x86: MSR: add methods for writing of an MSR on several CPUs

Provide for concurrent MSR writes on all the CPUs in the cpumask. Also,
add a temporary workaround for smp_call_function_many which skips the
CPU we're executing on.

Bart: zero out rv struct which is allocated on stack.

CC: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

arch/x86/include/asm/msr.h

@@ -227,6 +227,8 @@ do {                                                            \
 #ifdef CONFIG_SMP
 int rdmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h);
 int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h);
+void rdmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs);
+void wrmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs);
 int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h);
 int wrmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h);
 #else  /*  CONFIG_SMP  */
@@ -240,6 +242,16 @@ static inline int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
 	wrmsr(msr_no, l, h);
 	return 0;
 }
+static inline void rdmsr_on_cpus(const cpumask_t *m, u32 msr_no,
+				struct msr *msrs)
+{
+       rdmsr_on_cpu(0, msr_no, &(msrs[0].l), &(msrs[0].h));
+}
+static inline void wrmsr_on_cpus(const cpumask_t *m, u32 msr_no,
+				struct msr *msrs)
+{
+       wrmsr_on_cpu(0, msr_no, msrs[0].l, msrs[0].h);
+}
 static inline int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no,
 				    u32 *l, u32 *h)
 {

arch/x86/lib/msr.c

@@ -6,21 +6,37 @@
 struct msr_info {
 	u32 msr_no;
 	struct msr reg;
+	struct msr *msrs;
+	int off;
 	int err;
 };
 
 static void __rdmsr_on_cpu(void *info)
 {
 	struct msr_info *rv = info;
+	struct msr *reg;
+	int this_cpu = raw_smp_processor_id();
 
-	rdmsr(rv->msr_no, rv->reg.l, rv->reg.h);
+	if (rv->msrs)
+		reg = &rv->msrs[this_cpu - rv->off];
+	else
+		reg = &rv->reg;
+
+	rdmsr(rv->msr_no, reg->l, reg->h);
 }
 
 static void __wrmsr_on_cpu(void *info)
 {
 	struct msr_info *rv = info;
+	struct msr *reg;
+	int this_cpu = raw_smp_processor_id();
 
-	wrmsr(rv->msr_no, rv->reg.l, rv->reg.h);
+	if (rv->msrs)
+		reg = &rv->msrs[this_cpu - rv->off];
+	else
+		reg = &rv->reg;
+
+	wrmsr(rv->msr_no, reg->l, reg->h);
 }
 
 int rdmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
@@ -28,6 +44,8 @@ int rdmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
 	int err;
 	struct msr_info rv;
 
+	memset(&rv, 0, sizeof(rv));
+
 	rv.msr_no = msr_no;
 	err = smp_call_function_single(cpu, __rdmsr_on_cpu, &rv, 1);
 	*l = rv.reg.l;
@@ -35,12 +53,15 @@ int rdmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
 
 	return err;
 }
+EXPORT_SYMBOL(rdmsr_on_cpu);
 
 int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
 {
 	int err;
 	struct msr_info rv;
 
+	memset(&rv, 0, sizeof(rv));
+
 	rv.msr_no = msr_no;
 	rv.reg.l = l;
 	rv.reg.h = h;
@@ -48,6 +69,70 @@ int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
 
 	return err;
 }
+EXPORT_SYMBOL(wrmsr_on_cpu);
+
+/* rdmsr on a bunch of CPUs
+ *
+ * @mask:       which CPUs
+ * @msr_no:     which MSR
+ * @msrs:       array of MSR values
+ *
+ */
+void rdmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs)
+{
+	struct msr_info rv;
+	int this_cpu;
+
+	memset(&rv, 0, sizeof(rv));
+
+	rv.off    = cpumask_first(mask);
+	rv.msrs	  = msrs;
+	rv.msr_no = msr_no;
+
+	preempt_disable();
+	/*
+	 * FIXME: handle the CPU we're executing on separately for now until
+	 * smp_call_function_many has been fixed to not skip it.
+	 */
+	this_cpu = raw_smp_processor_id();
+	smp_call_function_single(this_cpu, __rdmsr_on_cpu, &rv, 1);
+
+	smp_call_function_many(mask, __rdmsr_on_cpu, &rv, 1);
+	preempt_enable();
+}
+EXPORT_SYMBOL(rdmsr_on_cpus);
+
+/*
+ * wrmsr on a bunch of CPUs
+ *
+ * @mask:       which CPUs
+ * @msr_no:     which MSR
+ * @msrs:       array of MSR values
+ *
+ */
+void wrmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs)
+{
+	struct msr_info rv;
+	int this_cpu;
+
+	memset(&rv, 0, sizeof(rv));
+
+	rv.off    = cpumask_first(mask);
+	rv.msrs   = msrs;
+	rv.msr_no = msr_no;
+
+	preempt_disable();
+	/*
+	 * FIXME: handle the CPU we're executing on separately for now until
+	 * smp_call_function_many has been fixed to not skip it.
+	 */
+	this_cpu = raw_smp_processor_id();
+	smp_call_function_single(this_cpu, __wrmsr_on_cpu, &rv, 1);
+
+	smp_call_function_many(mask, __wrmsr_on_cpu, &rv, 1);
+	preempt_enable();
+}
+EXPORT_SYMBOL(wrmsr_on_cpus);
 
 /* These "safe" variants are slower and should be used when the target MSR
    may not actually exist. */
@@ -70,6 +155,8 @@ int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
 	int err;
 	struct msr_info rv;
 
+	memset(&rv, 0, sizeof(rv));
+
 	rv.msr_no = msr_no;
 	err = smp_call_function_single(cpu, __rdmsr_safe_on_cpu, &rv, 1);
 	*l = rv.reg.l;
@@ -77,12 +164,15 @@ int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
 
 	return err ? err : rv.err;
 }
+EXPORT_SYMBOL(rdmsr_safe_on_cpu);
 
 int wrmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
 {
 	int err;
 	struct msr_info rv;
 
+	memset(&rv, 0, sizeof(rv));
+
 	rv.msr_no = msr_no;
 	rv.reg.l = l;
 	rv.reg.h = h;
@@ -90,8 +180,4 @@ int wrmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
 
 	return err ? err : rv.err;
 }
-
-EXPORT_SYMBOL(rdmsr_on_cpu);
-EXPORT_SYMBOL(wrmsr_on_cpu);
-EXPORT_SYMBOL(rdmsr_safe_on_cpu);
 EXPORT_SYMBOL(wrmsr_safe_on_cpu);