qemu-server/PVE/QemuServer/PCI.pm
Dominik Csapak 48ada6982f pci: mdev: adapt to NVIDIA's modern interface with kernel >= 6.8
Since kernel 6.8, NVIDIAs vGPU driver does not use the generic mdev
interface anymore, since they relied on a feature there which is not
available anymore. IIUC the kernel [0] recommends drivers to implement
their own device specific features since putting all in the generic one
does not make sense.

They now have an 'nvidia' folder in the device sysfs path, which
contains the files `creatable_vgpu_types`/`current_vgpu_type` to
control the virtual functions model, and then the whole virtual function
has to be passed through (although without resetting and changing to the
vfio-pci driver).

This patch implements changes so that from a config perspective, it
still is an mediated device, and we map the functionality iff the device
has no mediated devices but the new NVIDIAs sysfsapi and the model name
is 'nvidia-<..>'

It behaves a bit different than mdevs and normal pci passthrough, as we
have to choose the correct device immediately since it's bound to the
pciid, but we must not bind the device to vfio-pci as the NVIDIA driver
implements this functionality itself.

When cleaning up, we iterate over all reserved devices (since for a
mapping we can't know at this point which was chosen besides looking at
the reservations) and reset the vgpu model to '0', so it frees up the
reservation from NVIDIAs side. (We also do that in a loop, since it's
not always immediately ready after QEMU closes)

A general problem (but that was previously also the case) is that a
showcmd (for a not running guest) reserves the pciids, which might block
an execution of a different real vm. This is now a bit more problematic
as we (temporarily) set the vgpu type then.

0: https://docs.kernel.org/driver-api/vfio-pci-device-specific-driver-acceptance.html

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Tested-by: Christoph Heiss <c.heiss@proxmox.com>
Reviewed-by: Christoph Heiss <c.heiss@proxmox.com>
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
2024-10-24 18:43:52 +02:00

868 lines
27 KiB
Perl

package PVE::QemuServer::PCI;
use warnings;
use strict;
use PVE::JSONSchema;
use PVE::Mapping::PCI;
use PVE::SysFSTools;
use PVE::Tools;
use base 'Exporter';
our @EXPORT_OK = qw(
print_pci_addr
print_pcie_addr
print_pcie_root_port
parse_hostpci
);
our $MAX_HOSTPCI_DEVICES = 16;
my $PCIRE = qr/(?:[a-f0-9]{4,}:)?[a-f0-9]{2}:[a-f0-9]{2}(?:\.[a-f0-9])?/;
my $hostpci_fmt = {
host => {
default_key => 1,
optional => 1,
type => 'string',
pattern => qr/$PCIRE(;$PCIRE)*/,
format_description => 'HOSTPCIID[;HOSTPCIID2...]',
description => <<EODESCR,
Host PCI device pass through. The PCI ID of a host's PCI device or a list
of PCI virtual functions of the host. HOSTPCIID syntax is:
'bus:dev.func' (hexadecimal numbers)
You can use the 'lspci' command to list existing PCI devices.
Either this or the 'mapping' key must be set.
EODESCR
},
mapping => {
optional => 1,
type => 'string',
format_description => 'mapping-id',
format => 'pve-configid',
description => "The ID of a cluster wide mapping. Either this or the default-key 'host'"
." must be set.",
},
rombar => {
type => 'boolean',
description => "Specify whether or not the device's ROM will be visible in the"
." guest's memory map.",
optional => 1,
default => 1,
},
romfile => {
type => 'string',
pattern => '[^,;]+',
format_description => 'string',
description => "Custom pci device rom filename (must be located in /usr/share/kvm/).",
optional => 1,
},
pcie => {
type => 'boolean',
description => "Choose the PCI-express bus (needs the 'q35' machine model).",
optional => 1,
default => 0,
},
'x-vga' => {
type => 'boolean',
description => "Enable vfio-vga device support.",
optional => 1,
default => 0,
},
'legacy-igd' => {
type => 'boolean',
description => "Pass this device in legacy IGD mode, making it the primary and exclusive"
." graphics device in the VM. Requires 'pc-i440fx' machine type and VGA set to 'none'.",
optional => 1,
default => 0,
},
'mdev' => {
type => 'string',
format_description => 'string',
pattern => '[^/\.:]+',
optional => 1,
description => <<EODESCR
The type of mediated device to use.
