[PATCH v2 0/3] ACPI: RIMT: validate nested firmware extents

RIMT consumers trust three separate layers of firmware data: the root
node area, the payload and arrays contained in each node, and cross-node
ID mappings.  Validating only the outer table does not establish the
bounds of either nested layer.

Split the validation by contract.  Patch 1 validates the root and node
envelopes while preserving unknown-node traversal.  Patch 2 validates
supported node payloads, arrays and the platform-device string.  Patch 3
validates exact destination-node references, mapping depth and ID
arithmetic.

Changes since v1:
https://lore.kernel.org/all/[email protected]/
- split root, node-data and mapping-reference validation
- preserve unknown-node traversal compatibility
- require destination offsets to match exact node boundaries
- treat num_ids as a count, as specified by the RIMT definition
- keep the existing unchanged-ID result on mapping failure

Testing used Linux 3d6d817622b0 plus this series, GCC 13.3 and QEMU
8.2.2 for riscv64.  A temporary KUnit harness in rimt.c called the
actual static validation helpers; the harness is not part of this
series.  The acpi_rimt_bounds suite passed all three tests, covering
truncated table and node extents, invalid payload and array extents,
unterminated device names, interior node references, ID range and
addition errors, and a cyclic mapping chain.

This did not exercise a platform-provided RIMT table through the complete
boot-time discovery path.

Pengpeng Hou (3):
  ACPI: RIMT: validate table and node extents before traversal
  ACPI: RIMT: validate node payload and array extents
  ACPI: RIMT: validate mapping references and ID arithmetic

 drivers/acpi/riscv/rimt.c | 239 ++++++++++++++++++++++++++++++++++----
 1 file changed, 219 insertions(+), 20 deletions(-)


base-commit: 3d6d817622b0a9721e3cc404df3469171582be13
-- 
2.50.1 (Apple Git-155)