[PATCH 00/12] EDAC, MCE: F15h MCE decoding

[PATCH 00/12] EDAC, MCE: F15h MCE decoding

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Hi all,

this is the final set of MCE decoding changes. It adds support for MCE
signatures generated by the new BD core. I'll add them to the linux-next
mix in the following days and one fine day aim them at the next merge
window.

All suggestions are welcome, as always.

Thanks.

[PATCH 01/12] EDAC, MCE: Select extended error code mask

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

F15h enlarges the extended error code of an MCE to a 5-bit field
(MCi_STATUS[20:16]). Add a mask variable which default 0xf is overridden
on F15h.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   13 +++++++++----
 1 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index c018109..01853ee 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -5,6 +5,7 @@
 
 static struct amd_decoder_ops *fam_ops;
 
+static u8 xec_mask	 = 0xf;
 static u8 nb_err_cpumask = 0xf;
 
 static bool report_gart_errors;
@@ -172,7 +173,7 @@ static bool f14h_dc_mce(u16 ec)
 static void amd_decode_dc_mce(struct mce *m)
 {
 	u16 ec = m->status & 0xffff;
-	u8 xec = (m->status >> 16) & 0xf;
+	u8 xec = (m->status >> 16) & xec_mask;
 
 	pr_emerg(HW_ERR "Data Cache Error: ");
 
@@ -257,7 +258,7 @@ static bool f14h_ic_mce(u16 ec)
 static void amd_decode_ic_mce(struct mce *m)
 {
 	u16 ec = m->status & 0xffff;
-	u8 xec = (m->status >> 16) & 0xf;
+	u8 xec = (m->status >> 16) & xec_mask;
 
 	pr_emerg(HW_ERR "Instruction Cache Error: ");
 
@@ -277,7 +278,7 @@ static void amd_decode_ic_mce(struct mce *m)
 static void amd_decode_bu_mce(struct mce *m)
 {
 	u32 ec = m->status & 0xffff;
-	u32 xec = (m->status >> 16) & 0xf;
+	u32 xec = (m->status >> 16) & xec_mask;
 
 	pr_emerg(HW_ERR "Bus Unit Error");
 
@@ -319,7 +320,7 @@ wrong_bu_mce:
 static void amd_decode_ls_mce(struct mce *m)
 {
 	u16 ec = m->status & 0xffff;
-	u8 xec = (m->status >> 16) & 0xf;
+	u8 xec = (m->status >> 16) & xec_mask;
 
 	if (boot_cpu_data.x86 == 0x14) {
 		pr_emerg("You shouldn't be seeing an LS MCE on this cpu family,"
@@ -651,6 +652,10 @@ static int __init mce_amd_init(void)
 		fam_ops->nb_mce = nb_noop_mce;
 		break;
 
+	case 0x15:
+		xec_mask = 0x1f;
+		break;
+
 	default:
 		printk(KERN_WARNING "Huh? What family is that: %d?!\n",
 				    boot_cpu_data.x86);
-- 
1.7.3.1.50.g1e633

[PATCH 02/12] EDAC, MCE: Add F15h DC MCE decoder

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Add a decoder for F15h DC MCEs to support the new types of DC MCEs
introduced by the BD microarchitecture.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   79 +++++++++++++++++++++++++++++++++++++-----------
 drivers/edac/mce_amd.h |    2 +-
 2 files changed, 62 insertions(+), 19 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 01853ee..12bae3b 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -75,7 +75,7 @@ static const char *f10h_nb_mce_desc[] = {
 	"ECC Error in the Probe Filter directory"
 };
 
-static bool f12h_dc_mce(u16 ec)
+static bool f12h_dc_mce(u16 ec, u8 xec)
 {
 	bool ret = false;
 
@@ -93,7 +93,7 @@ static bool f12h_dc_mce(u16 ec)
 	return ret;
 }
 
-static bool f10h_dc_mce(u16 ec)
+static bool f10h_dc_mce(u16 ec, u8 xec)
 {
 	u8 r4  = (ec >> 4) & 0xf;
 	u8 ll  = ec & 0x3;
@@ -102,20 +102,20 @@ static bool f10h_dc_mce(u16 ec)
 		pr_cont("during data scrub.\n");
 		return true;
 	}
-	return f12h_dc_mce(ec);
+	return f12h_dc_mce(ec, xec);
 }
 
-static bool k8_dc_mce(u16 ec)
+static bool k8_dc_mce(u16 ec, u8 xec)
 {
 	if (BUS_ERROR(ec)) {
 		pr_cont("during system linefill.\n");
 		return true;
 	}
 
-	return f10h_dc_mce(ec);
+	return f10h_dc_mce(ec, xec);
 }
 
-static bool f14h_dc_mce(u16 ec)
+static bool f14h_dc_mce(u16 ec, u8 xec)
 {
 	u8 r4	 = (ec >> 4) & 0xf;
 	u8 ll	 = ec & 0x3;
@@ -170,6 +170,54 @@ static bool f14h_dc_mce(u16 ec)
 	return ret;
 }
 
