Skip to contentSkip to conversation

AMD PowerNow! in the Hypervisor (hypervisor changes)

PATCH In review Mark Langsdorf 2 participants 2 messages Jun 20, 2008

lore
Share & tools
lore

Mark Langsdorf

author
#0 lore raw

PATCH 2/2 AMD PowerNow! in the Hypervisor (hypervisor changes)

This patch set moves support for AMD's PowerNow! technology from dom0 into the hypervisor, now that there is support for transferring ACPI data to the hypervisor. It will only work for AMD processors that support the architectural P-state driver, such as 3rd generation Opterons, Phenoms, and Turion Ultras.

This patch creates the Architectural P-state driver inside of the Xen hypervisor and hooks it into the Xen code. It has been tested at AMD with a variety of para- and fully- virtualized guests for a week without regressions.

It should apply to current xen-unstable.hg.

-Mark Langsdorf
Operating System Research Center
AMD

Signed-off-by Mark Langsdorf <[email protected]>
Acked-by Conny Seidel <[email protected]>
xen/arch/x86/acpi/cpufreq/Makefile +1 -0
diff -r fe60bf79d96f xen/arch/x86/acpi/cpufreq/Makefile
@@ -1,3 +1,4 @@ obj-y += cpufreq.o
 obj-y += cpufreq.o
 obj-y += utility.o
 obj-y += cpufreq_ondemand.o
+obj-y += powernow.o
xen/arch/x86/acpi/cpufreq/powernow.c new file +305 -0

305 lines not on this page · raw message

xen/arch/x86/platform_hypercall.c +7 -1
diff -r fe60bf79d96f xen/arch/x86/platform_hypercall.c
@@ -354,6 +354,7 @@ ret_t do_platform_op(XEN_GUEST_HANDLE(xe
             int cpuid = get_cpu_id(xenpmpt->id);
             struct processor_pminfo *pmpt;
             struct processor_performance *pxpt;
+            struct cpuinfo_x86 *cpu = &cpu_data[cpuid];
 
             if ( cpuid < 0 )
             {

@@ -408,7 +409,12 @@ ret_t do_platform_op(XEN_GUEST_HANDLE(xe
                 cpu_count++;
             }
             if ( cpu_count == num_online_cpus() )
-                ret = acpi_cpufreq_init();
+            {
+                if (cpu->x86_vendor == X86_VENDOR_AMD)
+                    ret = powernow_cpufreq_init();
+                else
+                    ret = acpi_cpufreq_init();
+            }
             break;
         }
xen/include/acpi/cpufreq/processor_perf.h +1 -0
diff -r fe60bf79d96f xen/include/acpi/cpufreq/processor_perf.h
@@ -6,6 +6,7 @@
 
 int get_cpu_id(u8);
 int acpi_cpufreq_init(void);
+int powernow_cpufreq_init(void);
 void px_statistic_update(cpumask_t, uint8_t, uint8_t);
 int  px_statistic_init(int);
 void px_statistic_reset(int);
#1 lore raw
Quoted 6 lines · Mark Langsdorf

This patch set moves support for AMD's PowerNow! technology from dom0 into the hypervisor, now that there is support for transferring ACPI data to the hypervisor. It will only work for AMD processors that support the architectural P-state driver, such as 3rd generation Opterons, Phenoms, and Turion Ultras.

Quoted 6 lines

This patch creates the Architectural P-state driver inside of the Xen hypervisor and hooks it into the Xen code. It has been tested at AMD with a variety of para- and fully- virtualized guests for a week without regressions.

It should apply to current xen-unstable.hg.

The patch is malformed (according to 'patch'). Perhaps try resending as an attachment.

-- Keir