An instance of this type will be created on startup of the VM and
will be cleaned up when the VM stops.
EODESCR
},
'vendor-id' => {
type => 'string',
pattern => qr/^0x[0-9a-fA-F]{4}$/,
format_description => 'hex id',
optional => 1,
description => "Override PCI vendor ID visible to guest"
},
'device-id' => {
type => 'string',
pattern => qr/^0x[0-9a-fA-F]{4}$/,
format_description => 'hex id',
optional => 1,
description => "Override PCI device ID visible to guest"
},
'sub-vendor-id' => {
type => 'string',
pattern => qr/^0x[0-9a-fA-F]{4}$/,
format_description => 'hex id',
optional => 1,
description => "Override PCI subsystem vendor ID visible to guest"
},
'sub-device-id' => {
type => 'string',
pattern => qr/^0x[0-9a-fA-F]{4}$/,
format_description => 'hex id',
optional => 1,
description => "Override PCI subsystem device ID visible to guest"
}
};
PVE::JSONSchema::register_format('pve-qm-hostpci', $hostpci_fmt);
our $hostpcidesc = {
optional => 1,
type => 'string', format => 'pve-qm-hostpci',
description => "Map host PCI devices into guest.",
verbose_description => <<EODESCR,
Map host PCI devices into guest.
NOTE: This option allows direct access to host hardware. So it is no longer
possible to migrate such machines - use with special care.
CAUTION: Experimental! User reported problems with this option.
EODESCR
};
PVE::JSONSchema::register_standard_option("pve-qm-hostpci", $hostpcidesc);
my $pci_addr_map;
sub get_pci_addr_map {
$pci_addr_map = {
piix3 => { bus => 0, addr => 1, conflict_ok => qw(ehci) },
ehci => { bus => 0, addr => 1, conflict_ok => qw(piix3) }, # instead of piix3 on arm
vga => { bus => 0, addr => 2, conflict_ok => qw(legacy-igd) },
'legacy-igd' => { bus => 0, addr => 2, conflict_ok => qw(vga) }, # legacy-igd requires vga=none
balloon0 => { bus => 0, addr => 3 },
watchdog => { bus => 0, addr => 4 },
scsihw0 => { bus => 0, addr => 5, conflict_ok => qw(pci.3) },
'pci.3' => { bus => 0, addr => 5, conflict_ok => qw(scsihw0) }, # also used for virtio-scsi-single bridge
scsihw1 => { bus => 0, addr => 6 },
ahci0 => { bus => 0, addr => 7 },
qga0 => { bus => 0, addr => 8 },
spice => { bus => 0, addr => 9 },
virtio0 => { bus => 0, addr => 10 },
virtio1 => { bus => 0, addr => 11 },
virtio2 => { bus => 0, addr => 12 },
virtio3 => { bus => 0, addr => 13 },
virtio4 => { bus => 0, addr => 14 },
virtio5 => { bus => 0, addr => 15 },
hostpci0 => { bus => 0, addr => 16 },
hostpci1 => { bus => 0, addr => 17 },
net0 => { bus => 0, addr => 18 },
net1 => { bus => 0, addr => 19 },
net2 => { bus => 0, addr => 20 },
net3 => { bus => 0, addr => 21 },
net4 => { bus => 0, addr => 22 },
net5 => { bus => 0, addr => 23 },
vga1 => { bus => 0, addr => 24 },
vga2 => { bus => 0, addr => 25 },
vga3 => { bus => 0, addr => 26 },
hostpci2 => { bus => 0, addr => 27 },
hostpci3 => { bus => 0, addr => 28 },
#addr29 : usb-host (pve-usb.cfg)
'pci.1' => { bus => 0, addr => 30 },
'pci.2' => { bus => 0, addr => 31 },
'net6' => { bus => 1, addr => 1 },
'net7' => { bus => 1, addr => 2 },
'net8' => { bus => 1, addr => 3 },
'net9' => { bus => 1, addr => 4 },
'net10' => { bus => 1, addr => 5 },
'net11' => { bus => 1, addr => 6 },
