x86/MCE: Implement clearbank callback for AMD

Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Move initialization of mce_clear_banks into common code (would not get
initialized on AMD CPUs otherwise). Mark per-CPU struct mce_bank
pointers read-mostly.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Committed-by: Jan Beulich <jbeulich@suse.com>

diff --git a/xen/arch/x86/cpu/mcheck/amd_k8.c b/xen/arch/x86/cpu/mcheck/amd_k8.c
index 6516a82be53ee7e690f2911254469ffadb0bf742..8ff359cb87580dffb9a7b2c007522b57a30bc0c8 100644 (file)
--- a/xen/arch/x86/cpu/mcheck/amd_k8.c
+++ b/xen/arch/x86/cpu/mcheck/amd_k8.c
@@ -72,7 +72,23 @@
 /* Machine Check Handler for AMD K8 family series */
 static void k8_machine_check(struct cpu_user_regs *regs, long error_code)
 {
-       mcheck_cmn_handler(regs, error_code, mca_allbanks, NULL);
+       mcheck_cmn_handler(regs, error_code, mca_allbanks,
+                          __get_cpu_var(mce_clear_banks));
+}
+
+static int k8_need_clearbank_scan(enum mca_source who, uint64_t status)
+{
+       if (who != MCA_MCE_SCAN)
+               return 1;
+
+       /*
+        * For fatal error, it shouldn't be cleared so that sticky bank
+        * have a chance to be handled after reboot by polling.
+        */
+       if ((status & MCi_STATUS_UC) && (status & MCi_STATUS_PCC))
+               return 0;
+
+       return 1;
 }
 
 /* AMD K8 machine check */
@@ -85,6 +101,7 @@ enum mcheck_type amd_k8_mcheck_init(struct cpuinfo_x86 *c)
 
        mce_handler_init();
        x86_mce_vector_register(k8_machine_check);
+       mce_need_clearbank_register(k8_need_clearbank_scan);
 
        for (i = 0; i < nr_mce_banks; i++) {
                if (quirkflag == MCEQUIRK_K8_GART && i == 4) {

diff --git a/xen/arch/x86/cpu/mcheck/mce.c b/xen/arch/x86/cpu/mcheck/mce.c
index 75f0f73a1d7a9544a8878bc3d2552e4bcc541bdc..7d4743e8ed73066e257775d3e6abc570f6951f54 100644 (file)
--- a/xen/arch/x86/cpu/mcheck/mce.c
+++ b/xen/arch/x86/cpu/mcheck/mce.c
@@ -35,6 +35,10 @@ bool_t is_mc_panic;
 unsigned int __read_mostly nr_mce_banks;
 unsigned int __read_mostly firstbank;
 
+DEFINE_PER_CPU_READ_MOSTLY(struct mca_banks *, poll_bankmask);
+DEFINE_PER_CPU_READ_MOSTLY(struct mca_banks *, no_cmci_banks);
+DEFINE_PER_CPU_READ_MOSTLY(struct mca_banks *, mce_clear_banks);
+
 static void intpose_init(void);
 static void mcinfo_clear(struct mc_info *);
 struct mca_banks *mca_allbanks;
@@ -691,22 +695,29 @@ int mca_cap_init(void)
     return mca_allbanks ? 0:-ENOMEM;
 }
 
-static void cpu_poll_bankmask_free(unsigned int cpu)
+static void cpu_bank_free(unsigned int cpu)
 {
-    struct mca_banks *mb = per_cpu(poll_bankmask, cpu);
+    struct mca_banks *poll = per_cpu(poll_bankmask, cpu);
+    struct mca_banks *clr = per_cpu(mce_clear_banks, cpu);
 
-    mcabanks_free(mb);
+    mcabanks_free(poll);
+    mcabanks_free(clr);
 }
 
-static int cpu_poll_bankmask_alloc(unsigned int cpu)
+static int cpu_bank_alloc(unsigned int cpu)
 {
-    struct mca_banks *mb;
+    struct mca_banks *poll = mcabanks_alloc();
+    struct mca_banks *clr = mcabanks_alloc();
 