+static bool f15h_dc_mce(u16 ec, u8 xec)
+{
+	bool ret = true;
+
+	if (MEM_ERROR(ec)) {
+
+		switch (xec) {
+		case 0x0:
+			pr_cont("Data Array access error.\n");
+			break;
+
+		case 0x1:
+			pr_cont("UC error during a linefill from L2/NB.\n");
+			break;
+
+		case 0x2:
+		case 0x11:
+			pr_cont("STQ access error.\n");
+			break;
+
+		case 0x3:
+			pr_cont("SCB access error.\n");
+			break;
+
+		case 0x10:
+			pr_cont("Tag error.\n");
+			break;
+
+		case 0x12:
+			pr_cont("LDQ access error.\n");
+			break;
+
+		default:
+			ret = false;
+		}
+	} else if (BUS_ERROR(ec)) {
+
+		if (!xec)
+			pr_cont("during system linefill.\n");
+		else
+			pr_cont(" Internal %s condition.\n",
+				((xec == 1) ? "livelock" : "deadlock"));
+	} else
+		ret = false;
+
+	return ret;
+}
+
 static void amd_decode_dc_mce(struct mce *m)
 {
 	u16 ec = m->status & 0xffff;
@@ -183,20 +231,14 @@ static void amd_decode_dc_mce(struct mce *m)
 
 		if (tt == TT_DATA) {
 			pr_cont("%s TLB %s.\n", LL_MSG(ec),
-				(xec ? "multimatch" : "parity error"));
+				((xec == 2) ? "locked miss"
+					    : (xec ? "multimatch" : "parity")));
 			return;
 		}
-		else
-			goto wrong_dc_mce;
-	}
-
-	if (!fam_ops->dc_mce(ec))
-		goto wrong_dc_mce;
-
-	return;
-
-wrong_dc_mce:
-	pr_emerg(HW_ERR "Corrupted DC MCE info?\n");
+	} else if (fam_ops->dc_mce(ec, xec))
+		;
+	else
+		pr_emerg(HW_ERR "Corrupted DC MCE info?\n");
 }
 
 static bool k8_ic_mce(u16 ec)
@@ -654,6 +696,7 @@ static int __init mce_amd_init(void)
 
 	case 0x15:
 		xec_mask = 0x1f;
+		fam_ops->dc_mce = f15h_dc_mce;
 		break;
 
 	default:
diff --git a/drivers/edac/mce_amd.h b/drivers/edac/mce_amd.h
index 35f6e0e..c12394d 100644
--- a/drivers/edac/mce_amd.h
+++ b/drivers/edac/mce_amd.h
@@ -100,7 +100,7 @@ struct err_regs {
  * per-family decoder ops
  */
 struct amd_decoder_ops {
-	bool (*dc_mce)(u16);
+	bool (*dc_mce)(u16, u8);
 	bool (*ic_mce)(u16);
 	bool (*nb_mce)(u16, u8);
 };
-- 
1.7.3.1.50.g1e633

[PATCH 03/12] EDAC, MCE: Add F15h IC MCE decoder

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Add support for decoding F15h IC MCEs.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   53 +++++++++++++++++++++++++++++++++++++++++++++--
 drivers/edac/mce_amd.h |    2 +-
 2 files changed, 51 insertions(+), 4 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 12bae3b..158cd5f 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -75,6 +75,26 @@ static const char *f10h_nb_mce_desc[] = {
 	"ECC Error in the Probe Filter directory"
 };
 
+static const char * const f15h_ic_mce_desc[] = {
+	"UC during a demand linefill from L2",
+	"Parity error during data load from IC",
+	"Parity error for IC valid bit",
+	"Main tag parity error",
+	"Parity error in prediction queue",
+	"PFB data/address parity error",
+	"Parity error in the branch status reg",
+	"PFB promotion address error",
+	"Tag error during probe/victimization",
+	"Parity error for IC probe tag valid bit",
+	"PFB non-cacheable bit parity error",
+	"PFB valid bit parity error",			/* xec = 0xd */
+	"patch RAM",					/* xec = 010 */
+	"uop queue",
+	"insn buffer",
+	"predecode buffer",
+	"fetch address FIFO"
+};
+
 static bool f12h_dc_mce(u16 ec, u8 xec)
 {
 	bool ret = false;
@@ -241,7 +261,7 @@ static void amd_decode_dc_mce(struct mce *m)
 		pr_emerg(HW_ERR "Corrupted DC MCE info?\n");
 }
 
-static bool k8_ic_mce(u16 ec)
+static bool k8_ic_mce(u16 ec, u8 xec)
 {
 	u8 ll	 = ec & 0x3;
 	u8 r4	 = (ec >> 4) & 0xf;
@@ -276,7 +296,7 @@ static bool k8_ic_mce(u16 ec)
 	return ret;
 }
 
-static bool f14h_ic_mce(u16 ec)
+static bool f14h_ic_mce(u16 ec, u8 xec)
 {
 	u8 ll    = ec & 0x3;
 	u8 tt    = (ec >> 2) & 0x3;
@@ -297,6 +317,32 @@ static bool f14h_ic_mce(u16 ec)
 	return ret;
 }
 
+static bool f15h_ic_mce(u16 ec, u8 xec)
+{
+	bool ret = true;
+
+	if (!MEM_ERROR(ec))
+		return false;
+
+	switch (xec) {
+	case 0x0 ... 0xa:
+		pr_cont("%s.\n", f15h_ic_mce_desc[xec]);
+		break;
+
+	case 0xd:
+		pr_cont("%s.\n", f15h_ic_mce_desc[xec-2]);
+		break;
+
+	case 0x10 ... 0x14:
+		pr_cont("Decoder %s parity error.\n", f15h_ic_mce_desc[xec-4]);
+		break;
+
+	default:
+		ret = false;
+	}
+	return ret;
+}
+
 static void amd_decode_ic_mce(struct mce *m)
 {
 	u16 ec = m->status & 0xffff;
@@ -311,7 +357,7 @@ static void amd_decode_ic_mce(struct mce *m)
 		bool k8 = (boot_cpu_data.x86 == 0xf && (m->status & BIT_64(58)));
 
