RFC - Cleaned up hypervisor layer

From: H. Peter Anvin
Date: Fri May 07 2010 - 18:43:23 EST


This patch implements a cleaned up hypervisor layer, plus cleans up some
of the aspects of both Hank's MS-HyperV patch and the existing VMware
driver. In particular, the MS-HyperV information is private
information, per system, not per CPU, and can thus simply be stored in
an ordinary global variable. This also turns the hypervisor detection
into an information structure instead of just an identifier.

-hpa
diff --git a/arch/x86/include/asm/hyperv.h b/arch/x86/include/asm/hyperv.h
index 4604047..5df477a 100644
--- a/arch/x86/include/asm/hyperv.h
+++ b/arch/x86/include/asm/hyperv.h
@@ -1,5 +1,5 @@
-#ifndef _ASM_X86_KVM_HYPERV_H
-#define _ASM_X86_KVM_HYPERV_H
+#ifndef _ASM_X86_HYPERV_H
+#define _ASM_X86_HYPERV_H

#include <linux/types.h>

@@ -16,6 +16,7 @@

#define HYPERV_HYPERVISOR_PRESENT_BIT 0x80000000
#define HYPERV_CPUID_MIN 0x40000005
+#define HYPERV_CPUID_MAX 0x4000ffff

/*
* Feature identification. EAX indicates which features are available
diff --git a/arch/x86/include/asm/hypervisor.h b/arch/x86/include/asm/hypervisor.h
index b78c094..70abda7 100644
--- a/arch/x86/include/asm/hypervisor.h
+++ b/arch/x86/include/asm/hypervisor.h
@@ -17,10 +17,33 @@
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
*/
-#ifndef ASM_X86__HYPERVISOR_H
-#define ASM_X86__HYPERVISOR_H
+#ifndef _ASM_X86_HYPERVISOR_H
+#define _ASM_X86_HYPERVISOR_H

extern void init_hypervisor(struct cpuinfo_x86 *c);
extern void init_hypervisor_platform(void);

+/*
+ * x86 hypervisor information
+ */
+struct hypervisor_x86 {
+ /* Hypervisor name */
+ const char *name;
+
+ /* Detection routine */
+ bool (*detect)(void);
+
+ /* Adjust CPU feature bits (run once per CPU) */
+ void (*set_cpu_features)(struct cpuinfo_x86 *);
+
+ /* Platform setup (run once per boot) */
+ void (*init_platform)(void);
+};
+
+extern const struct hypervisor_x86 *x86_hyper;
+
+/* Recognized hypervisors */
+extern const struct hypervisor_x86 x86_hyper_vmware;
+extern const struct hypervisor_x86 x86_hyper_ms_hyperv;
+
#endif
diff --git a/arch/x86/include/asm/mshyperv.h b/arch/x86/include/asm/mshyperv.h
index 6cd8101..79ce568 100644
--- a/arch/x86/include/asm/mshyperv.h
+++ b/arch/x86/include/asm/mshyperv.h
@@ -1,7 +1,14 @@
-#ifndef ASM_X86__MSHYPER_H
-#define ASM_X86__MSHYPER_H
+#ifndef _ASM_X86_MSHYPER_H
+#define _ASM_X86_MSHYPER_H

-int ms_hyperv_platform(void);
-void __cpuinit ms_hyperv_set_feature_bits(struct cpuinfo_x86 *c);
+#include <linux/types.h>
+#include <asm/hyperv.h>
+
+struct ms_hyperv_info {
+ u32 features;
+ u32 hints;
+};
+
+extern struct ms_hyperv_info ms_hyperv;

#endif
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 597c041..e4f1dfb 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -113,9 +113,6 @@ struct cpuinfo_x86 {
/* Index into per_cpu list: */
u16 cpu_index;
#endif
- unsigned int x86_hyper_vendor;
- /* The layout of this field is hypervisor specific */
- unsigned int x86_hyper_features;
} __attribute__((__aligned__(SMP_CACHE_BYTES)));

