Index: head/usr.sbin/bhyve/bhyve.8 =================================================================== --- head/usr.sbin/bhyve/bhyve.8 (revision 362611) +++ head/usr.sbin/bhyve/bhyve.8 (revision 362612) @@ -1,714 +1,716 @@ .\" Copyright (c) 2013 Peter Grehan .\" All rights reserved. .\" .\" 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 AUTHORS AND CONTRIBUTORS ``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 AUTHORS OR CONTRIBUTORS 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. .\" .\" $FreeBSD$ .\" -.Dd May 18, 2020 +.Dd Jun 25, 2020 .Dt BHYVE 8 .Os .Sh NAME .Nm bhyve .Nd "run a guest operating system inside a virtual machine" .Sh SYNOPSIS .Nm -.Op Fl AabCeHhPSuWwxY +.Op Fl AabCDeHhPSuWwxY .Oo .Sm off .Fl c\~ .Oo .Op Cm cpus= .Ar numcpus .Oc .Op Cm ,sockets= Ar n .Op Cm ,cores= Ar n .Op Cm ,threads= Ar n .Oc .Sm on .Op Fl G Ar port .Op Fl g Ar gdbport .Oo Fl l .Sm off .Cm help | Ar lpcdev Op Cm \&, Ar conf .Sm on .Oc .Oo Fl m .Sm off .Ar memsize .Oo .Cm K No | Cm k No | Cm M No | Cm m No | Cm G No | Cm g No | Cm T No | Cm t .Oc .Sm on .Oc .Op Fl p Ar vcpu Ns Cm \&: Ns Ar hostcpu .Op Fl r Ar file .Oo Fl s .Sm off .Cm help | Ar slot Cm \&, Ar emulation Op Cm \&, Ar conf .Sm on .Oc .Op Fl U Ar uuid .Ar vmname .Sh DESCRIPTION .Nm is a hypervisor that runs guest operating systems inside a virtual machine. .Pp Parameters such as the number of virtual CPUs, amount of guest memory, and I/O connectivity can be specified with command-line parameters. .Pp If not using a boot ROM, the guest operating system must be loaded with .Xr bhyveload 8 or a similar boot loader before running .Nm , otherwise, it is enough to run .Nm with a boot ROM of choice. .Pp .Nm runs until the guest operating system reboots or an unhandled hypervisor exit is detected. .Sh OPTIONS .Bl -tag -width 10n .It Fl a The guest's local APIC is configured in xAPIC mode. The xAPIC mode is the default setting so this option is redundant. It will be deprecated in a future version. .It Fl A Generate ACPI tables. Required for .Fx Ns /amd64 guests. .It Fl b Enable a low-level console device supported by .Fx kernels compiled with .Cd "device bvmconsole" . This option will be deprecated in a future version. .It Fl c Op Ar setting ... Number of guest virtual CPUs and/or the CPU topology. The default value for each of .Ar numcpus , .Ar sockets , .Ar cores , and .Ar threads is 1. The current maximum number of guest virtual CPUs is 16. If .Ar numcpus is not specified then it will be calculated from the other arguments. The topology must be consistent in that the .Ar numcpus must equal the product of .Ar sockets , .Ar cores , and .Ar threads . If a .Ar setting is specified more than once the last one has precedence. .It Fl C Include guest memory in core file. +.It Fl D +Destroy the VM on guest initiated power-off. .It Fl e Force .Nm to exit when a guest issues an access to an I/O port that is not emulated. This is intended for debug purposes. .It Fl g Ar gdbport For .Fx kernels compiled with .Cd "device bvmdebug" , allow a remote kernel kgdb to be relayed to the guest kernel gdb stub via a local IPv4 address and this port. This option will be deprecated in a future version. .It Fl G Ar port Start a debug server that uses the GDB protocol to export guest state to a debugger. An IPv4 TCP socket will be bound to the supplied .Ar port to listen for debugger connections. Only a single debugger may be attached to the debug server at a time. If .Ar port begins with .Sq w , .Nm will pause execution at the first instruction waiting for a debugger to attach. .It Fl h Print help message and exit. .It Fl H Yield the virtual CPU thread when a HLT instruction is detected. If this option is not specified, virtual CPUs will use 100% of a host CPU. .It Fl l Op Ar help|lpcdev Ns Op , Ns Ar conf Allow devices behind the LPC PCI-ISA bridge to be configured. The only supported devices are the TTY-class devices .Ar com1 and .Ar com2 and the boot ROM device .Ar bootrom . .Pp .Ar help print a list of supported LPC devices. .It Fl m Ar memsize Ns Op Ar K|k|M|m|G|g|T|t Guest physical memory size in bytes. This must be the same size that was given to .Xr bhyveload 8 . .Pp The size argument may be suffixed with one of K, M, G or T (either upper or lower case) to indicate a multiple of kilobytes, megabytes, gigabytes, or terabytes. If no suffix is given, the value is assumed to be in megabytes. .Pp .Ar memsize defaults to 256M. .It Fl p Ar vcpu:hostcpu Pin guest's virtual CPU .Em vcpu to .Em hostcpu . .It Fl P Force the guest virtual CPU to exit when a PAUSE instruction is detected. .It Fl r Ar file Resume a guest from a snapshot. The guest memory contents are restored from .Ar file , and the guest device and vCPU state are restored from the file .Dq Ar file Ns .kern . .Pp Note that the current snapshot file format requires that the configuration of devices in the new VM match the VM from which the snapshot was taken by specifying the same .Op Fl s and .Op Fl l options. The count of vCPUs and memory configuration are read from the snapshot. .It Fl s Op Ar help|slot,emulation Ns Op , Ns Ar conf Configure a virtual PCI slot and function. .Pp .Nm provides PCI bus emulation and virtual devices that can be attached to slots on the bus. There are 32 available slots, with the option of providing up to 8 functions per slot. .Bl -tag -width 10n .It Ar help print a list of supported PCI devices. .It Ar slot .Ar pcislot[:function] .Ar bus:pcislot:function .Pp The .Ar pcislot value is 0 to 31. The optional .Ar function value is 0 to 7. The optional .Ar bus value is 0 to 255. If not specified, the .Ar function value defaults to 0. If not specified, the .Ar bus value defaults to 0. .It Ar emulation .Bl -tag -width 10n .It Li hostbridge | Li amd_hostbridge .Pp Provide a simple host bridge. This is usually configured at slot 0, and is required by most guest operating systems. The .Li amd_hostbridge emulation is identical but uses a PCI vendor ID of .Li AMD . .It Li passthru PCI pass-through device. .It Li virtio-net Virtio network