NetBSD/lib/libnvmm/libnvmm.3

740 lines
19 KiB
Groff

.\" $NetBSD: libnvmm.3,v 1.28 2021/12/10 11:08:45 msaitoh Exp $
.\"
.\" Copyright (c) 2018-2020 Maxime Villard, m00nbsd.net
.\" All rights reserved.
.\"
.\" This code is part of the NVMM hypervisor.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice, this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
.\" IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
.\" OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
.\" IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
.\" BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
.\" LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
.\" AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
.\" OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.Dd December 10, 2021
.Dt LIBNVMM 3
.Os
.Sh NAME
.Nm libnvmm
.Nd NetBSD Virtualization API
.Sh LIBRARY
.Lb libnvmm
.Sh SYNOPSIS
.In nvmm.h
.Ft int
.Fn nvmm_init "void"
.Ft int
.Fn nvmm_capability "struct nvmm_capability *cap"
.Ft int
.Fn nvmm_machine_create "struct nvmm_machine *mach"
.Ft int
.Fn nvmm_machine_destroy "struct nvmm_machine *mach"
.Ft int
.Fn nvmm_machine_configure "struct nvmm_machine *mach" "uint64_t op" \
"void *conf"
.Ft int
.Fn nvmm_vcpu_create "struct nvmm_machine *mach" "nvmm_cpuid_t cpuid" \
"struct nvmm_vcpu *vcpu"
.Ft int
.Fn nvmm_vcpu_destroy "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu"
.Ft int
.Fn nvmm_vcpu_configure "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu" \
"uint64_t op" "void *conf"
.Ft int
.Fn nvmm_vcpu_getstate "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu" \
"uint64_t flags"
.Ft int
.Fn nvmm_vcpu_setstate "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu" \
"uint64_t flags"
.Ft int
.Fn nvmm_vcpu_inject "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu"
.Ft int
.Fn nvmm_vcpu_run "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu"
.Ft int
.Fn nvmm_hva_map "struct nvmm_machine *mach" "uintptr_t hva" "size_t size"
.Ft int
.Fn nvmm_hva_unmap "struct nvmm_machine *mach" "uintptr_t hva" "size_t size"
.Ft int
.Fn nvmm_gpa_map "struct nvmm_machine *mach" "uintptr_t hva" "gpaddr_t gpa" \
"size_t size" "int prot"
.Ft int
.Fn nvmm_gpa_unmap "struct nvmm_machine *mach" "uintptr_t hva" "gpaddr_t gpa" \
"size_t size"
.Ft int
.Fn nvmm_gva_to_gpa "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu" \
"gvaddr_t gva" "gpaddr_t *gpa" "nvmm_prot_t *prot"
.Ft int
.Fn nvmm_gpa_to_hva "struct nvmm_machine *mach" "gpaddr_t gpa" \
"uintptr_t *hva" "nvmm_prot_t *prot"
.Ft int
.Fn nvmm_assist_io "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu"
.Ft int
.Fn nvmm_assist_mem "struct nvmm_machine *mach" "struct nvmm_vcpu *vcpu"
.Sh DESCRIPTION
.Nm
provides a library for emulator software to handle hardware-accelerated virtual
machines in
.Nx .
A virtual machine is described by an opaque structure,
.Cd nvmm_machine .
Emulator software should not attempt to modify this structure directly, and
should use the API provided by
.Nm
to manage virtual machines.
A virtual CPU is described by a public structure,
.Cd nvmm_vcpu .
.Pp
.Fn nvmm_init
initializes NVMM.
See
.Sx NVMM Initialization
below for details.
.Pp
.Fn nvmm_capability
gets the capabilities of NVMM.
See
.Sx NVMM Capability
below for details.
.Pp
.Fn nvmm_machine_create
creates a virtual machine in the kernel.
The
.Fa mach
structure is initialized, and describes the machine.
.Pp
.Fn nvmm_machine_destroy
destroys the virtual machine described in
.Fa mach .
.Pp
.Fn nvmm_machine_configure
configures, on the machine
.Fa mach ,
the parameter indicated in
.Fa op .
.Fa conf
describes the value of the parameter.
.Pp
.Fn nvmm_vcpu_create
creates a virtual CPU in the machine
.Fa mach ,
giving it the CPU id
.Fa cpuid ,
and initializes
.Fa vcpu .