-    mb = mcabanks_alloc();
-    if ( !mb )
+    if ( !poll || !clr )
+    {
+        mcabanks_free(poll);
+        mcabanks_free(clr);
         return -ENOMEM;
+    }
 
-    per_cpu(poll_bankmask, cpu) = mb;
+    per_cpu(poll_bankmask, cpu) = poll;
+    per_cpu(mce_clear_banks, cpu) = clr;
     return 0;
 }
 
@@ -719,11 +730,11 @@ static int cpu_callback(
     switch ( action )
     {
     case CPU_UP_PREPARE:
-        rc = cpu_poll_bankmask_alloc(cpu);
+        rc = cpu_bank_alloc(cpu);
         break;
     case CPU_UP_CANCELED:
     case CPU_DEAD:
-        cpu_poll_bankmask_free(cpu);
+        cpu_bank_free(cpu);
         break;
     default:
         break;
@@ -757,6 +768,10 @@ void mcheck_init(struct cpuinfo_x86 *c, bool_t bsp)
     if (mca_cap_init())
         return;
 
+    /* Early MCE initialisation for BSP. */
+    if ( bsp && cpu_bank_alloc(smp_processor_id()) )
+        BUG();
+
     switch (c->x86_vendor) {
     case X86_VENDOR_AMD:
         inited = amd_mcheck_init(c);
@@ -787,18 +802,14 @@ void mcheck_init(struct cpuinfo_x86 *c, bool_t bsp)
     set_in_cr4(X86_CR4_MCE);
 
     if ( bsp )
-    {
-        /* Early MCE initialisation for BSP. */
-        if ( cpu_poll_bankmask_alloc(0) )
-            BUG();
         register_cpu_notifier(&cpu_nfb);
-    }
     set_poll_bankmask(c);
 
     return;
  out:
-    if (smp_processor_id() == 0)
+    if ( bsp )
     {
+        cpu_bank_free(smp_processor_id());
         mcabanks_free(mca_allbanks);
         mca_allbanks = NULL;
     }

diff --git a/xen/arch/x86/cpu/mcheck/mce.h b/xen/arch/x86/cpu/mcheck/mce.h
index 11dae80b4c53775caee68b3dd020612162d4425a..21f65d2384109f2eaed5bf3451ce380b49ae58e4 100644 (file)
--- a/xen/arch/x86/cpu/mcheck/mce.h
+++ b/xen/arch/x86/cpu/mcheck/mce.h
@@ -122,6 +122,7 @@ struct mca_summary {
 
 DECLARE_PER_CPU(struct mca_banks *, poll_bankmask);
 DECLARE_PER_CPU(struct mca_banks *, no_cmci_banks);
+DECLARE_PER_CPU(struct mca_banks *, mce_clear_banks);
 
 extern bool_t cmci_support;
 extern bool_t is_mc_panic;

diff --git a/xen/arch/x86/cpu/mcheck/mce_intel.c b/xen/arch/x86/cpu/mcheck/mce_intel.c
index e70bd55b1d934be44261b8af6db9709d62174829..d80f69249686ece1fb1431393aa6e37eb731ca13 100644 (file)
--- a/xen/arch/x86/cpu/mcheck/mce_intel.c
+++ b/xen/arch/x86/cpu/mcheck/mce_intel.c
@@ -21,9 +21,7 @@
 #include "vmce.h"
 #include "mcaction.h"
 
-DEFINE_PER_CPU(struct mca_banks *, mce_banks_owned);
-DEFINE_PER_CPU(struct mca_banks *, no_cmci_banks);
-DEFINE_PER_CPU(struct mca_banks *, mce_clear_banks);
+static DEFINE_PER_CPU_READ_MOSTLY(struct mca_banks *, mce_banks_owned);
 bool_t __read_mostly cmci_support = 0;
 static bool_t __read_mostly ser_support = 0;
 static bool_t __read_mostly mce_force_broadcast;
@@ -789,36 +787,28 @@ static void intel_init_mce(void)
 
 static void cpu_mcabank_free(unsigned int cpu)
 {
-    struct mca_banks *mb1, *mb2, *mb3;
+    struct mca_banks *cmci = per_cpu(no_cmci_banks, cpu);
+    struct mca_banks *owned = per_cpu(mce_banks_owned, cpu);
 