interface. .It Li virtio-blk Virtio block storage interface. .It Li virtio-scsi Virtio SCSI interface. .It Li virtio-rnd Virtio RNG interface. .It Li virtio-console Virtio console interface, which exposes multiple ports to the guest in the form of simple char devices for simple IO between the guest and host userspaces. .It Li ahci AHCI controller attached to arbitrary devices. .It Li ahci-cd AHCI controller attached to an ATAPI CD/DVD. .It Li ahci-hd AHCI controller attached to a SATA hard-drive. .It Li e1000 Intel e82545 network interface. .It Li uart PCI 16550 serial device. .It Li lpc LPC PCI-ISA bridge with COM1 and COM2 16550 serial ports and a boot ROM. The LPC bridge emulation can only be configured on bus 0. .It Li fbuf Raw framebuffer device attached to VNC server. .It Li xhci eXtensible Host Controller Interface (xHCI) USB controller. .It Li nvme NVM Express (NVMe) controller. .It Li hda High Definition Audio Controller. .El .It Op Ar conf This optional parameter describes the backend for device emulations. If .Ar conf is not specified, the device emulation has no backend and can be considered unconnected. .Pp Network backends: .Bl -tag -width 10n .It Ar tapN Ns Oo , Ns Ar mac=xx:xx:xx:xx:xx:xx Oc Ns Oo , Ns Ar mtu=N Oc .It Ar vmnetN Ns Oo , Ns Ar mac=xx:xx:xx:xx:xx:xx Oc Ns Oo , Ns Ar mtu=N Oc .It Ar netgraph,path=ADDRESS,peerhook=HOOK Ns Oo , Ns Ar socket=NAME Oc Ns Oo , Ns Ar hook=HOOK Oc Ns Oo , Ns Ar mac=xx:xx:xx:xx:xx:xx Oc Ns Oo , Ns Ar mtu=N Oc .Pp If .Ar mac is not specified, the MAC address is derived from a fixed OUI and the remaining bytes from an MD5 hash of the slot and function numbers and the device name. .Pp The MAC address is an ASCII string in .Xr ethers 5 format. .Pp With virtio-net devices, the .Ar mtu parameter can be specified to inform the guest about the largest MTU that should be allowed, expressed in bytes. .Pp With netgraph backend, the .Ar path and .Ar peerhook parameters must be specified to set the destination node and corresponding hook. The optional parameters .Ar socket and .Ar hook may be used to set the .Xr ng_socket 4 node name and source hook. The .Ar ADDRESS , .Ar HOOK and .Ar NAME must comply with .Xr netgraph 4 addressing rules. .El .Pp Block storage devices: .Bl -tag -width 10n .It Pa /filename Ns Oo , Ns Ar block-device-options Oc .It Pa /dev/xxx Ns Oo , Ns Ar block-device-options Oc .El .Pp The .Ar block-device-options are: .Bl -tag -width 8n .It Li nocache Open the file with .Dv O_DIRECT . .It Li direct Open the file using .Dv O_SYNC . .It Li ro Force the file to be opened read-only. .It Li sectorsize= Ns Ar logical Ns Oo / Ns Ar physical Oc Specify the logical and physical sector sizes of the emulated disk. The physical sector size is optional and is equal to the logical sector size if not explicitly specified. .El .Pp SCSI devices: .Bl -tag -width 10n .It Pa /dev/cam/ctl Ns Oo Ar pp . Ns Ar vp Oc Ns Oo , Ns Ar scsi-device-options Oc .El .Pp The .Ar scsi-device-options are: .Bl -tag -width 10n .It Li iid= Ns Ar IID Initiator ID to use when sending requests to specified CTL port. The default value is 0. .El .Pp TTY devices: .Bl -tag -width 10n .It Li stdio Connect the serial port to the standard input and output of the .Nm process. .It Pa /dev/xxx Use the host TTY device for serial port I/O. .El .Pp Boot ROM device: .Bl -tag -width 10n .It Pa romfile Map .Ar romfile in the guest address space reserved for boot firmware. .El .Pp Pass-through devices: .Bl -tag -width 10n .It Ns Ar slot Ns / Ns Ar bus Ns / Ns Ar function Connect to a PCI device on the host at the selector described by .Ar slot , .Ar bus , and .Ar function numbers. .El .Pp Guest memory must be wired using the .Fl S option when a pass-through device is configured. .Pp The host device must have been reserved at boot-time using the .Va pptdevs loader variable as described in .Xr vmm 4 . .Pp Virtio console devices: .Bl -tag -width 10n .It Li port1= Ns Pa /path/to/port1.sock Ns ,anotherport= Ns Pa ... A maximum of 16 ports per device can be created. Every port is named and corresponds to a Unix domain socket created by .Nm . .Nm accepts at most one connection per port at a time. .Pp Limitations: .Bl -bullet -offset 2n .It Due to lack of destructors in .Nm , sockets on the filesystem must be cleaned up manually after .Nm exits. .It There is no way to use the "console port" feature, nor the console port resize at present. .It Emergency write is advertised, but no-op at present. .El .El .Pp Framebuffer devices: .Bl -tag -width 10n .It Xo .Oo rfb= Ns Oo Ar IP\&: Oc Ns Ar port Oc Ns Oo ,w= Ns Ar width Oc Ns Oo ,h= Ns .Ar height Oc Ns Oo ,vga= Ns Ar vgaconf Oc Ns Oo Ns ,wait Oc Ns Oo ,password= Ns .Ar password Oc .Xc .Bl -tag -width 8n .It Ar IPv4:port No or Ar [IPv6%zone]:port An .Ar IP address and a .Ar port VNC should listen on. The default is to listen on localhost IPv4 address and default VNC port 5900. An IPv6 address must be enclosed in square brackets and may contain an optional zone identifier. .It Ar width No and Ar height A display resolution, width and height, respectively. If not specified, a default resolution of 1024x768 pixels will be used. Minimal supported resolution is 640x480 pixels, and maximum is 1920x1200 pixels. .It Ar vgaconf Possible values for this option are .Dq io (default), .Dq on , and .Dq off . PCI graphics cards have a dual personality in that they are standard PCI devices with BAR addressing, but may also implicitly decode legacy VGA I/O space .Pq Ad 0x3c0-3df and memory space .Pq 64KB at Ad 0xA0000 . The default .Dq io option should be used for guests that attempt to issue BIOS calls which result in I/O port queries, and fail to boot if I/O decode is disabled. .Pp The .Dq on option should be used along with the CSM BIOS capability in UEFI to boot traditional BIOS guests that require the legacy VGA I/O and memory regions to be available. .Pp The .Dq off option should be used for the UEFI guests that assume that VGA adapter is present if they detect the I/O ports. An example of such a guest is .Ox in UEFI mode. .Pp Please