[pve-devel] [PATCH guest-common v5 1/1] add PCI/USB Mapping configs

[Dominik Csapak <d.csapak@proxmox.com>]

adds a config file for each type of resource (usb/pci) by using a 'map'
array propertystring for each node mapping

in each mapping we save the path(s) and some other information to detect
hardware changes (if possible) like the vendor/device id

both configs have custom header parser/formatter to omit the type (since
we only want one type per config here)

also each config has some helpers like find_on_current_node

the resulting config (e.g. for pci) would look like this:

---
some-pci-device
    description some device
    map node=node1,path=0000:00:01.0,id=1234:1234,iommugroup=4
    map node=node2,path=0000:01:01.0,id=1234:1234,iommugroup=5
    map node=node3,path=0000:02:01.0,id=1234:1234,iommugroup=6
---

some special notes:
* mdev is a per config entry, since it does not make sense to mix mdev
  and non-mdev devices
* to have multiple mappings for a node (for choosing a single one during
  vm startup), the config has a 'map' line for each such mapping

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
---
changes from v4:
* rename s/resource/mapping/i

 src/Makefile           |   3 +
 src/PVE/Mapping/PCI.pm | 226 +++++++++++++++++++++++++++++++++++++++++
 src/PVE/Mapping/USB.pm | 183 +++++++++++++++++++++++++++++++++
 3 files changed, 412 insertions(+)
 create mode 100644 src/PVE/Mapping/PCI.pm
 create mode 100644 src/PVE/Mapping/USB.pm

