
ctlvctl --spec-dir Documentation/ctlv/specs \
query ctlv0 mr-read mr.id=1
# Extend RTMR0 with 48 zero bytes (SHA-384).
digest=$(printf '%096d' 0)
ctlvctl --spec-dir Documentation/ctlv/specs \
exec ctlv0 mr-extend --json \
'{"id":1,"value":[{"hash-algo":5,"data":"hex:'"$digest"'"}]}'
# Retrieve the RTMR0 journal.
ctlvctl --spec-dir Documentation/ctlv/specs \
query ctlv0 mr-log-get --json '{"selection":[{"id":1}]}'
The 20 patches are organized as follows:
1-13 add CTLV, its generator and documentation
14-17 add the base attestation family and implementation
18-20 expose the native TDX service and add the TDX family
I am using this to understand what the IDE setup flow would look like,
but it is unclear whether the connect flow belongs in an attestation
framework. I understand the desire to pass the certificate slot number
as part of connect, and sysfs is probably not a suitable interface for
that. However, is CTLV the interface we want to use for connect?
I was thinking about this and "connect" really does not belong in the
attestation CTLV (same applies to "lock"). I was thinking about some
parallel PCITSM CTLV instance, but that does not really click as well.
Currently I'm thinking we can maintain existing "connect" in sysfs and
extend it with "lock" sysfs knob for Realm side. Both would create
ctlv-attestation-spdm instance. The same spdm instance may be created
for other non-PCI-backed device.
The proposed host flow is:
sysfs connect: select tsmN and prepare the connection
-> negotiate SPDM and discover certificates
-> register ctlv-attestation-spdm instance in DISCOVERED state
SPDM_GET
-> populated certificate slots and digests
EVIDENCE_GET(CERT_CHAIN, blob-instance=slot) /*optional*/
-> retrieve candidate certificate chain for appraisal
SPDM_AUTHENTICATE(device-generation, spdm-slot=selected_slot)
-> authenticate the selected chain
-> complete backend-required session and IDE setup
sysfs disconnect
-> remove the attestation instance
-> destroy the native connection
The slot is therefore selected after discovery, rather than supplied to
connect. The attestation instance represents the prepared connection; it
does not exist beforehand as a controller for a disconnected PCI function.
This requires a staged native backend. It is not how existing automatic
sysfs connect behaves today, and those backends would retain their
existing behavior.
For Realm assignment, as written above lock is sysfs interface.
Lock creates a separate assigned-TDI attestation instance for evidence
retrieval, EVIDENCE_ACCEPT, and DEVSEC_RUN
Protocol authentication is distinct from the userspace attestation
verdict: SPDM_AUTHENTICATE does not imply acceptance or TDISP RUN.
So the main difference from your implementation is the lifecycle boundary:
sysfs creates/removes the native contexts, while attestation handles
discovery, explicit authentication, evidence, and the evidence-gated
transition to RUN.
Makes sense?
My patches end up doing the following. I am also unsure whether this is
the intended workflow for the attestation framework.
# ./ctlvctl/ctlvctl --spec-dir=./specs/ ops ctlv0
# ctlv/ctlv-attestation/ctlv-attestation-pci, generation 1
families-dump query read fixed request 16 reply 112..65536
ops-dump query read fixed request 16 reply 56..65536
op-schema-get query read fixed request 40 reply 552..65536
subscriptions-add action read fixed request 6160
subscriptions-remove action read fixed request 6160
event-queue-get query read fixed request 16 reply 40..65536
event-queue-set action read fixed request 40
op-changed event record 112
event-overflow event record 112
device-removed event record 40
device-get action read fixed request 16 reply 12984
evidence-get action read-write dynamic disabled request 26800 reply 472
device-state-changed event record 64
mr-log-entry-appended event record 73416
evidence-accept action read-write dynamic disabled request 64
evidence-accept-clear action read-write dynamic disabled request 16
devsec-state-get query read dynamic disabled request 16 reply 40..1024
devsec-run action read-write dynamic disabled request 64 reply 40
evidence-state-changed event record 112
state-get action read fixed request 16 reply 184
connect action read-write dynamic enabled request 88 reply 40
disconnect action read-write dynamic disabled request 40 reply 40
lock action read-write dynamic disabled request 64 reply 40
unlock action read-write dynamic disabled request 40 reply 40
evidence-refresh action read-write dynamic disabled request 88 reply 40
# ./ctlvctl/ctlvctl --spec-dir=./specs/ exec ctlv0 state-get
# state-get, 1 fragment, generation 0
pci.generation 0
pci.connected 0
pci.locked 0
pci.accepted 0
#
# ./ctlvctl/ctlvctl --spec-dir=./specs/ schema ctlv0 connect
connect (action)
request:
root nest
pci.generation required uint [0, 4294967295]
pci.tsm-id required uint [0, 2147483647]
pci.cert-slot optional uint [0, 7]
reply:
root nest
pci.generation required uint [0, 4294967295]
#
# ./ctlvctl/ctlvctl --spec-dir=./specs/ exec ctlv0 connect pci.generation=0 pci.tsm-id=0
-aneesh