refer to the .Nm .Fx wiki page .Pq Lk https://wiki.freebsd.org/bhyve for configuration notes of particular guests. .It wait Instruct .Nm to only boot upon the initiation of a VNC connection, simplifying the installation of operating systems that require immediate keyboard input. This can be removed for post-installation use. .It password This type of authentication is known to be cryptographically weak and is not intended for use on untrusted networks. Many implementations will want to use stronger security, such as running the session over an encrypted channel provided by IPsec or SSH. .El .El .Pp xHCI USB devices: .Bl -tag -width 10n .It Li tablet A USB tablet device which provides precise cursor synchronization when using VNC. .El .Pp NVMe devices: .Bl -tag -width 10n .It Li devpath Accepted device paths are: .Ar /dev/blockdev or .Ar /path/to/image or .Ar ram=size_in_MiB . .It Li maxq Max number of queues. .It Li qsz Max elements in each queue. .It Li ioslots Max number of concurrent I/O requests. .It Li sectsz Sector size (defaults to blockif sector size). .It Li ser Serial number with maximum 20 characters. .El .Pp HD Audio devices: .Bl -tag -width 10n .It Li play Playback device, typically .Ar /dev/dsp0 . .It Li rec Recording device, typically .Ar /dev/dsp0 . .El .El .It Fl S Wire guest memory. .It Fl u RTC keeps UTC time. .It Fl U Ar uuid Set the universally unique identifier .Pq UUID in the guest's System Management BIOS System Information structure. By default a UUID is generated from the host's hostname and .Ar vmname . .It Fl w Ignore accesses to unimplemented Model Specific Registers (MSRs). This is intended for debug purposes. .It Fl W Force virtio PCI device emulations to use MSI interrupts instead of MSI-X interrupts. .It Fl x The guest's local APIC is configured in x2APIC mode. .It Fl Y Disable MPtable generation. .It Ar vmname Alphanumeric name of the guest. This should be the same as that created by .Xr bhyveload 8 . .El .Sh DEBUG SERVER The current debug server provides limited support for debuggers. .Ss Registers Each virtual CPU is exposed to the debugger as a thread. .Pp General purpose registers can be queried for each virtual CPU, but other registers such as floating-point and system registers cannot be queried. .Ss Memory Memory (including memory mapped I/O regions) can be read and written by the debugger. Memory operations use virtual addresses that are resolved to physical addresses via the current virtual CPU's active address translation. .Ss Control The running guest can be interrupted by the debugger at any time .Pq for example, by pressing Ctrl-C in the debugger . .Pp Single stepping is only supported on Intel CPUs supporting the MTRAP VM exit. .Pp Breakpoints are supported on Intel CPUs that support single stepping. Note that continuing from a breakpoint while interrupts are enabled in the guest may not work as expected due to timer interrupts firing while single stepping over the breakpoint. .Sh SIGNAL HANDLING .Nm deals with the following signals: .Pp .Bl -tag -width indent -compact .It SIGTERM Trigger ACPI poweroff for a VM .El .Sh EXIT STATUS Exit status indicates how the VM was terminated: .Pp .Bl -tag -width indent -compact .It 0 rebooted .It 1 powered off .It 2 halted .It 3 triple fault .It 4 exited due to an error .El .Sh EXAMPLES If not using a boot ROM, the guest operating system must have been loaded with .Xr bhyveload 8 or a similar boot loader before .Xr bhyve 4 can be run. Otherwise, the boot loader is not needed. .Pp To run a virtual machine with 1GB of memory, two virtual CPUs, a virtio block device backed by the .Pa /my/image filesystem image, and a serial port for the console: .Bd -literal -offset indent bhyve -c 2 -s 0,hostbridge -s 1,lpc -s 2,virtio-blk,/my/image \\ -l com1,stdio -A -H -P -m 1G vm1 .Ed .Pp Run a 24GB single-CPU virtual machine with three network ports, one of which has a MAC address specified: .Bd -literal -offset indent bhyve -s 0,hostbridge -s 1,lpc -s 2:0,virtio-net,tap0 \\ -s 2:1,virtio-net,tap1 \\ -s 2:2,virtio-net,tap2,mac=00:be:fa:76:45:00 \\ -s 3,virtio-blk,/my/image -l com1,stdio \\ -A -H -P -m 24G bigvm .Ed .Pp Run an 8GB quad-CPU virtual machine with 8 AHCI SATA disks, an AHCI ATAPI CD-ROM, a single virtio network port, an AMD hostbridge, and the console port connected to an .Xr nmdm 4 null-modem device. .Bd -literal -offset indent bhyve -c 4 \\ -s 0,amd_hostbridge -s 1,lpc \\ -s 1:0,ahci,hd:/images/disk.1,hd:/images/disk.2,\\ hd:/images/disk.3,hd:/images/disk.4,\\ hd:/images/disk.5,hd:/images/disk.6,\\ hd:/images/disk.7,hd:/images/disk.8,\\ cd:/images/install.iso \\ -s 3,virtio-net,tap0 \\ -l com1,/dev/nmdm0A \\ -A -H -P -m 8G .Ed .Pp Run a UEFI virtual machine with a display resolution of 800 by 600 pixels that can be accessed via VNC at: 0.0.0.0:5900. .Bd -literal -offset indent bhyve -c 2 -m 4G -w -H \\ -s 0,hostbridge \\ -s 3,ahci-cd,/path/to/uefi-OS-install.iso \\ -s 4,ahci-hd,disk.img \\ -s 5,virtio-net,tap0 \\ -s 29,fbuf,tcp=0.0.0.0:5900,w=800,h=600,wait \\ -s 30,xhci,tablet \\ -s 31,lpc -l com1,stdio \\ -l bootrom,/usr/local/share/uefi-firmware/BHYVE_UEFI.fd \\ uefivm .Ed .Pp Run a UEFI virtual machine with a VNC display that is bound to all IPv6 addresses on port 5900. .Bd -literal -offset indent bhyve -c 2 -m 4G -w -H \\ -s 0,hostbridge \\ -s 4,ahci-hd,disk.img \\ -s 5,virtio-net,tap0 \\ -s 29,fbuf,tcp=[::]:5900,w=800,h=600 \\ -s 30,xhci,tablet \\ -s 31,lpc -l com1,stdio \\ -l bootrom,/usr/local/share/uefi-firmware/BHYVE_UEFI.fd \\ uefivm .Ed .Sh SEE ALSO .Xr bhyve 4 , .Xr netgraph 4 , .Xr ng_socket 4 , .Xr nmdm 4 , .Xr vmm 4 , .Xr ethers 5 , .Xr bhyvectl 8 , .Xr bhyveload 8 .Pp .Rs .%A Intel .%B 64 and IA-32 Architectures Software Developer’s Manual .