x86, amd: Restrict usage of c1e_idle()

Currently c1e_idle returns true for all CPUs greater than or equal to
family 0xf model 0x40. This covers too many CPUs.

Meanwhile a respective erratum for the underlying problem was filed
(#400). This patch adds the logic to check whether erratum #400
applies to a given CPU.
Especially for CPUs where SMI/HW triggered C1e is not supported,
c1e_idle() doesn't need to be used. We can check this by looking at
the respective OSVW bit for erratum #400.

Cc: <stable@kernel.org> # .32.x .33.x
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100319110922.GA19614@alberich.amd.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>

1 files changed, 24 insertions, 8 deletions

diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index ad9540676fcc..28ad9f4d8b94 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -526,21 +526,37 @@ static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c)
 }
 
 /*
- * Check for AMD CPUs, which have potentially C1E support
+ * Check for AMD CPUs, where APIC timer interrupt does not wake up CPU from C1e.
+ * For more information see
+ * - Erratum #400 for NPT family 0xf and family 0x10 CPUs
+ * - Erratum #365 for family 0x11 (not affected because C1e not in use)
  */
 static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c)
 {
+	u64 val;
 	if (c->x86_vendor != X86_VENDOR_AMD)
-		return 0;
-
-	if (c->x86 < 0x0F)
-		return 0;
+		goto no_c1e_idle;
 
 	/* Family 0x0f models < rev F do not have C1E */
-	if (c->x86 == 0x0f && c->x86_model < 0x40)
-		return 0;
+	if (c->x86 == 0x0F && c->x86_model >= 0x40)
+		return 1;
 
-	return 1;
+	if (c->x86 == 0x10) {
+		/*
+		 * check OSVW bit for CPUs that are not affected
+		 * by erratum #400
+		 */
+		rdmsrl(MSR_AMD64_OSVW_ID_LENGTH, val);
+		if (val >= 2) {
+			rdmsrl(MSR_AMD64_OSVW_STATUS, val);
+			if (!(val & BIT(1)))
+				goto no_c1e_idle;
+		}
+		return 1;
+	}
+
+no_c1e_idle:
+	return 0;
 }
 
 static cpumask_var_t c1e_mask;