[0/5] AMD64 EDAC GPU Updates

Message

M K, Muralidhara May 15, 2023, 11:35 a.m. UTC

From: Muralidhara M K <[e-mail]>

This set adds GPU support to AMD64 EDAC starting with the MI200
(Aldebaran) series.
The AMD Instinct™ MI200 series accelerators are the data center GPUs.

Patch 1:
Adds PCI IDs without changing the existing AMD NB code.

Patch 2:
Recognizes the new UMC SMCA bank type in the decoder module as just
another UMC bank. Decode details are done in the AMD64 EDAC module.

Patch 3:
Adds code documentation on how the GPU memory is laid out in EDAC.

Patch 4:
Adds the bulk of the GPU-specific code.

Patch 5:
Handle differences in "AMD Node" enumeration for GPU dies.

Muralidhara M K (2):
  EDAC/amd64: Document heterogeneous system enumeration
  EDAC/amd64: Add support for AMD heterogeneous Family 19h Model 30h-3Fh

Yazen Ghannam (3):
  x86/amd_nb: Add MI200 PCI IDs
  x86/MCE/AMD, EDAC/mce_amd: Decode UMC_V2 ECC errors
  EDAC/amd64: Cache and use GPU node map

 Documentation/driver-api/edac.rst | 120 ++++++++++
 arch/x86/kernel/amd_nb.c          |   5 +
 arch/x86/kernel/cpu/mce/amd.c     |   6 +-
 drivers/edac/amd64_edac.c         | 386 +++++++++++++++++++++++++++---
 drivers/edac/amd64_edac.h         |   1 +
 drivers/edac/mce_amd.c            |   3 +-
 include/linux/pci_ids.h           |   1 +
 7 files changed, 488 insertions(+), 34 deletions(-)