 		pr_cont("during %s.\n", (k8 ? "system linefill" : "NB data read"));
-	} else if (fam_ops->ic_mce(ec))
+	} else if (fam_ops->ic_mce(ec, xec))
 		;
 	else
 		pr_emerg(HW_ERR "Corrupted IC MCE info?\n");
@@ -697,6 +743,7 @@ static int __init mce_amd_init(void)
 	case 0x15:
 		xec_mask = 0x1f;
 		fam_ops->dc_mce = f15h_dc_mce;
+		fam_ops->ic_mce = f15h_ic_mce;
 		break;
 
 	default:
diff --git a/drivers/edac/mce_amd.h b/drivers/edac/mce_amd.h
index c12394d..8398847 100644
--- a/drivers/edac/mce_amd.h
+++ b/drivers/edac/mce_amd.h
@@ -101,7 +101,7 @@ struct err_regs {
  */
 struct amd_decoder_ops {
 	bool (*dc_mce)(u16, u8);
-	bool (*ic_mce)(u16);
+	bool (*ic_mce)(u16, u8);
 	bool (*nb_mce)(u16, u8);
 };
 
-- 
1.7.3.1.50.g1e633

[PATCH 04/12] EDAC, MCE: Add F15h CU MCE decoder

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

MCE bank 2 is redefined from a BU to a CU (Combined Unit) bank on F15h.
Add a decoder function for CU MCEs.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   62 +++++++++++++++++++++++++++++++++++++++++++++++-
 1 files changed, 61 insertions(+), 1 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 158cd5f..7eb2706 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -95,6 +95,23 @@ static const char * const f15h_ic_mce_desc[] = {
 	"fetch address FIFO"
 };
 
+static const char * const f15h_cu_mce_desc[] = {
+	"Fill ECC error on data fills",			/* xec = 0x4 */
+	"Fill parity error on insn fills",
+	"Prefetcher request FIFO parity error",
+	"PRQ address parity error",
+	"PRQ data parity error",
+	"WCC Tag ECC error",
+	"WCC Data ECC error",
+	"WCB Data parity error",
+	"VB Data/ECC error",
+	"L2 Tag ECC error",				/* xec = 0x10 */
+	"Hard L2 Tag ECC error",
+	"Multiple hits on L2 tag",
+	"XAB parity error",
+	"PRB address parity error"
+};
+
 static bool f12h_dc_mce(u16 ec, u8 xec)
 {
 	bool ret = false;
@@ -405,6 +422,46 @@ wrong_bu_mce:
 	pr_emerg(HW_ERR "Corrupted BU MCE info?\n");
 }
 
+static void amd_decode_cu_mce(struct mce *m)
+{
+	u16 ec = m->status & 0xffff;
+	u8 xec = (m->status >> 16) & xec_mask;
+
+	pr_emerg(HW_ERR "Combined Unit Error: ");
+
+	if (TLB_ERROR(ec)) {
+		if (xec == 0x0)
+			pr_cont("Data parity TLB read error.\n");
+		else if (xec == 0x1)
+			pr_cont("Poison data provided for TLB fill.\n");
+		else
+			goto wrong_cu_mce;
+	} else if (BUS_ERROR(ec)) {
+		if (xec > 2)
+			goto wrong_cu_mce;
+
+		pr_cont("Error during attempted NB data read.\n");
+	} else if (MEM_ERROR(ec)) {
+		switch (xec) {
+		case 0x4 ... 0xc:
+			pr_cont("%s.\n", f15h_cu_mce_desc[xec - 0x4]);
+			break;
+
+		case 0x10 ... 0x14:
+			pr_cont("%s.\n", f15h_cu_mce_desc[xec - 0x7]);
+			break;
+
+		default:
+			goto wrong_cu_mce;
+		}
+	}
+
+	return;
+
+wrong_cu_mce:
+	pr_emerg(HW_ERR "Corrupted CU MCE info?\n");
+}
+
 static void amd_decode_ls_mce(struct mce *m)
 {
 	u16 ec = m->status & 0xffff;
@@ -665,7 +722,10 @@ int amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
 		break;
 
 	case 2:
-		amd_decode_bu_mce(m);
+		if (boot_cpu_data.x86 == 0x15)
+			amd_decode_cu_mce(m);
+		else
+			amd_decode_bu_mce(m);
 		break;
 
 	case 3:
-- 
1.7.3.1.50.g1e633

[PATCH 05/12] EDAC, MCE: No F15h LS MCE decoder

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

F15h BD doesn't generate LS MCEs so warn about it.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 7eb2706..44e6cda 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -467,7 +467,7 @@ static void amd_decode_ls_mce(struct mce *m)
 	u16 ec = m->status & 0xffff;
 	u8 xec = (m->status >> 16) & xec_mask;
 
-	if (boot_cpu_data.x86 == 0x14) {
+	if (boot_cpu_data.x86 >= 0x14) {
 		pr_emerg("You shouldn't be seeing an LS MCE on this cpu family,"
 			 " please report on LKML.\n");
 		return;
-- 
1.7.3.1.50.g1e633

[PATCH 06/12] EDAC, MCE: Add an F15h NB MCE decoder

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

by (almost) reusing the F10h one since the signatures are the same.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   10 ++++++++++
 1 files changed, 10 insertions(+), 0 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 44e6cda..efbcb53 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -556,6 +556,15 @@ static bool f10h_nb_mce(u16 ec, u8 xec)
 		goto out;
 		break;
 