#define X86_VENDOR_INTEL 0
@@ -129,10 +126,6 @@ struct cpuinfo_x86 {

#define X86_VENDOR_UNKNOWN 0xff

-#define X86_HYPER_VENDOR_NONE 0
-#define X86_HYPER_VENDOR_VMWARE 1
-#define X86_HYPER_VENDOR_MSFT 2
-
/*
* capabilities of CPUs
*/
diff --git a/arch/x86/include/asm/vmware.h b/arch/x86/include/asm/vmware.h
deleted file mode 100644
index e49ed6d..0000000
--- a/arch/x86/include/asm/vmware.h
+++ /dev/null
@@ -1,27 +0,0 @@
-/*
- * Copyright (C) 2008, VMware, Inc.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
- * NON INFRINGEMENT. See the GNU General Public License for more
- * details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- */
-#ifndef ASM_X86__VMWARE_H
-#define ASM_X86__VMWARE_H
-
-extern void vmware_platform_setup(void);
-extern int vmware_platform(void);
-extern void vmware_set_feature_bits(struct cpuinfo_x86 *c);
-
-#endif
diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c
index de3f4e0..888777a 100644
--- a/arch/x86/kernel/cpu/hypervisor.c
+++ b/arch/x86/kernel/cpu/hypervisor.c
@@ -22,40 +22,52 @@
*/

#include <asm/processor.h>
-#include <asm/vmware.h>
-#include <asm/mshyperv.h>
#include <asm/hypervisor.h>

-static inline void __cpuinit
-detect_hypervisor_vendor(struct cpuinfo_x86 *c)
+/*
+ * Hypervisor detect order. This is specified explicitly here because
+ * some hypervisors might implement compatibility modes for other
+ * hypervisors and therefore need to be detected in specific sequence.
+ */
+static const __initconst struct hypervisor_x86 * const hypervisors[] =
{
- if (vmware_platform())
- c->x86_hyper_vendor = X86_HYPER_VENDOR_VMWARE;
- else if (ms_hyperv_platform())
- c->x86_hyper_vendor = X86_HYPER_VENDOR_MSFT;
- else
- c->x86_hyper_vendor = X86_HYPER_VENDOR_NONE;
-}
+ &x86_hyper_vmware,
+ &x86_hyper_ms_hyperv,
+};
+
+const struct hypervisor_x86 *x86_hyper;

-static inline void __cpuinit
-hypervisor_set_feature_bits(struct cpuinfo_x86 *c)
+static inline void __init
+detect_hypervisor_vendor(void)
{
- if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE)
- vmware_set_feature_bits(c);
- else if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_MSFT)
- ms_hyperv_set_feature_bits(c);
- return;
+ const struct hypervisor_x86 *h, * const *p;
+
+ for (p = hypervisors; p < hypervisors + ARRAY_SIZE(hypervisors); p++) {
+ h = *p;
+ if (h->detect()) {
+ x86_hyper = h;
+ printk(KERN_INFO "Hypervisor detected: %s\n", h->name);
+ break;
+ }
+ }
}

void __cpuinit init_hypervisor(struct cpuinfo_x86 *c)
{
- detect_hypervisor_vendor(c);
- hypervisor_set_feature_bits(c);
+ if (x86_hyper && x86_hyper->set_cpu_features)
+ x86_hyper->set_cpu_features(c);
}

void __init init_hypervisor_platform(void)
{
+
+ detect_hypervisor_vendor();
+
+ if (!x86_hyper)
+ return;
+
init_hypervisor(&boot_cpu_data);
- if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE)
- vmware_platform_setup();
+
+ if (x86_hyper->init_platform)
+ x86_hyper->init_platform();
}
diff --git a/arch/x86/kernel/cpu/mshyperv.c b/arch/x86/kernel/cpu/mshyperv.c
index 2443b61..2a1a256 100644
--- a/arch/x86/kernel/cpu/mshyperv.c
+++ b/arch/x86/kernel/cpu/mshyperv.c
@@ -23,45 +23,42 @@

#include <linux/types.h>
#include <asm/processor.h>
+#include <asm/hypervisor.h>
#include <asm/hyperv.h>
#include <asm/mshyperv.h>

+struct ms_hyperv_info ms_hyperv;

-int ms_hyperv_platform(void)
+static bool __init ms_hyperv_platform(void)
{
- u32 eax, ebx, ecx, edx;
- char hyp_signature[13];
+ u32 eax;
+ u32 hyp_signature[3];

- cpuid(1, &eax, &ebx, &ecx, &edx);
- if (!(ecx & HYPERV_HYPERVISOR_PRESENT_BIT))
- return 0;
+ if (!boot_cpu_has(X86_FEATURE_HYPERVISOR))
+ return false;

- cpuid(HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS, &eax, &ebx, &ecx, &edx);
- *(u32 *)(hyp_signature + 0) = ebx;
- *(u32 *)(hyp_signature + 4) = ecx;
- *(u32 *)(hyp_signature + 8) = edx;
+ cpuid(HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS,
+ &eax, &hyp_signature[0], &hyp_signature[1], &hyp_signature[2]);