'net12' => { bus => 1, addr => 7 },
'net13' => { bus => 1, addr => 8 },
'net14' => { bus => 1, addr => 9 },
'net15' => { bus => 1, addr => 10 },
'net16' => { bus => 1, addr => 11 },
'net17' => { bus => 1, addr => 12 },
'net18' => { bus => 1, addr => 13 },
'net19' => { bus => 1, addr => 14 },
'net20' => { bus => 1, addr => 15 },
'net21' => { bus => 1, addr => 16 },
'net22' => { bus => 1, addr => 17 },
'net23' => { bus => 1, addr => 18 },
'net24' => { bus => 1, addr => 19 },
'net25' => { bus => 1, addr => 20 },
'net26' => { bus => 1, addr => 21 },
'net27' => { bus => 1, addr => 22 },
'net28' => { bus => 1, addr => 23 },
'net29' => { bus => 1, addr => 24 },
'net30' => { bus => 1, addr => 25 },
'net31' => { bus => 1, addr => 26 },
'xhci' => { bus => 1, addr => 27 },
'pci.4' => { bus => 1, addr => 28 },
'rng0' => { bus => 1, addr => 29 },
'pci.2-igd' => { bus => 1, addr => 30 }, # replaces pci.2 in case a legacy IGD device is passed through
'virtio6' => { bus => 2, addr => 1 },
'virtio7' => { bus => 2, addr => 2 },
'virtio8' => { bus => 2, addr => 3 },
'virtio9' => { bus => 2, addr => 4 },
'virtio10' => { bus => 2, addr => 5 },
'virtio11' => { bus => 2, addr => 6 },
'virtio12' => { bus => 2, addr => 7 },
'virtio13' => { bus => 2, addr => 8 },
'virtio14' => { bus => 2, addr => 9 },
'virtio15' => { bus => 2, addr => 10 },
'ivshmem' => { bus => 2, addr => 11 },
'audio0' => { bus => 2, addr => 12 },
hostpci4 => { bus => 2, addr => 13 },
hostpci5 => { bus => 2, addr => 14 },
hostpci6 => { bus => 2, addr => 15 },
hostpci7 => { bus => 2, addr => 16 },
hostpci8 => { bus => 2, addr => 17 },
hostpci9 => { bus => 2, addr => 18 },
hostpci10 => { bus => 2, addr => 19 },
hostpci11 => { bus => 2, addr => 20 },
hostpci12 => { bus => 2, addr => 21 },
hostpci13 => { bus => 2, addr => 22 },
hostpci14 => { bus => 2, addr => 23 },
hostpci15 => { bus => 2, addr => 24 },
'virtioscsi0' => { bus => 3, addr => 1 },
'virtioscsi1' => { bus => 3, addr => 2 },
'virtioscsi2' => { bus => 3, addr => 3 },
'virtioscsi3' => { bus => 3, addr => 4 },
'virtioscsi4' => { bus => 3, addr => 5 },
'virtioscsi5' => { bus => 3, addr => 6 },
'virtioscsi6' => { bus => 3, addr => 7 },
'virtioscsi7' => { bus => 3, addr => 8 },
'virtioscsi8' => { bus => 3, addr => 9 },
'virtioscsi9' => { bus => 3, addr => 10 },
'virtioscsi10' => { bus => 3, addr => 11 },
'virtioscsi11' => { bus => 3, addr => 12 },
'virtioscsi12' => { bus => 3, addr => 13 },
'virtioscsi13' => { bus => 3, addr => 14 },
'virtioscsi14' => { bus => 3, addr => 15 },
'virtioscsi15' => { bus => 3, addr => 16 },
'virtioscsi16' => { bus => 3, addr => 17 },
'virtioscsi17' => { bus => 3, addr => 18 },
'virtioscsi18' => { bus => 3, addr => 19 },
'virtioscsi19' => { bus => 3, addr => 20 },
'virtioscsi20' => { bus => 3, addr => 21 },
'virtioscsi21' => { bus => 3, addr => 22 },
'virtioscsi22' => { bus => 3, addr => 23 },
'virtioscsi23' => { bus => 3, addr => 24 },
'virtioscsi24' => { bus => 3, addr => 25 },
'virtioscsi25' => { bus => 3, addr => 26 },
'virtioscsi26' => { bus => 3, addr => 27 },
'virtioscsi27' => { bus => 3, addr => 28 },
'virtioscsi28' => { bus => 3, addr => 29 },
'virtioscsi29' => { bus => 3, addr => 30 },