+	case 0x19:
+		if (boot_cpu_data.x86 == 0x15)
+			pr_cont("Compute Unit Data Error.\n");
+		else
+			ret = false;
+
+		goto out;
+		break;
+
 	case 0x1c ... 0x1f:
 		offset = 24;
 		break;
@@ -804,6 +813,7 @@ static int __init mce_amd_init(void)
 		xec_mask = 0x1f;
 		fam_ops->dc_mce = f15h_dc_mce;
 		fam_ops->ic_mce = f15h_ic_mce;
+		fam_ops->nb_mce = f10h_nb_mce;
 		break;
 
 	default:
-- 
1.7.3.1.50.g1e633

[PATCH 07/12] EDAC, MCE: Add F15 EX MCE decoder

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Integrate the single FIROB signature into an expanded table along with
the new BD MCE types.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   41 ++++++++++++++++++++++++++++++++++-------
 1 files changed, 34 insertions(+), 7 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index efbcb53..afda6de5 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -112,6 +112,22 @@ static const char * const f15h_cu_mce_desc[] = {
 	"PRB address parity error"
 };
 
+static const char * const fr_ex_mce_desc[] = {
+	"CPU Watchdog timer expire",
+	"Wakeup array dest tag",
+	"AG payload array",
+	"EX payload array",
+	"IDRF array",
+	"Retire dispatch queue",
+	"Mapper checkpoint array",
+	"Physical register file EX0 port",
+	"Physical register file EX1 port",
+	"Physical register file AG0 port",
+	"Physical register file AG1 port",
+	"Flag register file",
+	"DE correctable error could not be corrected"
+};
+
 static bool f12h_dc_mce(u16 ec, u8 xec)
 {
 	bool ret = false;
@@ -651,15 +667,26 @@ EXPORT_SYMBOL_GPL(amd_decode_nb_mce);
 
 static void amd_decode_fr_mce(struct mce *m)
 {
-	if (boot_cpu_data.x86 == 0xf ||
-	    boot_cpu_data.x86 == 0x11)
+	struct cpuinfo_x86 *c = &boot_cpu_data;
+	u8 xec = (m->status >> 16) & xec_mask;
+
+	if (c->x86 == 0xf || c->x86 == 0x11)
 		goto wrong_fr_mce;
 
-	/* we have only one error signature so match all fields at once. */
-	if ((m->status & 0xffff) == 0x0f0f) {
-		pr_emerg(HW_ERR "FR Error: CPU Watchdog timer expire.\n");
-		return;
-	}
+	if (c->x86 != 0x15 && xec != 0x0)
+		goto wrong_fr_mce;
+
+	pr_emerg(HW_ERR "%s Error: ",
+		 (c->x86 == 0x15 ? "Execution Unit" : "FIROB"));
+
+	if (xec == 0x0 || xec == 0xc)
+		pr_cont("%s.\n", fr_ex_mce_desc[xec]);
+	else if (xec < 0xd)
+		pr_cont("%s parity error.\n", fr_ex_mce_desc[xec]);
+	else
+		goto wrong_fr_mce;
+
+	return;
 
 wrong_fr_mce:
 	pr_emerg(HW_ERR "Corrupted FR MCE info?\n");
-- 
1.7.3.1.50.g1e633

[PATCH 08/12] EDAC, MCE: Add F15h FP MCE decoder

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Add decoder for FP MCEs.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   44 ++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 44 insertions(+), 0 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index afda6de5..c14abe3 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -692,6 +692,46 @@ wrong_fr_mce:
 	pr_emerg(HW_ERR "Corrupted FR MCE info?\n");
 }
 
+static void amd_decode_fp_mce(struct mce *m)
+{
+	u8 xec = (m->status >> 16) & xec_mask;
+
+	pr_emerg(HW_ERR "Floating Point Unit Error: ");
+
+	switch (xec) {
+	case 0x1:
+		pr_cont("Free List");
+		break;
+
+	case 0x2:
+		pr_cont("Physical Register File");
+		break;
+
+	case 0x3:
+		pr_cont("Retire Queue");
+		break;
+
+	case 0x4:
+		pr_cont("Scheduler table");
+		break;
+
+	case 0x5:
+		pr_cont("Status Register File");
+		break;
+
+	default:
+		goto wrong_fp_mce;
+		break;
+	}
+
+	pr_cont(" parity error.\n");
+
+	return;
+
+wrong_fp_mce:
+	pr_emerg(HW_ERR "Corrupted FP MCE info?\n");
+}
+
 static inline void amd_decode_err_code(u16 ec)
 {
 	if (TLB_ERROR(ec)) {
@@ -777,6 +817,10 @@ int amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
 		amd_decode_fr_mce(m);
 		break;
 
+	case 6:
+		amd_decode_fp_mce(m);
+		break;
+
 	default:
 		break;
 	}
-- 
1.7.3.1.50.g1e633

[PATCH 09/12] EDAC, MCE: Overhaul error fields extraction macros

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Make macro names shorter thus making code shorter and more clear.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/amd64_edac.c |    4 +-
 drivers/edac/mce_amd.c    |   83 +++++++++++++++++++-------------------------
 drivers/edac/mce_amd.h    |   10 +++---
 3 files changed, 43 insertions(+), 54 deletions(-)

diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c
index 8521401..dfbaa97 100644
--- a/drivers/edac/amd64_edac.c
+++ b/drivers/edac/amd64_edac.c
@@ -2055,8 +2055,8 @@ static void amd64_handle_ue(struct mem_ctl_info *mci,
 static inline void __amd64_decode_bus_error(struct mem_ctl_info *mci,
 					    struct err_regs *info)
 {
-	u32 ec  = ERROR_CODE(info->nbsl);
-	u32 xec = EXT_ERROR_CODE(info->nbsl);
+	u16 ec = EC(info->nbsl);
+	u8 xec = XEC(info->nbsl, 0x1f);
 	int ecc_type = (info->nbsh >> 13) & 0x3;
 