.Pp
.Fn nvmm_vcpu_destroy
destroys the virtual CPU identified by
.Fa vcpu
in the machine
.Fa mach .
.Pp
.Fn nvmm_vcpu_configure
configures, on the VCPU
.Fa vcpu
of machine
.Fa mach ,
the parameter indicated in
.Fa op .
.Fa conf
describes the value of the parameter.
.Pp
.Fn nvmm_vcpu_getstate
gets the state of the virtual CPU identified by
.Fa vcpu
in the machine
.Fa mach .
.Fa flags
is the bitmap of the components that are to be retrieved.
The components are located in
.Fa vcpu->state .
See
.Sx VCPU State Area
below for details.
.Pp
.Fn nvmm_vcpu_setstate
sets the state of the virtual CPU identified by
.Fa vcpu
in the machine
.Fa mach .
.Fa flags
is the bitmap of the components that are to be set.
The components are located in
.Fa vcpu->state .
See
.Sx VCPU State Area
below for details.
.Pp
.Fn nvmm_vcpu_inject
injects into the CPU identified by
.Fa vcpu
of the machine
.Fa mach
an event described by
.Fa vcpu->event .
See
.Sx Event Injection
below for details.
.Pp
.Fn nvmm_vcpu_run
runs the CPU identified by
.Fa vcpu
in the machine
.Fa mach ,
until a VM exit is triggered.
The
.Fa vcpu->exit
structure is filled to indicate the exit reason, and the associated parameters
if any.
.Pp
.Fn nvmm_hva_map
maps at address
.Fa hva
a buffer of size
.Fa size
in the calling process' virtual address space.
This buffer is allowed to be subsequently mapped in a virtual machine.
.Pp
.Fn nvmm_hva_unmap
unmaps the buffer of size
.Fa size
at address
.Fa hva
from the calling process' virtual address space.
.Pp
.Fn nvmm_gpa_map
maps into the guest physical memory beginning on address
.Fa gpa
the buffer of size
.Fa size
located at address
.Fa hva
of the calling process' virtual address space.
The
.Fa hva
parameter must point to a buffer that was previously mapped with
.Fn nvmm_hva_map .
.Pp
.Fn nvmm_gpa_unmap
removes the guest physical memory area beginning on address
.Fa gpa
and of size
.Fa size
from the machine
.Fa mach .
.Pp
.Fn nvmm_gva_to_gpa
translates, on the CPU
.Fa vcpu
from the machine
.Fa mach ,
the guest virtual address given in
.Fa gva
into a guest physical address returned in
.Fa gpa .
The associated page permissions are returned in
.Fa prot .
.Fa gva
must be page-aligned.
.Pp
.Fn nvmm_gpa_to_hva
translates, on the machine
.Fa mach ,
the guest physical address indicated in
.Fa gpa
into a host virtual address returned in
.Fa hva .
The associated page permissions are returned in
.Fa prot .
.Fa gpa
must be page-aligned.
.Pp
.Fn nvmm_assist_io
emulates the I/O operation described in
.Fa vcpu->exit
on CPU
.Fa vcpu
from machine
.Fa mach .
See
.Sx I/O Assist
below for details.
.Pp
.Fn nvmm_assist_mem
emulates the Mem operation described in
.Fa vcpu->exit
on CPU
.Fa vcpu
from machine
.Fa mach .
See
.Sx Mem Assist
below for details.
.Ss NVMM Initialization
NVMM initialization is performed by the
.Fn nvmm_init
function, which must be invoked by emulator software before any other NVMM
function.
.Pp
.Fn nvmm_init
opens the NVMM device, and expects to have the proper permissions to do so.
In a default configuration, this implies being part of the "nvmm" group.
If using a special configuration, emulator software should arrange to have the
proper permissions before invoking
.Fn nvmm_init ,
and can drop them after the call has completed.
.Pp
It is to be noted that
.Fn nvmm_init
may perform non-re-entrant operations, and should be called only once.
.Ss NVMM Capability
The
.Cd nvmm_capability
structure helps emulator software identify the capabilities offered by NVMM on
the host:
.Bd -literal
struct nvmm_capability {
uint64_t version;
uint64_t state_size;
uint64_t max_machines;
uint64_t max_vcpus;
uint64_t max_ram;
struct {
...
