* [PATCH 1/5] mwait-idle: add SKX support
2016-06-08 13:47 [PATCH 0/5] x86: mwait-idle updates Jan Beulich
@ 2016-06-08 13:49 ` Jan Beulich
2016-06-08 13:50 ` [PATCH 2/5] mwait-idle: add KBL support Jan Beulich
` (4 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Jan Beulich @ 2016-06-08 13:49 UTC (permalink / raw)
To: xen-devel; +Cc: Andrew Cooper
[-- Attachment #1: Type: text/plain, Size: 1374 bytes --]
SKX is similar to BDX
Signed-off-by: Len Brown <len.brown@intel.com>
[Linux commit: f9e71657c2c0a8f1c50884ab45794be2854e158e]
Signed-off-by: Jan Beulich <jbeulich@suse.com>
--- a/xen/arch/x86/cpu/mwait-idle.c
+++ b/xen/arch/x86/cpu/mwait-idle.c
@@ -530,6 +530,28 @@ static struct cpuidle_state skl_cstates[
{}
};
+static const struct cpuidle_state skx_cstates[] = {
+ {
+ .name = "C1-SKX",
+ .flags = MWAIT2flg(0x00),
+ .exit_latency = 2,
+ .target_residency = 2,
+ },
+ {
+ .name = "C1E-SKX",
+ .flags = MWAIT2flg(0x01),
+ .exit_latency = 10,
+ .target_residency = 20,
+ },
+ {
+ .name = "C6-SKX",
+ .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 133,
+ .target_residency = 600,
+ },
+ {}
+};
+
static const struct cpuidle_state atom_cstates[] = {
{
.name = "C1E-ATM",
@@ -757,6 +779,11 @@ static const struct idle_cpu idle_cpu_sk
.disable_promotion_to_c1e = 1,
};
+static const struct idle_cpu idle_cpu_skx = {
+ .state_table = skx_cstates,
+ .disable_promotion_to_c1e = 1,
+};
+
static const struct idle_cpu idle_cpu_avn = {
.state_table = avn_cstates,
.disable_promotion_to_c1e = 1,
@@ -798,6 +825,7 @@ static const struct x86_cpu_id intel_idl
ICPU(0x56, bdw),
ICPU(0x4e, skl),
ICPU(0x5e, skl),
+ ICPU(0x55, skx),
ICPU(0x57, knl),
{}
};
[-- Attachment #2: x86-mwait-idle-SKX.patch --]
[-- Type: text/plain, Size: 1399 bytes --]
mwait-idle: add SKX support
SKX is similar to BDX
Signed-off-by: Len Brown <len.brown@intel.com>
[Linux commit: f9e71657c2c0a8f1c50884ab45794be2854e158e]
Signed-off-by: Jan Beulich <jbeulich@suse.com>
--- a/xen/arch/x86/cpu/mwait-idle.c
+++ b/xen/arch/x86/cpu/mwait-idle.c
@@ -530,6 +530,28 @@ static struct cpuidle_state skl_cstates[
{}
};
+static const struct cpuidle_state skx_cstates[] = {
+ {
+ .name = "C1-SKX",
+ .flags = MWAIT2flg(0x00),
+ .exit_latency = 2,
+ .target_residency = 2,
+ },
+ {
+ .name = "C1E-SKX",
+ .flags = MWAIT2flg(0x01),
+ .exit_latency = 10,
+ .target_residency = 20,
+ },
+ {
+ .name = "C6-SKX",
+ .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 133,
+ .target_residency = 600,
+ },
+ {}
+};
+
static const struct cpuidle_state atom_cstates[] = {
{
.name = "C1E-ATM",
@@ -757,6 +779,11 @@ static const struct idle_cpu idle_cpu_sk
.disable_promotion_to_c1e = 1,
};
+static const struct idle_cpu idle_cpu_skx = {
+ .state_table = skx_cstates,
+ .disable_promotion_to_c1e = 1,
+};
+
static const struct idle_cpu idle_cpu_avn = {
.state_table = avn_cstates,
.disable_promotion_to_c1e = 1,
@@ -798,6 +825,7 @@ static const struct x86_cpu_id intel_idl
ICPU(0x56, bdw),
ICPU(0x4e, skl),
ICPU(0x5e, skl),
+ ICPU(0x55, skx),
ICPU(0x57, knl),
{}
};
[-- Attachment #3: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 3/5] mwait-idle: add BXT support
2016-06-08 13:47 [PATCH 0/5] x86: mwait-idle updates Jan Beulich
2016-06-08 13:49 ` [PATCH 1/5] mwait-idle: add SKX support Jan Beulich
2016-06-08 13:50 ` [PATCH 2/5] mwait-idle: add KBL support Jan Beulich
@ 2016-06-08 13:50 ` Jan Beulich
2016-06-08 13:51 ` [PATCH 4/5] mwait-idle: add a missing __init annotation Jan Beulich
` (2 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Jan Beulich @ 2016-06-08 13:50 UTC (permalink / raw)
To: xen-devel; +Cc: Andrew Cooper
[-- Attachment #1: Type: text/plain, Size: 4979 bytes --]
Broxton has all the HSW C-states, except C3.
