File 5b3cab8f-2-VMX-remote-access-to-MSR-lists.patch of Package xen.8389
# Commit 80599f0b770199116aa753bfdfac9bfe2e8ea86a
# Date 2018-07-04 12:12:15 +0100
# Author Andrew Cooper <andrew.cooper3@citrix.com>
# Committer Andrew Cooper <andrew.cooper3@citrix.com>
x86/vmx: Support remote access to the MSR lists
At the moment, all modifications of the MSR lists are in current context.
However, future changes may need to put MSR_EFER into the lists from domctl
hypercall context.
Plumb a struct vcpu parameter down through the infrastructure, and use
vmx_vmcs_{enter,exit}() for safe access to the VMCS in vmx_add_msr(). Use
assertions to ensure that access is either in current context, or while the
vcpu is paused.
Note these expectations beside the fields in arch_vmx_struct, and reorder the
fields to avoid unnecessary padding.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
--- a/xen/arch/x86/cpu/vpmu_intel.c
+++ b/xen/arch/x86/cpu/vpmu_intel.c
@@ -455,12 +455,12 @@ static int core2_vpmu_alloc_resource(str
if ( is_hvm_vcpu(v) )
{
wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0);
- if ( vmx_add_host_load_msr(MSR_CORE_PERF_GLOBAL_CTRL) )
+ if ( vmx_add_host_load_msr(v, MSR_CORE_PERF_GLOBAL_CTRL) )
goto out_err;
- if ( vmx_add_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL) )
+ if ( vmx_add_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL) )
goto out_err;
- vmx_write_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL, 0);
+ vmx_write_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL, 0);
}
core2_vpmu_cxt = xzalloc_bytes(sizeof(*core2_vpmu_cxt) +
@@ -613,7 +613,7 @@ static int core2_vpmu_do_wrmsr(unsigned
return -EINVAL;
if ( is_hvm_vcpu(v) )
- vmx_read_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL,
+ vmx_read_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL,
&core2_vpmu_cxt->global_ctrl);
else
rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, core2_vpmu_cxt->global_ctrl);
@@ -682,7 +682,7 @@ static int core2_vpmu_do_wrmsr(unsigned
return -EINVAL;
if ( is_hvm_vcpu(v) )
- vmx_read_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL,
+ vmx_read_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL,
&core2_vpmu_cxt->global_ctrl);
else
rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, core2_vpmu_cxt->global_ctrl);
@@ -701,7 +701,7 @@ static int core2_vpmu_do_wrmsr(unsigned
else
{
if ( is_hvm_vcpu(v) )
- vmx_write_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
+ vmx_write_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
else
wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
}
@@ -735,7 +735,7 @@ static int core2_vpmu_do_rdmsr(unsigned
break;
case MSR_CORE_PERF_GLOBAL_CTRL:
if ( is_hvm_vcpu(v) )
- vmx_read_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
+ vmx_read_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
else
rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, *msr_content);
break;
--- a/xen/arch/x86/hvm/vmx/vmcs.c
+++ b/xen/arch/x86/hvm/vmx/vmcs.c
@@ -1308,13 +1308,15 @@ static struct vmx_msr_entry *locate_msr_
return start;
}
-struct vmx_msr_entry *vmx_find_msr(uint32_t msr, enum vmx_msr_list_type type)
+struct vmx_msr_entry *vmx_find_msr(const struct vcpu *v, uint32_t msr,
+ enum vmx_msr_list_type type)
{
- struct vcpu *curr = current;
- struct arch_vmx_struct *vmx = &curr->arch.hvm_vmx;
+ const struct arch_vmx_struct *vmx = &v->arch.hvm_vmx;
struct vmx_msr_entry *start = NULL, *ent, *end;
unsigned int total;
+ ASSERT(v == current || !vcpu_runnable(v));
+
switch ( type )
{
case VMX_MSR_HOST:
@@ -1340,12 +1342,14 @@ struct vmx_msr_entry *vmx_find_msr(uint3
return ((ent < end) && (ent->index == msr)) ? ent : NULL;
}
-int vmx_add_msr(uint32_t msr, enum vmx_msr_list_type type)
+int vmx_add_msr(struct vcpu *v, uint32_t msr, enum vmx_msr_list_type type)
