similar to: [PATCH 0/5] x86/vmware: Steal time accounting support

This patchset includes several VMware guest improvements: Alexey Makhalov (3): x86/vmware: Use tsc_khz value for calibrate_cpu() x86/vmware: Add basic paravirt ops support x86/vmware: Add paravirt sched clock Documentation/kernel-parameters.txt | 4 +++ arch/x86/kernel/cpu/vmware.c | 51 +++++++++++++++++++++++++++++++++++++ 2 files changed, 55 insertions(+) -- 2.10.1

This patchset includes several VMware guest improvements: Alexey Makhalov (3): x86/vmware: Use tsc_khz value for calibrate_cpu() x86/vmware: Add basic paravirt ops support x86/vmware: Add paravirt sched clock Documentation/kernel-parameters.txt | 4 +++ arch/x86/kernel/cpu/vmware.c | 51 +++++++++++++++++++++++++++++++++++++ 2 files changed, 55 insertions(+) -- 2.10.1

Set pv_time_ops.sched_clock to vmware_sched_clock(). It is simplified version of native_sched_clock() without ring buffer of mult/shift/offset triplets and preempt toggling. Since VMware hypervisor provides constant tsc we can use constant mult/shift/offset triplet calculated at boot time. no-vmw-sched-clock kernel parameter is added to switch back to the native_sched_clock() implementation.

After aa297292d708, there are separate native calibrations for cpu_khz and tsc_khz. The code sets x86_platform.calibrate_cpu to native_calibrate_cpu() which looks in cpuid leaf 0x16 or msrs for the cpu frequency. Since we keep the tsc_khz constant (even after vmotion), the cpu_khz and tsc_khz may start diverging. tsc_init() now does cpu_khz = x86_platform.calibrate_cpu(); tsc_khz =

After aa297292d708, there are separate native calibrations for cpu_khz and tsc_khz. The code sets x86_platform.calibrate_cpu to native_calibrate_cpu() which looks in cpuid leaf 0x16 or msrs for the cpu frequency. Since we keep the tsc_khz constant (even after vmotion), the cpu_khz and tsc_khz may start diverging. tsc_init() now does cpu_khz = x86_platform.calibrate_cpu(); tsc_khz =

Thanks Thomas for the valuable comments. Changelog for the updated patchset: v1->v2 - Update pvinfo.name. v2->v3 - Address comments from Thomas G, * Created separate function: vmware_sched_clock_setup() (patch 3/3) * Updated commit descriptions for 1/3 and 3/3 Alexey Makhalov (3): x86/vmware: Use tsc_khz value for calibrate_cpu() x86/vmware: Add basic paravirt ops support

Thanks Thomas for the valuable comments. Changelog for the updated patchset: v1->v2 - Update pvinfo.name. v2->v3 - Address comments from Thomas G, * Created separate function: vmware_sched_clock_setup() (patch 3/3) * Updated commit descriptions for 1/3 and 3/3 Alexey Makhalov (3): x86/vmware: Use tsc_khz value for calibrate_cpu() x86/vmware: Add basic paravirt ops support

vmware_select_hypercall() is used only by the __init functions, and should be annotated with __init as well. Signed-off-by: Alexey Makhalov <amakhalov at vmware.com> Reviewed-by: Thomas Hellstrom <thellstrom at vmware.com> --- arch/x86/kernel/cpu/vmware.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c

On Wed, Nov 02, 2016 at 05:08:33AM -0400, Pan Xinhui wrote: > diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h > index 0f400c0..38c3bb7 100644 > --- a/arch/x86/include/asm/paravirt_types.h > +++ b/arch/x86/include/asm/paravirt_types.h > @@ -310,6 +310,8 @@ struct pv_lock_ops { > > void (*wait)(u8 *ptr, u8 val); > void