%V Volume 3 .Re .Sh HISTORY .Nm first appeared in .Fx 10.0 . .Sh AUTHORS .An Neel Natu Aq Mt neel@freebsd.org .An Peter Grehan Aq Mt grehan@freebsd.org Index: head/usr.sbin/bhyve/bhyverun.c =================================================================== --- head/usr.sbin/bhyve/bhyverun.c (revision 362611) +++ head/usr.sbin/bhyve/bhyverun.c (revision 362612) @@ -1,1445 +1,1452 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2011 NetApp, Inc. * All rights reserved. * * 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 NETAPP, INC ``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 NETAPP, INC OR CONTRIBUTORS 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. * * $FreeBSD$ */ #include __FBSDID("$FreeBSD$"); #include #ifndef WITHOUT_CAPSICUM #include #endif #include #ifdef BHYVE_SNAPSHOT #include #include #endif #include #ifdef BHYVE_SNAPSHOT #include #endif #include #include #include #ifndef WITHOUT_CAPSICUM #include #endif #include #include #include #include #include #ifdef BHYVE_SNAPSHOT #include #endif #include #include #include #include #include #include #include #include #ifdef BHYVE_SNAPSHOT #include #include #include #endif #include #ifndef WITHOUT_CAPSICUM #include #endif #include #include #include "bhyverun.h" #include "acpi.h" #include "atkbdc.h" #include "bootrom.h" #include "inout.h" #include "dbgport.h" #include "debug.h" #include "fwctl.h" #include "gdb.h" #include "ioapic.h" #include "kernemu_dev.h" #include "mem.h" #include "mevent.h" #include "mptbl.h" #include "pci_emul.h" #include "pci_irq.h" #include "pci_lpc.h" #include "smbiostbl.h" #ifdef BHYVE_SNAPSHOT #include "snapshot.h" #endif #include "xmsr.h" #include "spinup_ap.h" #include "rtc.h" #include "vmgenc.h" #define GUEST_NIO_PORT 0x488 /* guest upcalls via i/o port */ #define MB (1024UL * 1024) #define GB (1024UL * MB) static const char * const vmx_exit_reason_desc[] = { [EXIT_REASON_EXCEPTION] = "Exception or non-maskable interrupt (NMI)", [EXIT_REASON_EXT_INTR] = "External interrupt", [EXIT_REASON_TRIPLE_FAULT] = "Triple fault", [EXIT_REASON_INIT] = "INIT signal", [EXIT_REASON_SIPI] = "Start-up IPI (SIPI)", [EXIT_REASON_IO_SMI] = "I/O system-management interrupt (SMI)", [EXIT_REASON_SMI] = "Other SMI", [EXIT_REASON_INTR_WINDOW] = "Interrupt window", [EXIT_REASON_NMI_WINDOW] = "NMI window", [EXIT_REASON_TASK_SWITCH] = "Task switch", [EXIT_REASON_CPUID] = "CPUID", [EXIT_REASON_GETSEC] = "GETSEC", [EXIT_REASON_HLT] = "HLT", [EXIT_REASON_INVD] = "INVD", [EXIT_REASON_INVLPG] = "INVLPG", [EXIT_REASON_RDPMC] = "RDPMC", [EXIT_REASON_RDTSC] = "RDTSC", [EXIT_REASON_RSM] = "RSM", [EXIT_REASON_VMCALL] = "VMCALL", [EXIT_REASON_VMCLEAR] = "VMCLEAR", [EXIT_REASON_VMLAUNCH] = "VMLAUNCH", [EXIT_REASON_VMPTRLD] = "VMPTRLD", [EXIT_REASON_VMPTRST] = "VMPTRST", [EXIT_REASON_VMREAD] = "VMREAD", [EXIT_REASON_VMRESUME] = "VMRESUME", [EXIT_REASON_VMWRITE] = "VMWRITE", [EXIT_REASON_VMXOFF] = "VMXOFF", [EXIT_REASON_VMXON] = "VMXON", [EXIT_REASON_CR_ACCESS] = "Control-register accesses", [EXIT_REASON_DR_ACCESS] = "MOV DR", [EXIT_REASON_INOUT] = "I/O instruction", [EXIT_REASON_RDMSR] = "RDMSR", [EXIT_REASON_WRMSR] = "WRMSR", [EXIT_REASON_INVAL_VMCS] = "VM-entry failure due to invalid guest state", [EXIT_REASON_INVAL_MSR] = "VM-entry failure due to MSR loading", [EXIT_REASON_MWAIT] = "MWAIT", [EXIT_REASON_MTF] = "Monitor trap flag", [EXIT_REASON_MONITOR] = "MONITOR", [EXIT_REASON_PAUSE] = "PAUSE", [EXIT_REASON_MCE_DURING_ENTRY] = "VM-entry failure due to machine-check event", [EXIT_REASON_TPR] = "TPR below threshold", [EXIT_REASON_APIC_ACCESS] = "APIC access", [EXIT_REASON_VIRTUALIZED_EOI] = "Virtualized EOI", [EXIT_REASON_GDTR_IDTR] = "Access to GDTR or IDTR", [EXIT_REASON_LDTR_TR] = "Access to LDTR or TR", [EXIT_REASON_EPT_FAULT] = "EPT violation", [EXIT_REASON_EPT_MISCONFIG] = "EPT misconfiguration", [EXIT_REASON_INVEPT] = "INVEPT", [EXIT_REASON_RDTSCP] = "RDTSCP", [EXIT_REASON_VMX_PREEMPT] = "VMX-preemption timer expired", [EXIT_REASON_INVVPID] = "INVVPID", [EXIT_REASON_WBINVD] = "WBINVD", [EXIT_REASON_XSETBV] = "XSETBV", [EXIT_REASON_APIC_WRITE] = "APIC write", [EXIT_REASON_RDRAND] = "RDRAND", [EXIT_REASON_INVPCID] = "INVPCID", [EXIT_REASON_VMFUNC] = "VMFUNC", [EXIT_REASON_ENCLS] = "ENCLS", [EXIT_REASON_RDSEED] = "RDSEED", [EXIT_REASON_PM_LOG_FULL] = "Page-modification log full", [EXIT_REASON_XSAVES] = "XSAVES", [EXIT_REASON_XRSTORS] = "XRSTORS" }; typedef int (*vmexit_handler_t)(struct vmctx *, struct vm_exit *, int *vcpu); extern int vmexit_task_switch(struct vmctx *, struct vm_exit *, int *vcpu); const char *vmname; int guest_ncpus; uint16_t cores, maxcpus, sockets, threads; char *guest_uuid_str; int raw_stdio = 0; static int gdb_port = 0; static int guest_vmexit_on_hlt, guest_vmexit_on_pause; static int virtio_msix = 1; static int x2apic_mode = 0; /* default is xAPIC */ +static int destroy_on_poweroff = 0; static int strictio; static int strictmsr = 1; static int acpi; static char *progname; static const int BSP = 0; static cpuset_t cpumask; static void vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip); static struct vm_exit vmexit[VM_MAXCPU]; struct bhyvestats { uint64_t vmexit_bogus; uint64_t vmexit_reqidle; uint64_t vmexit_hlt; uint64_t vmexit_pause; uint64_t vmexit_mtrap; uint64_t vmexit_inst_emul; uint64_t cpu_switch_rotate; uint64_t cpu_switch_direct; } stats; struct mt_vmm_info { pthread_t mt_thr; struct vmctx *mt_ctx; int mt_vcpu; } mt_vmm_info[VM_MAXCPU]; static cpuset_t *vcpumap[VM_MAXCPU] = { NULL }; static void usage(int code) { fprintf(stderr, - "Usage: %s [-abehuwxACHPSWY]\n" + "Usage: %s [-abehuwxACDHPSWY]\n" " %*s [-c [[cpus=]numcpus][,sockets=n][,cores=n][,threads=n]]\n" " %*s [-g ] [-l ]\n" " %*s [-m mem] [-p vcpu:hostcpu] [-s ] [-U uuid] \n" " -a: local apic is in xAPIC mode (deprecated)\n" " -A: create ACPI tables\n" " -c: number of cpus and/or topology specification\n" " -C: include guest memory in core file\n" + " -D: destroy on power-off\n" " -e: exit on unhandled I/O access\n" " -g: gdb port\n" " -h: help\n" " -H: vmexit from the guest on hlt\n" " -l: LPC device configuration\n" " -m: memory size in MB\n" #ifdef BHYVE_SNAPSHOT " -r: path to checkpoint file\n" #endif " -p: pin 'vcpu' to 'hostcpu'\n" " -P: vmexit from the guest on pause\n" " -s: PCI slot config\n" " -S: guest memory cannot be swapped\n" " -u: RTC keeps UTC time\n" " -U: uuid\n" " -w: ignore unimplemented