Add support for reading flow hash configuration via Netlink ethtool.
$ ynl --family ethtool --dump rss-get [{ "header": { "dev-index": 1, "dev-name": "enp1s0" }, "hfunc": 1, "hkey": b"...", "indir": [0, 1, ...], "flow-hash": { "ether": {"l2da"}, "ah-esp4": {"ip-src", "ip-dst"}, "ah-esp6": {"ip-src", "ip-dst"}, "ah4": {"ip-src", "ip-dst"}, "ah6": {"ip-src", "ip-dst"}, "esp4": {"ip-src", "ip-dst"}, "esp6": {"ip-src", "ip-dst"}, "ip4": {"ip-src", "ip-dst"}, "ip6": {"ip-src", "ip-dst"}, "sctp4": {"ip-src", "ip-dst"}, "sctp6": {"ip-src", "ip-dst"}, "udp4": {"ip-src", "ip-dst"}, "udp6": {"ip-src", "ip-dst"} "tcp4": {"l4-b-0-1", "l4-b-2-3", "ip-src", "ip-dst"}, "tcp6": {"l4-b-0-1", "l4-b-2-3", "ip-src", "ip-dst"}, }, }]
Jakub Kicinski (5): ethtool: rss: make sure dump takes the rss lock tools: ynl: decode enums in auto-ints ethtool: mark ETHER_FLOW as usable for Rx hash ethtool: rss: report which fields are configured for hashing selftests: drv-net: test RSS header field configuration
Documentation/netlink/specs/ethtool.yaml | 151 ++++++++++++++++++ Documentation/networking/ethtool-netlink.rst | 9 +- include/uapi/linux/ethtool.h | 4 +- .../uapi/linux/ethtool_netlink_generated.h | 34 ++++ net/ethtool/ioctl.c | 7 +- net/ethtool/rss.c | 145 +++++++++++++---- tools/net/ynl/pyynl/lib/ynl.py | 2 + .../selftests/drivers/net/hw/rss_api.py | 47 ++++++ 8 files changed, 364 insertions(+), 35 deletions(-)
After commit 040cef30b5e6 ("net: ethtool: move get_rxfh callback under the rss_lock") we're expected to take rss_lock around get. Switch dump to using the new prep helper and move the locking into it.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- net/ethtool/rss.c | 40 +++++++++++++++++++++------------------- 1 file changed, 21 insertions(+), 19 deletions(-)
diff --git a/net/ethtool/rss.c b/net/ethtool/rss.c index 4e8ca2c38175..37a7b20fcd07 100644 --- a/net/ethtool/rss.c +++ b/net/ethtool/rss.c @@ -64,6 +64,7 @@ rss_prepare_get(const struct rss_req_info *request, struct net_device *dev, ret = ethnl_ops_begin(dev); if (ret < 0) return ret; + mutex_lock(&dev->ethtool->rss_lock);
data->indir_size = 0; data->hkey_size = 0; @@ -77,7 +78,7 @@ rss_prepare_get(const struct rss_req_info *request, struct net_device *dev, rss_config = kzalloc(total_size, GFP_KERNEL); if (!rss_config) { ret = -ENOMEM; - goto out_ops; + goto out_unlock; }
if (data->indir_size) @@ -92,11 +93,12 @@ rss_prepare_get(const struct rss_req_info *request, struct net_device *dev,
ret = ops->get_rxfh(dev, &rxfh); if (ret) - goto out_ops; + goto out_unlock;
data->hfunc = rxfh.hfunc; data->input_xfrm = rxfh.input_xfrm; -out_ops: +out_unlock: + mutex_unlock(&dev->ethtool->rss_lock); ethnl_ops_complete(dev); return ret; } @@ -108,12 +110,16 @@ rss_prepare_ctx(const struct rss_req_info *request, struct net_device *dev, struct ethtool_rxfh_context *ctx; u32 total_size, indir_bytes; u8 *rss_config; + int ret;
data->no_key_fields = !dev->ethtool_ops->rxfh_per_ctx_key;
