X-Git-Url: https://git.proxmox.com/?a=blobdiff_plain;f=PVE%2FQemuServer.pm;h=c3b682cdd99c8199b16771dc32115dabb9627850;hb=dea4b04c1cf77414dea50c54f2f3f89b7ff71388;hp=6445508286f1ea8046ef99cf14f8ab2f03fc9eef;hpb=818c3b8d91ad65b004042699277aa2f6543527ca;p=qemu-server.git diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm index 6445508..c3b682c 100644 --- a/PVE/QemuServer.pm +++ b/PVE/QemuServer.pm @@ -26,7 +26,7 @@ use Time::HiRes qw(gettimeofday); use URI::Escape; use UUID; -use PVE::Cluster qw(cfs_register_file cfs_read_file cfs_write_file cfs_lock_file); +use PVE::Cluster qw(cfs_register_file cfs_read_file cfs_write_file); use PVE::DataCenterConfig; use PVE::Exception qw(raise raise_param_exc); use PVE::GuestHelpers qw(safe_string_ne safe_num_ne safe_boolean_ne); @@ -37,7 +37,7 @@ use PVE::RPCEnvironment; use PVE::Storage; use PVE::SysFSTools; use PVE::Systemd; -use PVE::Tools qw(run_command lock_file lock_file_full file_read_firstline file_get_contents dir_glob_foreach get_host_arch $IPV6RE); +use PVE::Tools qw(run_command file_read_firstline file_get_contents dir_glob_foreach get_host_arch $IPV6RE); use PVE::QMPClient; use PVE::QemuConfig; @@ -48,7 +48,7 @@ use PVE::QemuServer::Drive qw(is_valid_drivename drive_is_cloudinit drive_is_cdr use PVE::QemuServer::Machine; use PVE::QemuServer::Memory; use PVE::QemuServer::Monitor qw(mon_cmd); -use PVE::QemuServer::PCI qw(print_pci_addr print_pcie_addr print_pcie_root_port); +use PVE::QemuServer::PCI qw(print_pci_addr print_pcie_addr print_pcie_root_port parse_hostpci); use PVE::QemuServer::USB qw(parse_usb_device); my $have_sdn; @@ -768,7 +768,6 @@ while (my ($k, $v) = each %$confdesc) { my $MAX_USB_DEVICES = 5; my $MAX_NETS = 32; -my $MAX_HOSTPCI_DEVICES = 16; my $MAX_SERIAL_PORTS = 4; my $MAX_PARALLEL_PORTS = 3; my $MAX_NUMA = 8; @@ -884,6 +883,12 @@ my $net_fmt = { description => 'Whether this interface should be disconnected (like pulling the plug).', optional => 1, }, + mtu => { + type => 'integer', + minimum => 1, maximum => 65520, + description => "Force MTU, for VirtIO only. Set to '1' to use the bridge MTU", + optional => 1, + }, }; my $netdesc = { @@ -1005,76 +1010,6 @@ my $usbdesc = { }; PVE::JSONSchema::register_standard_option("pve-qm-usb", $usbdesc); -my $PCIRE = qr/([a-f0-9]{4}:)?[a-f0-9]{2}:[a-f0-9]{2}(?:\.[a-f0-9])?/; -my $hostpci_fmt = { - host => { - default_key => 1, - type => 'string', - pattern => qr/$PCIRE(;$PCIRE)*/, - format_description => 'HOSTPCIID[;HOSTPCIID2...]', - description => < { - 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, - }, - 'mdev' => { - type => 'string', - format_description => 'string', - pattern => '[^/\.:]+', - optional => 1, - description => < 1, - type => 'string', format => 'pve-qm-hostpci', - description => "Map host PCI devices into guest.", - verbose_description => < 1, type => 'string', @@ -1113,8 +1048,8 @@ for (my $i = 0; $i < $MAX_SERIAL_PORTS; $i++) { $confdesc->{"serial$i"} = $serialdesc; } -for (my $i = 0; $i < $MAX_HOSTPCI_DEVICES; $i++) { - $confdesc->{"hostpci$i"} = $hostpcidesc; +for (my $i = 0; $i < $PVE::QemuServer::PCI::MAX_HOSTPCI_DEVICES; $i++) { + $confdesc->{"hostpci$i"} = $PVE::QemuServer::PCI::hostpcidesc; } for my $key (keys %{$PVE::QemuServer::Drive::drivedesc_hash}) { @@ -1593,6 +1528,22 @@ sub print_netdevice_full { } $tmpstr .