'virtioscsi30' => { bus => 3, addr => 31 },
'scsihw2' => { bus => 4, addr => 1 },
'scsihw3' => { bus => 4, addr => 2 },
'scsihw4' => { bus => 4, addr => 3 },
} if !defined($pci_addr_map);
return $pci_addr_map;
}
sub generate_mdev_uuid {
my ($vmid, $index) = @_;
return sprintf("%08d-0000-0000-0000-%012d", $index, $vmid);
}
my $get_addr_mapping_from_id = sub {
my ($map, $id) = @_;
my $d = $map->{$id};
return if !defined($d) || !defined($d->{bus}) || !defined($d->{addr});
return { bus => $d->{bus}, addr => sprintf("0x%x", $d->{addr}) };
};
sub print_pci_addr {
my ($id, $bridges, $arch, $machine) = @_;
my $res = '';
# using same bus slots on all HW, so we need to check special cases here:
my $busname = 'pci';
if ($arch eq 'aarch64' && $machine =~ /^virt/) {
die "aarch64/virt cannot use IDE devices\n" if $id =~ /^ide/;
$busname = 'pcie';
}
my $map = get_pci_addr_map();
if (my $d = $get_addr_mapping_from_id->($map, $id)) {
$res = ",bus=$busname.$d->{bus},addr=$d->{addr}";
$bridges->{$d->{bus}} = 1 if $bridges;
}
return $res;
}
my $pcie_addr_map;
sub get_pcie_addr_map {
$pcie_addr_map = {
vga => { bus => 'pcie.0', addr => 1 },
hostpci0 => { bus => "ich9-pcie-port-1", addr => 0 },
hostpci1 => { bus => "ich9-pcie-port-2", addr => 0 },
hostpci2 => { bus => "ich9-pcie-port-3", addr => 0 },
hostpci3 => { bus => "ich9-pcie-port-4", addr => 0 },
hostpci4 => { bus => "ich9-pcie-port-5", addr => 0 },
hostpci5 => { bus => "ich9-pcie-port-6", addr => 0 },
hostpci6 => { bus => "ich9-pcie-port-7", addr => 0 },
hostpci7 => { bus => "ich9-pcie-port-8", addr => 0 },
hostpci8 => { bus => "ich9-pcie-port-9", addr => 0 },
hostpci9 => { bus => "ich9-pcie-port-10", addr => 0 },
hostpci10 => { bus => "ich9-pcie-port-11", addr => 0 },
hostpci11 => { bus => "ich9-pcie-port-12", addr => 0 },
hostpci12 => { bus => "ich9-pcie-port-13", addr => 0 },
hostpci13 => { bus => "ich9-pcie-port-14", addr => 0 },
hostpci14 => { bus => "ich9-pcie-port-15", addr => 0 },
hostpci15 => { bus => "ich9-pcie-port-16", addr => 0 },
# win7 is picky about pcie assignments
hostpci0bus0 => { bus => "pcie.0", addr => 16 },
hostpci1bus0 => { bus => "pcie.0", addr => 17 },
hostpci2bus0 => { bus => "pcie.0", addr => 18 },
hostpci3bus0 => { bus => "pcie.0", addr => 19 },
ivshmem => { bus => 'pcie.0', addr => 20 },
hostpci4bus0 => { bus => "pcie.0", addr => 9 },
hostpci5bus0 => { bus => "pcie.0", addr => 10 },
hostpci6bus0 => { bus => "pcie.0", addr => 11 },
hostpci7bus0 => { bus => "pcie.0", addr => 12 },
hostpci8bus0 => { bus => "pcie.0", addr => 13 },
hostpci9bus0 => { bus => "pcie.0", addr => 14 },
hostpci10bus0 => { bus => "pcie.0", addr => 15 },
hostpci11bus0 => { bus => "pcie.0", addr => 21 },
hostpci12bus0 => { bus => "pcie.0", addr => 22 },
hostpci13bus0 => { bus => "pcie.0", addr => 23 },
hostpci14bus0 => { bus => "pcie.0", addr => 24 },
hostpci15bus0 => { bus => "pcie.0", addr => 25 },
} if !defined($pcie_addr_map);
return $pcie_addr_map;
}
sub print_pcie_addr {
my ($id) = @_;
my $res = '';
my $map = get_pcie_addr_map($id);
if (my $d = $get_addr_mapping_from_id->($map, $id)) {
$res = ",bus=$d->{bus},addr=$d->{addr}";
}
return $res;
