[SRU][N][PATCH 0/8] Ubuntu 24.04 - GPU cannot be installed with DL380a Gen12 (2P, SRF-SP)

[Michael Reed]

From: Michael Reed <Michael.Reed at canonical.com>

BugLink: https://bugs.launchpad.net/bugs/2081079

SRU Justification:

[Impact]
Description:
Failed to install GPU with Ubuntu 24.04 on a DL380a Gen12 with Intel Sierra Forest 2P

There is a random write to VF BAR0's memory region that causes the kernel got MCE error.

Version-Release number :
Ubuntu 24.04

Additional info:

We have tracked this issue with RHEL9.4, it's caused by the following pathes.

cb4a6ccf3583 perf/x86/intel/uncore: Support Sierra Forest and Grand Ridge (v6.8-rc1)
388d76175bd9 perf/x86/intel/uncore: Support IIO free-running counters on GNR (v6.8-rc1)
632c4bf6d007 perf/x86/intel/uncore: Support Granite Rapids (v6.8-rc1)
b560e0cd882b perf/x86/uncore: Use u64 to replace unsigned for the uncore offsets array (v6.8-rc1)
cf35791476fc perf/x86/intel/uncore: Generic uncore_get_uncores and MMIO format of SPR (v6.8-rc1)

[Test Plan]
How reproducible:
Each time

Steps to reproduce
- PCI segment, Intel VT-d and SR-IOV , all enabled in the BIOS
- Run a fresh install on a DL380a server with 2P with GPU in slot17

Expected results
No MCE and run installation w/o problem

Actual results
The kernel got MCE errors.

[Fix]
Intel gave us a patch set that resolves the issue.
https://lore.kernel.org/lkml/20240614134631.1092359-1-kan.liang@linux.intel.com/#r

The following patches are required.

f8a86a9bb5f7 perf/x86/intel/uncore: Support HBM and CXL PMON counters (v6.11-rc1)
15a4bd51853b perf/x86/uncore: Cleanup unused unit structure (v6.11-rc1)
f76a8420444b perf/x86/uncore: Apply the unit control RB tree to PCI uncore units (v6.11-rc1)
b1d9ea2e1ca4 perf/x86/uncore: Apply the unit control RB tree to MSR uncore units (v6.11-rc1)
80580dae65b9 perf/x86/uncore: Apply the unit control RB tree to MMIO uncore units (v6.11-rc1)
585463fee642 perf/x86/uncore: Retrieve the unit ID from the unit control RB tree (v6.11-rc1)
c74443d92f68 perf/x86/uncore: Support per PMU cpumask (v6.11-rc1)
0007f3932592 perf/x86/uncore: Save the unit control address of all units (v6.11-rc1)

[Where problems could occur]

[Other Info]

https://code.launchpad.net/~mreed8855/ubuntu/+source/linux/+git/noble/+ref/lp_2081079_dl380a_gen12

Kan Liang (8):
  perf/x86/uncore: Save the unit control address of all units
  perf/x86/uncore: Support per PMU cpumask
  perf/x86/uncore: Retrieve the unit ID from the unit control RB tree
  perf/x86/uncore: Apply the unit control RB tree to MMIO uncore units
  perf/x86/uncore: Apply the unit control RB tree to MSR uncore units
  perf/x86/uncore: Apply the unit control RB tree to PCI uncore units
  perf/x86/uncore: Cleanup unused unit structure
  perf/x86/intel/uncore: Support HBM and CXL PMON counters

 arch/x86/events/intel/uncore.c           |  97 ++++---
 arch/x86/events/intel/uncore.h           |   8 +-
 arch/x86/events/intel/uncore_discovery.c | 306 +++++++++++++++--------
 arch/x86/events/intel/uncore_discovery.h |  22 +-
 arch/x86/events/intel/uncore_snbep.c     | 128 ++++++++--
 5 files changed, 388 insertions(+), 173 deletions(-)

-- 
2.34.1