{
- struct vcpu *curr = current;
- struct arch_vmx_struct *vmx = &curr->arch.hvm_vmx;
+ struct arch_vmx_struct *vmx = &v->arch.hvm_vmx;
struct vmx_msr_entry **ptr, *start = NULL, *ent, *end;
unsigned int total;
+ int rc;
+
+ ASSERT(v == current || !vcpu_runnable(v));
switch ( type )
{
@@ -1364,13 +1368,18 @@ int vmx_add_msr(uint32_t msr, enum vmx_m
return -EINVAL;
}
+ vmx_vmcs_enter(v);
+
/* Allocate memory on first use. */
if ( unlikely(!*ptr) )
{
paddr_t addr;
if ( (*ptr = alloc_xenheap_page()) == NULL )
- return -ENOMEM;
+ {
+ rc = -ENOMEM;
+ goto out;
+ }
addr = virt_to_maddr(*ptr);
@@ -1392,10 +1401,16 @@ int vmx_add_msr(uint32_t msr, enum vmx_m
ent = locate_msr_entry(start, end, msr);
if ( (ent < end) && (ent->index == msr) )
- return 0;
+ {
+ rc = 0;
+ goto out;
+ }
if ( total == (PAGE_SIZE / sizeof(*ent)) )
- return -ENOSPC;
+ {
+ rc = -ENOSPC;
+ goto out;
+ }
memmove(ent + 1, ent, sizeof(*ent) * (end - ent));
@@ -1416,7 +1431,12 @@ int vmx_add_msr(uint32_t msr, enum vmx_m
break;
}
- return 0;
+ rc = 0;
+
+ out:
+ vmx_vmcs_exit(v);
+
+ return rc;
}
void vmx_set_eoi_exit_bitmap(struct vcpu *v, u8 vector)
--- a/xen/arch/x86/hvm/vmx/vmx.c
+++ b/xen/arch/x86/hvm/vmx/vmx.c
@@ -2903,7 +2903,7 @@ static int is_last_branch_msr(u32 ecx)
static int vmx_msr_read_intercept(unsigned int msr, uint64_t *msr_content)
{
- const struct vcpu *curr = current;
+ struct vcpu *curr = current;
HVM_DBG_LOG(DBG_LEVEL_MSR, "ecx=%#x", msr);
@@ -2962,7 +2962,7 @@ static int vmx_msr_read_intercept(unsign
goto done;
}
- if ( vmx_read_guest_msr(msr, msr_content) == 0 )
+ if ( vmx_read_guest_msr(curr, msr, msr_content) == 0 )
break;
if ( is_last_branch_msr(msr) )
@@ -3135,7 +3135,7 @@ static int vmx_msr_write_intercept(unsig
for ( ; (rc == 0) && lbr->count; lbr++ )
for ( i = 0; (rc == 0) && (i < lbr->count); i++ )
- if ( (rc = vmx_add_guest_msr(lbr->base + i)) == 0 )
+ if ( (rc = vmx_add_guest_msr(v, lbr->base + i)) == 0 )
{
vmx_clear_msr_intercept(v, lbr->base + i, VMX_MSR_RW);
if ( lbr_tsx_fixup_needed )
@@ -3178,7 +3178,7 @@ static int vmx_msr_write_intercept(unsig
switch ( long_mode_do_msr_write(msr, msr_content) )
{
case HNDL_unhandled:
- if ( (vmx_write_guest_msr(msr, msr_content) != 0) &&
+ if ( (vmx_write_guest_msr(v, msr, msr_content) != 0) &&
!is_last_branch_msr(msr) )
switch ( wrmsr_hypervisor_regs(msr, msr_content) )
{
@@ -4196,7 +4196,7 @@ static void lbr_tsx_fixup(void)
struct vmx_msr_entry *msr_area = curr->arch.hvm_vmx.msr_area;
struct vmx_msr_entry *msr;
- if ( (msr = vmx_find_msr(lbr_from_start, VMX_MSR_GUEST)) != NULL )
+ if ( (msr = vmx_find_msr(curr, lbr_from_start, VMX_MSR_GUEST)) != NULL )
{
/*
* Sign extend into bits 61:62 while preserving bit 63
@@ -4206,15 +4206,15 @@ static void lbr_tsx_fixup(void)
msr->data |= ((LBR_FROM_SIGNEXT_2MSB & msr->data) << 2);
}
- if ( (msr = vmx_find_msr(lbr_lastint_from, VMX_MSR_GUEST)) != NULL )
+ if ( (msr = vmx_find_msr(curr, lbr_lastint_from, VMX_MSR_GUEST)) != NULL )
msr->data |= ((LBR_FROM_SIGNEXT_2MSB & msr->data) << 2);
}
-static void sign_extend_msr(u32 msr, int type)
+static void sign_extend_msr(struct vcpu *v, u32 msr, int type)
{
struct vmx_msr_entry *entry;
- if ( (entry = vmx_find_msr(msr, type)) != NULL )
+ if ( (entry = vmx_find_msr(v, msr, type)) != NULL )
{
if ( entry->data & VADDR_TOP_BIT )
entry->data |= CANONICAL_MASK;
@@ -4225,6 +4225,8 @@ static void sign_extend_msr(u32 msr, int
static void bdw_erratum_bdf14_fixup(void)
{
+ struct vcpu *curr = current;
+
/*
* Occasionally, on certain Broadwell CPUs MSR_IA32_LASTINTTOIP has
* been observed to have the top three bits corrupted as though the
@@ -4234,8 +4236,8 @@ static void bdw_erratum_bdf14_fixup(void
* erratum BDF14. Fix up MSR_IA32_LASTINT{FROM,TO}IP by
* sign-extending into bits 48:63.