}
# Generates the device strings for additional pcie root ports. The first 4 pcie
# root ports are defined in the pve-q35*.cfg files.
sub print_pcie_root_port {
my ($i) = @_;
my $res = '';
my $root_port_addresses = {
4 => "10.0",
5 => "10.1",
6 => "10.2",
7 => "10.3",
8 => "10.4",
9 => "10.5",
10 => "10.6",
11 => "10.7",
12 => "11.0",
13 => "11.1",
14 => "11.2",
15 => "11.3",
};
if (defined($root_port_addresses->{$i})) {
my $id = $i + 1;
$res = "pcie-root-port,id=ich9-pcie-port-${id}";
$res .= ",addr=$root_port_addresses->{$i}";
$res .= ",x-speed=16,x-width=32,multifunction=on,bus=pcie.0";
$res .= ",port=${id},chassis=${id}";
}
return $res;
}
# returns the parsed pci config but parses the 'host' part into
# a list if lists into the 'id' property like this:
#
# {
# mdev => 1,
# rombar => ...
# ...
# ids => [
# # this contains a list of alternative devices,
# [
# # which are itself lists of ids for one multifunction device
# {
# id => "0000:00:00.0",
# vendor => "...",
# },
# {
# id => "0000:00:00.1",
# vendor => "...",
# },
# ],
# [
# ...
# ],
# ...
# ],
# }
sub parse_hostpci {
my ($value) = @_;
return if !$value;
my $res = PVE::JSONSchema::parse_property_string($hostpci_fmt, $value);
my $alternatives = [];
my $host = delete $res->{host};
my $mapping = delete $res->{mapping};
die "Cannot set both 'host' and 'mapping'.\n" if defined($host) && defined($mapping);
if ($mapping) {
# we have no ordinary pci id, must be a mapping
my $devices = PVE::Mapping::PCI::find_on_current_node($mapping);
die "PCI device mapping not found for '$mapping'\n" if !$devices || !scalar($devices->@*);
for my $device ($devices->@*) {
eval { PVE::Mapping::PCI::assert_valid($mapping, $device) };
die "PCI device mapping invalid (hardware probably changed): $@\n" if $@;
push $alternatives->@*, [split(/;/, $device->{path})];
}
} elsif ($host) {
push $alternatives->@*, [split(/;/, $host)];
} else {
die "Either 'host' or 'mapping' must be set.\n";
}
$res->{ids} = [];
for my $alternative ($alternatives->@*) {
my $ids = [];
foreach my $id ($alternative->@*) {
my $devs = PVE::SysFSTools::lspci($id);
die "no PCI device found for '$id'\n" if !scalar($devs->@*);
push $ids->@*, @$devs;
}
if (scalar($ids->@*) > 1) {
$res->{'has-multifunction'} = 1;
die "cannot use mediated device with multifunction device\n" if $res->{mdev} || $res->{nvidia};
} elsif ($res->{mdev}) {
if ($ids->[0]->{nvidia} && $res->{mdev} =~ m/^nvidia-(\d+)$/) {
$res->{nvidia} = $1;
delete $res->{mdev};
}
}
push $res->{ids}->@*, $ids;
}
return $res;
