Date: Fri, 3 Oct 2025 16:03:49 GMT From: "Jonathan T. Looney" <jtl@FreeBSD.org> To: src-committers@FreeBSD.org, dev-commits-src-all@FreeBSD.org, dev-commits-src-main@FreeBSD.org Subject: git: 5f1f7d8457d4 - main - x86: Keep cumulative MCA statistics in the kernel Message-ID: <202510031603.593G3nRt070824@gitrepo.freebsd.org>
next in thread | raw e-mail | index | archive | help
The branch main has been updated by jtl: URL: https://cgit.FreeBSD.org/src/commit/?id=5f1f7d8457d4fc28c6cff7e26a629a2d6ee3fc61 commit 5f1f7d8457d4fc28c6cff7e26a629a2d6ee3fc61 Author: Jonathan T. Looney <jtl@FreeBSD.org> AuthorDate: 2025-09-24 14:35:18 +0000 Commit: Jonathan T. Looney <jtl@FreeBSD.org> CommitDate: 2025-10-03 15:29:21 +0000 x86: Keep cumulative MCA statistics in the kernel Keeping cumulative MCA statistics in the kernel provides a way for users to get an accurate count of various kinds of errors reported by the CPU. After ca8929d2a3e9b9df31d2e487377f99d7c39aa01d, it is possible that the kernel will drop the record of some MCA interrupts. Moreover, this provides a cheaper interface to obtain statistics if that is the only reason a user is processing MCA logs. Reviewed by: markj Sponsored by: Netflix MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D52709 --- sys/x86/include/mca.h | 25 +++++++++++++++++++++++++ sys/x86/x86/mca.c | 43 ++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 65 insertions(+), 3 deletions(-) diff --git a/sys/x86/include/mca.h b/sys/x86/include/mca.h index 183480625f6d..553b5d765f17 100644 --- a/sys/x86/include/mca.h +++ b/sys/x86/include/mca.h @@ -44,6 +44,31 @@ struct mca_record { int mr_cpu; }; +enum mca_stat_types { + MCA_T_NONE = 0, + MCA_T_UNCLASSIFIED, + MCA_T_UCODE_ROM_PARITY, + MCA_T_EXTERNAL, + MCA_T_FRC, + MCA_T_INTERNAL_PARITY, + MCA_T_SMM_HANDLER, + MCA_T_INTERNAL_TIMER, + MCA_T_GENERIC_IO, + MCA_T_INTERNAL, + MCA_T_MEMORY, + MCA_T_TLB, + MCA_T_MEMCONTROLLER_GEN, + MCA_T_MEMCONTROLLER_RD, + MCA_T_MEMCONTROLLER_WR, + MCA_T_MEMCONTROLLER_AC, + MCA_T_MEMCONTROLLER_MS, + MCA_T_MEMCONTROLLER_OTHER, + MCA_T_CACHE, + MCA_T_BUS, + MCA_T_UNKNOWN, + MCA_T_COUNT /* Must stay last */ +}; + #ifdef _KERNEL void cmc_intr(void); diff --git a/sys/x86/x86/mca.c b/sys/x86/x86/mca.c index 4ba49469d3a2..1851df8d00a0 100644 --- a/sys/x86/x86/mca.c +++ b/sys/x86/x86/mca.c @@ -134,6 +134,12 @@ static struct task mca_resize_task; static struct timeout_task mca_scan_task; static struct mtx mca_lock; +/* Statistics on number of MCA events by type, updated atomically. */ +static uint64_t mca_stats[MCA_T_COUNT]; +SYSCTL_OPAQUE(_hw_mca, OID_AUTO, stats, CTLFLAG_RD | CTLFLAG_SKIP, + mca_stats, MCA_T_COUNT * sizeof(mca_stats[0]), + "S", "Array of MCA events by type"); + static unsigned int mca_ia32_ctl_reg(int bank) { @@ -356,21 +362,27 @@ mca_error_request(uint16_t mca_error) } static const char * -mca_error_mmtype(uint16_t mca_error) +mca_error_mmtype(uint16_t mca_error, enum mca_stat_types *event_type) { switch ((mca_error & 0x70) >> 4) { case 0x0: + *event_type = MCA_T_MEMCONTROLLER_GEN; return ("GEN"); case 0x1: + *event_type = MCA_T_MEMCONTROLLER_RD; return ("RD"); case 0x2: + *event_type = MCA_T_MEMCONTROLLER_WR; return ("WR"); case 0x3: + *event_type = MCA_T_MEMCONTROLLER_AC; return ("AC"); case 0x4: + *event_type = MCA_T_MEMCONTROLLER_MS; return ("MS"); } + *event_type = MCA_T_MEMCONTROLLER_OTHER; return ("???"); } @@ -426,6 +438,7 @@ static void mca_log(const