udev/net: also support [SR-IOV] section in .link files

The same section is already supported by .network files. But such
low-level inteerface setting should be done by udevd, instead of
networkd. Let's also support the same semantics by .link files.

Prompted by https://github.com/systemd/systemd/issues/20474#issuecomment-901901360.
This commit is contained in:
Yu Watanabe 2022-01-14 17:24:49 +09:00
parent b4fd34d81c
commit bd29dfef8b
6 changed files with 174 additions and 89 deletions

View file

@ -949,6 +949,95 @@
</variablelist>
</refsect1>
<refsect1 id='sr-iov'>
<title>[SR-IOV] Section Options</title>
<para>The [SR-IOV] section accepts the following keys. Specify several [SR-IOV] sections to
configure several SR-IOVs. SR-IOV provides the ability to partition a single physical PCI resource
into virtual PCI functions which can then be injected into a VM. In the case of network VFs, SR-IOV
improves north-south network performance (that is, traffic with endpoints outside the host machine)
by allowing traffic to bypass the host machines network stack.</para>
<variablelist class='network-directives'>
<varlistentry>
<term><varname>VirtualFunction=</varname></term>
<listitem>
<para>Specifies a Virtual Function (VF), lightweight PCIe function designed solely to move
data in and out. Takes an integer in the range 0…2147483646. This option is compulsory.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>VLANId=</varname></term>
<listitem>
<para>Specifies VLAN ID of the virtual function. Takes an integer in the range 1…4095.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>QualityOfService=</varname></term>
<listitem>
<para>Specifies quality of service of the virtual function. Takes an integer in the range
1…4294967294.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>VLANProtocol=</varname></term>
<listitem>
<para>Specifies VLAN protocol of the virtual function. Takes <literal>802.1Q</literal> or
<literal>802.1ad</literal>.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MACSpoofCheck=</varname></term>
<listitem>
<para>Takes a boolean. Controls the MAC spoof checking. When unset, the kernel's default will
be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>QueryReceiveSideScaling=</varname></term>
<listitem>
<para>Takes a boolean. Toggle the ability of querying the receive side scaling (RSS)
configuration of the virtual function (VF). The VF RSS information like RSS hash key may be
considered sensitive on some devices where this information is shared between VF and the
physical function (PF). When unset, the kernel's default will be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>Trust=</varname></term>
<listitem>
<para>Takes a boolean. Allows one to set trust mode of the virtual function (VF). When set,
VF users can set a specific feature which may impact security and/or performance. When unset,
the kernel's default will be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>LinkState=</varname></term>
<listitem>
<para>Allows one to set the link state of the virtual function (VF). Takes a boolean or a
special value <literal>auto</literal>. Setting to <literal>auto</literal> means a
reflection of the physical function (PF) link state, <literal>yes</literal> lets the VF to
communicate with other VFs on this host even if the PF link state is down,
<literal>no</literal> causes the hardware to drop any packets sent by the VF. When unset,
the kernel's default will be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MACAddress=</varname></term>
<listitem>
<para>Specifies the MAC address for the virtual function.</para>
</listitem>
</varlistentry>
</variablelist>
</refsect1>
<refsect1>
<title>Examples</title>