}
# parses all hostpci devices from a config and does some sanity checks
# returns a hash like this:
# {
# hostpci0 => {
# # hash from parse_hostpci function
# },
# hostpci1 => { ... },
# ...
# }
sub parse_hostpci_devices {
my ($conf) = @_;
my $q35 = PVE::QemuServer::Machine::machine_type_is_q35($conf);
my $legacy_igd = 0;
my $parsed_devices = {};
for (my $i = 0; $i < $MAX_HOSTPCI_DEVICES; $i++) {
my $id = "hostpci$i";
my $d = parse_hostpci($conf->{$id});
next if !$d;
# check syntax
die "q35 machine model is not enabled" if !$q35 && $d->{pcie};
if ($d->{'legacy-igd'}) {
die "only one device can be assigned in legacy-igd mode\n"
if $legacy_igd;
$legacy_igd = 1;
die "legacy IGD assignment requires VGA mode to be 'none'\n"
if !defined($conf->{'vga'}) || $conf->{'vga'} ne 'none';
die "legacy IGD assignment requires rombar to be enabled\n"
if defined($d->{rombar}) && !$d->{rombar};
die "legacy IGD assignment is not compatible with x-vga\n"
if $d->{'x-vga'};
die "legacy IGD assignment is not compatible with mdev\n"
if $d->{mdev} || $d->{nvidia};
die "legacy IGD assignment is not compatible with q35\n"
if $q35;
die "legacy IGD assignment is not compatible with multifunction devices\n"
if $d->{'has-multifunction'};
die "legacy IGD assignment is not compatible with alternate devices\n"
if scalar($d->{ids}->@*) > 1;
# check first device for valid id
die "legacy IGD assignment only works for devices on host bus 00:02.0\n"
if $d->{ids}->[0]->[0]->{id} !~ m/02\.0$/;
}
$parsed_devices->{$id} = $d;
}
return $parsed_devices;
}
# set vgpu type of a vf of an nvidia gpu with kernel 6.8 or newer
my sub create_nvidia_device {
my ($id, $model) = @_;
$id = '0000:'.$id if $id !~ m/^0000:/;
my $creation = "/sys/bus/pci/devices/$id/nvidia/current_vgpu_type";
die "no nvidia sysfs api for '$id'\n" if ! -f $creation;
my $current = PVE::Tools::file_read_firstline($creation);
if ($current ne "0") {
return 1 if $current eq $model;
# reset vgpu type so we can see all available and set the real device
die "unable to reset vgpu type for '$id'\n" if !PVE::SysFSTools::file_write($creation, "0");
}
my $types = PVE::SysFSTools::get_mdev_types($id);
my $selected;
for my $type_definition ($types->@*) {
next if $type_definition->{type} ne "nvidia-$model";
$selected = $type_definition;
}
if (!defined($selected) || $selected->{available} < 1) {
die "vgpu type '$model' not available for '$id'\n";
}
if (!PVE::SysFSTools::file_write($creation, $model)) {
die "could not set vgpu type to '$model' for '$id'\n";
}
return 1;
}
# takes the hash returned by parse_hostpci_devices and for all non mdev gpus,
# selects one of the given alternatives by trying to reserve it
#
# mdev devices must be chosen later when we actually allocate it, but we
# flatten the inner list since there can only be one device per alternative anyway
my sub choose_hostpci_devices {
my ($devices, $vmid) = @_;
# if the vm is running, we must be in 'showcmd', so don't actually reserve or create anything
my $is_running = PVE::QemuServer::Helpers::vm_running_locally($vmid) ? 1 : 0;
my $used = {};
my $add_used_device = sub {
my ($devices) = @_;
for my $used_device ($devices->@*) {
my $used_id = $used_device->{id};
die "device '$used_id' assigned more than once\n" if $used->{$used_id};
$used->{$used_id} = 1;
}
};
for (my $i = 0; $i < $MAX_HOSTPCI_DEVICES; $i++) {
my $device = $devices->{"hostpci$i"};
next if !$device;
if ($device->{mdev} && !$device->{nvidia}) {
$device->{ids} = [ map { $_->[0] } $device->{ids}->@* ];
next;
}
if (scalar($device->{ids}->@* == 1)) {
# we only have one alternative, use that
$device->{ids} = $device->{ids}->[0];
$add_used_device->($device->{ids});
if ($device->{nvidia} && !$is_running) {
reserve_pci_usage($device->{ids}->[0]->{id}, $vmid, 10, undef);
create_nvidia_device($device->{ids}->[0]->{id}, $device->{nvidia});
}
next;
}
my $found = 0;
for my $alternative ($device->{ids}->@*) {
my $ids = [map { $_->{id} } @$alternative];
next if grep { defined($used->{$_}) } @$ids; # already used
if (!$is_running) {
eval { reserve_pci_usage($ids, $vmid, 10, undef) };
next if $@;
}
if ($device->{nvidia} && !$is_running) {
eval { create_nvidia_device($ids->[0], $device->{nvidia}) };
if (my $err = $@) {
warn $err;
remove_pci_reservation($vmid, $ids);
next;
}
}
# found one that is not used or reserved
$add_used_device->($alternative);
$device->{ids} = $alternative;
$found = 1;
last;
}
die "could not find a free device for 'hostpci$i'\n" if !$found;
}
return $devices;
}
sub print_hostpci_devices {
my ($vmid, $conf, $devices, $vga, $winversion, $bridges, $arch, $machine_type, $bootorder) = @_;
my $kvm_off = 0;
my $gpu_passthrough = 0;
my $legacy_igd = 0;
my $pciaddr;
my $pci_devices = choose_hostpci_devices(parse_hostpci_devices($conf), $vmid);
for (my $i = 0; $i < $MAX_HOSTPCI_DEVICES; $i++) {
my $id = "hostpci$i";
my $d = $pci_devices->{$id};
next if !$d;
$legacy_igd = 1 if $d->{'legacy-igd'};
if (my $pcie = $d->{pcie}) {
# win7 wants to have the pcie devices directly on the pcie bus
# instead of in the root port
if ($winversion == 7) {
$pciaddr = print_pcie_addr("${id}bus0");