-    mb1 = per_cpu(mce_clear_banks, cpu);
-    mb2 = per_cpu(no_cmci_banks, cpu);
-    mb3 = per_cpu(mce_banks_owned, cpu);
-
-    mcabanks_free(mb1);
-    mcabanks_free(mb2);
-    mcabanks_free(mb3);
+    mcabanks_free(cmci);
+    mcabanks_free(owned);
 }
 
 static int cpu_mcabank_alloc(unsigned int cpu)
 {
-    struct mca_banks *mb1, *mb2, *mb3;
+    struct mca_banks *cmci = mcabanks_alloc();
+    struct mca_banks *owned = mcabanks_alloc();
 
-    mb1 = mcabanks_alloc();
-    mb2 = mcabanks_alloc();
-    mb3 = mcabanks_alloc();
-    if (!mb1 || !mb2 || !mb3)
+    if (!cmci || !owned)
         goto out;
 
-    per_cpu(mce_clear_banks, cpu) = mb1;
-    per_cpu(no_cmci_banks, cpu) = mb2;
-    per_cpu(mce_banks_owned, cpu) = mb3;
+    per_cpu(no_cmci_banks, cpu) = cmci;
+    per_cpu(mce_banks_owned, cpu) = owned;
 
     return 0;
 out:
-    mcabanks_free(mb1);
-    mcabanks_free(mb2);
-    mcabanks_free(mb3);
+    mcabanks_free(cmci);
+    mcabanks_free(owned);
     return -ENOMEM;
 }
 

diff --git a/xen/arch/x86/cpu/mcheck/mctelem.h b/xen/arch/x86/cpu/mcheck/mctelem.h
index 04edf98656bf64bb5ad2e8282ff8fffbbc4a0fda..e5514d98918f2b5e9b5cbc45fa11af7544f02ebb 100644 (file)
--- a/xen/arch/x86/cpu/mcheck/mctelem.h
+++ b/xen/arch/x86/cpu/mcheck/mctelem.h
@@ -23,7 +23,7 @@
  * urgent uses, intended for use from machine check exception handlers,
  * and non-urgent uses intended for use from error pollers.
  * Associated with each logout entry of whatever class is a data area
- * sized per the single argument to mctelem_init.  mcelem_init should be
+ * sized per the single argument to mctelem_init.  mctelem_init should be
  * called from MCA init code before anybody has the chance to change the
  * machine check vector with mcheck_mca_logout or to use mcheck_mca_logout.
  *
@@ -45,7 +45,7 @@
  * which will return a cookie referencing the oldest (first committed)
  * entry of the requested class.  Access the associated data using
  * mctelem_dataptr and when finished use mctelem_consume_oldest_end - in the
- * begin .. end bracket you are guaranteed that the entry canot be freed
+ * begin .. end bracket you are guaranteed that the entry can't be freed
  * even if it is ack'd elsewhere).  Once the ultimate consumer of the
  * telemetry has processed it to stable storage it should acknowledge
  * the telemetry quoting the cookie id, at which point we will free

diff --git a/xen/arch/x86/cpu/mcheck/non-fatal.c b/xen/arch/x86/cpu/mcheck/non-fatal.c
index 16fbae68279d6bc2fe6c187181ed5a88ae742bde..2e79fbdc1c6a71cf0dce980b3843bfba372377b5 100644 (file)
--- a/xen/arch/x86/cpu/mcheck/non-fatal.c
+++ b/xen/arch/x86/cpu/mcheck/non-fatal.c
@@ -23,7 +23,6 @@
 #include "mce.h"
 #include "vmce.h"
 
-DEFINE_PER_CPU(struct mca_banks *, poll_bankmask);
 static struct timer mce_timer;
 
 #define MCE_PERIOD MILLISECS(8000)