= ",bootindex=$net->{bootindex}" if $net->{bootindex} ; + if (my $mtu = $net->{mtu}) { + if ($net->{model} eq 'virtio' && $net->{bridge}) { + my $bridge_mtu = PVE::Network::read_bridge_mtu($net->{bridge}); + if ($mtu == 1) { + $mtu = $bridge_mtu; + } elsif ($mtu < 576) { + die "netdev $netid: MTU '$mtu' is smaller than the IP minimum MTU '576'\n"; + } elsif ($mtu > $bridge_mtu) { + die "netdev $netid: MTU '$mtu' is bigger than the bridge MTU '$bridge_mtu'\n"; + } + $tmpstr .= ",host_mtu=$mtu"; + } else { + warn "WARN: netdev $netid: ignoring MTU '$mtu', not using VirtIO or no bridge configured.\n"; + } + } + if ($use_old_bios_files) { my $romfile; if ($device eq 'virtio-net-pci') { @@ -1734,23 +1685,6 @@ sub parse_numa { return $res; } -sub parse_hostpci { - my ($value) = @_; - - return undef if !$value; - - my $res = PVE::JSONSchema::parse_property_string($hostpci_fmt, $value); - - my @idlist = split(/;/, $res->{host}); - delete $res->{host}; - foreach my $id (@idlist) { - my $devs = PVE::SysFSTools::lspci($id); - die "no PCI device found for '$id'\n" if !scalar(@$devs); - push @{$res->{pciid}}, @$devs; - } - return $res; -} - # netX: e1000=XX:XX:XX:XX:XX:XX,bridge=vmbr0,rate= sub parse_net { my ($data) = @_; @@ -2751,6 +2685,32 @@ sub conf_has_audio { }; } +sub audio_devs { + my ($audio, $audiopciaddr, $machine_version) = @_; + + my $devs = []; + + my $id = $audio->{dev_id}; + my $audiodev = ""; + if (min_version($machine_version, 4, 2)) { + $audiodev = ",audiodev=$audio->{backend_id}"; + } + + if ($audio->{dev} eq 'AC97') { + push @$devs, '-device', "AC97,id=${id}${audiopciaddr}$audiodev"; + } elsif ($audio->{dev} =~ /intel\-hda$/) { + push @$devs, '-device', "$audio->{dev},id=${id}${audiopciaddr}"; + push @$devs, '-device', "hda-micro,id=${id}-codec0,bus=${id}.0,cad=0$audiodev"; + push @$devs, '-device', "hda-duplex,id=${id}-codec1,bus=${id}.0,cad=1$audiodev"; + } else { + die "unkown audio device '$audio->{dev}', implement me!"; + } + + push @$devs, '-audiodev', "$audio->{backend},id=$audio->{backend_id}"; + + return $devs; +} + sub vga_conf_has_spice { my ($vga) = @_; @@ -3061,7 +3021,8 @@ sub config_to_command { die "uefi base image '$ovmf_code' not found\n" if ! -f $ovmf_code; my ($path, $format); - if (my $d = parse_drive('efidisk0', $conf->{efidisk0})) { + if (my $efidisk = $conf->{efidisk0}) { + my $d = parse_drive('efidisk0', $efidisk); my ($storeid, $volname) = PVE::Storage::parse_volume_id($d->{file}, 1); $format = $d->{format}; if ($storeid) { @@ -3140,77 +3101,9 @@ sub config_to_command { push @$devices, '-device', $kbd if defined($kbd); } - my $kvm_off = 0; - my $gpu_passthrough; - - # host pci devices - for (my $i = 0; $i < $MAX_HOSTPCI_DEVICES; $i++) { - my $id = "hostpci$i"; - my $d = parse_hostpci($conf->{$id}); - next if !$d; - - if (my $pcie = $d->{pcie}) { - die "q35 machine model is not enabled" if !$q35; - # 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 { - $pciaddr = print_pci_addr($id, $bridges, $arch, $machine_type); - } - - 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 $pcidevices = $d->{pciid}; - my $multifunction = 1 if @$pcidevices > 1; - - my $sysfspath; - if ($d->{mdev} && scalar(@$pcidevices) == 1) { - my $pci_id = $pcidevices->[0]->{id}; - my $uuid = PVE::SysFSTools::generate_mdev_uuid($vmid, $i); - $sysfspath = "/sys/bus/pci/devices/$pci_id/$uuid"; - } elsif ($d->{mdev}) { - warn "ignoring mediated device '$id' with multifunction device\n"; - } - - my $j=0; - foreach my $pcidevice (@$pcidevices) { - 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 $d->{romfile}; - } - - push @$devices, '-device', $devicestr; - $j++; - } - } + # host