 	/* Bail early out if this was an 'observed' error */
diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index c14abe3..53d4dc0 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -133,13 +133,13 @@ static bool f12h_dc_mce(u16 ec, u8 xec)
 	bool ret = false;
 
 	if (MEM_ERROR(ec)) {
-		u8 ll = ec & 0x3;
+		u8 ll = LL(ec);
 		ret = true;
 
 		if (ll == LL_L2)
 			pr_cont("during L1 linefill from L2.\n");
 		else if (ll == LL_L1)
-			pr_cont("Data/Tag %s error.\n", RRRR_MSG(ec));
+			pr_cont("Data/Tag %s error.\n", R4_MSG(ec));
 		else
 			ret = false;
 	}
@@ -148,10 +148,7 @@ static bool f12h_dc_mce(u16 ec, u8 xec)
 
 static bool f10h_dc_mce(u16 ec, u8 xec)
 {
-	u8 r4  = (ec >> 4) & 0xf;
-	u8 ll  = ec & 0x3;
-
-	if (r4 == R4_GEN && ll == LL_L1) {
+	if (R4(ec) == R4_GEN && LL(ec) == LL_L1) {
 		pr_cont("during data scrub.\n");
 		return true;
 	}
@@ -170,15 +167,12 @@ static bool k8_dc_mce(u16 ec, u8 xec)
 
 static bool f14h_dc_mce(u16 ec, u8 xec)
 {
-	u8 r4	 = (ec >> 4) & 0xf;
-	u8 ll	 = ec & 0x3;
-	u8 tt	 = (ec >> 2) & 0x3;
-	u8 ii	 = tt;
+	u8 r4	 = R4(ec);
 	bool ret = true;
 
 	if (MEM_ERROR(ec)) {
 
-		if (tt != TT_DATA || ll != LL_L1)
+		if (TT(ec) != TT_DATA || LL(ec) != LL_L1)
 			return false;
 
 		switch (r4) {
@@ -198,7 +192,7 @@ static bool f14h_dc_mce(u16 ec, u8 xec)
 		}
 	} else if (BUS_ERROR(ec)) {
 
-		if ((ii != II_MEM && ii != II_IO) || ll != LL_LG)
+		if ((II(ec) != II_MEM && II(ec) != II_IO) || LL(ec) != LL_LG)
 			return false;
 
 		pr_cont("System read data error on a ");
@@ -273,16 +267,14 @@ static bool f15h_dc_mce(u16 ec, u8 xec)
 
 static void amd_decode_dc_mce(struct mce *m)
 {
-	u16 ec = m->status & 0xffff;
-	u8 xec = (m->status >> 16) & xec_mask;
+	u16 ec = EC(m->status);
+	u8 xec = XEC(m->status, xec_mask);
 
 	pr_emerg(HW_ERR "Data Cache Error: ");
 
 	/* TLB error signatures are the same across families */
 	if (TLB_ERROR(ec)) {
-		u8 tt = (ec >> 2) & 0x3;
-
-		if (tt == TT_DATA) {
+		if (TT(ec) == TT_DATA) {
 			pr_cont("%s TLB %s.\n", LL_MSG(ec),
 				((xec == 2) ? "locked miss"
 					    : (xec ? "multimatch" : "parity")));
@@ -296,8 +288,7 @@ static void amd_decode_dc_mce(struct mce *m)
 
 static bool k8_ic_mce(u16 ec, u8 xec)
 {
-	u8 ll	 = ec & 0x3;
-	u8 r4	 = (ec >> 4) & 0xf;
+	u8 ll	 = LL(ec);
 	bool ret = true;
 
 	if (!MEM_ERROR(ec))
@@ -306,7 +297,7 @@ static bool k8_ic_mce(u16 ec, u8 xec)
 	if (ll == 0x2)
 		pr_cont("during a linefill from L2.\n");
 	else if (ll == 0x1) {
-		switch (r4) {
+		switch (R4(ec)) {
 		case R4_IRD:
 			pr_cont("Parity error during data load.\n");
 			break;
@@ -331,13 +322,11 @@ static bool k8_ic_mce(u16 ec, u8 xec)
 
 static bool f14h_ic_mce(u16 ec, u8 xec)
 {
-	u8 ll    = ec & 0x3;
-	u8 tt    = (ec >> 2) & 0x3;
-	u8 r4  = (ec >> 4) & 0xf;
+	u8 r4    = R4(ec);
 	bool ret = true;
 
 	if (MEM_ERROR(ec)) {
-		if (tt != 0 || ll != 1)
+		if (TT(ec) != 0 || LL(ec) != 1)
 			ret = false;
 
 		if (r4 == R4_IRD)
@@ -378,8 +367,8 @@ static bool f15h_ic_mce(u16 ec, u8 xec)
 
 static void amd_decode_ic_mce(struct mce *m)
 {
-	u16 ec = m->status & 0xffff;
-	u8 xec = (m->status >> 16) & xec_mask;
+	u16 ec = EC(m->status);
+	u8 xec = XEC(m->status, xec_mask);
 
 	pr_emerg(HW_ERR "Instruction Cache Error: ");
 
@@ -398,8 +387,8 @@ static void amd_decode_ic_mce(struct mce *m)
 
 static void amd_decode_bu_mce(struct mce *m)
 {
-	u32 ec = m->status & 0xffff;
-	u32 xec = (m->status >> 16) & xec_mask;
+	u16 ec = EC(m->status);
+	u8 xec = XEC(m->status, xec_mask);
 
 	pr_emerg(HW_ERR "Bus Unit Error");
 
