[PATCH v2 11/16] x86/mce: Skip AMD threshold init if no threshold banks found

From: Yazen Ghannam
Date: Thu Apr 04 2024 - 11:16:30 EST


AMD systems optionally support MCA Thresholding. This feature is
discovered by checking capability bits in the MCA_MISC* registers.

Currently, MCA Thresholding is set up in two passes. The first is during
CPU init where available banks are detected, and the "bank_map" variable
is updated. The second is during sysfs/device init when the thresholding
data structures are allocated and hardware is fully configured.

During device init, the "threshold_banks" array is allocated even if no
available banks were discovered. Furthermore, the thresholding reset
flow checks if the top-level "threshold_banks" array is non-NULL, but it
doesn't check if individual "threshold_bank" structures are non-NULL.
This is avoided because the hardware interrupt is not enabled in this
case. But this issue becomes present if enabling the interrupt when the
thresholding data structures are not initialized.

Check "bank_map" to determine if the thresholding structures should be
allocated and initialized. Also, remove "mce_flags.amd_threshold" which
is redundant when checking "bank_map".

Signed-off-by: Yazen Ghannam <yazen.ghannam@xxxxxxx>
---

Notes:
Link:
https://lkml.kernel.org/r/20231118193248.1296798-16-yazen.ghannam@xxxxxxx

v1->v2:
* Update mce_vendor_flags reserved bits. (Yazen)

arch/x86/kernel/cpu/mce/amd.c | 2 +-
arch/x86/kernel/cpu/mce/core.c | 1 -
arch/x86/kernel/cpu/mce/internal.h | 5 +----
3 files changed, 2 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/cpu/mce/amd.c b/arch/x86/kernel/cpu/mce/amd.c
index 40912c5e35d1..08ee647cb6ce 100644
--- a/arch/x86/kernel/cpu/mce/amd.c
+++ b/arch/x86/kernel/cpu/mce/amd.c
@@ -1455,7 +1455,7 @@ int mce_threshold_create_device(unsigned int cpu)
struct threshold_bank **bp;
int err;

- if (!mce_flags.amd_threshold)
+ if (!this_cpu_read(bank_map))
return 0;

bp = this_cpu_read(threshold_banks);
diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index 308766868f39..17cf5a9df3cd 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -2024,7 +2024,6 @@ static void __mcheck_cpu_init_early(struct cpuinfo_x86 *c)
mce_flags.overflow_recov = !!cpu_has(c, X86_FEATURE_OVERFLOW_RECOV);
mce_flags.succor = !!cpu_has(c, X86_FEATURE_SUCCOR);
mce_flags.smca = !!cpu_has(c, X86_FEATURE_SMCA);
- mce_flags.amd_threshold = 1;
}
}

diff --git a/arch/x86/kernel/cpu/mce/internal.h b/arch/x86/kernel/cpu/mce/internal.h
index 96b108175ca2..9802f7c6cc93 100644
--- a/arch/x86/kernel/cpu/mce/internal.h
+++ b/arch/x86/kernel/cpu/mce/internal.h
@@ -214,9 +214,6 @@ struct mce_vendor_flags {
/* Zen IFU quirk */
zen_ifu_quirk : 1,

- /* AMD-style error thresholding banks present. */
- amd_threshold : 1,
-
/* Pentium, family 5-style MCA */
p5 : 1,

@@ -229,7 +226,7 @@ struct mce_vendor_flags {
/* Skylake, Cascade Lake, Cooper Lake REP;MOVS* quirk */
skx_repmov_quirk : 1,

- __reserved_0 : 55;
+ __reserved_0 : 56;
};

extern struct mce_vendor_flags mce_flags;
--
2.34.1