x86, msr: Add support for non-contiguous cpumasks

The current rd/wrmsr_on_cpus helpers assume that the supplied
cpumasks are contiguous. However, there are machines out there
like some K8 multinode Opterons which have a non-contiguous core
enumeration on each node (e.g. cores 0,2 on node 0 instead of 0,1), see
http://www.gossamer-threads.com/lists/linux/kernel/1160268.

This patch fixes out-of-bounds writes (see URL above) by adding per-CPU
msr structs which are used on the respective cores.

Additionally, two helpers, msrs_{alloc,free}, are provided for use by
the callers of the MSR accessors.

Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Mauro Carvalho Chehab <mchehab@redhat.com>
Cc: Aristeu Rozanski <aris@redhat.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Doug Thompson <dougthompson@xmission.com>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
LKML-Reference: <20091211171440.GD31998@aftab>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>

2 files changed, 25 insertions, 4 deletions

diff --git a/arch/x86/include/asm/msr.h b/arch/x86/include/asm/msr.h
index 5bef931..2d228fc 100644
--- a/arch/x86/include/asm/msr.h
+++ b/arch/x86/include/asm/msr.h
@@ -244,6 +244,9 @@ do {                                                            \
 
 #define write_rdtscp_aux(val) wrmsr(0xc0000103, (val), 0)
 
+struct msr *msrs_alloc(void);
+void msrs_free(struct msr *msrs);
+
 #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);
diff --git a/arch/x86/lib/msr.c b/arch/x86/lib/msr.c
index 41628b1..8728341 100644
--- a/arch/x86/lib/msr.c
+++ b/arch/x86/lib/msr.c
@@ -7,7 +7,6 @@ struct msr_info {
 	u32 msr_no;
 	struct msr reg;
 	struct msr *msrs;
-	int off;
 	int err;
 };
 
@@ -18,7 +17,7 @@ static void __rdmsr_on_cpu(void *info)
 	int this_cpu = raw_smp_processor_id();
 
 	if (rv->msrs)
-		reg = &rv->msrs[this_cpu - rv->off];
+		reg = per_cpu_ptr(rv->msrs, this_cpu);
 	else
 		reg = &rv->reg;
 
@@ -32,7 +31,7 @@ static void __wrmsr_on_cpu(void *info)
 	int this_cpu = raw_smp_processor_id();
 
 	if (rv->msrs)
-		reg = &rv->msrs[this_cpu - rv->off];
+		reg = per_cpu_ptr(rv->msrs, this_cpu);
 	else
 		reg = &rv->reg;
 
@@ -80,7 +79,6 @@ static void __rwmsr_on_cpus(const struct cpumask *mask, u32 msr_no,
 
 	memset(&rv, 0, sizeof(rv));
 
-	rv.off    = cpumask_first(mask);
 	rv.msrs	  = msrs;
 	rv.msr_no = msr_no;
 
@@ -120,6 +118,26 @@ void wrmsr_on_cpus(const struct cpumask *mask, u32 msr_no, struct msr *msrs)
 }
 EXPORT_SYMBOL(wrmsr_on_cpus);
 
+struct msr *msrs_alloc(void)
+{
+	struct msr *msrs = NULL;
+
+	msrs = alloc_percpu(struct msr);
+	if (!msrs) {
+		pr_warning("%s: error allocating msrs\n", __func__);
+		return NULL;
+	}
+
+	return msrs;
+}
+EXPORT_SYMBOL(msrs_alloc);
+
+void msrs_free(struct msr *msrs)
+{
+	free_percpu(msrs);
+}
+EXPORT_SYMBOL(msrs_free);
+
 /* These "safe" variants are slower and should be used when the target MSR
    may not actually exist. */
 static void __rdmsr_safe_on_cpu(void *info)