From: Rhythm Mahajan <rhythm.m.mahajan@xxxxxxxxxx>
The commit: 3f235279828c ("x86/cpu: Restore AMD's DE_CFG MSR after resume")
which was backported from the upstream commit: 2632daebafd0 renamed the
MSR_F10H_DECFG_LFENCE_SERIALIZE macro to MSR_AMD64_DE_CFG_LFENCE_SERIALIZE.
The fix for 4.14 and 4.9 changed MSR_F10H_DECFG_LFENCE_SERIALIZE to
MSR_AMD64_DE_CFG_LFENCE_SERIALIZE_BIT in the init_amd() function, but should
have used MSR_AMD64_DE_CFG_LFENCE_SERIALIZE. This causes a discrepency in the
LFENCE serialization check in the init_amd() function.
This causes a ~16% sysbench memory regression, when running:
sysbench --test=memory run
Fixes: 3f235279828c ("x86/cpu: Restore AMD's DE_CFG MSR after resume")
Signed-off-by: Rhythm Mahajan <rhythm.m.mahajan@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
arch/x86/kernel/cpu/amd.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -950,7 +950,7 @@ static void init_amd(struct cpuinfo_x86
* serializing.
*/
ret = rdmsrl_safe(MSR_AMD64_DE_CFG, &val);
- if (!ret && (val & MSR_AMD64_DE_CFG_LFENCE_SERIALIZE_BIT)) {
+ if (!ret && (val & MSR_AMD64_DE_CFG_LFENCE_SERIALIZE)) {
/* A serializing LFENCE stops RDTSC speculation */
set_cpu_cap(c, X86_FEATURE_LFENCE_RDTSC);
} else {