} arch;
};
.Ed
.Pp
For example, the
.Cd max_machines
field indicates the maximum number of virtual machines supported, while
.Cd max_vcpus
indicates the maximum number of VCPUs supported per virtual machine.
.Ss Machine Ownership
When a process creates a virtual machine via
.Fn nvmm_machine_create ,
it is considered the owner of this machine.
No other processes than the owner can operate a virtual machine.
.Pp
When an owner exits, all the virtual machines associated with it are destroyed,
if they were not already destroyed by the owner itself via
.Fn nvmm_machine_destroy .
.Pp
Virtual machines are not inherited across
.Xr fork 2
operations.
.Ss Machine Configuration
Emulator software can configure several parameters of a virtual machine by using
.Fn nvmm_machine_configure .
Currently, no parameters are implemented.
.Ss VCPU Configuration
Emulator software can configure several parameters of a VCPU by using
.Fn nvmm_vcpu_configure ,
which can take the following operations:
.Bd -literal
#define NVMM_VCPU_CONF_CALLBACKS 0
...
.Ed
.Pp
The higher fields depend on the architecture.
.Ss Guest-Host Mappings
Each virtual machine has an associated guest physical memory.
Emulator software is allowed to modify this guest physical memory by mapping
it into some parts of its virtual address space.
.Pp
Emulator software should follow the following steps to achieve that:
.Pp
.Bl -bullet -offset indent -compact
.It
Call
.Fn nvmm_hva_map
to create in the host's virtual address space an area of memory that can
be shared with a guest.
Typically, the
.Fa hva
parameter will be a pointer to an area that was previously mapped via
.Fn mmap .
.Fn nvmm_hva_map
will replace the content of the area, and will make it read-write (but not
executable).
.It
Make available in the guest an area of guest physical memory, by calling
.Fn nvmm_gpa_map
and passing in the
.Fa hva
parameter the value that was previously given to
.Fn nvmm_hva_map .
.Fn nvmm_gpa_map
does not replace the content of any memory, it only creates a direct link
from
.Fa gpa
into
.Fa hva .
.Fn nvmm_gpa_unmap
removes this link without modifying
.Fa hva .
.El
.Pp
The guest will then be able to use the guest physical address passed in the
.Fa gpa
parameter of
.Fn nvmm_gpa_map .
Each change the guest makes in
.Fa gpa
will be reflected in the host's
.Fa hva ,
and vice versa.
.Pp
It is illegal for emulator software to use
.Fn munmap
on an area that was mapped via
.Fn nvmm_hva_map .
.Ss VCPU State Area
A VCPU state area is a structure that entirely defines the content of the
registers of a VCPU.
Only one such structure exists, for x86:
.Bd -literal
struct nvmm_x64_state {
struct nvmm_x64_state_seg segs[NVMM_X64_NSEG];
uint64_t gprs[NVMM_X64_NGPR];
uint64_t crs[NVMM_X64_NCR];
uint64_t drs[NVMM_X64_NDR];
uint64_t msrs[NVMM_X64_NMSR];
struct nvmm_x64_state_intr intr;
struct fxsave fpu;
};
#define nvmm_vcpu_state nvmm_x64_state
.Ed
.Pp
Refer to functional examples to see precisely how to use this structure.
.Pp
A VCPU state area is divided in sub-states.
A
.Fa flags
parameter is used to set and get the VCPU state; it acts as a bitmap which
indicates which sub-states to set or get.
.Pp
During VM exits, a partial VCPU state area is provided in
.Va exitstate ,
see
.Sx Exit Reasons
below for details.
.Ss VCPU Programming Model
A VCPU is described by a public structure,
.Cd nvmm_vcpu :
.Bd -literal
struct nvmm_vcpu {
nvmm_cpuid_t cpuid;
struct nvmm_vcpu_state *state;
struct nvmm_vcpu_event *event;
struct nvmm_vcpu_exit *exit;
};
.Ed
.Pp
This structure is used both publicly by emulator software and internally by
.Nm .
Emulator software should not modify the pointers of this structure, because
they are initialized to special values by
.Nm .