BXT C-state timing is slightly different.
Here we trust the IRTL MSRs as authority
on maximum C-state latency, and override the driver's tables
with the values found in the associated IRTL MSRs.
Further we set the target_residency to 1x maximum latency,
trusting the hardware demotion logic.
Signed-off-by: Len Brown <len.brown@intel.com>
[Linux commit: 5dcef694860100fd16885f052591b1268b764d21]
Signed-off-by: Jan Beulich <jbeulich@suse.com>
--- a/xen/arch/x86/cpu/mwait-idle.c
+++ b/xen/arch/x86/cpu/mwait-idle.c
@@ -612,6 +612,52 @@ static const struct cpuidle_state knl_cs
{}
};
+static struct cpuidle_state bxt_cstates[] = {
+ {
+ .name = "C1-BXT",
+ .flags = MWAIT2flg(0x00),
+ .exit_latency = 2,
+ .target_residency = 2,
+ },
+ {
+ .name = "C1E-BXT",
+ .flags = MWAIT2flg(0x01),
+ .exit_latency = 10,
+ .target_residency = 20,
+ },
+ {
+ .name = "C6-BXT",
+ .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 133,
+ .target_residency = 133,
+ },
+ {
+ .name = "C7s-BXT",
+ .flags = MWAIT2flg(0x31) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 155,
+ .target_residency = 155,
+ },
+ {
+ .name = "C8-BXT",
+ .flags = MWAIT2flg(0x40) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 1000,
+ .target_residency = 1000,
+ },
+ {
+ .name = "C9-BXT",
+ .flags = MWAIT2flg(0x50) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 2000,
+ .target_residency = 2000,
+ },
+ {
+ .name = "C10-BXT",
+ .flags = MWAIT2flg(0x60) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 10000,
+ .target_residency = 10000,
+ },
+ {}
+};
+
static void mwait_idle(void)
{
unsigned int cpu = smp_processor_id();
@@ -793,11 +839,16 @@ static const struct idle_cpu idle_cpu_kn
.state_table = knl_cstates,
};
+static const struct idle_cpu idle_cpu_bxt = {
+ .state_table = bxt_cstates,
+ .disable_promotion_to_c1e = 1,
+};
+
#define ICPU(model, cpu) \
{ X86_VENDOR_INTEL, 6, model, X86_FEATURE_MONITOR, \
&idle_cpu_##cpu}
-static const struct x86_cpu_id intel_idle_ids[] __initconst = {
+static const struct x86_cpu_id intel_idle_ids[] __initconstrel = {
ICPU(0x1a, nehalem),
ICPU(0x1e, nehalem),
ICPU(0x1f, nehalem),
@@ -829,6 +880,7 @@ static const struct x86_cpu_id intel_idl
ICPU(0x9e, skl),
ICPU(0x55, skx),
ICPU(0x57, knl),
+ ICPU(0x5c, bxt),
{}
};
@@ -860,6 +912,72 @@ static void __init ivt_idle_state_table_
}
/*
+ * Translate IRTL (Interrupt Response Time Limit) MSR to usec
+ */
+
+static const unsigned int __initconst irtl_ns_units[] = {
+ 1, 32, 1024, 32768, 1048576, 33554432, 0, 0 };
+
+static unsigned long long __init irtl_2_usec(unsigned long long irtl)
+{
+ unsigned long long ns;
+
+ ns = irtl_ns_units[(irtl >> 10) & 0x3];
+
+ return (irtl & 0x3FF) * ns / 1000;
+}
+/*
+ * bxt_idle_state_table_update(void)
+ *
+ * On BXT, we trust the IRTL to show the definitive maximum latency
+ * We use the same value for target_residency.
