diff mbox series

[v2,2/2] x86/MCE/AMD, EDAC/mce_amd: Support non-uniform MCA bank type enumeration

Message ID 20211216162905.4132657-3-yazen.ghannam@amd.com (mailing list archive)
State New, archived
Headers show
Series AMD SMCA Updates | expand

Commit Message

Yazen Ghannam Dec. 16, 2021, 4:29 p.m. UTC
AMD systems currently lay out MCA bank types such that the type of bank
number "i" is either the same across all CPUs or is Reserved/Read-as-Zero.

For example:

  Bank # | CPUx | CPUy
    0      LS     LS
    1      RAZ    UMC
    2      CS     CS
    3      SMU    RAZ

Future AMD systems will lay out MCA bank types such that the type of
bank number "i" may be different across CPUs.

For example:

  Bank # | CPUx | CPUy
    0      LS     LS
    1      RAZ    UMC
    2      CS     NBIO
    3      SMU    RAZ

Change the structures that cache MCA bank types to be per-CPU and update
smca_get_bank_type() to handle this change.

Move some SMCA-specific structures to amd.c from mce.h, since they no
longer need to be global.

Break out the "count" for bank types from struct smca_hwid, since this
should provide a per-CPU count rather than a system-wide count.

Apply the "const" qualifier to the struct smca_hwid_mcatypes array. The
values in this array should not change at runtime.

Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
---
Link:
https://lkml.kernel.org/r/20211203020017.728440-4-yazen.ghannam@amd.com

v1->v2:
* Rework due to dropping patch 1 from v1.

 arch/x86/include/asm/mce.h              | 18 +-------
 arch/x86/kernel/cpu/mce/amd.c           | 59 +++++++++++++++----------
 drivers/edac/mce_amd.c                  | 11 +----
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c |  2 +-
 4 files changed, 39 insertions(+), 51 deletions(-)
diff mbox series

Patch

diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index 52d2b35cac2d..cc73061e7255 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -329,22 +329,6 @@  enum smca_bank_types {
 	N_SMCA_BANK_TYPES
 };
 
-#define HWID_MCATYPE(hwid, mcatype) (((hwid) << 16) | (mcatype))
-
-struct smca_hwid {
-	unsigned int bank_type;	/* Use with smca_bank_types for easy indexing. */
-	u32 hwid_mcatype;	/* (hwid,mcatype) tuple */
-	u8 count;		/* Number of instances. */
-};
-
-struct smca_bank {
-	struct smca_hwid *hwid;
-	u32 id;			/* Value of MCA_IPID[InstanceId]. */
-	u8 sysfs_id;		/* Value used for sysfs name. */
-};
-
-extern struct smca_bank smca_banks[MAX_NR_BANKS];
-
 extern const char *smca_get_long_name(enum smca_bank_types t);
 extern bool amd_mce_is_memory_error(struct mce *m);
 
@@ -352,7 +336,7 @@  extern int mce_threshold_create_device(unsigned int cpu);
 extern int mce_threshold_remove_device(unsigned int cpu);
 
 void mce_amd_feature_init(struct cpuinfo_x86 *c);
-enum smca_bank_types smca_get_bank_type(unsigned int bank);
+enum smca_bank_types smca_get_bank_type(unsigned int cpu, unsigned int bank);
 #else
 
 static inline int mce_threshold_create_device(unsigned int cpu)		{ return 0; };
diff --git a/arch/x86/kernel/cpu/mce/amd.c b/arch/x86/kernel/cpu/mce/amd.c
index ffe8ea8bc877..345ccdf329bc 100644
--- a/arch/x86/kernel/cpu/mce/amd.c
+++ b/arch/x86/kernel/cpu/mce/amd.c
@@ -71,6 +71,22 @@  static const char * const smca_umc_block_names[] = {
 	"misc_umc"
 };
 
+#define HWID_MCATYPE(hwid, mcatype) (((hwid) << 16) | (mcatype))
+
+struct smca_hwid {
+	unsigned int bank_type;	/* Use with smca_bank_types for easy indexing. */
+	u32 hwid_mcatype;	/* (hwid,mcatype) tuple */
+};
+
+struct smca_bank {
+	const struct smca_hwid *hwid;
+	u32 id;			/* Value of MCA_IPID[InstanceId]. */
+	u8 sysfs_id;		/* Value used for sysfs name. */
+};
+
+static DEFINE_PER_CPU_READ_MOSTLY(struct smca_bank[MAX_NR_BANKS], smca_banks);
+static DEFINE_PER_CPU_READ_MOSTLY(u8[N_SMCA_BANK_TYPES], smca_bank_counts);
+
 struct smca_bank_name {
 	const char *name;	/* Short name for sysfs */
 	const char *long_name;	/* Long name for pretty-printing */
@@ -126,14 +142,14 @@  const char *smca_get_long_name(enum smca_bank_types t)
 }
 EXPORT_SYMBOL_GPL(smca_get_long_name);
 