@@ -408,23 +397,23 @@ static void amd_decode_bu_mce(struct mce *m)
 	else if (xec == 0x3)
 		pr_cont(" in the victim data buffers.\n");
 	else if (xec == 0x2 && MEM_ERROR(ec))
-		pr_cont(": %s error in the L2 cache tags.\n", RRRR_MSG(ec));
+		pr_cont(": %s error in the L2 cache tags.\n", R4_MSG(ec));
 	else if (xec == 0x0) {
 		if (TLB_ERROR(ec))
 			pr_cont(": %s error in a Page Descriptor Cache or "
 				"Guest TLB.\n", TT_MSG(ec));
 		else if (BUS_ERROR(ec))
 			pr_cont(": %s/ECC error in data read from NB: %s.\n",
-				RRRR_MSG(ec), PP_MSG(ec));
+				R4_MSG(ec), PP_MSG(ec));
 		else if (MEM_ERROR(ec)) {
-			u8 rrrr = (ec >> 4) & 0xf;
+			u8 r4 = R4(ec);
 
-			if (rrrr >= 0x7)
+			if (r4 >= 0x7)
 				pr_cont(": %s error during data copyback.\n",
-					RRRR_MSG(ec));
-			else if (rrrr <= 0x1)
+					R4_MSG(ec));
+			else if (r4 <= 0x1)
 				pr_cont(": %s parity/ECC error during data "
-					"access from L2.\n", RRRR_MSG(ec));
+					"access from L2.\n", R4_MSG(ec));
 			else
 				goto wrong_bu_mce;
 		} else
@@ -440,8 +429,8 @@ wrong_bu_mce:
 
 static void amd_decode_cu_mce(struct mce *m)
 {
-	u16 ec = m->status & 0xffff;
-	u8 xec = (m->status >> 16) & xec_mask;
+	u16 ec = EC(m->status);
+	u8 xec = XEC(m->status, xec_mask);
 
 	pr_emerg(HW_ERR "Combined Unit Error: ");
 
@@ -480,8 +469,8 @@ wrong_cu_mce:
 
 static void amd_decode_ls_mce(struct mce *m)
 {
-	u16 ec = m->status & 0xffff;
-	u8 xec = (m->status >> 16) & xec_mask;
+	u16 ec = EC(m->status);
+	u8 xec = XEC(m->status, xec_mask);
 
 	if (boot_cpu_data.x86 >= 0x14) {
 		pr_emerg("You shouldn't be seeing an LS MCE on this cpu family,"
@@ -492,12 +481,12 @@ static void amd_decode_ls_mce(struct mce *m)
 	pr_emerg(HW_ERR "Load Store Error");
 
 	if (xec == 0x0) {
-		u8 r4 = (ec >> 4) & 0xf;
+		u8 r4 = R4(ec);
 
 		if (!BUS_ERROR(ec) || (r4 != R4_DRD && r4 != R4_DWR))
 			goto wrong_ls_mce;
 
-		pr_cont(" during %s.\n", RRRR_MSG(ec));
+		pr_cont(" during %s.\n", R4_MSG(ec));
 	} else
 		goto wrong_ls_mce;
 
@@ -605,8 +594,8 @@ static bool nb_noop_mce(u16 ec, u8 xec)
 
 void amd_decode_nb_mce(int node_id, struct mce *m, u32 nbcfg)
 {
-	u8 xec   = (m->status >> 16) & 0x1f;
-	u16 ec   = m->status & 0xffff;
+	u16 ec   = EC(m->status);
+	u8 xec   = XEC(m->status, 0x1f);
 	u32 nbsh = (u32)(m->status >> 32);
 
 	pr_emerg(HW_ERR "Northbridge Error, node %d: ", node_id);
@@ -668,7 +657,7 @@ EXPORT_SYMBOL_GPL(amd_decode_nb_mce);
 static void amd_decode_fr_mce(struct mce *m)
 {
 	struct cpuinfo_x86 *c = &boot_cpu_data;
-	u8 xec = (m->status >> 16) & xec_mask;
+	u8 xec = XEC(m->status, xec_mask);
 
 	if (c->x86 == 0xf || c->x86 == 0x11)
 		goto wrong_fr_mce;
@@ -694,7 +683,7 @@ wrong_fr_mce:
 
 static void amd_decode_fp_mce(struct mce *m)
 {
-	u8 xec = (m->status >> 16) & xec_mask;
+	u8 xec = XEC(m->status, xec_mask);
 
 	pr_emerg(HW_ERR "Floating Point Unit Error: ");
 
@@ -739,11 +728,11 @@ static inline void amd_decode_err_code(u16 ec)
 			 TT_MSG(ec), LL_MSG(ec));
 	} else if (MEM_ERROR(ec)) {
 		pr_emerg(HW_ERR "Transaction: %s, Type: %s, Cache Level: %s\n",
-			 RRRR_MSG(ec), TT_MSG(ec), LL_MSG(ec));
+			 R4_MSG(ec), TT_MSG(ec), LL_MSG(ec));
 	} else if (BUS_ERROR(ec)) {
 		pr_emerg(HW_ERR "Transaction: %s (%s), %s, Cache Level: %s, "
 			 "Participating Processor: %s\n",
-			  RRRR_MSG(ec), II_MSG(ec), TO_MSG(ec), LL_MSG(ec),
+			  R4_MSG(ec), II_MSG(ec), TO_MSG(ec), LL_MSG(ec),
 			  PP_MSG(ec));
 	} else
 		pr_emerg(HW_ERR "Huh? Unknown MCE error 0x%x\n", ec);
diff --git a/drivers/edac/mce_amd.h b/drivers/edac/mce_amd.h
index 8398847..45dda47 100644
--- a/drivers/edac/mce_amd.h
+++ b/drivers/edac/mce_amd.h
@@ -7,8 +7,8 @@
 