View file

@ -300,94 +300,7 @@
</variablelist>
</refsect1>
<refsect1>
<title>[SR-IOV] Section Options</title>
<para>The [SR-IOV] section accepts the following keys. Specify several [SR-IOV] sections to
configure several SR-IOVs. SR-IOV provides the ability to partition a single physical PCI resource
into virtual PCI functions which can then be injected into a VM. In the case of network VFs, SR-IOV
improves north-south network performance (that is, traffic with endpoints outside the host machine)
by allowing traffic to bypass the host machines network stack.</para>
<variablelist class='network-directives'>
<varlistentry>
<term><varname>VirtualFunction=</varname></term>
<listitem>
<para>Specifies a Virtual Function (VF), lightweight PCIe function designed solely to move
data in and out. Takes an integer in the range 0…2147483646. This option is compulsory.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>VLANId=</varname></term>
<listitem>
<para>Specifies VLAN ID of the virtual function. Takes an integer in the range 1…4095.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>QualityOfService=</varname></term>
<listitem>
<para>Specifies quality of service of the virtual function. Takes an integer in the range
1…4294967294.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>VLANProtocol=</varname></term>
<listitem>
<para>Specifies VLAN protocol of the virtual function. Takes <literal>802.1Q</literal> or
<literal>802.1ad</literal>.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MACSpoofCheck=</varname></term>
<listitem>
<para>Takes a boolean. Controls the MAC spoof checking. When unset, the kernel's default will
be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>QueryReceiveSideScaling=</varname></term>
<listitem>
<para>Takes a boolean. Toggle the ability of querying the receive side scaling (RSS)
configuration of the virtual function (VF). The VF RSS information like RSS hash key may be
considered sensitive on some devices where this information is shared between VF and the
physical function (PF). When unset, the kernel's default will be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>Trust=</varname></term>
<listitem>
<para>Takes a boolean. Allows one to set trust mode of the virtual function (VF). When set, VF
users can set a specific feature which may impact security and/or performance. When unset,
the kernel's default will be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>LinkState=</varname></term>
<listitem>
<para>Allows one to set the link state of the virtual function (VF). Takes a boolean or a
special value <literal>auto</literal>. Setting to <literal>auto</literal> means a
reflection of the physical function (PF) link state, <literal>yes</literal> lets the VF to
communicate with other VFs on this host even if the PF link state is down,
<literal>no</literal> causes the hardware to drop any packets sent by the VF. When unset,
the kernel's default will be used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MACAddress=</varname></term>
<listitem>
<para>Specifies the MAC address for the virtual function.</para>
</listitem>
</varlistentry>
</variablelist>
</refsect1>
<xi:include href="systemd.link.xml" xpointer="sr-iov" />
<refsect1>
<title>[Network] Section Options</title>

View file

@ -8,6 +8,7 @@ _Pragma("GCC diagnostic ignored \"-Wimplicit-fallthrough\"")
#include "ethtool-util.h"
#include "link-config.h"
#include "net-condition.h"
#include "netif-sriov.h"
#include "socket-util.h"
%}
struct ConfigPerfItem;
@ -101,3 +102,12 @@ Link.RxMaxCoalescedHighFrames, config_parse_coalesce_u32,
Link.TxCoalesceHighSec, config_parse_coalesce_sec, 0, offsetof(LinkConfig, coalesce.tx_coalesce_usecs_high)
Link.TxMaxCoalescedHighFrames, config_parse_coalesce_u32, 0, offsetof(LinkConfig, coalesce.tx_max_coalesced_frames_high)
Link.CoalescePacketRateSampleIntervalSec, config_parse_coalesce_sec, 0, offsetof(LinkConfig, coalesce.rate_sample_interval)
SR-IOV.VirtualFunction, config_parse_sr_iov_uint32, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.VLANId, config_parse_sr_iov_uint32, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.QualityOfService, config_parse_sr_iov_uint32, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.VLANProtocol, config_parse_sr_iov_vlan_proto, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.MACSpoofCheck, config_parse_sr_iov_boolean, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.QueryReceiveSideScaling, config_parse_sr_iov_boolean, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.Trust, config_parse_sr_iov_boolean, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.LinkState, config_parse_sr_iov_link_state, 0, offsetof(LinkConfig, sr_iov_by_section)
SR-IOV.MACAddress, config_parse_sr_iov_mac, 0, offsetof(LinkConfig, sr_iov_by_section)

