[PATCH 0/11] KVM updates for 2.6.21

[PATCH 0/11] KVM updates for 2.6.21

[Avi Kivity]

This patchset contains an assortment of guest and host fixes for kvm.  
Andrew, please queue for 2.6.21.

-- 
error compiling committee.c: too many arguments to function

[PATCH 1/11] KVM: optimize inline assembly

[Avi Kivity]

From: Ingo Molnar <mingo@elte.hu>

forms like "0(%rsp)" generate an instruction with an unnecessary one byte
displacement under certain circumstances.  replace with the equivalent 
"(%rsp)".

Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -1784,10 +1784,10 @@ again:
 		"kvm_vmx_return: "
 		/* Save guest registers, load host registers, keep flags */
 #ifdef CONFIG_X86_64
-		"xchg %3,     0(%%rsp) \n\t"
+		"xchg %3,     (%%rsp) \n\t"
 		"mov %%rax, %c[rax](%3) \n\t"
 		"mov %%rbx, %c[rbx](%3) \n\t"
-		"pushq 0(%%rsp); popq %c[rcx](%3) \n\t"
+		"pushq (%%rsp); popq %c[rcx](%3) \n\t"
 		"mov %%rdx, %c[rdx](%3) \n\t"
 		"mov %%rsi, %c[rsi](%3) \n\t"
 		"mov %%rdi, %c[rdi](%3) \n\t"
@@ -1802,24 +1802,24 @@ again:
 		"mov %%r15, %c[r15](%3) \n\t"
 		"mov %%cr2, %%rax   \n\t"
 		"mov %%rax, %c[cr2](%3) \n\t"
-		"mov 0(%%rsp), %3 \n\t"
+		"mov (%%rsp), %3 \n\t"
 
 		"pop  %%rcx; pop  %%r15; pop  %%r14; pop  %%r13; pop  %%r12;"
 		"pop  %%r11; pop  %%r10; pop  %%r9;  pop  %%r8;"
 		"pop  %%rbp; pop  %%rdi; pop  %%rsi;"
 		"pop  %%rdx; pop  %%rbx; pop  %%rax \n\t"
 #else
-		"xchg %3, 0(%%esp) \n\t"
+		"xchg %3, (%%esp) \n\t"
 		"mov %%eax, %c[rax](%3) \n\t"
 		"mov %%ebx, %c[rbx](%3) \n\t"
-		"pushl 0(%%esp); popl %c[rcx](%3) \n\t"
+		"pushl (%%esp); popl %c[rcx](%3) \n\t"
 		"mov %%edx, %c[rdx](%3) \n\t"
 		"mov %%esi, %c[rsi](%3) \n\t"
 		"mov %%edi, %c[rdi](%3) \n\t"
 		"mov %%ebp, %c[rbp](%3) \n\t"
 		"mov %%cr2, %%eax  \n\t"
 		"mov %%eax, %c[cr2](%3) \n\t"
-		"mov 0(%%esp), %3 \n\t"
+		"mov (%%esp), %3 \n\t"
 
 		"pop %%ecx; popa \n\t"
 #endif

[PATCH 2/11] KVM: Fix asm constraint for lldt instruction

[Avi Kivity]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 571 bytes --]

From: S.Çağlar Onur <caglar@pardus.org.tr>

lldt does not accept immediate operands, which "g" allows.

Signed-off-by: S.Çağlar Onur <caglar@pardus.org.tr>
Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -558,7 +558,7 @@ static inline void load_gs(u16 sel)
 #ifndef load_ldt
 static inline void load_ldt(u16 sel)
 {
-	asm ("lldt %0" : : "g"(sel));
+	asm ("lldt %0" : : "rm"(sel));
 }
 #endif
 

Re: [PATCH 2/11] KVM: Fix asm constraint for lldt instruction

[Avi Kivity]

Avi Kivity wrote:
> From: S.Çağlar Onur <caglar@pardus.org.tr>
> Signed-off-by: Avi Kivity <avi@qumranet.com>
>  
>   

The mangled name is actually S.Çağlar Onur <caglar@pardus.org.tr>. 
Apologies to S.Çağlar and Andrew. Will fix my script to include charset 
info.

-- 
error compiling committee.c: too many arguments to function

[PATCH 3/11] KVM: Fix gva_to_gpa()

[Avi Kivity]

gva_to_gpa() needs to be updated to the new walk_addr() calling convention,
otherwise it may oops under some circumstances.

Use the opportunity to remove all the code duplication in gva_to_gpa(), which
essentially repeats the calculations in walk_addr().

Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/paging_tmpl.h
===================================================================
--- linux-2.6.orig/drivers/kvm/paging_tmpl.h
+++ linux-2.6/drivers/kvm/paging_tmpl.h
@@ -443,31 +443,17 @@ static int FNAME(page_fault)(struct kvm_
 static gpa_t FNAME(gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t vaddr)
 {
 	struct guest_walker walker;
-	pt_element_t guest_pte;
-	gpa_t gpa;
+	gpa_t gpa = UNMAPPED_GVA;
+	int r;
 
-	FNAME(walk_addr)(&walker, vcpu, vaddr, 0, 0, 0);
-	guest_pte = *walker.ptep;
-	FNAME(release_walker)(&walker);
-
-	if (!is_present_pte(guest_pte))
-		return UNMAPPED_GVA;
-
-	if (walker.level == PT_DIRECTORY_LEVEL) {
-		ASSERT((guest_pte & PT_PAGE_SIZE_MASK));
-		ASSERT(PTTYPE == 64 || is_pse(vcpu));
+	r = FNAME(walk_addr)(&walker, vcpu, vaddr, 0, 0, 0);
 
-		gpa = (guest_pte & PT_DIR_BASE_ADDR_MASK) | (vaddr &
-			(PT_LEVEL_MASK(PT_PAGE_TABLE_LEVEL) | ~PAGE_MASK));
-
-		if (PTTYPE == 32 && is_cpuid_PSE36())
-			gpa |= (guest_pte & PT32_DIR_PSE36_MASK) <<
-					(32 - PT32_DIR_PSE36_SHIFT);
-	} else {
-		gpa = (guest_pte & PT_BASE_ADDR_MASK);
-		gpa |= (vaddr & ~PAGE_MASK);
+	if (r) {
+		gpa = (gpa_t)walker.gfn << PAGE_SHIFT;
+		gpa |= vaddr & ~PAGE_MASK;
 	}
 
+	FNAME(release_walker)(&walker);
 	return gpa;
 }
 

[PATCH 4/11] KVM: vmx: handle triple faults by returning EXIT_REASON_SHUTDOWN to userspace

[Avi Kivity]

Just like svm.

Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -1373,6 +1373,11 @@ static int handle_external_interrupt(str
 	return 1;
 }
 
+static int handle_triple_fault(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+	kvm_run->exit_reason = KVM_EXIT_SHUTDOWN;
+	return 0;
+}
 
 static int get_io_count(struct kvm_vcpu *vcpu, u64 *count)
 {
@@ -1633,6 +1638,7 @@ static int (*kvm_vmx_exit_handlers[])(st
 				      struct kvm_run *kvm_run) = {
 	[EXIT_REASON_EXCEPTION_NMI]           = handle_exception,
 	[EXIT_REASON_EXTERNAL_INTERRUPT]      = handle_external_interrupt,
+	[EXIT_REASON_TRIPLE_FAULT]            = handle_triple_fault,
 	[EXIT_REASON_IO_INSTRUCTION]          = handle_io,
 	[EXIT_REASON_CR_ACCESS]               = handle_cr,
 	[EXIT_REASON_DR_ACCESS]               = handle_dr,
Index: linux-2.6/drivers/kvm/vmx.h
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.h
+++ linux-2.6/drivers/kvm/vmx.h
@@ -180,6 +180,7 @@ enum vmcs_field {
 
 #define EXIT_REASON_EXCEPTION_NMI       0
 #define EXIT_REASON_EXTERNAL_INTERRUPT  1
+#define EXIT_REASON_TRIPLE_FAULT        2
 
 #define EXIT_REASON_PENDING_INTERRUPT   7
 

[PATCH 5/11] KVM: fix what looks like an obvious typo in the file drivers/kvm/svm.c

[Avi Kivity]

From: Robert P. J. Day <rpjday@mindspring.com>

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/svm.c
===================================================================
--- linux-2.6.orig/drivers/kvm/svm.c
+++ linux-2.6/drivers/kvm/svm.c
@@ -1163,7 +1163,7 @@ static int svm_set_msr(struct kvm_vcpu *
 	case MSR_K6_STAR:
 		vcpu->svm->vmcb->save.star = data;
 		break;
-#ifdef CONFIG_X86_64_
+#ifdef CONFIG_X86_64
 	case MSR_LSTAR:
 		vcpu->svm->vmcb->save.lstar = data;
 		break;

[PATCH 6/11] KVM: Fix mmu going crazy of guest sets cr0.wp == 0

[Avi Kivity]

The kvm mmu relies on cr0.wp being set even if the guest does not set it.  The
vmx code correctly forces cr0.wp at all times, the svm code does not, so it
can't boot solaris without this patch.

Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/svm.c
===================================================================
--- linux-2.6.orig/drivers/kvm/svm.c
+++ linux-2.6/drivers/kvm/svm.c
@@ -723,7 +723,7 @@ static void svm_set_cr0(struct kvm_vcpu 
 	}
 #endif
 	vcpu->svm->cr0 = cr0;
-	vcpu->svm->vmcb->save.cr0 = cr0 | CR0_PG_MASK;
+	vcpu->svm->vmcb->save.cr0 = cr0 | CR0_PG_MASK | CR0_WP_MASK;
 	vcpu->cr0 = cr0;
 }
 

[PATCH 7/11] KVM: SVM: Hack initial cpu csbase to be consistent with intel

[Avi Kivity]

This allows us to run the mmu testsuite on amd.

Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/svm.c
===================================================================
--- linux-2.6.orig/drivers/kvm/svm.c
+++ linux-2.6/drivers/kvm/svm.c
@@ -528,7 +528,13 @@ static void init_vmcb(struct vmcb *vmcb)
 	save->cs.attrib = SVM_SELECTOR_READ_MASK | SVM_SELECTOR_P_MASK |
 		SVM_SELECTOR_S_MASK | SVM_SELECTOR_CODE_MASK;
 	save->cs.limit = 0xffff;
-	save->cs.base = 0xffff0000;
+	/*
+	 * cs.base should really be 0xffff0000, but vmx can't handle that, so
+	 * be consistent with it.
+	 *
+	 * Replace when we have real mode working for vmx.
+	 */
+	save->cs.base = 0xf0000;
 
 	save->gdtr.limit = 0xffff;
 	save->idtr.limit = 0xffff;

[PATCH 8/11] KVM: Two-way apic tpr synchronization

[Avi Kivity]

From: Dor Laor <dor.laor@qumranet.com>

We report the value of cr8 to userspace on an exit.  Also let userspace change
cr8 when we re-enter the guest.  The lets 64-bit guest code maintain the tpr
correctly.

Thanks for Yaniv Kamay for the idea.

Signed-off-by: Dor Laor <dor.laor@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -1360,6 +1360,9 @@ static int kvm_dev_ioctl_run(struct kvm 
 	if (!vcpu)
 		return -ENOENT;
 
+	/* re-sync apic's tpr */
+	vcpu->cr8 = kvm_run->cr8;
+
 	if (kvm_run->emulated) {
 		kvm_arch_ops->skip_emulated_instruction(vcpu);
 		kvm_run->emulated = 0;
Index: linux-2.6/include/linux/kvm.h
===================================================================
--- linux-2.6.orig/include/linux/kvm.h
+++ linux-2.6/include/linux/kvm.h
@@ -65,6 +65,8 @@ struct kvm_run {
 	__u8 ready_for_interrupt_injection;
 	__u8 if_flag;
 	__u16 padding2;
+
+	/* in (pre_kvm_run), out (post_kvm_run) */
 	__u64 cr8;
 	__u64 apic_base;
 

[PATCH 9/11] KVM: VMX: Reload ds and es even in 64-bit mode

[Avi Kivity]

Or 32-bit userspace will get confused.

Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -1863,9 +1863,7 @@ again:
 	fx_restore(vcpu->host_fx_image);
 	vcpu->interrupt_window_open = (vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) & 3) == 0;
 
-#ifndef CONFIG_X86_64
 	asm ("mov %0, %%ds; mov %0, %%es" : : "r"(__USER_DS));
-#endif
 
 	/*
 	 * Profile KVM exit RIPs:

[PATCH 10/11] KVM: Fix mismatch between 32-bit and 64-bit abi

[Avi Kivity]

Unfortunately requiring a version bump.

Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/include/linux/kvm.h
===================================================================
--- linux-2.6.orig/include/linux/kvm.h
+++ linux-2.6/include/linux/kvm.h
@@ -11,7 +11,7 @@
 #include <asm/types.h>
 #include <linux/ioctl.h>
 
-#define KVM_API_VERSION 2
+#define KVM_API_VERSION 3
 
 /*
  * Architectural interrupt line count, and the size of the bitmap needed
@@ -187,6 +187,7 @@ struct kvm_translation {
 	__u8  valid;
 	__u8  writeable;
 	__u8  usermode;
+	__u8  pad[5];
 };
 
 /* for KVM_INTERRUPT */

[PATCH 11/11] KVM: fix vcpu freeing bug

[Avi Kivity]

From: Ingo Molnar <mingo@elte.hu>

vcpu_load() can return NULL and it sometimes does in failure paths (for 
example when the userspace ABI version is too old) - causing a 
preemption count underflow in the ->vcpu_free() later on. So check for 
NULL.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Avi Kivity <avi@qumranet.com>

Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -272,7 +272,9 @@ static void kvm_free_physmem(struct kvm 
 
 static void kvm_free_vcpu(struct kvm_vcpu *vcpu)
 {
-	vcpu_load(vcpu->kvm, vcpu_slot(vcpu));
+	if (!vcpu_load(vcpu->kvm, vcpu_slot(vcpu)))
+		return;
+
 	kvm_mmu_destroy(vcpu);
 	vcpu_put(vcpu);
 	kvm_arch_ops->vcpu_free(vcpu);