} else {
# add more root ports if needed, 4 are present by default
# by pve-q35 cfgs, rest added here on demand.
if ($i > 3) {
push @$devices, '-device', print_pcie_root_port($i);
}
$pciaddr = print_pcie_addr($id);
}
} else {
my $pci_name = $d->{'legacy-igd'} ? 'legacy-igd' : $id;
$pciaddr = print_pci_addr($pci_name, $bridges, $arch, $machine_type);
}
my $num_devices = scalar($d->{ids}->@*);
my $multifunction = $num_devices > 1 && !$d->{mdev};
my $xvga = '';
if ($d->{'x-vga'}) {
$xvga = ',x-vga=on' if !($conf->{bios} && $conf->{bios} eq 'ovmf');
$kvm_off = 1;
$vga->{type} = 'none' if !defined($conf->{vga});
$gpu_passthrough = 1;
}
my $sysfspath;
if ($d->{mdev}) {
my $uuid = generate_mdev_uuid($vmid, $i);
$sysfspath = "/sys/bus/mdev/devices/$uuid";
}
for (my $j = 0; $j < $num_devices; $j++) {
my $pcidevice = $d->{ids}->[$j];
my $devicestr = "vfio-pci";
if ($sysfspath) {
$devicestr .= ",sysfsdev=$sysfspath";
} else {
$devicestr .= ",host=$pcidevice->{id}";
}
my $mf_addr = $multifunction ? ".$j" : '';
$devicestr .= ",id=${id}${mf_addr}${pciaddr}${mf_addr}";
if ($j == 0) {
$devicestr .= ',rombar=0' if defined($d->{rombar}) && !$d->{rombar};
$devicestr .= "$xvga";
$devicestr .= ",multifunction=on" if $multifunction;
$devicestr .= ",romfile=/usr/share/kvm/$d->{romfile}" if $d->{romfile};
$devicestr .= ",bootindex=$bootorder->{$id}" if $bootorder->{$id};
for my $option (qw(vendor-id device-id sub-vendor-id sub-device-id)) {
$devicestr .= ",x-pci-$option=$d->{$option}" if $d->{$option};
}
}
push @$devices, '-device', $devicestr;
last if $d->{mdev};
}
}
return ($kvm_off, $gpu_passthrough, $legacy_igd, $pci_devices);
}
sub prepare_pci_device {
my ($vmid, $pciid, $index, $device) = @_;
my $info = PVE::SysFSTools::pci_device_info("$pciid");
die "cannot prepare PCI pass-through, IOMMU not present\n" if !PVE::SysFSTools::check_iommu_support();
die "no pci device info for device '$pciid'\n" if !$info;
if ($device->{nvidia}) {
# nothing to do
} elsif (my $mdev = $device->{mdev}) {
my $uuid = generate_mdev_uuid($vmid, $index);
PVE::SysFSTools::pci_create_mdev_device($pciid, $uuid, $mdev);
} else {
die "can't unbind/bind PCI group to VFIO '$pciid'\n"
if !PVE::SysFSTools::pci_dev_group_bind_to_vfio($pciid);
die "can't reset PCI device '$pciid'\n"
if $info->{has_fl_reset} && !PVE::SysFSTools::pci_dev_reset($info);
}
return $info;
}
my $RUNDIR = '/run/qemu-server';
my $PCIID_RESERVATION_FILE = "${RUNDIR}/pci-id-reservations";
my $PCIID_RESERVATION_LOCK = "${PCIID_RESERVATION_FILE}.lock";