pci device passthrough + my ($kvm_off, $gpu_passthrough, $legacy_igd) = PVE::QemuServer::PCI::print_hostpci_devices( + $conf, $devices, $winversion, $q35, $bridges, $arch, $machine_type); # usb devices my $usb_dev_features = {}; @@ -3250,22 +3143,10 @@ sub config_to_command { } } - if (my $audio = conf_has_audio($conf)) { - + if (min_version($machine_version, 4, 0) && (my $audio = conf_has_audio($conf))) { my $audiopciaddr = print_pci_addr("audio0", $bridges, $arch, $machine_type); - - my $id = $audio->{dev_id}; - if ($audio->{dev} eq 'AC97') { - push @$devices, '-device', "AC97,id=${id}${audiopciaddr}"; - } elsif ($audio->{dev} =~ /intel\-hda$/) { - push @$devices, '-device', "$audio->{dev},id=${id}${audiopciaddr}"; - push @$devices, '-device', "hda-micro,id=${id}-codec0,bus=${id}.0,cad=0"; - push @$devices, '-device', "hda-duplex,id=${id}-codec1,bus=${id}.0,cad=1"; - } else { - die "unkown audio device '$audio->{dev}', implement me!"; - } - - push @$devices, '-audiodev', "$audio->{backend},id=$audio->{backend_id}"; + my $audio_devs = audio_devs($audio, $audiopciaddr, $machine_version); + push @$devices, @$audio_devs; } my $sockets = 1; @@ -3376,20 +3257,16 @@ sub config_to_command { my $rng = parse_rng($conf->{rng0}) if $conf->{rng0}; if ($rng && &$version_guard(4, 1, 2)) { + check_rng_source($rng->{source}); + my $max_bytes = $rng->{max_bytes} // $rng_fmt->{max_bytes}->{default}; my $period = $rng->{period} // $rng_fmt->{period}->{default}; - my $limiter_str = ""; if ($max_bytes) { $limiter_str = ",max-bytes=$max_bytes,period=$period"; } - # mostly relevant for /dev/hwrng, but doesn't hurt to check others too - die "cannot create VirtIO RNG device: source file '$rng->{source}' doesn't exist\n" - if ! -e $rng->{source}; - my $rng_addr = print_pci_addr("rng0", $bridges, $arch, $machine_type); - push @$devices, '-object', "rng-random,filename=$rng->{source},id=rng0"; push @$devices, '-device', "virtio-rng-pci,rng=rng0$limiter_str$rng_addr"; } @@ -3581,7 +3458,13 @@ sub config_to_command { for my $k (sort {$b cmp $a} keys %$bridges) { next if $q35 && $k < 4; # q35.cfg already includes bridges up to 3 - $pciaddr = print_pci_addr("pci.$k", undef, $arch, $machine_type); + + my $k_name = $k; + if ($k == 2 && $legacy_igd) { + $k_name = "$k-igd"; + } + $pciaddr = print_pci_addr("pci.$k_name", undef, $arch, $machine_type); + my $devstr = "pci-bridge,id=pci.$k,chassis_nr=$k$pciaddr"; if ($q35) { # add after -readconfig pve-q35.cfg @@ -3623,6 +3506,24 @@ sub config_to_command { return wantarray ? ($cmd, $vollist, $spice_port) : $cmd; } +sub check_rng_source { + my ($source) = @_; + + # mostly relevant for /dev/hwrng, but doesn't hurt to check others too + die "cannot create VirtIO RNG device: source file '$source' doesn't exist\n" + if ! -e $source; + + my $rng_current = '/sys/devices/virtual/misc/hw_random/rng_current'; + if ($source eq '/dev/hwrng' && file_read_firstline($rng_current) eq 'none') { + # Needs to abort, otherwise QEMU crashes on first rng access. + # Note that rng_current cannot be changed to 'none' manually, so + # once the VM is past this point, it is no longer an issue. + die "Cannot start VM with passed-through RNG device: '/dev/hwrng'" + . " exists, but '$rng_current' is set to 'none'. Ensure that" + . " a compatible hardware-RNG is attached to the host.