struct mca_record *rec) { uint16_t mca_error; + enum mca_stat_types event_type; if (mca_mute(rec)) return; @@ -473,34 +486,44 @@ mca_log(const struct mca_record *rec) if (rec->mr_status & MC_STATUS_OVER) printf("OVER "); mca_error = rec->mr_status & MC_STATUS_MCA_ERROR; + event_type = MCA_T_COUNT; switch (mca_error) { /* Simple error codes. */ case 0x0000: printf("no error"); + event_type = MCA_T_NONE; break; case 0x0001: printf("unclassified error"); + event_type = MCA_T_UNCLASSIFIED; break; case 0x0002: printf("ucode ROM parity error"); + event_type = MCA_T_UCODE_ROM_PARITY; break; case 0x0003: printf("external error"); + event_type = MCA_T_EXTERNAL; break; case 0x0004: printf("FRC error"); + event_type = MCA_T_FRC; break; case 0x0005: printf("internal parity error"); + event_type = MCA_T_INTERNAL_PARITY; break; case 0x0006: printf("SMM handler code access violation"); + event_type = MCA_T_SMM_HANDLER; break; case 0x0400: printf("internal timer error"); + event_type = MCA_T_INTERNAL_TIMER; break; case 0x0e0b: printf("generic I/O error"); + event_type = MCA_T_GENERIC_IO; if (rec->mr_cpu_vendor_id == CPU_VENDOR_INTEL && (rec->mr_status & MC_STATUS_MISCV)) { printf(" (pci%d:%d:%d:%d)", @@ -513,6 +536,7 @@ mca_log(const struct mca_record *rec) default: if ((mca_error & 0xfc00) == 0x0400) { printf("internal error %x", mca_error & 0x03ff); + event_type = MCA_T_INTERNAL; break; } @@ -521,6 +545,7 @@ mca_log(const struct mca_record *rec) /* Memory hierarchy error. */ if ((mca_error & 0xeffc) == 0x000c) { printf("%s memory error", mca_error_level(mca_error)); + event_type = MCA_T_MEMORY; break; } @@ -528,12 +553,14 @@ mca_log(const struct mca_record *rec) if ((mca_error & 0xeff0) == 0x0010) { printf("%sTLB %s error", mca_error_ttype(mca_error), mca_error_level(mca_error)); + event_type = MCA_T_TLB; break; } /* Memory controller error. */ if ((mca_error & 0xef80) == 0x0080) { - printf("%s channel ", mca_error_mmtype(mca_error)); + printf("%s channel ", mca_error_mmtype(mca_error, + &event_type)); if ((mca_error & 0x000f) != 0x000f) printf("%d", mca_error & 0x000f); else @@ -548,12 +575,14 @@ mca_log(const struct mca_record *rec) mca_error_ttype(mca_error), mca_error_level(mca_error), mca_error_request(mca_error)); + event_type = MCA_T_CACHE; break; } /* Extended memory error. */ if ((mca_error & 0xef80) == 0x0280) { - printf("%s channel ", mca_error_mmtype(mca_error)); + printf("%s channel ", mca_error_mmtype(mca_error, + &event_type)); if ((mca_error & 0x000f) != 0x000f) printf("%d", mca_error & 0x000f); else @@ -565,6 +594,7 @@ mca_log(const struct mca_record *rec) /* Bus and/or Interconnect error. */ if ((mca_error & 0xe800) == 0x0800) { printf("BUS%s ", mca_error_level(mca_error)); + event_type = MCA_T_BUS; switch ((mca_error & 0x0600) >> 9) { case 0: printf("Source"); @@ -600,6 +630,7 @@ mca_log(const struct mca_record *rec) } printf("unknown error %x", mca_error); + event_type = MCA_T_UNKNOWN; break; } printf("\n"); @@ -615,6 +646,12 @@ mca_log(const struct mca_record *rec) } if (rec->mr_status & MC_STATUS_MISCV) printf("MCA: Misc 0x%llx\n", (long long)rec->mr_misc); + if (event_type < 0 || event_type >= MCA_T_COUNT) { + KASSERT(0, ("%s: invalid event type (%d)", __func__, + event_type)); + event_type = MCA_T_UNKNOWN; + } + atomic_add_64(&mca_stats[event_type], 1); } static bool
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?202510031603.593G3nRt070824>