+ */
+static void __init bxt_idle_state_table_update(void)
+{
+ unsigned long long msr;
+
+ rdmsrl(MSR_PKGC6_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[2].exit_latency = usec;
+ bxt_cstates[2].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC7_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[3].exit_latency = usec;
+ bxt_cstates[3].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC8_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[4].exit_latency = usec;
+ bxt_cstates[4].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC9_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[5].exit_latency = usec;
+ bxt_cstates[5].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC10_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[6].exit_latency = usec;
+ bxt_cstates[6].target_residency = usec;
+ }
+}
+
+/*
* sklh_idle_state_table_update(void)
*
* On SKL-H (model 0x5e) disable C8 and C9 if:
@@ -907,6 +1025,9 @@ static void __init mwait_idle_state_tabl
case 0x3e: /* IVT */
ivt_idle_state_table_update();
break;
+ case 0x5c: /* BXT */
+ bxt_idle_state_table_update();
+ break;
case 0x5e: /* SKL-H */
sklh_idle_state_table_update();
break;
--- a/xen/include/asm-x86/msr-index.h
+++ b/xen/include/asm-x86/msr-index.h
@@ -501,4 +501,12 @@
#define MSR_CC6_DEMOTION_POLICY_CONFIG 0x00000668
#define MSR_MC6_DEMOTION_POLICY_CONFIG 0x00000669
+/* Interrupt Response Limit */
+#define MSR_PKGC3_IRTL 0x0000060a
+#define MSR_PKGC6_IRTL 0x0000060b
+#define MSR_PKGC7_IRTL 0x0000060c
+#define MSR_PKGC8_IRTL 0x00000633
+#define MSR_PKGC9_IRTL 0x00000634
+#define MSR_PKGC10_IRTL 0x00000635
+
#endif /* __ASM_MSR_INDEX_H */
[-- Attachment #2: x86-mwait-idle-BXT.patch --]
[-- Type: text/plain, Size: 5006 bytes --]
mwait-idle: add BXT support
Broxton has all the HSW C-states, except C3.
BXT C-state timing is slightly different.
Here we trust the IRTL MSRs as authority
on maximum C-state latency, and override the driver's tables
with the values found in the associated IRTL MSRs.
Further we set the target_residency to 1x maximum latency,
trusting the hardware demotion logic.
Signed-off-by: Len Brown <len.brown@intel.com>
[Linux commit: 5dcef694860100fd16885f052591b1268b764d21]
Signed-off-by: Jan Beulich <jbeulich@suse.com>
--- a/xen/arch/x86/cpu/mwait-idle.c
+++ b/xen/arch/x86/cpu/mwait-idle.c
@@ -612,6 +612,52 @@ static const struct cpuidle_state knl_cs
{}
};
+static struct cpuidle_state bxt_cstates[] = {
+ {
+ .name = "C1-BXT",
+ .flags = MWAIT2flg(0x00),
+ .exit_latency = 2,
+ .target_residency = 2,
+ },
+ {
+ .name = "C1E-BXT",
+ .flags = MWAIT2flg(0x01),
+ .exit_latency = 10,
+ .target_residency = 20,
+ },
+ {
+ .name = "C6-BXT",
+ .flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 133,
+ .target_residency = 133,
+ },
+ {
+ .name = "C7s-BXT",
+ .flags = MWAIT2flg(0x31) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 155,
+ .target_residency = 155,
+ },
+ {
+ .name = "C8-BXT",
+ .flags = MWAIT2flg(0x40) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 1000,
+ .target_residency = 1000,
+ },
+ {
+ .name = "C9-BXT",
+ .flags = MWAIT2flg(0x50) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 2000,
+ .target_residency = 2000,
+ },
+ {
