x86/MCE: Make correctable error detection look at the Deferred bit

AMD systems may log Deferred errors. These are errors that are uncorrected
but which do not need immediate action. The MCA_STATUS[UC] bit may not be
set for Deferred errors.

Flag the error as not correctable when MCA_STATUS[Deferred] is set and
do not feed it into the Correctable Errors Collector.

[ bp: Massage commit message. ]

Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20171212165143.27475-1-Yazen.Ghannam@amd.com

diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
index 321c7a80be66c5ae71c077cc23cab446678203e2..1b2c114733768744cc0c15d594e031bbeedfb203 100644 (file)
--- a/arch/x86/kernel/cpu/mcheck/mce.c
+++ b/arch/x86/kernel/cpu/mcheck/mce.c
@@ -528,6 +528,17 @@ bool mce_is_memory_error(struct mce *m)
 }
 EXPORT_SYMBOL_GPL(mce_is_memory_error);
 
+static bool mce_is_correctable(struct mce *m)
+{
+       if (m->cpuvendor == X86_VENDOR_AMD && m->status & MCI_STATUS_DEFERRED)
+               return false;
+
+       if (m->status & MCI_STATUS_UC)
+               return false;
+
+       return true;
+}
+
 static bool cec_add_mce(struct mce *m)
 {
        if (!m)
@@ -535,7 +546,7 @@ static bool cec_add_mce(struct mce *m)
 
        /* We eat only correctable DRAM errors with usable addresses. */
        if (mce_is_memory_error(m) &&
-           !(m->status & MCI_STATUS_UC) &&
+           mce_is_correctable(m)  &&
            mce_usable_address(m))
                if (!cec_add_elem(m->addr >> PAGE_SHIFT))
                        return true;