+ mutex_lock(&dev->ethtool->rss_lock); ctx = xa_load(&dev->ethtool->rss_ctx, request->rss_context); - if (!ctx) - return -ENOENT; + if (!ctx) { + ret = -ENOENT; + goto out_unlock; + }
data->indir_size = ctx->indir_size; data->hkey_size = ctx->key_size; @@ -123,8 +129,10 @@ rss_prepare_ctx(const struct rss_req_info *request, struct net_device *dev, indir_bytes = data->indir_size * sizeof(u32); total_size = indir_bytes + data->hkey_size; rss_config = kzalloc(total_size, GFP_KERNEL); - if (!rss_config) - return -ENOMEM; + if (!rss_config) { + ret = -ENOMEM; + goto out_unlock; + }
data->indir_table = (u32 *)rss_config; memcpy(data->indir_table, ethtool_rxfh_context_indir(ctx), indir_bytes); @@ -135,7 +143,10 @@ rss_prepare_ctx(const struct rss_req_info *request, struct net_device *dev, data->hkey_size); }
- return 0; + ret = 0; +out_unlock: + mutex_unlock(&dev->ethtool->rss_lock); + return ret; }
static int @@ -156,7 +167,6 @@ rss_prepare_data(const struct ethnl_req_info *req_base, struct rss_req_info *request = RSS_REQINFO(req_base); struct net_device *dev = reply_base->dev; const struct ethtool_ops *ops; - int ret;
ops = dev->ethtool_ops; if (!ops->get_rxfh) @@ -166,11 +176,7 @@ rss_prepare_data(const struct ethnl_req_info *req_base, if (request->rss_context && !ops->create_rxfh_context) return -EOPNOTSUPP;
- mutex_lock(&dev->ethtool->rss_lock); - ret = rss_prepare(request, dev, data, info); - mutex_unlock(&dev->ethtool->rss_lock); - - return ret; + return rss_prepare(request, dev, data, info); }
static int @@ -294,11 +300,7 @@ rss_dump_one_ctx(struct sk_buff *skb, struct netlink_callback *cb, if (ret < 0) goto err_cancel;
- /* Context 0 is not currently storred or cached in the XArray */ - if (!rss_context) - ret = rss_prepare_get(&req, dev, &data, info); - else - ret = rss_prepare_ctx(&req, dev, &data, info); + ret = rss_prepare(&req, dev, &data, info); if (ret) goto err_cancel;
Use enum decoding on auto-ints. Looks like we only had enum auto-ints for input values until now. Upcoming RSS work will need this to declare an attribute with flags as a uint.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- CC: donald.hunter@gmail.com --- tools/net/ynl/pyynl/lib/ynl.py | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/tools/net/ynl/pyynl/lib/ynl.py b/tools/net/ynl/pyynl/lib/ynl.py index 61deb5923067..7529bce174ff 100644 --- a/tools/net/ynl/pyynl/lib/ynl.py +++ b/tools/net/ynl/pyynl/lib/ynl.py @@ -762,6 +762,8 @@ genl_family_name_to_id = None decoded = True elif attr_spec.is_auto_scalar: decoded = attr.as_auto_scalar(attr_spec['type'], attr_spec.byte_order) + if 'enum' in attr_spec: + decoded = self._decode_enum(decoded, attr_spec) elif attr_spec["type"] in NlAttr.type_formats: decoded = attr.as_scalar(attr_spec['type'], attr_spec.byte_order) if 'enum' in attr_spec:
Jakub Kicinski kuba@kernel.org writes:
Use enum decoding on auto-ints. Looks like we only had enum auto-ints for input values until now. Upcoming RSS work will need this to declare an attribute with flags as a uint.
