Some Intel CPUs implement a "bitfix filter" to suppress reporting of the
same corrected errors repeatedly in the case where aging silicon
develops stuck bits.
But when systems run for weeks, or months, the filters may become clogged
with transient errors.
When the filter is full (indicated by a "yellow" signature in a machine
check bank) clear the filter. This may make space for additional hard
errors.
Save a time stamp when clearing a filter. Log with "WARN" severity if
the filter overflows quickly (in tem minutes or less).
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
Changes since v1: https://lore.kernel.org/all/20260825181526.13203-1-tony.luck@intel.com/
Boris:
Change from a platform driver to an EDAC driver to avoid
spreading RAS code into yet another corner of the source tree.
Ilpo:
Use <linux/bitfield.h> macros FIELD_GET() and FIELD_PREP() for
bit wrangling.
Did NOT add comma to last initializer for dmr_mcbanks[]. The
list is complete, this CPU model won't add extra banks that
support the bitfix filter.
Sashiko: https://sashiko.dev/#/patchset/20260825181526.13203-1-tony.luck%40intel.com
Ignored complaint about lost #GP faults. The driver
checks that the feature is enumerated, so #GP can't happen.
Ignored complaint about overwriting bits 63:1 in the reset
MSRs. Only bit 0 is defined in these MSRs.
Applied suggestion to save bank number in upper 32 bits of bff_id.
a/IS_ERR()/xa_is_err()/ for check of return from xa_store()
Miscellaneous:
Added entry in MAINTAINERS file
Applied some cleanup suggestions from internal AI review.
Qiuxu Zhuo (1):
cacheinfo: Export get_cpu_cacheinfo_id() for loadable modules
Tony Luck (6):
x86/mce: Add enumeration for Intel bitfix filter reset
EDAC/intel-bff: Add stub Intel bitfix filter driver
EDAC/intel-bff: Add Diamond Rapids support
EDAC/intel-bff: Reset bitfix filter when it overflows
EDAC/intel-bff: Compute unique ID for overflowed filter
EDAC/intel-bff: Report frequent filter overflows
include/linux/cacheinfo.h | 12 +-
arch/x86/include/asm/mce.h | 5 +
arch/x86/include/asm/msr-index.h | 2 +
drivers/base/cacheinfo.c | 17 ++
drivers/edac/intel-bff.c | 295 +++++++++++++++++++++++++++++++
MAINTAINERS | 6 +
drivers/edac/Kconfig | 13 ++
drivers/edac/Makefile | 2 +
8 files changed, 341 insertions(+), 11 deletions(-)
create mode 100644 drivers/edac/intel-bff.c
base-commit: 8d3ae59288f1e7d58d76558a6ee96d533bc5019f
--
2.55.0