MSRs\n" " -W: force virtio to use single-vector MSI\n" " -x: local apic is in x2APIC mode\n" " -Y: disable MPtable generation\n", progname, (int)strlen(progname), "", (int)strlen(progname), "", (int)strlen(progname), ""); exit(code); } /* * XXX This parser is known to have the following issues: * 1. It accepts null key=value tokens ",,". * 2. It accepts whitespace after = and before value. * 3. Values out of range of INT are silently wrapped. * 4. It doesn't check non-final values. * 5. The apparently bogus limits of UINT16_MAX are for future expansion. * * The acceptance of a null specification ('-c ""') is by design to match the * manual page syntax specification, this results in a topology of 1 vCPU. */ static int topology_parse(const char *opt) { uint64_t ncpus; int c, chk, n, s, t, tmp; char *cp, *str; bool ns, scts; c = 1, n = 1, s = 1, t = 1; ns = false, scts = false; str = strdup(opt); if (str == NULL) goto out; while ((cp = strsep(&str, ",")) != NULL) { if (sscanf(cp, "%i%n", &tmp, &chk) == 1) { n = tmp; ns = true; } else if (sscanf(cp, "cpus=%i%n", &tmp, &chk) == 1) { n = tmp; ns = true; } else if (sscanf(cp, "sockets=%i%n", &tmp, &chk) == 1) { s = tmp; scts = true; } else if (sscanf(cp, "cores=%i%n", &tmp, &chk) == 1) { c = tmp; scts = true; } else if (sscanf(cp, "threads=%i%n", &tmp, &chk) == 1) { t = tmp; scts = true; #ifdef notyet /* Do not expose this until vmm.ko implements it */ } else if (sscanf(cp, "maxcpus=%i%n", &tmp, &chk) == 1) { m = tmp; #endif /* Skip the empty argument case from -c "" */ } else if (cp[0] == '\0') continue; else goto out; /* Any trailing garbage causes an error */ if (cp[chk] != '\0') goto out; } free(str); str = NULL; /* * Range check 1 <= n <= UINT16_MAX all values */ if (n < 1 || s < 1 || c < 1 || t < 1 || n > UINT16_MAX || s > UINT16_MAX || c > UINT16_MAX || t > UINT16_MAX) return (-1); /* If only the cpus was specified, use that as sockets */ if (!scts) s = n; /* * Compute sockets * cores * threads avoiding overflow * The range check above insures these are 16 bit values * If n was specified check it against computed ncpus */ ncpus = (uint64_t)s * c * t; if (ncpus > UINT16_MAX || (ns && n != ncpus)) return (-1); guest_ncpus = ncpus; sockets = s; cores = c; threads = t; return(0); out: free(str); return (-1); } static int pincpu_parse(const char *opt) { int vcpu, pcpu; if (sscanf(opt, "%d:%d", &vcpu, &pcpu) != 2) { fprintf(stderr, "invalid format: %s\n", opt); return (-1); } if (vcpu < 0 || vcpu >= VM_MAXCPU) { fprintf(stderr, "vcpu '%d' outside valid range from 0 to %d\n", vcpu, VM_MAXCPU - 1); return (-1); } if (pcpu < 0 || pcpu >= CPU_SETSIZE) { fprintf(stderr, "hostcpu '%d' outside valid range from " "0 to %d\n", pcpu, CPU_SETSIZE - 1); return (-1); } if (vcpumap[vcpu] == NULL) { if ((vcpumap[vcpu] = malloc(sizeof(cpuset_t))) == NULL) { perror("malloc"); return (-1); } CPU_ZERO(vcpumap[vcpu]); } CPU_SET(pcpu, vcpumap[vcpu]); return (0); } void vm_inject_fault(void *arg, int vcpu, int vector, int errcode_valid, int errcode) { struct vmctx *ctx; int error, restart_instruction; ctx = arg; restart_instruction = 1; error = vm_inject_exception(ctx, vcpu, vector, errcode_valid, errcode, restart_instruction); assert(error == 0); } void * paddr_guest2host(struct vmctx *ctx, uintptr_t gaddr, size_t len) { return (vm_map_gpa(ctx, gaddr, len)); } #ifdef BHYVE_SNAPSHOT uintptr_t paddr_host2guest(struct vmctx *ctx, void *addr) { return (vm_rev_map_gpa(ctx, addr)); } #endif int fbsdrun_vmexit_on_pause(void) { return (guest_vmexit_on_pause); } int fbsdrun_vmexit_on_hlt(void) { return (guest_vmexit_on_hlt); } int fbsdrun_virtio_msix(void) { return (virtio_msix); } static void * fbsdrun_start_thread(void *param) { char tname[MAXCOMLEN + 1]; struct mt_vmm_info *mtp; int vcpu; mtp = param; vcpu = mtp->mt_vcpu; snprintf(tname, sizeof(tname), "vcpu %d", vcpu); pthread_set_name_np(mtp->mt_thr, tname); #ifdef BHYVE_SNAPSHOT checkpoint_cpu_add(vcpu); #endif if (gdb_port != 0) gdb_cpu_add(vcpu); vm_loop(mtp->mt_ctx, vcpu, vmexit[vcpu].rip); /* not reached */ exit(1); return (NULL); } void fbsdrun_addcpu(struct vmctx *ctx, int fromcpu, int newcpu, uint64_t rip) { int error; assert(fromcpu == BSP); /* * The 'newcpu' must be activated in the context of 'fromcpu'. If * vm_activate_cpu() is delayed until newcpu's pthread starts running * then vmm.ko is out-of-sync with bhyve and this can create a race * with vm_suspend(). */ error = vm_activate_cpu(ctx, newcpu); if (error != 0) err(EX_OSERR, "could not activate CPU %d", newcpu); CPU_SET_ATOMIC(newcpu, &cpumask); /* * Set up the vmexit struct to allow execution to start * at the given RIP */ vmexit[newcpu].rip = rip; vmexit[newcpu].inst_length = 0; mt_vmm_info[newcpu].mt_ctx = ctx; mt_vmm_info[newcpu].mt_vcpu = newcpu; error = pthread_create(&mt_vmm_info[newcpu].mt_thr, NULL, fbsdrun_start_thread, &mt_vmm_info[newcpu]); assert(error == 0); } static int fbsdrun_deletecpu(struct vmctx *ctx, int vcpu) { if (!CPU_ISSET(vcpu, &cpumask)) { fprintf(stderr, "Attempting to delete unknown cpu %d\n", vcpu); exit(4); } CPU_CLR_ATOMIC(vcpu, &cpumask); return (CPU_EMPTY(&cpumask)); } static int vmexit_handle_notify(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu, uint32_t eax) { #if BHYVE_DEBUG /* * put guest-driven debug here */ #endif return (VMEXIT_CONTINUE); } static int vmexit_inout(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) { int error; int bytes, port, in, out; int vcpu; vcpu = *pvcpu; port = vme->u.inout.port; bytes = vme->u.inout.bytes; in = vme->u.inout.in; out = !in; /* Extra-special case of host notifications */ if (out && port == GUEST_NIO_PORT) { error = vmexit_handle_notify(ctx, vme, pvcpu, vme->u.inout.eax); return (error); } error = emulate_inout(ctx, vcpu, vme, strictio); if (error) { fprintf(stderr, "Unhandled %s%c 0x%04x at 0x%lx\n", in ? "in" : "out", bytes == 1 ? 'b' : (bytes == 2 ? 