diff --git a/src/Makefile b/src/Makefile
index 57a360b..cbc40c1 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -14,6 +14,9 @@ install: PVE
 	install -m 0644 PVE/Replication.pm ${PERL5DIR}/PVE/
 	install -m 0644 PVE/StorageTunnel.pm ${PERL5DIR}/PVE/
 	install -m 0644 PVE/Tunnel.pm ${PERL5DIR}/PVE/
+	install -d ${PERL5DIR}/PVE/Mapping
+	install -m 0644 PVE/Mapping/PCI.pm ${PERL5DIR}/PVE/Mapping/
+	install -m 0644 PVE/Mapping/USB.pm ${PERL5DIR}/PVE/Mapping/
 	install -d ${PERL5DIR}/PVE/VZDump
 	install -m 0644 PVE/VZDump/Plugin.pm ${PERL5DIR}/PVE/VZDump/
 	install -m 0644 PVE/VZDump/Common.pm ${PERL5DIR}/PVE/VZDump/
diff --git a/src/PVE/Mapping/PCI.pm b/src/PVE/Mapping/PCI.pm
new file mode 100644
index 0000000..5b9d6b3
--- /dev/null
+++ b/src/PVE/Mapping/PCI.pm
@@ -0,0 +1,226 @@
+package PVE::Mapping::PCI;
+
+use strict;
+use warnings;
+
+use PVE::Cluster qw(cfs_register_file cfs_read_file cfs_lock_file cfs_write_file);
+use PVE::INotify;
+use PVE::JSONSchema qw(get_standard_option parse_property_string);
+use PVE::SectionConfig;
+use PVE::SysFSTools;
+
+use base qw(PVE::SectionConfig);
+
+my $FILENAME = 'mapping/pci.cfg';
+
+cfs_register_file($FILENAME,
+		  sub { __PACKAGE__->parse_config(@_); },
+		  sub { __PACKAGE__->write_config(@_); });
+
+
+# so we don't have to repeat the type every time
+sub parse_section_header {
+    my ($class, $line) = @_;
+
+    if ($line =~ m/^(\S+)\s*$/) {
+	my $id = $1;
+	my $errmsg = undef; # set if you want to skip whole section
+	eval { PVE::JSONSchema::pve_verify_configid($id) };
+	$errmsg = $@ if $@;
+	my $config = {}; # to return additional attributes
+	return ('pci', $id, $errmsg, $config);
+    }
+    return undef;
+}
+
+sub format_section_header {
+    my ($class, $type, $sectionId, $scfg, $done_hash) = @_;
+
+    return "$sectionId\n";
+}
+
+sub type {
+    return 'pci';
+}
+
+my $PCI_RE = "[a-f0-9]{4,}:[a-f0-9]{2}:[a-f0-9]{2}(?:\.[a-f0-9])?";
+
+my $map_fmt = {
+    node => get_standard_option('pve-node'),
+    id =>{
+	description => "The vendor and device ID that is expected. Used for".
+	" detecting hardware changes",
+	type => 'string',
+	pattern => qr/^[0-9A-Fa-f]{4}:[0-9A-Fa-f]{4}$/,
+    },
+    'subsystem-id' => {
+	description => "The subsystem vendor and device ID that is expected. Used".
+	" for detecting hardware changes.",
+	type => 'string',
+	pattern => qr/^[0-9A-Fa-f]{4}:[0-9A-Fa-f]{4}$/,
+	optional => 1,
+    },
+    path => {
+	description => "The path to the device. If the function is omitted, the whole device is"
+	." mapped. In that case use the attributes of the first device. You can give"
+	." multiple paths as a semicolon seperated list, the first available will then"
+	." be chosen on guest start.",
+	type => 'string',
+	pattern => "(?:${PCI_RE};)*${PCI_RE}",
+    },
+    iommugroup => {
+	type => 'integer',
+	description => "The IOMMU group in which the device is to be expected in.".
+	"Used for detecting hardware changes.",
+	optional => 1,
+    },
+    description => {
+	description => "Description of the node specific device.",
+	type => 'string',
+	optional => 1,
+	maxLength => 4096,
+    },
+};
+
+my $defaultData = {
+    propertyList => {
+	id => {
+	    type => 'string',
+	    description => "The ID of the logical PCI mapping.",
+	    format => 'pve-configid',
+	},
+	description => {
+	    description => "Description of the logical PCI device.",
+	    type => 'string',
+	    optional => 1,
+	    maxLength => 4096,
+	},
+	mdev => {
+	    type => 'boolean',
+	    optional => 1,
+	},
+	map => {
+	    type => 'array',
+	    description => 'A list of maps for the cluster nodes.',
+	    optional => 1,
+	    items => {
+		type => 'string',
+		format => $map_fmt,
+	    },
+	},
+    },
+};
+
+sub private {
+    return $defaultData;
+}
+
+sub options {
+    return {
+	description => { optional => 1 },
+	mdev => { optional => 1 },
+	map => {},
+    };
+}
+
+# checks if the given config is valid for the current node
+sub assert_valid {
+    my ($name, $cfg) = @_;
+
+    my @paths = split(';', $cfg->{path} // '');
+
+    my $idx = 0;
+    for my $path (@paths) {
+
+	my $multifunction = 0;
+	if ($path !~ m/\.[a-f0-9]/i) {
+	    # whole device, add .0 (must exist)
+	    $path = "$path.0";
+	    $multifunction = 1;
+	}
+
+	my $info = PVE::SysFSTools::pci_device_info($path, 1);
+	die "pci device '$path' not found\n" if !defined($info);
+
+	my $correct_props = {
+	    id => "$info->{vendor}:$info->{device}",
+	    iommugroup => $info->{iommugroup},
+	};
+
+	if (defined($info->{'subsystem_vendor'}) && defined($info->{'subsystem_device'})) {
+	    $correct_props->{'subsystem-id'} = "$info->{'subsystem_vendor'}:$info->{'subsystem_device'}";
+	}
+
+	for my $prop (sort keys %$correct_props) {
+	    next if $prop eq 'iommugroup' && $idx > 0; # check iommu only on the first device
+
+	    next if !defined($correct_props->{$prop}) && !defined($cfg->{$prop});
+	    die "no '$prop' for device '$path'\n"
+		if defined($correct_props->{$prop}) && !defined($cfg->{$prop});
+	    die "'$prop' configured but should not be\n"
+		if !defined($correct_props->{$prop}) && defined($cfg->{$prop});
+
+	    my $correct_prop = $correct_props->{$prop};
+	    $correct_prop =~ s/0x//g;
+	    my $configured_prop = $cfg->{$prop};
+	    $configured_prop =~ s/0x//g;
+
+	    die "'$prop' does not match for '$name' ($correct_prop != $configured_prop)\n"