*/
- sign_extend_msr(MSR_IA32_LASTINTFROMIP, VMX_MSR_GUEST);
- sign_extend_msr(MSR_IA32_LASTINTTOIP, VMX_MSR_GUEST);
+ sign_extend_msr(curr, MSR_IA32_LASTINTFROMIP, VMX_MSR_GUEST);
+ sign_extend_msr(curr, MSR_IA32_LASTINTTOIP, VMX_MSR_GUEST);
}
static void lbr_fixup(void)
--- a/xen/include/asm-x86/hvm/vmx/vmcs.h
+++ b/xen/include/asm-x86/hvm/vmx/vmcs.h
@@ -130,10 +130,17 @@ struct arch_vmx_struct {
uint64_t sfmask;
struct vmx_msr_bitmap *msr_bitmap;
- unsigned int msr_count;
+
+ /*
+ * Most accesses to the MSR host/guest load/save lists are in current
+ * context. However, the data can be modified by toolstack/migration
+ * actions. Remote access is only permitted for paused vcpus, and is
+ * protected under the domctl lock.
+ */
struct vmx_msr_entry *msr_area;
- unsigned int host_msr_count;
struct vmx_msr_entry *host_msr_area;
+ unsigned int msr_count;
+ unsigned int host_msr_count;
unsigned long eoi_exitmap_changed;
DECLARE_BITMAP(eoi_exit_bitmap, NR_VECTORS);
@@ -534,23 +541,25 @@ enum vmx_msr_list_type {
VMX_MSR_GUEST, /* MSRs saved on VMExit, loaded on VMEntry. */
};
-int vmx_add_msr(uint32_t msr, enum vmx_msr_list_type type);
+int vmx_add_msr(struct vcpu *v, uint32_t msr, enum vmx_msr_list_type type);
-static inline int vmx_add_host_load_msr(uint32_t msr)
+static inline int vmx_add_guest_msr(struct vcpu *v, uint32_t msr)
{
- return vmx_add_msr(msr, VMX_MSR_HOST);
+ return vmx_add_msr(v, msr, VMX_MSR_GUEST);
}
-static inline int vmx_add_guest_msr(uint32_t msr)
+static inline int vmx_add_host_load_msr(struct vcpu *v, uint32_t msr)
{
- return vmx_add_msr(msr, VMX_MSR_GUEST);
+ return vmx_add_msr(v, msr, VMX_MSR_HOST);
}
-struct vmx_msr_entry *vmx_find_msr(uint32_t msr, enum vmx_msr_list_type type);
+struct vmx_msr_entry *vmx_find_msr(const struct vcpu *v, uint32_t msr,
+ enum vmx_msr_list_type type);
-static inline int vmx_read_guest_msr(uint32_t msr, uint64_t *val)
+static inline int vmx_read_guest_msr(const struct vcpu *v, uint32_t msr,
+ uint64_t *val)
{
- const struct vmx_msr_entry *ent = vmx_find_msr(msr, VMX_MSR_GUEST);
+ const struct vmx_msr_entry *ent = vmx_find_msr(v, msr, VMX_MSR_GUEST);
if ( !ent )
return -ESRCH;
@@ -560,9 +569,10 @@ static inline int vmx_read_guest_msr(uin
return 0;
}
-static inline int vmx_write_guest_msr(uint32_t msr, uint64_t val)
+static inline int vmx_write_guest_msr(struct vcpu *v, uint32_t msr,
+ uint64_t val)
{
- struct vmx_msr_entry *ent = vmx_find_msr(msr, VMX_MSR_GUEST);
+ struct vmx_msr_entry *ent = vmx_find_msr(v, msr, VMX_MSR_GUEST);
if ( !ent )
return -ESRCH;
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -796,7 +796,7 @@ static inline struct domain *next_domain
#define _VPF_parked 8
#define VPF_parked (1UL<<_VPF_parked)
-static inline int vcpu_runnable(struct vcpu *v)
+static inline bool vcpu_runnable(const struct vcpu *v)
{
return !(v->pause_flags |
atomic_read(&v->pause_count) |