-enum smca_bank_types smca_get_bank_type(unsigned int bank)
+enum smca_bank_types smca_get_bank_type(unsigned int cpu, unsigned int bank)
 {
 	struct smca_bank *b;
 
 	if (bank >= MAX_NR_BANKS)
 		return N_SMCA_BANK_TYPES;
 
-	b = &smca_banks[bank];
+	b = &per_cpu(smca_banks, cpu)[bank];
 	if (!b->hwid)
 		return N_SMCA_BANK_TYPES;
 
@@ -141,7 +157,7 @@  enum smca_bank_types smca_get_bank_type(unsigned int bank)
 }
 EXPORT_SYMBOL_GPL(smca_get_bank_type);
 
-static struct smca_hwid smca_hwid_mcatypes[] = {
+static const struct smca_hwid smca_hwid_mcatypes[] = {
 	/* { bank_type, hwid_mcatype } */
 
 	/* Reserved type */
@@ -219,9 +235,6 @@  static struct smca_hwid smca_hwid_mcatypes[] = {
 	{ SMCA_GMI_PHY,	 HWID_MCATYPE(0x269, 0x0)	},
 };
 
-struct smca_bank smca_banks[MAX_NR_BANKS];
-EXPORT_SYMBOL_GPL(smca_banks);
-
 /*
  * In SMCA enabled processors, we can have multiple banks for a given IP type.
  * So to define a unique name for each bank, we use a temp c-string to append
@@ -277,8 +290,9 @@  static void smca_set_misc_banks_map(unsigned int bank, unsigned int cpu)
 
 static void smca_configure(unsigned int bank, unsigned int cpu)
 {
+	u8 *bank_counts = this_cpu_ptr(smca_bank_counts);
+	const struct smca_hwid *s_hwid;
 	unsigned int i, hwid_mcatype;
-	struct smca_hwid *s_hwid;
 	u32 high, low;
 	u32 smca_config = MSR_AMD64_SMCA_MCx_CONFIG(bank);
 
@@ -314,10 +328,6 @@  static void smca_configure(unsigned int bank, unsigned int cpu)
 
 	smca_set_misc_banks_map(bank, cpu);
 
-	/* Return early if this bank was already initialized. */
-	if (smca_banks[bank].hwid && smca_banks[bank].hwid->hwid_mcatype != 0)
-		return;
-
 	if (rdmsr_safe(MSR_AMD64_SMCA_MCx_IPID(bank), &low, &high)) {
 		pr_warn("Failed to read MCA_IPID for bank %d\n", bank);
 		return;
@@ -328,10 +338,11 @@  static void smca_configure(unsigned int bank, unsigned int cpu)
 
 	for (i = 0; i < ARRAY_SIZE(smca_hwid_mcatypes); i++) {
 		s_hwid = &smca_hwid_mcatypes[i];
+
 		if (hwid_mcatype == s_hwid->hwid_mcatype) {
-			smca_banks[bank].hwid = s_hwid;
-			smca_banks[bank].id = low;
-			smca_banks[bank].sysfs_id = s_hwid->count++;
+			this_cpu_ptr(smca_banks)[bank].hwid = s_hwid;
+			this_cpu_ptr(smca_banks)[bank].id = low;
+			this_cpu_ptr(smca_banks)[bank].sysfs_id = bank_counts[s_hwid->bank_type]++;
 			break;
 		}
 	}
@@ -617,7 +628,7 @@  prepare_threshold_block(unsigned int bank, unsigned int block, u32 addr,
 
 bool amd_filter_mce(struct mce *m)
 {
-	enum smca_bank_types bank_type = smca_get_bank_type(m->bank);
+	enum smca_bank_types bank_type = smca_get_bank_type(m->extcpu, m->bank);
 	struct cpuinfo_x86 *c = &boot_cpu_data;
 
 	/* See Family 17h Models 10h-2Fh Erratum #1114. */
@@ -655,7 +666,7 @@  static void disable_err_thresholding(struct cpuinfo_x86 *c, unsigned int bank)
 	} else if (c->x86 == 0x17 &&
 		   (c->x86_model >= 0x10 && c->x86_model <= 0x2F)) {
 