View file

@ -22,6 +22,7 @@
#include "log-link.h"
#include "memory-util.h"
#include "net-condition.h"
#include "netif-sriov.h"
#include "netif-util.h"
#include "netlink-util.h"
#include "parse-util.h"
@ -60,6 +61,8 @@ static LinkConfig* link_config_free(LinkConfig *config) {
free(config->wol_password_file);
erase_and_free(config->wol_password);
ordered_hashmap_free_with_destructor(config->sr_iov_by_section, sr_iov_free);
return mfree(config);
}
@ -257,7 +260,9 @@ int link_load_one(LinkConfigContext *ctx, const char *filename) {
STRV_MAKE_CONST(filename),
(const char* const*) CONF_PATHS_STRV("systemd/network"),
dropin_dirname,
"Match\0Link\0",
"Match\0"
"Link\0"
"SR-IOV\0",
config_item_perf_lookup, link_config_gperf_lookup,
CONFIG_PARSE_WARN, config, NULL);
if (r < 0)
@ -285,6 +290,10 @@ int link_load_one(LinkConfigContext *ctx, const char *filename) {
if (r < 0)
return r;
r = sr_iov_drop_invalid_sections(config->sr_iov_by_section);
if (r < 0)
return r;
log_debug("Parsed configuration file %s", filename);
LIST_PREPEND(configs, ctx->configs, TAKE_PTR(config));
@ -830,6 +839,53 @@ static int link_apply_alternative_names(Link *link, sd_netlink **rtnl) {
return 0;
}
static int sr_iov_configure(Link *link, sd_netlink **rtnl, SRIOV *sr_iov) {
_cleanup_(sd_netlink_message_unrefp) sd_netlink_message *req = NULL;
int r;
assert(link);
assert(rtnl);
assert(link->ifindex > 0);
if (!*rtnl) {
r = sd_netlink_open(rtnl);
if (r < 0)
return r;
}
r = sd_rtnl_message_new_link(*rtnl, &req, RTM_SETLINK, link->ifindex);
if (r < 0)
return r;
r = sr_iov_set_netlink_message(sr_iov, req);
if (r < 0)
return r;
r = sd_netlink_call(*rtnl, req, 0, NULL);
if (r < 0)
return r;
return 0;
}
static int link_apply_sr_iov_config(Link *link, sd_netlink **rtnl) {
SRIOV *sr_iov;
int r;
assert(link);
assert(link->config);
ORDERED_HASHMAP_FOREACH(sr_iov, link->config->sr_iov_by_section) {
r = sr_iov_configure(link, rtnl, sr_iov);
if (r < 0)
log_link_warning_errno(link, r,
"Failed to configure SR-IOV virtual function %"PRIu32", ignoring: %m",
sr_iov->vf);
}
return 0;
}
int link_apply_config(LinkConfigContext *ctx, sd_netlink **rtnl, Link *link) {
int r;
@ -861,6 +917,10 @@ int link_apply_config(LinkConfigContext *ctx, sd_netlink **rtnl, Link *link) {
if (r < 0)
return r;
r = link_apply_sr_iov_config(link, rtnl);
if (r < 0)
return r;
return 0;
}

View file

@ -7,6 +7,7 @@
#include "condition.h"
#include "conf-parser.h"
#include "ethtool-util.h"
#include "hashmap.h"
#include "list.h"
#include "net-condition.h"
#include "netif-naming-scheme.h"
@ -76,6 +77,8 @@ struct LinkConfig {
int autoneg_flow_control;
netdev_coalesce_param coalesce;
OrderedHashmap *sr_iov_by_section;
LIST_FIELDS(LinkConfig, configs);
};

View file

@ -80,3 +80,13 @@ RxMaxCoalescedHighFrames=
TxCoalesceHighSec=
TxMaxCoalescedHighFrames=
CoalescePacketRateSampleIntervalSec=
[SR-IOV]
VirtualFunction=
MACSpoofCheck=
VLANId=
VLANProtocol=
QualityOfService=
QueryReceiveSideScaling=
Trust=
LinkState=
MACAddress=