.Pp
A call to
.Fn nvmm_vcpu_getstate
will fetch the desired parts of the VCPU state and put them in
.Fa vcpu->state .
A call to
.Fn nvmm_vcpu_setstate
will install in the VCPU the desired parts of
.Fa vcpu->state .
A call to
.Fn nvmm_vcpu_inject
will inject in the VCPU the event in
.Fa vcpu->event .
A call to
.Fn nvmm_vcpu_run
will fill
.Fa vcpu->exit
with the VCPU exit information.
.Pp
If emulator software uses several threads, a VCPU should be associated with
only one thread, and only this thread should perform VCPU modifications.
Emulator software should not modify the state of a VCPU with several
different threads.
.Ss Exit Reasons
The
.Cd nvmm_vcpu_exit
structure is used to handle VM exits:
.Bd -literal
/* Generic. */
#define NVMM_VCPU_EXIT_NONE 0x0000000000000000ULL
#define NVMM_VCPU_EXIT_INVALID 0xFFFFFFFFFFFFFFFFULL
/* x86: operations. */
#define NVMM_VCPU_EXIT_MEMORY 0x0000000000000001ULL
#define NVMM_VCPU_EXIT_IO 0x0000000000000002ULL
/* x86: changes in VCPU state. */
#define NVMM_VCPU_EXIT_SHUTDOWN 0x0000000000001000ULL
#define NVMM_VCPU_EXIT_INT_READY 0x0000000000001001ULL
#define NVMM_VCPU_EXIT_NMI_READY 0x0000000000001002ULL
#define NVMM_VCPU_EXIT_HALTED 0x0000000000001003ULL
#define NVMM_VCPU_EXIT_TPR_CHANGED 0x0000000000001004ULL
/* x86: instructions. */
#define NVMM_VCPU_EXIT_RDMSR 0x0000000000002000ULL
#define NVMM_VCPU_EXIT_WRMSR 0x0000000000002001ULL
#define NVMM_VCPU_EXIT_MONITOR 0x0000000000002002ULL
#define NVMM_VCPU_EXIT_MWAIT 0x0000000000002003ULL
#define NVMM_VCPU_EXIT_CPUID 0x0000000000002004ULL
struct nvmm_vcpu_exit {
uint64_t reason;
union {
...
} u;
struct {
...
} exitstate;
};
.Ed
.Pp
The
.Va reason
field indicates the reason of the VM exit.
Additional parameters describing the exit can be present in
.Va u .
.Va exitstate
contains a partial, implementation-specific VCPU state, usable as a fast-path
to retrieve certain state values.
.Pp
It is possible that a VM exit was caused by a reason internal to the host
kernel, and that emulator software should not be concerned with.
In this case, the exit reason is set to
.Cd NVMM_VCPU_EXIT_NONE .
This gives a chance for emulator software to halt the VM in its tracks.
.Pp
Refer to functional examples to see precisely how to handle VM exits.
.Ss Event Injection
It is possible to inject an event into a VCPU.
An event can be a hardware interrupt, a software interrupt, or a software
exception, defined by:
.Bd -literal
#define NVMM_VCPU_EVENT_EXCP 0
#define NVMM_VCPU_EVENT_INTR 1
struct nvmm_vcpu_event {
u_int type;
uint8_t vector;
union {
struct {
uint64_t error;
} excp;
} u;
};
.Ed
.Pp
This describes an event of type
.Va type ,
to be sent to vector number
.Va vector ,
with a possible additional
.Va error
code that is implementation-specific.
.Pp
It is possible that the VCPU is in a state where it cannot receive this
event, if:
.Pp
.Bl -bullet -offset indent -compact
.It
the event is a hardware interrupt, and the VCPU runs with interrupts disabled,
or
.It
the event is a non-maskable interrupt (NMI), and the VCPU is already in an
in-NMI context.
.El
.Pp
Emulator software can manage interrupt and NMI window-exiting via the
.Va intr
component of the VCPU state.
When such window-exiting is enabled, NVMM will cause a VM exit with reason
.Cd NVMM_VCPU_EXIT_INT_READY
or
.Cd NVMM_VCPU_EXIT_NMI_READY
to indicate that the guest is now able to handle the corresponding class
of interrupts.