# a list of PCI ID to VMID reservations, the validity is protected against leakage by either a PID,
# for succesfully started VM processes, or a expiration time for the initial time window between
# reservation and actual VM process start-up.
my $parse_pci_reservation_unlocked = sub {
my $pciids = {};
if (my $fh = IO::File->new($PCIID_RESERVATION_FILE, "r")) {
while (my $line = <$fh>) {
if ($line =~ m/^($PCIRE)\s(\d+)\s(time|pid)\:(\d+)$/) {
$pciids->{$1} = {
vmid => $2,
"$3" => $4,
};
}
}
}
return $pciids;
};
my $write_pci_reservation_unlocked = sub {
my ($reservations) = @_;
my $data = "";
for my $pci_id (sort keys $reservations->%*) {
my ($vmid, $pid, $time) = $reservations->{$pci_id}->@{'vmid', 'pid', 'time'};
if (defined($pid)) {
$data .= "$pci_id $vmid pid:$pid\n";
} else {
$data .= "$pci_id $vmid time:$time\n";
}
}
PVE::Tools::file_set_contents($PCIID_RESERVATION_FILE, $data);
};
# removes all PCI device reservations held by the `vmid`
sub remove_pci_reservation {
my ($vmid, $pciids) = @_;
PVE::Tools::lock_file($PCIID_RESERVATION_LOCK, 2, sub {
my $reservation_list = $parse_pci_reservation_unlocked->();
for my $id (keys %$reservation_list) {
next if defined($pciids) && !grep { $_ eq $id } $pciids->@*;
my $reservation = $reservation_list->{$id};
next if $reservation->{vmid} != $vmid;
delete $reservation_list->{$id};
}
$write_pci_reservation_unlocked->($reservation_list);
});
die $@ if $@;
}
# return all currently reserved ids from the given vmid
sub get_reservations {
my ($vmid) = @_;
my $reservations = $parse_pci_reservation_unlocked->();
my $list = [];
for my $pci_id (sort keys $reservations->%*) {
push $list->@*, $pci_id if $reservations->{$pci_id}->{vmid} == $vmid;
}
return $list;
}
sub reserve_pci_usage {
my ($requested_ids, $vmid, $timeout, $pid) = @_;
$requested_ids = [ $requested_ids ] if !ref($requested_ids);
return if !scalar(@$requested_ids); # do nothing for empty list
PVE::Tools::lock_file($PCIID_RESERVATION_LOCK, 5, sub {
my $reservation_list = $parse_pci_reservation_unlocked->();
my $ctime = time();
for my $id ($requested_ids->@*) {
my $reservation = $reservation_list->{$id};
if ($reservation && $reservation->{vmid} != $vmid) {
# check time based reservation
die "PCI device '$id' is currently reserved for use by VMID '$reservation->{vmid}'\n"
if defined($reservation->{time}) && $reservation->{time} > $ctime;
if (my $reserved_pid = $reservation->{pid}) {
# check running vm
my $running_pid = PVE::QemuServer::Helpers::vm_running_locally($reservation->{vmid});
if (defined($running_pid) && $running_pid == $reserved_pid) {
die "PCI device '$id' already in use by VMID '$reservation->{vmid}'\n";
} else {
warn "leftover PCI reservation found for $id, lets take it...\n";
}
}
} elsif ($reservation) {
# already reserved by the same vmid
if (my $reserved_time = $reservation->{time}) {
if (defined($timeout)) {
# use the longer timeout
my $old_timeout = $reservation->{time} - 5 - $ctime;
$timeout = $old_timeout if $old_timeout > $timeout;
}
} elsif (my $reserved_pid = $reservation->{pid}) {
my $running_pid = PVE::QemuServer::Helpers::vm_running_locally($reservation->{vmid});
if (defined($running_pid) && $running_pid == $reservation->{pid}) {
if (defined($pid)) {
die "PCI device '$id' already in use by running VMID '$reservation->{vmid}'\n";
} elsif (defined($timeout)) {
# ignore timeout reservation for running vms, can happen with e.g.
# qm showcmd
return;
}
}
}
}
$reservation_list->{$id} = { vmid => $vmid };
if (defined($pid)) { # VM started up, we can reserve now with the actual PID
$reservation_list->{$id}->{pid} = $pid;
} elsif (defined($timeout)) { # tempoaray reserve as we don't now the PID yet
$reservation_list->{$id}->{time} = $ctime + $timeout + 5;
}
}
$write_pci_reservation_unlocked->($reservation_list);
});
die $@ if $@;
}
1;