'w' : 'l'), port, vmexit->rip); return (VMEXIT_ABORT); } else { return (VMEXIT_CONTINUE); } } static int vmexit_rdmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) { uint64_t val; uint32_t eax, edx; int error; val = 0; error = emulate_rdmsr(ctx, *pvcpu, vme->u.msr.code, &val); if (error != 0) { fprintf(stderr, "rdmsr to register %#x on vcpu %d\n", vme->u.msr.code, *pvcpu); if (strictmsr) { vm_inject_gp(ctx, *pvcpu); return (VMEXIT_CONTINUE); } } eax = val; error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RAX, eax); assert(error == 0); edx = val >> 32; error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RDX, edx); assert(error == 0); return (VMEXIT_CONTINUE); } static int vmexit_wrmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) { int error; error = emulate_wrmsr(ctx, *pvcpu, vme->u.msr.code, vme->u.msr.wval); if (error != 0) { fprintf(stderr, "wrmsr to register %#x(%#lx) on vcpu %d\n", vme->u.msr.code, vme->u.msr.wval, *pvcpu); if (strictmsr) { vm_inject_gp(ctx, *pvcpu); return (VMEXIT_CONTINUE); } } return (VMEXIT_CONTINUE); } static int vmexit_spinup_ap(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) { (void)spinup_ap(ctx, *pvcpu, vme->u.spinup_ap.vcpu, vme->u.spinup_ap.rip); return (VMEXIT_CONTINUE); } #define DEBUG_EPT_MISCONFIG #ifdef DEBUG_EPT_MISCONFIG #define VMCS_GUEST_PHYSICAL_ADDRESS 0x00002400 static uint64_t ept_misconfig_gpa, ept_misconfig_pte[4]; static int ept_misconfig_ptenum; #endif static const char * vmexit_vmx_desc(uint32_t exit_reason) { if (exit_reason >= nitems(vmx_exit_reason_desc) || vmx_exit_reason_desc[exit_reason] == NULL) return ("Unknown"); return (vmx_exit_reason_desc[exit_reason]); } static int vmexit_vmx(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { fprintf(stderr, "vm exit[%d]\n", *pvcpu); fprintf(stderr, "\treason\t\tVMX\n"); fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip); fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length); fprintf(stderr, "\tstatus\t\t%d\n", vmexit->u.vmx.status); fprintf(stderr, "\texit_reason\t%u (%s)\n", vmexit->u.vmx.exit_reason, vmexit_vmx_desc(vmexit->u.vmx.exit_reason)); fprintf(stderr, "\tqualification\t0x%016lx\n", vmexit->u.vmx.exit_qualification); fprintf(stderr, "\tinst_type\t\t%d\n", vmexit->u.vmx.inst_type); fprintf(stderr, "\tinst_error\t\t%d\n", vmexit->u.vmx.inst_error); #ifdef DEBUG_EPT_MISCONFIG if (vmexit->u.vmx.exit_reason == EXIT_REASON_EPT_MISCONFIG) { vm_get_register(ctx, *pvcpu, VMCS_IDENT(VMCS_GUEST_PHYSICAL_ADDRESS), &ept_misconfig_gpa); vm_get_gpa_pmap(ctx, ept_misconfig_gpa, ept_misconfig_pte, &ept_misconfig_ptenum); fprintf(stderr, "\tEPT misconfiguration:\n"); fprintf(stderr, "\t\tGPA: %#lx\n", ept_misconfig_gpa); fprintf(stderr, "\t\tPTE(%d): %#lx %#lx %#lx %#lx\n", ept_misconfig_ptenum, ept_misconfig_pte[0], ept_misconfig_pte[1], ept_misconfig_pte[2], ept_misconfig_pte[3]); } #endif /* DEBUG_EPT_MISCONFIG */ return (VMEXIT_ABORT); } static int vmexit_svm(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { fprintf(stderr, "vm exit[%d]\n", *pvcpu); fprintf(stderr, "\treason\t\tSVM\n"); fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip); fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length); fprintf(stderr, "\texitcode\t%#lx\n", vmexit->u.svm.exitcode); fprintf(stderr, "\texitinfo1\t%#lx\n", vmexit->u.svm.exitinfo1); fprintf(stderr, "\texitinfo2\t%#lx\n", vmexit->u.svm.exitinfo2); return (VMEXIT_ABORT); } static int vmexit_bogus(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { assert(vmexit->inst_length == 0); stats.vmexit_bogus++; return (VMEXIT_CONTINUE); } static int vmexit_reqidle(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { assert(vmexit->inst_length == 0); stats.vmexit_reqidle++; return (VMEXIT_CONTINUE); } static int vmexit_hlt(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { stats.vmexit_hlt++; /* * Just continue execution with the next instruction. We use * the HLT VM exit as a way to be friendly with the host * scheduler. */ return (VMEXIT_CONTINUE); } static int vmexit_pause(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { stats.vmexit_pause++; return (VMEXIT_CONTINUE); } static int vmexit_mtrap(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { assert(vmexit->inst_length == 0); stats.vmexit_mtrap++; #ifdef BHYVE_SNAPSHOT checkpoint_cpu_suspend(*pvcpu); #endif if (gdb_port != 0) gdb_cpu_mtrap(*pvcpu); #ifdef BHYVE_SNAPSHOT checkpoint_cpu_resume(*pvcpu); #endif return (VMEXIT_CONTINUE); } static int vmexit_inst_emul(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { int err, i, cs_d; struct vie *vie; enum vm_cpu_mode mode; stats.vmexit_inst_emul++; vie = &vmexit->u.inst_emul.vie; if (!vie->decoded) { /* * Attempt to decode in userspace as a fallback. This allows * updating instruction decode in bhyve without rebooting the * kernel (rapid prototyping), albeit with much slower * emulation. */ vie_restart(vie); mode = vmexit->u.inst_emul.paging.cpu_mode; cs_d = vmexit->u.inst_emul.cs_d; (void)vmm_decode_instruction(mode, cs_d, vie); } err = emulate_mem(ctx, *pvcpu, vmexit->u.inst_emul.gpa, vie, &vmexit->u.inst_emul.paging); if (err) { if (err == ESRCH) { EPRINTLN("Unhandled memory access to 0x%lx\n", vmexit->u.inst_emul.gpa); } fprintf(stderr, "Failed to emulate instruction sequence [ "); for (i = 0; i < vie->num_valid; i++) fprintf(stderr, "%02x", vie->inst[i]); FPRINTLN(stderr, " ] at 0x%lx", vmexit->rip); return (VMEXIT_ABORT); } return (VMEXIT_CONTINUE); } static pthread_mutex_t resetcpu_mtx = PTHREAD_MUTEX_INITIALIZER; static pthread_cond_t resetcpu_cond = PTHREAD_COND_INITIALIZER; static int vmexit_suspend(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { enum vm_suspend_how how; how = vmexit->u.suspended.how; fbsdrun_deletecpu(ctx, *pvcpu); if (*pvcpu != BSP) { pthread_mutex_lock(&resetcpu_mtx); pthread_cond_signal(&resetcpu_cond); pthread_mutex_unlock(&resetcpu_mtx); pthread_exit(NULL); } pthread_mutex_lock(&resetcpu_mtx); while (!CPU_EMPTY(&cpumask)) { pthread_cond_wait(&resetcpu_cond, &resetcpu_mtx); } pthread_mutex_unlock(&resetcpu_mtx); switch (how) { case VM_SUSPEND_RESET: exit(0); case VM_SUSPEND_POWEROFF: + if (destroy_on_poweroff) + vm_destroy(ctx); exit(1); case VM_SUSPEND_HALT: exit(2); case