 #define BIT_64(n)			(U64_C(1) << (n))
 
-#define ERROR_CODE(x)			((x) & 0xffff)
-#define EXT_ERROR_CODE(x)		(((x) >> 16) & 0x1f)
+#define EC(x)				((x) & 0xffff)
+#define XEC(x, mask)			(((x) >> 16) & mask)
 
 #define LOW_SYNDROME(x)			(((x) >> 15) & 0xff)
 #define HIGH_SYNDROME(x)		(((x) >> 24) & 0xff)
@@ -21,15 +21,15 @@
 #define TT_MSG(x)			tt_msgs[TT(x)]
 #define II(x)				(((x) >> 2) & 0x3)
 #define II_MSG(x)			ii_msgs[II(x)]
-#define LL(x)				(((x) >> 0) & 0x3)
+#define LL(x)				((x) & 0x3)
 #define LL_MSG(x)			ll_msgs[LL(x)]
 #define TO(x)				(((x) >> 8) & 0x1)
 #define TO_MSG(x)			to_msgs[TO(x)]
 #define PP(x)				(((x) >> 9) & 0x3)
 #define PP_MSG(x)			pp_msgs[PP(x)]
 
-#define RRRR(x)				(((x) >> 4) & 0xf)
-#define RRRR_MSG(x)			((RRRR(x) < 9) ?  rrrr_msgs[RRRR(x)] : "Wrong R4!")
+#define R4(x)				(((x) >> 4) & 0xf)
+#define R4_MSG(x)			((R4(x) < 9) ?  rrrr_msgs[R4(x)] : "Wrong R4!")
 
 #define K8_NBSH				0x4C
 
-- 
1.7.3.1.50.g1e633

[PATCH 10/12] EDAC, MCE: Shorten error report formatting

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Shorten up MCi_STATUS flags and add BD's new deferred and poison types.
Also, simplify formatting.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   54 ++++++++++++++++++++++++++++-------------------
 1 files changed, 32 insertions(+), 22 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 53d4dc0..0a19d26 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -723,19 +723,22 @@ wrong_fp_mce:
 
 static inline void amd_decode_err_code(u16 ec)
 {
-	if (TLB_ERROR(ec)) {
-		pr_emerg(HW_ERR "Transaction: %s, Cache Level: %s\n",
-			 TT_MSG(ec), LL_MSG(ec));
-	} else if (MEM_ERROR(ec)) {
-		pr_emerg(HW_ERR "Transaction: %s, Type: %s, Cache Level: %s\n",
-			 R4_MSG(ec), TT_MSG(ec), LL_MSG(ec));
-	} else if (BUS_ERROR(ec)) {
-		pr_emerg(HW_ERR "Transaction: %s (%s), %s, Cache Level: %s, "
-			 "Participating Processor: %s\n",
-			  R4_MSG(ec), II_MSG(ec), TO_MSG(ec), LL_MSG(ec),
-			  PP_MSG(ec));
-	} else
-		pr_emerg(HW_ERR "Huh? Unknown MCE error 0x%x\n", ec);
+
+	pr_emerg(HW_ERR "cache level: %s", LL_MSG(ec));
+
+	if (BUS_ERROR(ec))
+		pr_cont(", mem/io: %s", II_MSG(ec));
+	else
+		pr_cont(", tx: %s", TT_MSG(ec));
+
+	if (MEM_ERROR(ec) || BUS_ERROR(ec)) {
+		pr_cont(", mem-tx: %s", R4_MSG(ec));
+
+		if (BUS_ERROR(ec))
+			pr_cont(", part-proc: %s (%s)", PP_MSG(ec), TO_MSG(ec));
+	}
+
+	pr_cont("\n");
 }
 
 /*
@@ -757,25 +760,32 @@ static bool amd_filter_mce(struct mce *m)
 int amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
 {
 	struct mce *m = (struct mce *)data;
+	struct cpuinfo_x86 *c = &boot_cpu_data;
 	int node, ecc;
 
 	if (amd_filter_mce(m))
 		return NOTIFY_STOP;
 
-	pr_emerg(HW_ERR "MC%d_STATUS: ", m->bank);
+	pr_emerg(HW_ERR "MC%d_STATUS[%s|%s|%s|%s|%s",
+		m->bank,
+		((m->status & MCI_STATUS_OVER)	? "Over"  : "-"),
+		((m->status & MCI_STATUS_UC)	? "UE"	  : "CE"),
+		((m->status & MCI_STATUS_MISCV)	? "MiscV" : "-"),
+		((m->status & MCI_STATUS_PCC)	? "PCC"	  : "-"),
+		((m->status & MCI_STATUS_ADDRV)	? "AddrV" : "-"));
 
-	pr_cont("%sorrected error, other errors lost: %s, "
-		 "CPU context corrupt: %s",
-		 ((m->status & MCI_STATUS_UC) ? "Unc"  : "C"),
-		 ((m->status & MCI_STATUS_OVER) ? "yes"  : "no"),
-		 ((m->status & MCI_STATUS_PCC) ? "yes" : "no"));
+	if (c->x86 == 0x15)
+		pr_cont("|%s|%s",
+			((m->status & BIT(44))	? "Deferred" : "-"),
+			((m->status & BIT(43))	? "Poison"   : "-"));
 