+		if $correct_prop ne $configured_prop;
+	}
+	$idx++;
+    }
+
+    return 1;
+};
+
+sub config {
+    return cfs_read_file($FILENAME);
+}
+
+sub lock_pci_config {
+    my ($code, $errmsg) = @_;
+
+    cfs_lock_file($FILENAME, undef, $code);
+    if (my $err = $@) {
+	$errmsg ? die "$errmsg: $err" : die $err;
+    }
+}
+
+sub write_pci_config {
+    my ($cfg) = @_;
+
+    cfs_write_file($FILENAME, $cfg);
+}
+
+sub find_on_current_node {
+    my ($id) = @_;
+
+    my $cfg = PVE::Mapping::PCI::config();
+    my $node = PVE::INotify::nodename();
+
+    # ignore errors
+    return get_node_mapping($cfg, $id, $node);
+}
+
+sub get_node_mapping {
+    my ($cfg, $id, $nodename) = @_;
+
+    return undef if !defined($cfg->{ids}->{$id});
+
+    my $res = [];
+    for my $map ($cfg->{ids}->{$id}->{map}->@*) {
+	my $entry = eval { parse_property_string($map_fmt, $map) };
+	warn $@ if $@;
+	if ($entry && $entry->{node} eq $nodename) {
+	    push $res->@*, $entry;
+	}
+    }
+
+    return $res;
+}
+
+PVE::Mapping::PCI->register();
+PVE::Mapping::PCI->init();
+
+1;
diff --git a/src/PVE/Mapping/USB.pm b/src/PVE/Mapping/USB.pm
new file mode 100644
index 0000000..483e92b
--- /dev/null
+++ b/src/PVE/Mapping/USB.pm
@@ -0,0 +1,183 @@
+package PVE::Mapping::USB;
+
+use strict;
+use warnings;
+
+use PVE::Cluster qw(cfs_register_file cfs_read_file cfs_lock_file cfs_write_file);
+use PVE::INotify;
+use PVE::JSONSchema qw(get_standard_option parse_property_string);
+use PVE::SectionConfig;
+use PVE::SysFSTools;
+
+use base qw(PVE::SectionConfig);
+
+my $FILENAME = 'mapping/usb.cfg';
+
+cfs_register_file($FILENAME,
+		  sub { __PACKAGE__->parse_config(@_); },
+		  sub { __PACKAGE__->write_config(@_); });
+
+
+# so we don't have to repeat the type every time
+sub parse_section_header {
+    my ($class, $line) = @_;
+
+    if ($line =~ m/^(\S+)\s*$/) {
+	my $id = $1;
+	my $errmsg = undef; # set if you want to skip whole section
+	eval { PVE::JSONSchema::pve_verify_configid($id) };
+	$errmsg = $@ if $@;
+	my $config = {}; # to return additional attributes
+	return ('usb', $id, $errmsg, $config);
+    }
+    return undef;
+}
+
+sub format_section_header {
+    my ($class, $type, $sectionId, $scfg, $done_hash) = @_;
+
+    return "$sectionId\n";
+}
+
+sub type {
+    return 'usb';
+}
+
+my $map_fmt = {
+    node => get_standard_option('pve-node'),
+    'id' => {
+	description => "The vendor and device ID that is expected. If a USB path".
+	" is given, it is only used for detecting hardware changes",
+	type => 'string',
+	pattern => qr/^[0-9A-Fa-f]{4}:[0-9A-Fa-f]{4}$/,
+    },
+    path => {
+	description => "The path to the usb device.",
+	type => 'string',
+	optional => 1,
+	pattern => qr/^(\d+)\-(\d+(\.\d+)*)$/,
+    },
+    description => {
+	description => "Description of the node specific device.",
+	type => 'string',
+	optional => 1,
+	maxLength => 4096,
+    },
+};
+
+my $defaultData = {
+    propertyList => {
+	id => {
+	    type => 'string',
+	    description => "The ID of the logical PCI mapping.",
+	    format => 'pve-configid',
+	},
+	description => {
+	    description => "Description of the logical PCI device.",
+	    type => 'string',
+	    optional => 1,
+	    maxLength => 4096,
+	},
+	map => {
+	    type => 'array',
+	    description => 'A list of maps for the cluster nodes.',
+	    items => {
+		type => 'string',
+		format => $map_fmt,
+	    },
+	},
+    },
+};
+sub private {
+    return $defaultData;
+}
+
+sub options {
+    return {
+	description => { optional => 1 },
+	map => {},
+    };
+}
+
+# checks if the given device is valid for the current node
+sub assert_valid {
+    my ($name, $cfg) = @_;
+
+    my $id = $cfg->{id};
+
+    my $usb_list = PVE::SysFSTools::scan_usb();
+
+    my $info;
+    if (my $path = $cfg->{path}) {
+	for my $dev (@$usb_list) {
+	    next if !$dev->{usbpath} || !$dev->{busnum};
+	    my $usbpath = "$dev->{busnum}-$dev->{usbpath}";
+	    next if $usbpath ne $path;
+	    $info = $dev;
+	}
+	die "usb device '$path' not found\n" if !defined($info);
+
+	my $realId = "$info->{vendid}:$info->{prodid}";
+	die "'id' does not match for '$name' ($realId != $id)\n"
+	    if $realId ne $id;
+    } else {
+	for my $dev (@$usb_list) {
+	    my $realId = "$dev->{vendid}:$dev->{prodid}";
+	    next if $realId ne $id;
+	    $info = $dev;
+	}
+	die "usb device '$id' not found\n" if !defined($info);
+    }
+
+    return 1;
+};
+
+sub config {
+    return cfs_read_file($FILENAME);
+}
+
+sub lock_usb_config {
+    my ($code, $errmsg) = @_;
+
+    cfs_lock_file($FILENAME, undef, $code);
+    if (my $err = $@) {
+	$errmsg ? die "$errmsg: $err" : die $err;
+    }
+}
+
+sub write_usb_config {
+    my ($cfg) = @_;
+
+    cfs_write_file($FILENAME, $cfg);
+}
+
+sub find_on_current_node {
+    my ($id) = @_;
+
+    my $cfg = config();
+    my $node = PVE::INotify::nodename();
+
+    return get_node_mapping($cfg, $id, $node);
+}
+
+sub get_node_mapping {
+    my ($cfg, $id, $nodename) = @_;
+
+    return undef if !defined($cfg->{ids}->{$id});
+
+    my $res = [];
+    for my $map ($cfg->{ids}->{$id}->{map}->@*) {
+	my $entry = eval { parse_property_string($map_fmt, $map) };
+	warn $@ if $@;
+	if ($entry && $entry->{node} eq $nodename) {
+	    push $res->@*, $entry;
+	}
+    }
+
+    return $res;
+}
+
+PVE::Mapping::USB->register();
+PVE::Mapping::USB->init();
+
+1;
-- 
2.30.2