.Ss Assist Callbacks
In order to assist emulation of certain operations,
.Nm
requires emulator software to register, via
.Fn nvmm_vcpu_configure ,
a set of callbacks described in the following structure:
.Bd -literal
struct nvmm_assist_callbacks {
void (*io)(struct nvmm_io *);
void (*mem)(struct nvmm_mem *);
};
.Ed
.Pp
These callbacks are used by
.Nm
each time
.Fn nvmm_assist_io
or
.Fn nvmm_assist_mem
are invoked.
Emulator software that does not intend to use either of these assists can put
.Dv NULL
in the callbacks.
.Ss I/O Assist
When a VM exit occurs with reason
.Cd NVMM_VCPU_EXIT_IO ,
it is necessary for emulator software to emulate the associated I/O operation.
.Nm
provides an easy way for emulator software to perform that.
.Pp
.Fn nvmm_assist_io
will call the registered
.Fa io
callback function and give it a
.Cd nvmm_io
structure as argument.
This structure describes an I/O transaction:
.Bd -literal
struct nvmm_io {
struct nvmm_machine *mach;
struct nvmm_vcpu *vcpu;
uint16_t port;
bool in;
size_t size;
uint8_t *data;
};
.Ed
.Pp
The callback can emulate the operation using this descriptor, following two
unique cases:
.Pp
.Bl -bullet -offset indent -compact
.It
The operation is an input.
In this case, the callback should fill
.Va data
with the desired value.
.It
The operation is an output.
In this case, the callback should read
.Va data
to retrieve the desired value.
.El
.Pp
In either case,
.Va port
will indicate the I/O port,
.Va in
will indicate if the operation is an input, and
.Va size
will indicate the size of the access.
.Ss Mem Assist
When a VM exit occurs with reason
.Cd NVMM_VCPU_EXIT_MEMORY ,
it is necessary for emulator software to emulate the associated memory
operation.
.Nm
provides an easy way for emulator software to perform that, similar to the I/O
Assist.
.Pp
.Fn nvmm_assist_mem
will call the registered
.Fa mem
callback function and give it a
.Cd nvmm_mem
structure as argument.
This structure describes a Mem transaction:
.Bd -literal
struct nvmm_mem {
struct nvmm_machine *mach;
struct nvmm_vcpu *vcpu;
gpaddr_t gpa;
bool write;
size_t size;
uint8_t *data;
};
.Ed
.Pp
The callback can emulate the operation using this descriptor, following two
unique cases:
.Pp
.Bl -bullet -offset indent -compact
.It
The operation is a read.
In this case, the callback should fill
.Va data
with the desired value.
.It
The operation is a write.
In this case, the callback should read
.Va data
to retrieve the desired value.
.El
.Pp
In either case,
.Va gpa
will indicate the guest physical address,
.Va write
will indicate if the access is a write, and
.Va size
will indicate the size of the access.
.Sh RETURN VALUES
Upon successful completion, each of these functions returns zero.
Otherwise, a value of \-1 is returned and the global
variable
.Va errno
is set to indicate the error.
.Sh FILES
.Bl -tag -width XXXX -compact
.It Lk https://www.netbsd.org/~maxv/nvmm/nvmm-demo.zip
Functional example (demonstrator).
Contains an emulator that uses the
.Nm
API, and a small kernel that exercises this emulator.
.It Pa src/sys/dev/nvmm/
Source code of the kernel NVMM driver.
.It Pa src/lib/libnvmm/
Source code of the
.Nm
library.
.El
.Sh ERRORS
These functions will fail if:
.Bl -tag -width [ENOBUFS]
.It Bq Er EEXIST
An attempt was made to create a machine or a VCPU that already exists.
.It Bq Er EFAULT
An attempt was made to emulate a memory-based operation in a guest, and the
guest page tables did not have the permissions necessary for the operation
to complete successfully.
.It Bq Er EINVAL
An inappropriate parameter was used.
.It Bq Er ENOBUFS
The maximum number of machines or VCPUs was reached.
.It Bq Er ENOENT
A query was made on a machine or a VCPU that does not exist.
.It Bq Er EPERM
An attempt was made to access a machine that does not belong to the process.
.El
.Sh SEE ALSO
.Xr nvmm 4 ,
.Xr nvmmctl 8
.Sh AUTHORS
NVMM was designed and implemented by
.An Maxime Villard .