\n"; + } +} + sub spice_port { my ($vmid) = @_; @@ -4036,6 +3937,14 @@ sub qemu_netdevadd { my $netdev = print_netdev_full($vmid, $conf, $arch, $device, $deviceid, 1); my %options = split(/[=,]/, $netdev); + if (defined(my $vhost = $options{vhost})) { + $options{vhost} = JSON::boolean(PVE::JSONSchema::parse_boolean($vhost)); + } + + if (defined(my $queues = $options{queues})) { + $options{queues} = $queues + 0; + } + mon_cmd($vmid, "netdev_add", %options); return 1; } @@ -4292,40 +4201,46 @@ sub foreach_volid { my $volhash = {}; my $test_volid = sub { - my ($volid, $is_cdrom, $replicate, $shared, $snapname, $size) = @_; + my ($key, $drive, $snapname) = @_; + my $volid = $drive->{file}; return if !$volid; $volhash->{$volid}->{cdrom} //= 1; - $volhash->{$volid}->{cdrom} = 0 if !$is_cdrom; + $volhash->{$volid}->{cdrom} = 0 if !drive_is_cdrom($drive); + my $replicate = $drive->{replicate} // 1; $volhash->{$volid}->{replicate} //= 0; $volhash->{$volid}->{replicate} = 1 if $replicate; $volhash->{$volid}->{shared} //= 0; - $volhash->{$volid}->{shared} = 1 if $shared; + $volhash->{$volid}->{shared} = 1 if $drive->{shared}; $volhash->{$volid}->{referenced_in_config} //= 0; $volhash->{$volid}->{referenced_in_config} = 1 if !defined($snapname); $volhash->{$volid}->{referenced_in_snapshot}->{$snapname} = 1 if defined($snapname); - $volhash->{$volid}->{size} = $size if $size; + + my $size = $drive->{size}; + $volhash->{$volid}->{size} //= $size if $size; + + $volhash->{$volid}->{is_vmstate} //= 0; + $volhash->{$volid}->{is_vmstate} = 1 if $key eq 'vmstate'; + + $volhash->{$volid}->{is_unused} //= 0; + $volhash->{$volid}->{is_unused} = 1 if $key =~ /^unused\d+$/; }; - PVE::QemuConfig->foreach_volume($conf, sub { - my ($ds, $drive) = @_; - $test_volid->($drive->{file}, drive_is_cdrom($drive), $drive->{replicate} // 1, $drive->{shared}, undef, $drive->{size}); - }); + my $include_opts = { + extra_keys => ['vmstate'], + include_unused => 1, + }; + PVE::QemuConfig->foreach_volume_full($conf, $include_opts, $test_volid); foreach my $snapname (keys %{$conf->{snapshots}}) { my $snap = $conf->{snapshots}->{$snapname}; - $test_volid->($snap->{vmstate}, 0, 1, $snapname); - $volhash->{$snap->{vmstate}}->{is_vmstate} = 1 if $snap->{vmstate}; - PVE::QemuConfig->foreach_volume($snap, sub { - my ($ds, $drive) = @_; - $test_volid->($drive->{file}, drive_is_cdrom($drive), $drive->{replicate} // 1, $drive->{shared}, $snapname); - }); + PVE::QemuConfig->foreach_volume_full($snap, $include_opts, $test_volid, $snapname); } foreach my $volid (keys %$volhash) { @@ -5027,7 +4942,7 @@ sub vm_start_nolock { } # host pci devices - for (my $i = 0; $i < $MAX_HOSTPCI_DEVICES; $i++) { + for (my $i = 0; $i < $PVE::QemuServer::PCI::MAX_HOSTPCI_DEVICES; $i++) { my $d = parse_hostpci($conf->{"hostpci$i"}); next if !$d; my $pcidevices = $d->{pciid}; @@ -5368,7 +5283,11 @@ sub _do_vm_stop { return; } } else { - if ($force) { + if (!check_running($vmid, $nocheck)) { + warn "Unexpected: VM shutdown command failed, but VM not running anymore..\n"; + return; + } + if ($force) { warn "VM quit/powerdown failed - terminating now with SIGTERM\n"; kill 15, $pid; } else { @@ -5616,28 +5535,12 @@ sub tar_restore_cleanup { sub restore_file_archive { my ($archive, $vmid, $user, $opts) = @_; - my $format = $opts->{format}; - my $comp; - - if ($archive =~ m/\.tgz$/ || $archive =~ m/\.tar\.gz$/) { - $format = 'tar' if !$format; - $comp = 'gzip'; - } elsif ($archive =~ m/\.tar$/) { - $format = 'tar' if !$format; - } elsif ($archive =~ m/.tar.lzo$/) { - $format = 'tar' if !$format; - $comp = 'lzop'; - } elsif ($archive =~ m/\.vma$/) { - $format = 'vma' if !