- if ((eax < HYPERV_CPUID_MIN) || (memcmp("Microsoft Hv", hyp_signature, 12)))
- return 0;
- return 1;
+ return eax >= HYPERV_CPUID_MIN &&
+ eax <= HYPERV_CPUID_MAX &&
+ !memcmp("Microsoft Hv", hyp_signature, 12);
}

-void __cpuinit ms_hyperv_set_feature_bits(struct cpuinfo_x86 *c)
+static void __init ms_hyperv_init_platform(void)
{
- u32 eax, ebx, ecx, edx;
-
- c->x86_hyper_features = 0;
/*
- * Extract the features, recommendations etc.
- * The first 9 bits will be used to track hypervisor features.
- * The next 6 bits will be used to track the hypervisor
- * recommendations.
+ * Extract the features and hints
*/
- cpuid(HYPERV_CPUID_FEATURES, &eax, &ebx, &ecx, &edx);
- c->x86_hyper_features |= (eax & 0x1ff);
+ ms_hyperv.features = cpuid_eax(HYPERV_CPUID_FEATURES);
+ ms_hyperv.hints = cpuid_eax(HYPERV_CPUID_ENLIGHTMENT_INFO);

- cpuid(HYPERV_CPUID_ENLIGHTMENT_INFO, &eax, &ebx, &ecx, &edx);
- c->x86_hyper_features |= ((eax & 0x3f) << 9);
- printk(KERN_INFO "Detected HyperV with features: %x\n",
- c->x86_hyper_features);
+ printk(KERN_INFO "HyperV: features 0x%x, hints 0x%x\n",
+ ms_hyperv.features, ms_hyperv.hints);
}
+
+const __refconst struct hypervisor_x86 x86_hyper_ms_hyperv = {
+ .name = "Microsoft HyperV",
+ .detect = ms_hyperv_platform,
+ .init_platform = ms_hyperv_init_platform,
+};
diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c
index 1cbed97..46a5b5d 100644
--- a/arch/x86/kernel/cpu/vmware.c
+++ b/arch/x86/kernel/cpu/vmware.c
@@ -23,8 +23,8 @@

#include <linux/dmi.h>
#include <asm/div64.h>
-#include <asm/vmware.h>
#include <asm/x86_init.h>
+#include <asm/hypervisor.h>

#define CPUID_VMWARE_INFO_LEAF 0x40000000
#define VMWARE_HYPERVISOR_MAGIC 0x564D5868
@@ -64,7 +64,7 @@ static unsigned long vmware_get_tsc_khz(void)
return tsc_hz;
}

-void __init vmware_platform_setup(void)
+static void __init vmware_platform_setup(void)
{
uint32_t eax, ebx, ecx, edx;

@@ -82,24 +82,21 @@ void __init vmware_platform_setup(void)
* serial key should be enough, as this will always have a VMware
* specific string when running under VMware hypervisor.
*/
-int vmware_platform(void)
+static bool __init vmware_platform(void)
{
if (cpu_has_hypervisor) {
- unsigned int eax, ebx, ecx, edx;
- char hyper_vendor_id[13];
-
- cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &ebx, &ecx, &edx);
- memcpy(hyper_vendor_id + 0, &ebx, 4);
- memcpy(hyper_vendor_id + 4, &ecx, 4);
- memcpy(hyper_vendor_id + 8, &edx, 4);
- hyper_vendor_id[12] = '\0';
- if (!strcmp(hyper_vendor_id, "VMwareVMware"))
- return 1;
+ unsigned int eax;
+ unsigned int hyper_vendor_id[3];
+
+ cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &hyper_vendor_id[0],
+ &hyper_vendor_id[1], &hyper_vendor_id[2]);
+ if (!memcmp(hyper_vendor_id, "VMwareVMware", 12))
+ return true;
} else if (dmi_available && dmi_name_in_serial("VMware") &&
__vmware_platform())
- return 1;
+ return true;

- return 0;
+ return false;
}

/*
@@ -114,8 +111,15 @@ int vmware_platform(void)
* so that the kernel could just trust the hypervisor with providing a
* reliable virtual TSC that is suitable for timekeeping.
*/
-void __cpuinit vmware_set_feature_bits(struct cpuinfo_x86 *c)
+static void __cpuinit vmware_set_cpu_features(struct cpuinfo_x86 *c)
{
set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC);
set_cpu_cap(c, X86_FEATURE_TSC_RELIABLE);
}
+
+const __refconst struct hypervisor_x86 x86_hyper_vmware = {
+ .name = "VMware",
+ .detect = vmware_platform,
+ .set_cpu_features = vmware_set_cpu_features,
+ .init_platform = vmware_platform_setup,
+};