VM_SUSPEND_TRIPLEFAULT: exit(3); default: fprintf(stderr, "vmexit_suspend: invalid reason %d\n", how); exit(100); } return (0); /* NOTREACHED */ } static int vmexit_debug(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { #ifdef BHYVE_SNAPSHOT checkpoint_cpu_suspend(*pvcpu); #endif if (gdb_port != 0) gdb_cpu_suspend(*pvcpu); #ifdef BHYVE_SNAPSHOT checkpoint_cpu_resume(*pvcpu); #endif return (VMEXIT_CONTINUE); } static int vmexit_breakpoint(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) { if (gdb_port == 0) { fprintf(stderr, "vm_loop: unexpected VMEXIT_DEBUG\n"); exit(4); } gdb_cpu_breakpoint(*pvcpu, vmexit); return (VMEXIT_CONTINUE); } static vmexit_handler_t handler[VM_EXITCODE_MAX] = { [VM_EXITCODE_INOUT] = vmexit_inout, [VM_EXITCODE_INOUT_STR] = vmexit_inout, [VM_EXITCODE_VMX] = vmexit_vmx, [VM_EXITCODE_SVM] = vmexit_svm, [VM_EXITCODE_BOGUS] = vmexit_bogus, [VM_EXITCODE_REQIDLE] = vmexit_reqidle, [VM_EXITCODE_RDMSR] = vmexit_rdmsr, [VM_EXITCODE_WRMSR] = vmexit_wrmsr, [VM_EXITCODE_MTRAP] = vmexit_mtrap, [VM_EXITCODE_INST_EMUL] = vmexit_inst_emul, [VM_EXITCODE_SPINUP_AP] = vmexit_spinup_ap, [VM_EXITCODE_SUSPENDED] = vmexit_suspend, [VM_EXITCODE_TASK_SWITCH] = vmexit_task_switch, [VM_EXITCODE_DEBUG] = vmexit_debug, [VM_EXITCODE_BPT] = vmexit_breakpoint, }; static void vm_loop(struct vmctx *ctx, int vcpu, uint64_t startrip) { int error, rc; enum vm_exitcode exitcode; cpuset_t active_cpus; if (vcpumap[vcpu] != NULL) { error = pthread_setaffinity_np(pthread_self(), sizeof(cpuset_t), vcpumap[vcpu]); assert(error == 0); } error = vm_active_cpus(ctx, &active_cpus); assert(CPU_ISSET(vcpu, &active_cpus)); error = vm_set_register(ctx, vcpu, VM_REG_GUEST_RIP, startrip); assert(error == 0); while (1) { error = vm_run(ctx, vcpu, &vmexit[vcpu]); if (error != 0) break; exitcode = vmexit[vcpu].exitcode; if (exitcode >= VM_EXITCODE_MAX || handler[exitcode] == NULL) { fprintf(stderr, "vm_loop: unexpected exitcode 0x%x\n", exitcode); exit(4); } rc = (*handler[exitcode])(ctx, &vmexit[vcpu], &vcpu); switch (rc) { case VMEXIT_CONTINUE: break; case VMEXIT_ABORT: abort(); default: exit(4); } } fprintf(stderr, "vm_run error %d, errno %d\n", error, errno); } static int num_vcpus_allowed(struct vmctx *ctx) { int tmp, error; error = vm_get_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, &tmp); /* * The guest is allowed to spinup more than one processor only if the * UNRESTRICTED_GUEST capability is available. */ if (error == 0) return (VM_MAXCPU); else return (1); } void fbsdrun_set_capabilities(struct vmctx *ctx, int cpu) { int err, tmp; if (fbsdrun_vmexit_on_hlt()) { err = vm_get_capability(ctx, cpu, VM_CAP_HALT_EXIT, &tmp); if (err < 0) { fprintf(stderr, "VM exit on HLT not supported\n"); exit(4); } vm_set_capability(ctx, cpu, VM_CAP_HALT_EXIT, 1); if (cpu == BSP) handler[VM_EXITCODE_HLT] = vmexit_hlt; } if (fbsdrun_vmexit_on_pause()) { /* * pause exit support required for this mode */ err = vm_get_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, &tmp); if (err < 0) { fprintf(stderr, "SMP mux requested, no pause support\n"); exit(4); } vm_set_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, 1); if (cpu == BSP) handler[VM_EXITCODE_PAUSE] = vmexit_pause; } if (x2apic_mode) err = vm_set_x2apic_state(ctx, cpu, X2APIC_ENABLED); else err = vm_set_x2apic_state(ctx, cpu, X2APIC_DISABLED); if (err) { fprintf(stderr, "Unable to set x2apic state (%d)\n", err); exit(4); } vm_set_capability(ctx, cpu, VM_CAP_ENABLE_INVPCID, 1); } static struct vmctx * do_open(const char *vmname) { struct vmctx *ctx; int error; bool reinit, romboot; #ifndef WITHOUT_CAPSICUM cap_rights_t rights; const cap_ioctl_t *cmds; size_t ncmds; #endif reinit = romboot = false; if (lpc_bootrom()) romboot = true; error = vm_create(vmname); if (error) { if (errno == EEXIST) { if (romboot) { reinit = true; } else { /* * The virtual machine has been setup by the * userspace bootloader. */ } } else { perror("vm_create"); exit(4); } } else { if (!romboot) { /* * If the virtual machine was just created then a * bootrom must be configured to boot it. */ fprintf(stderr, "virtual machine cannot be booted\n"); exit(4); } } ctx = vm_open(vmname); if (ctx == NULL) { perror("vm_open"); exit(4); } #ifndef WITHOUT_CAPSICUM cap_rights_init(&rights, CAP_IOCTL, CAP_MMAP_RW); if (caph_rights_limit(vm_get_device_fd(ctx), &rights) == -1) errx(EX_OSERR, "Unable to apply rights for sandbox"); vm_get_ioctls(&ncmds); cmds = vm_get_ioctls(NULL); if (cmds == NULL) errx(EX_OSERR, "out of memory"); if (caph_ioctls_limit(vm_get_device_fd(ctx), cmds, ncmds) == -1) errx(EX_OSERR, "Unable to apply rights for sandbox"); free((cap_ioctl_t *)cmds); #endif if (reinit) { error = vm_reinit(ctx); if (error) { perror("vm_reinit"); exit(4); } } error = vm_set_topology(ctx, sockets, cores, threads, maxcpus); if (error) errx(EX_OSERR, "vm_set_topology"); return (ctx); } void spinup_vcpu(struct vmctx *ctx, int vcpu) { int error; uint64_t rip; error = vm_get_register(ctx, vcpu, VM_REG_GUEST_RIP, &rip); assert(error == 0); fbsdrun_set_capabilities(ctx, vcpu); error = vm_set_capability(ctx, vcpu, VM_CAP_UNRESTRICTED_GUEST, 1); assert(error == 0); fbsdrun_addcpu(ctx, BSP, vcpu, rip); } int main(int argc, char *argv[]) { int c, error, dbg_port, err, bvmcons; int max_vcpus, mptgen, memflags; int rtc_localtime; bool gdb_stop; struct vmctx *ctx; uint64_t rip; size_t memsize; char *optstr; #ifdef BHYVE_SNAPSHOT char *restore_file; struct restore_state rstate; int vcpu; restore_file = NULL; #endif bvmcons = 0; progname = basename(argv[0]); dbg_port = 0; gdb_stop = false; guest_ncpus = 1; sockets = cores = threads = 1; maxcpus = 0; memsize = 256 * MB; mptgen = 1; rtc_localtime = 1; memflags = 0; #ifdef BHYVE_SNAPSHOT - optstr = "abehuwxACHIPSWYp:g:G:c:s:m:l:U:r:"; + optstr = "abehuwxACDHIPSWYp:g:G:c:s:m:l:U:r:"; #else - optstr = "abehuwxACHIPSWYp:g:G:c:s:m:l:U:"; + optstr = "abehuwxACDHIPSWYp:g:G:c:s:m:l:U:"; #endif while ((c = getopt(argc, argv, optstr)) != -1) { switch (c) { case 'a': x2apic_mode = 0; break; case 'A': acpi = 1; break; case 'b': bvmcons = 