+ .name = "C10-BXT",
+ .flags = MWAIT2flg(0x60) | CPUIDLE_FLAG_TLB_FLUSHED,
+ .exit_latency = 10000,
+ .target_residency = 10000,
+ },
+ {}
+};
+
static void mwait_idle(void)
{
unsigned int cpu = smp_processor_id();
@@ -793,11 +839,16 @@ static const struct idle_cpu idle_cpu_kn
.state_table = knl_cstates,
};
+static const struct idle_cpu idle_cpu_bxt = {
+ .state_table = bxt_cstates,
+ .disable_promotion_to_c1e = 1,
+};
+
#define ICPU(model, cpu) \
{ X86_VENDOR_INTEL, 6, model, X86_FEATURE_MONITOR, \
&idle_cpu_##cpu}
-static const struct x86_cpu_id intel_idle_ids[] __initconst = {
+static const struct x86_cpu_id intel_idle_ids[] __initconstrel = {
ICPU(0x1a, nehalem),
ICPU(0x1e, nehalem),
ICPU(0x1f, nehalem),
@@ -829,6 +880,7 @@ static const struct x86_cpu_id intel_idl
ICPU(0x9e, skl),
ICPU(0x55, skx),
ICPU(0x57, knl),
+ ICPU(0x5c, bxt),
{}
};
@@ -860,6 +912,72 @@ static void __init ivt_idle_state_table_
}
/*
+ * Translate IRTL (Interrupt Response Time Limit) MSR to usec
+ */
+
+static const unsigned int __initconst irtl_ns_units[] = {
+ 1, 32, 1024, 32768, 1048576, 33554432, 0, 0 };
+
+static unsigned long long __init irtl_2_usec(unsigned long long irtl)
+{
+ unsigned long long ns;
+
+ ns = irtl_ns_units[(irtl >> 10) & 0x3];
+
+ return (irtl & 0x3FF) * ns / 1000;
+}
+/*
+ * bxt_idle_state_table_update(void)
+ *
+ * On BXT, we trust the IRTL to show the definitive maximum latency
+ * We use the same value for target_residency.
+ */
+static void __init bxt_idle_state_table_update(void)
+{
+ unsigned long long msr;
+
+ rdmsrl(MSR_PKGC6_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[2].exit_latency = usec;
+ bxt_cstates[2].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC7_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[3].exit_latency = usec;
+ bxt_cstates[3].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC8_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[4].exit_latency = usec;
+ bxt_cstates[4].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC9_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[5].exit_latency = usec;
+ bxt_cstates[5].target_residency = usec;
+ }
+
+ rdmsrl(MSR_PKGC10_IRTL, msr);
+ if (msr) {
+ unsigned int usec = irtl_2_usec(msr);
+
+ bxt_cstates[6].exit_latency = usec;
+ bxt_cstates[6].target_residency = usec;
+ }
+}
+
+/*
* sklh_idle_state_table_update(void)
*
* On SKL-H (model 0x5e) disable C8 and C9 if:
@@ -907,6 +1025,9 @@ static void __init mwait_idle_state_tabl
case 0x3e: /* IVT */
ivt_idle_state_table_update();
break;
+ case 0x5c: /* BXT */
+ bxt_idle_state_table_update();
+ break;
case 0x5e: /* SKL-H */
sklh_idle_state_table_update();
break;
--- a/xen/include/asm-x86/msr-index.h
+++ b/xen/include/asm-x86/msr-index.h
@@ -501,4 +501,12 @@
#define MSR_CC6_DEMOTION_POLICY_CONFIG 0x00000668
#define MSR_MC6_DEMOTION_POLICY_CONFIG 0x00000669
+/* Interrupt Response Limit */
+#define MSR_PKGC3_IRTL 0x0000060a
+#define MSR_PKGC6_IRTL 0x0000060b
+#define MSR_PKGC7_IRTL 0x0000060c
+#define MSR_PKGC8_IRTL 0x00000633
+#define MSR_PKGC9_IRTL 0x00000634
+#define MSR_PKGC10_IRTL 0x00000635
+
#endif /* __ASM_MSR_INDEX_H */
[-- Attachment #3: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 8+ messages in thread