$format; - } elsif ($archive =~ m/\.vma\.gz$/) { - $format = 'vma' if !$format; - $comp = 'gzip'; - } elsif ($archive =~ m/\.vma\.lzo$/) { - $format = 'vma' if !$format; - $comp = 'lzop'; - } else { - $format = 'vma' if !$format; # default - } + return restore_vma_archive($archive, $vmid, $user, $opts) + if $archive eq '-'; + + my $info = PVE::Storage::archive_info($archive); + my $format = $opts->{format} // $info->{format}; + my $comp = $info->{compression}; # try to detect archive format if ($format eq 'tar') { @@ -5728,12 +5631,13 @@ my $parse_backup_hints = sub { my $drive = parse_drive($virtdev, $2); if (drive_is_cloudinit($drive)) { my ($storeid, $volname) = PVE::Storage::parse_volume_id($drive->{file}); + $storeid = $options->{storage} if defined ($options->{storage}); my $scfg = PVE::Storage::storage_config($storecfg, $storeid); my $format = qemu_img_format($scfg, $volname); # has 'raw' fallback $virtdev_hash->{$virtdev} = { format => $format, - storeid => $options->{storage} // $storeid, + storeid => $storeid, size => PVE::QemuServer::Cloudinit::CLOUDINIT_DISK_SIZE, is_cloudinit => 1, }; @@ -5900,7 +5804,7 @@ sub update_disk_config { my ($vmid, $conf, $volid_hash) = @_; my $changes; - my $prefix = "VM $vmid:"; + my $prefix = "VM $vmid"; # used and unused disks my $referenced = {}; @@ -5928,11 +5832,11 @@ sub update_disk_config { return if drive_is_cdrom($drive); return if !$volid_hash->{$volid}; - my ($updated, $old_size, $new_size) = PVE::QemuServer::Drive::update_disksize($drive, $volid_hash); + my ($updated, $msg) = PVE::QemuServer::Drive::update_disksize($drive, $volid_hash->{$volid}->{size}); if (defined($updated)) { $changes = 1; $conf->{$opt} = print_drive($updated); - print "$prefix size of disk '$volid' ($opt) updated from $old_size to $new_size\n"; + print "$prefix ($opt): $msg\n"; } }); @@ -6033,6 +5937,8 @@ sub restore_proxmox_backup_archive { my $fingerprint = $scfg->{fingerprint}; my $repo = "$username\@$server:$datastore"; + + # This is only used for `pbs-restore`! my $password = PVE::Storage::PBSPlugin::pbs_get_password($scfg, $storeid); local $ENV{PBS_PASSWORD} = $password; local $ENV{PBS_FINGERPRINT} = $fingerprint if defined($fingerprint); @@ -6135,6 +6041,7 @@ sub restore_proxmox_backup_archive { my $path = PVE::Storage::path($storecfg, $volid); + # This is the ONLY user of the PBS_ env vars set on top of this function! my $pbs_restore_cmd = [ '/usr/bin/pbs-restore', '--repository', $repo, @@ -6144,6 +6051,8 @@ sub restore_proxmox_backup_archive { '--verbose', ]; + push @$pbs_restore_cmd, '--format', $d->{format} if $d->{format}; + if (PVE::Storage::volume_has_feature($storecfg, 'sparseinit', $volid)) { push @$pbs_restore_cmd, '--skip-zero'; } @@ -6224,14 +6133,9 @@ sub restore_vma_archive { } if ($comp) { - my $cmd; - if ($comp eq 'gzip') { - $cmd = ['zcat', $readfrom]; - } elsif ($comp eq 'lzop') { - $cmd = ['lzop', '-d', '-c', $readfrom]; - } else { - die "unknown compression method '$comp'\n"; - } + my $info = PVE::Storage::decompressor_info('vma', $comp); + my $cmd = $info->{decompressor}; + push @$cmd, $readfrom; $add_pipe->($cmd); } @@ -7178,7 +7082,7 @@ sub complete_backup_archives { my $res = []; foreach my $id (keys %$data) { foreach my $item (@{$data->{$id}}) { - next if $item->{format} !~ m/^vma\.(gz|lzo)$/; + next if $item->{format} !~ m/^vma\.(${\PVE::Storage::Plugin::COMPRESSOR_RE})$/; push @$res, $item->{volid} if defined($item->{volid}); } }