1; + break; + case 'D': + destroy_on_poweroff = 1; break; case 'p': if (pincpu_parse(optarg) != 0) { errx(EX_USAGE, "invalid vcpu pinning " "configuration '%s'", optarg); } break; case 'c': if (topology_parse(optarg) != 0) { errx(EX_USAGE, "invalid cpu topology " "'%s'", optarg); } break; case 'C': memflags |= VM_MEM_F_INCORE; break; case 'g': dbg_port = atoi(optarg); break; case 'G': if (optarg[0] == 'w') { gdb_stop = true; optarg++; } gdb_port = atoi(optarg); break; case 'l': if (strncmp(optarg, "help", strlen(optarg)) == 0) { lpc_print_supported_devices(); exit(0); } else if (lpc_device_parse(optarg) != 0) { errx(EX_USAGE, "invalid lpc device " "configuration '%s'", optarg); } break; #ifdef BHYVE_SNAPSHOT case 'r': restore_file = optarg; break; #endif case 's': if (strncmp(optarg, "help", strlen(optarg)) == 0) { pci_print_supported_devices(); exit(0); } else if (pci_parse_slot(optarg) != 0) exit(4); else break; case 'S': memflags |= VM_MEM_F_WIRED; break; case 'm': error = vm_parse_memsize(optarg, &memsize); if (error) errx(EX_USAGE, "invalid memsize '%s'", optarg); break; case 'H': guest_vmexit_on_hlt = 1; break; case 'I': /* * The "-I" option was used to add an ioapic to the * virtual machine. * * An ioapic is now provided unconditionally for each * virtual machine and this option is now deprecated. */ break; case 'P': guest_vmexit_on_pause = 1; break; case 'e': strictio = 1; break; case 'u': rtc_localtime = 0; break; case 'U': guest_uuid_str = optarg; break; case 'w': strictmsr = 0; break; case 'W': virtio_msix = 0; break; case 'x': x2apic_mode = 1; break; case 'Y': mptgen = 0; break; case 'h': usage(0); default: usage(1); } } argc -= optind; argv += optind; #ifdef BHYVE_SNAPSHOT if (argc > 1 || (argc == 0 && restore_file == NULL)) usage(1); if (restore_file != NULL) { error = load_restore_file(restore_file, &rstate); if (error) { fprintf(stderr, "Failed to read checkpoint info from " "file: '%s'.\n", restore_file); exit(1); } } if (argc == 1) { vmname = argv[0]; } else { vmname = lookup_vmname(&rstate); if (vmname == NULL) { fprintf(stderr, "Cannot find VM name in restore file. " "Please specify one.\n"); exit(1); } } #else if (argc != 1) usage(1); vmname = argv[0]; #endif ctx = do_open(vmname); #ifdef BHYVE_SNAPSHOT if (restore_file != NULL) { guest_ncpus = lookup_guest_ncpus(&rstate); memflags = lookup_memflags(&rstate); memsize = lookup_memsize(&rstate); } if (guest_ncpus < 1) { fprintf(stderr, "Invalid guest vCPUs (%d)\n", guest_ncpus); exit(1); } #endif max_vcpus = num_vcpus_allowed(ctx); if (guest_ncpus > max_vcpus) { fprintf(stderr, "%d vCPUs requested but only %d available\n", guest_ncpus, max_vcpus); exit(4); } fbsdrun_set_capabilities(ctx, BSP); vm_set_memflags(ctx, memflags); err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL); if (err) { fprintf(stderr, "Unable to setup memory (%d)\n", errno); exit(4); } error = init_msr(); if (error) { fprintf(stderr, "init_msr error %d", error); exit(4); } init_mem(); init_inout(); kernemu_dev_init(); init_bootrom(ctx); atkbdc_init(ctx); pci_irq_init(ctx); ioapic_init(ctx); rtc_init(ctx, rtc_localtime); sci_init(ctx); /* * Exit if a device emulation finds an error in its initilization */ if (init_pci(ctx) != 0) { perror("device emulation initialization error"); exit(4); } /* * Initialize after PCI, to allow a bootrom file to reserve the high * region. */ if (acpi) vmgenc_init(ctx); if (dbg_port != 0) init_dbgport(dbg_port); if (gdb_port != 0) init_gdb(ctx, gdb_port, gdb_stop); if (bvmcons) init_bvmcons(); if (lpc_bootrom()) { if (vm_set_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, 1)) { fprintf(stderr, "ROM boot failed: unrestricted guest " "capability not available\n"); exit(4); } error = vcpu_reset(ctx, BSP); assert(error == 0); } #ifdef BHYVE_SNAPSHOT if (restore_file != NULL) { fprintf(stdout, "Pausing pci devs...\r\n"); if (vm_pause_user_devs(ctx) != 0) { fprintf(stderr, "Failed to pause PCI device state.\n"); exit(1); } fprintf(stdout, "Restoring vm mem...\r\n"); if (restore_vm_mem(ctx, &rstate) != 0) { fprintf(stderr, "Failed to restore VM memory.\n"); exit(1); } fprintf(stdout, "Restoring pci devs...\r\n"); if (vm_restore_user_devs(ctx, &rstate) != 0) { fprintf(stderr, "Failed to restore PCI device state.\n"); exit(1); } fprintf(stdout, "Restoring kernel structs...\r\n"); if (vm_restore_kern_structs(ctx, &rstate) != 0) { fprintf(stderr, "Failed to restore kernel structs.\n"); exit(1); } fprintf(stdout, "Resuming pci devs...\r\n"); if (vm_resume_user_devs(ctx) != 0) { fprintf(stderr, "Failed to resume PCI device state.\n"); exit(1); } } #endif error = vm_get_register(ctx, BSP, VM_REG_GUEST_RIP, &rip); assert(error == 0); /* * build the guest tables, MP etc. */ if (mptgen) { error = mptable_build(ctx, guest_ncpus); if (error) { perror("error to build the guest tables"); exit(4); } } error = smbios_build(ctx); assert(error == 0); if (acpi) { error = acpi_build(ctx, guest_ncpus); assert(error == 0); } if (lpc_bootrom()) fwctl_init(); /* * Change the proc title to include the VM name. */ setproctitle("%s", vmname); #ifndef WITHOUT_CAPSICUM caph_cache_catpages(); if (caph_limit_stdout() == -1 || caph_limit_stderr() == -1) errx(EX_OSERR, "Unable to apply rights for sandbox"); if (caph_enter() == -1) errx(EX_OSERR, "cap_enter() failed"); #endif #ifdef BHYVE_SNAPSHOT if (restore_file != NULL) destroy_restore_state(&rstate); /* * checkpointing thread for communication with bhyvectl */ if (init_checkpoint_thread(ctx) < 0) printf("Failed to start checkpoint thread!\r\n"); if (restore_file != NULL) vm_restore_time(ctx); #endif /* * Add CPU 0 */ fbsdrun_addcpu(ctx, BSP, BSP, rip); #ifdef BHYVE_SNAPSHOT /* * If we restore a VM, start all vCPUs now (including APs), otherwise, * let the guest OS to spin them up later via vmexits. */ if (restore_file != NULL) { for (vcpu = 0; vcpu < guest_ncpus; vcpu++) { if (vcpu == BSP) continue; fprintf(stdout, "spinning up vcpu no %d...\r\n", vcpu); spinup_vcpu(ctx, vcpu); } } #endif /* * Head off to the main event dispatch loop */ mevent_dispatch(); exit(4); }