-		if (smca_get_bank_type(bank) != SMCA_IF)
+		if (smca_get_bank_type(smp_processor_id(), bank) != SMCA_IF)
 			return;
 
 		msrs[0] = MSR_AMD64_SMCA_MCx_MISC(bank);
@@ -723,7 +734,7 @@  bool amd_mce_is_memory_error(struct mce *m)
 	u8 xec = (m->status >> 16) & 0x1f;
 
 	if (mce_flags.smca)
-		return smca_get_bank_type(m->bank) == SMCA_UMC && xec == 0x0;
+		return smca_get_bank_type(m->extcpu, m->bank) == SMCA_UMC && xec == 0x0;
 
 	return m->bank == 4 && xec == 0x8;
 }
@@ -1039,7 +1050,7 @@  static struct kobj_type threshold_ktype = {
 	.release		= threshold_block_release,
 };
 
-static const char *get_name(unsigned int bank, struct threshold_block *b)
+static const char *get_name(unsigned int cpu, unsigned int bank, struct threshold_block *b)
 {
 	enum smca_bank_types bank_type;
 
@@ -1050,7 +1061,7 @@  static const char *get_name(unsigned int bank, struct threshold_block *b)
 		return th_names[bank];
 	}
 
-	bank_type = smca_get_bank_type(bank);
+	bank_type = smca_get_bank_type(cpu, bank);
 	if (bank_type >= N_SMCA_BANK_TYPES)
 		return NULL;
 
@@ -1060,12 +1071,12 @@  static const char *get_name(unsigned int bank, struct threshold_block *b)
 		return NULL;
 	}
 
-	if (smca_banks[bank].hwid->count == 1)
+	if (per_cpu(smca_bank_counts, cpu)[bank_type] == 1)
 		return smca_get_name(bank_type);
 
 	snprintf(buf_mcatype, MAX_MCATYPE_NAME_LEN,
-		 "%s_%x", smca_get_name(bank_type),
-			  smca_banks[bank].sysfs_id);
+		 "%s_%u", smca_get_name(bank_type),
+			  per_cpu(smca_banks, cpu)[bank].sysfs_id);
 	return buf_mcatype;
 }
 
@@ -1121,7 +1132,7 @@  static int allocate_threshold_blocks(unsigned int cpu, struct threshold_bank *tb
 	else
 		tb->blocks = b;
 
-	err = kobject_init_and_add(&b->kobj, &threshold_ktype, tb->kobj, get_name(bank, b));
+	err = kobject_init_and_add(&b->kobj, &threshold_ktype, tb->kobj, get_name(cpu, bank, b));
 	if (err)
 		goto out_free;
 recurse:
@@ -1176,7 +1187,7 @@  static int threshold_create_bank(struct threshold_bank **bp, unsigned int cpu,
 	struct device *dev = this_cpu_read(mce_device);
 	struct amd_northbridge *nb = NULL;
 	struct threshold_bank *b = NULL;
-	const char *name = get_name(bank, NULL);
+	const char *name = get_name(cpu, bank, NULL);
 	int err = 0;
 
 	if (!dev)
diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index cfd3f7ae9251..cc5c63feb26a 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -1166,20 +1166,13 @@  static void decode_mc6_mce(struct mce *m)
 /* Decode errors according to Scalable MCA specification */
 static void decode_smca_error(struct mce *m)
 {
-	struct smca_hwid *hwid;
-	enum smca_bank_types bank_type;
+	enum smca_bank_types bank_type = smca_get_bank_type(m->extcpu, m->bank);
 	const char *ip_name;
 	u8 xec = XEC(m->status, xec_mask);
 
-	if (m->bank >= ARRAY_SIZE(smca_banks))
+	if (bank_type >= N_SMCA_BANK_TYPES)
 		return;
 
-	hwid = smca_banks[m->bank].hwid;
-	if (!hwid)
-		return;
-
-	bank_type = hwid->bank_type;
-
 	if (bank_type == SMCA_RESERVED) {
 		pr_emerg(HW_ERR "Bank %d is reserved.\n", m->bank);
 		return;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 08133de21fdd..75dad0214dc7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -2647,7 +2647,7 @@  static int amdgpu_bad_page_notifier(struct notifier_block *nb,
 	 * and error occurred in DramECC (Extended error code = 0) then only
 	 * process the error, else bail out.
 	 */
-	if (!m || !((smca_get_bank_type(m->bank) == SMCA_UMC_V2) &&
+	if (!m || !((smca_get_bank_type(m->extcpu, m->bank) == SMCA_UMC_V2) &&
 		    (XEC(m->status, 0x3f) == 0x0)))
 		return NOTIFY_DONE;