 	/* do the two bits[14:13] together */
 	ecc = (m->status >> 45) & 0x3;
 	if (ecc)
-		pr_cont(", %sECC Error", ((ecc == 2) ? "C" : "U"));
+		pr_cont("|%sECC", ((ecc == 2) ? "C" : "U"));
+
+	pr_cont("]: 0x%016llx\n", m->status);
 
-	pr_cont("\n");
 
 	switch (m->bank) {
 	case 0:
@@ -787,7 +797,7 @@ int amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
 		break;
 
 	case 2:
-		if (boot_cpu_data.x86 == 0x15)
+		if (c->x86 == 0x15)
 			amd_decode_cu_mce(m);
 		else
 			amd_decode_bu_mce(m);
-- 
1.7.3.1.50.g1e633

[PATCH 11/12] EDAC, MCE: Allow F15h bank 6 MCE injection

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

F15h adds a sixth MCE bank: adjust bank number check in the injection
code.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd_inj.c |    9 +++++----
 1 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/edac/mce_amd_inj.c b/drivers/edac/mce_amd_inj.c
index 8d0688f..6db494f 100644
--- a/drivers/edac/mce_amd_inj.c
+++ b/drivers/edac/mce_amd_inj.c
@@ -88,10 +88,11 @@ static ssize_t edac_inject_bank_store(struct kobject *kobj,
 		return -EINVAL;
 	}
 
-	if (value > 5) {
-		printk(KERN_ERR "Non-existant MCE bank: %lu\n", value);
-		return -EINVAL;
-	}
+	if (value > 5)
+		if (boot_cpu_data.x86 != 0x15 || value > 6) {
+			printk(KERN_ERR "Non-existant MCE bank: %lu\n", value);
+			return -EINVAL;
+		}
 
 	i_mce.bank = value;
 
-- 
1.7.3.1.50.g1e633

[PATCH 12/12] EDAC, MCE: Enable MCE decoding on F15h

[Borislav Petkov]

From: Borislav Petkov <borislav.petkov@amd.com>

Now that everything is inplace, enable MCE decoding on F15h. Make
initcall routine a bit more readable.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
---
 drivers/edac/mce_amd.c |   14 ++++++++------
 1 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c
index 0a19d26..59ad34c 100644
--- a/drivers/edac/mce_amd.c
+++ b/drivers/edac/mce_amd.c
@@ -836,18 +836,21 @@ static struct notifier_block amd_mce_dec_nb = {
 
 static int __init mce_amd_init(void)
 {
-	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD)
+	struct cpuinfo_x86 *c = &boot_cpu_data;
+
+	if (c->x86_vendor != X86_VENDOR_AMD)
 		return 0;
 
-	if ((boot_cpu_data.x86 < 0xf || boot_cpu_data.x86 > 0x12) &&
-	    (boot_cpu_data.x86 != 0x14 || boot_cpu_data.x86_model > 0xf))
+	if ((c->x86 < 0xf || c->x86 > 0x12) &&
+	    (c->x86 != 0x14 || c->x86_model > 0xf) &&
+	    (c->x86 != 0x15 || c->x86_model > 0xf))
 		return 0;
 
 	fam_ops = kzalloc(sizeof(struct amd_decoder_ops), GFP_KERNEL);
 	if (!fam_ops)
 		return -ENOMEM;
 
-	switch (boot_cpu_data.x86) {
+	switch (c->x86) {
 	case 0xf:
 		fam_ops->dc_mce = k8_dc_mce;
 		fam_ops->ic_mce = k8_ic_mce;
@@ -887,8 +890,7 @@ static int __init mce_amd_init(void)
 		break;
 
 	default:
-		printk(KERN_WARNING "Huh? What family is that: %d?!\n",
-				    boot_cpu_data.x86);
+		printk(KERN_WARNING "Huh? What family is that: %d?!\n", c->x86);
 		kfree(fam_ops);
 		return -EINVAL;
 	}
-- 
1.7.3.1.50.g1e633

RE: [PATCH 01/12] EDAC, MCE: Select extended error code mask

[Luck, Tony]

> F15h enlarges the extended error code of an MCE to a 5-bit field
> (MCi_STATUS[20:16]). Add a mask variable which default 0xf is overridden
> on F15h.

> +static u8 xec_mask	 = 0xf;

If the field is 5 bits, shouldn't this mask be 0x1f?

-Tony

RE: [PATCH 01/12] EDAC, MCE: Select extended error code mask

[Luck, Tony]

> (MCi_STATUS[20:16]). Add a mask variable which default 0xf is overridden
> on F15h.

I failed to read the "overridden" part ...
Ignore the noise - I blame the cold medication I'm taking.

-Tony

Re: [PATCH 01/12] EDAC, MCE: Select extended error code mask

[Borislav Petkov]

On Wed, Nov 10, 2010 at 02:15:48PM -0500, Luck, Tony wrote:
> > (MCi_STATUS[20:16]). Add a mask variable which default 0xf is overridden
> > on F15h.
> 
> I failed to read the "overridden" part ...
> Ignore the noise - I blame the cold medication I'm taking.

And yet it doesn't stop you from looking at my wobbly decoding stuff.
Wait a sec, shouldn't you be relaxing instead? Ah well, kernel people...

Thanks for looking into it.

-- 
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
General Managers: Alberto Bozzo, Andrew Bowd
Registration: Dornach, Gemeinde Aschheim, Landkreis Muenchen
Registergericht Muenchen, HRB Nr. 43632