On Wed, Nov 02, 2016 at 05:08:33AM -0400, Pan Xinhui wrote: > diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h > index 0f400c0..38c3bb7 100644 > --- a/arch/x86/include/asm/paravirt_types.h > +++ b/arch/x86/include/asm/paravirt_types.h > @@ -310,6 +310,8 @@ struct pv_lock_ops { > > void (*wait)(u8 *ptr, u8 val); > void

On Tue, 25 Oct 2016, Alexey Makhalov wrote: > no-vmw-sched-clock kernel parameter is added to switch back to the > native_sched_clock() implementation. You are not switching back. The parameter is used to disable the paravirt sched clock. > #ifdef CONFIG_PARAVIRT > +static struct cyc2ns_data vmware_cyc2ns __ro_after_init; > + > +static int vmw_sched_clock __initdata = 1; >

On Thu, 27 Oct 2016, Alexey Makhalov wrote: > Set pv_time_ops.sched_clock to vmware_sched_clock(). Please do not describe WHAT the patch does, describe why. Describe the problem you are solving. I can see from the patch > + pv_time_ops.sched_clock = vmware_sched_clock; that you set pv_time_ops.sched_clock to vmware_sched_clock(). > It is simplified > version of

Set pv_time_ops.sched_clock to vmware_sched_clock(). It is simplified version of native_sched_clock() without ring buffer of mult/shift/offset triplets and preempt toggling. Since VMware hypervisor provides constant tsc we can use constant mult/shift/offset triplet calculated at boot time. no-vmw-sched-clock kernel parameter is added to disable the paravirt sched clock. Signed-off-by: Alexey

Move cyc2ns setup logic to separate function. This separation will allow to use cyc2ns mult/shift pair not only for the sched_clock but also for other clocks such as steal_clock. Signed-off-by: Alexey Makhalov <amakhalov at vmware.com> Reviewed-by: Thomas Hellstrom <thellstrom at vmware.com> --- arch/x86/kernel/cpu/vmware.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5

change from v4: spilt x86 kvm vcpu preempted check into two patches. add documentation patch. add x86 vcpu preempted check patch under xen add s390 vcpu preempted check patch change from v3: add x86 vcpu preempted check patch change from v2: no code change, fix typos, update some comments change from v1: a simplier definition of default vcpu_is_preempted skip mahcine type check on ppc,

change from v4: spilt x86 kvm vcpu preempted check into two patches. add documentation patch. add x86 vcpu preempted check patch under xen add s390 vcpu preempted check patch change from v3: add x86 vcpu preempted check patch change from v2: no code change, fix typos, update some comments change from v1: a simplier definition of default vcpu_is_preempted skip mahcine type check on ppc,

change from v6: fix typos and remove uncessary comments. change from v5: spilt x86/kvm patch into guest/host part. introduce kvm_write_guest_offset_cached. fix some typos. rebase patch onto 4.9.2 change from v4: spilt x86 kvm vcpu preempted check into two patches. add documentation patch. add x86 vcpu preempted check patch under xen add s390 vcpu preempted check patch change from v3:

change from v6: fix typos and remove uncessary comments. change from v5: spilt x86/kvm patch into guest/host part. introduce kvm_write_guest_offset_cached. fix some typos. rebase patch onto 4.9.2 change from v4: spilt x86 kvm vcpu preempted check into two patches. add documentation patch. add x86 vcpu preempted check patch under xen add s390 vcpu preempted check patch change from v3:

From: Yang Zhang <yang.zhang.wz at gmail.com> Some latency-intensive workload have seen obviously performance drop when running inside VM. The main reason is that the overhead is amplified when running inside VM. The most cost I have seen is inside idle path. This patch introduces a new mechanism to poll for a while before entering idle state. If schedule is needed during poll, then we

From: Yang Zhang <yang.zhang.wz at gmail.com> Some latency-intensive workload have seen obviously performance drop when running inside VM. The main reason is that the overhead is amplified when running inside VM. The most cost I have seen is inside idle path. This patch introduces a new mechanism to poll for a while before entering idle state. If schedule is needed during poll, then we