Signed-off-by: Jakub Kicinski kuba@kernel.org
Reviewed-by: Donald Hunter donald.hunter@gmail.com
Looks like some drivers (ena, enetc, fbnic.. there's probably more) consider ETHER_FLOW to be legitimate target for flow hashing. I'm not sure how intentional that is from the uAPI perspective vs just an effect of ethtool IOCTL doing minimal input validation. But Netlink will do strict validation, so we need to decide whether we allow this use case or not. I don't see a strong reason against it, and rejecting it would potentially regress a number of drivers. So update the comments and flow_type_hashable().
Signed-off-by: Jakub Kicinski kuba@kernel.org --- include/uapi/linux/ethtool.h | 4 ++-- net/ethtool/ioctl.c | 1 + 2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/include/uapi/linux/ethtool.h b/include/uapi/linux/ethtool.h index 707c1844010c..9e9afdd1238a 100644 --- a/include/uapi/linux/ethtool.h +++ b/include/uapi/linux/ethtool.h @@ -2314,7 +2314,7 @@ enum { IPV6_USER_FLOW = 0x0e, /* spec only (usr_ip6_spec; nfc only) */ IPV4_FLOW = 0x10, /* hash only */ IPV6_FLOW = 0x11, /* hash only */ - ETHER_FLOW = 0x12, /* spec only (ether_spec) */ + ETHER_FLOW = 0x12, /* hash or spec (ether_spec) */
/* Used for GTP-U IPv4 and IPv6. * The format of GTP packets only includes @@ -2371,7 +2371,7 @@ enum { /* Flag to enable RSS spreading of traffic matching rule (nfc only) */ #define FLOW_RSS 0x20000000
-/* L3-L4 network traffic flow hash options */ +/* L2-L4 network traffic flow hash options */ #define RXH_L2DA (1 << 1) #define RXH_VLAN (1 << 2) #define RXH_L3_PROTO (1 << 3) diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c index 139f95620cdd..67f6d900a4ee 100644 --- a/net/ethtool/ioctl.c +++ b/net/ethtool/ioctl.c @@ -981,6 +981,7 @@ static int ethtool_rxnfc_copy_to_user(void __user *useraddr, static bool flow_type_hashable(u32 flow_type) { switch (flow_type) { + case ETHER_FLOW: case TCP_V4_FLOW: case UDP_V4_FLOW: case SCTP_V4_FLOW:
Implement ETHTOOL_GRXFH over Netlink. The number of flow types is reasonable (around 20) so report all of them at once for simplicity.
Do not maintain the flow ID mapping with ioctl at the uAPI level. This gives us a chance to clean up the confusion that come from RxNFC vs RxFH (flow direction vs hashing) in the ioctl. Try to align with the names used in ethtool CLI, they seem to have stood the test of time just fine. One annoyance is that we still call L4 ports the weird names, but I guess they also apply to IPSec (where they cover the SPI) so it is what it is.
$ ynl --family ethtool --dump rss-get { "header": { "dev-index": 1, "dev-name": "enp1s0" }, "hfunc": 1, "hkey": b"...", "indir": [0, 1, ...], "flow-hash": { "ether": {"l2da"}, "ah-esp4": {"ip-src", "ip-dst"}, "ah-esp6": {"ip-src", "ip-dst"}, "ah4": {"ip-src", "ip-dst"}, "ah6": {"ip-src", "ip-dst"}, "esp4": {"ip-src", "ip-dst"}, "esp6": {"ip-src", "ip-dst"}, "ip4": {"ip-src", "ip-dst"}, "ip6": {"ip-src", "ip-dst"}, "sctp4": {"ip-src", "ip-dst"}, "sctp6": {"ip-src", "ip-dst"}, "udp4": {"ip-src", "ip-dst"}, "udp6": {"ip-src", "ip-dst"} "tcp4": {"l4-b-0-1", "l4-b-2-3", "ip-src", "ip-dst"}, "tcp6": {"l4-b-0-1", "l4-b-2-3", "ip-src", "ip-dst"}, }, }
Signed-off-by: Jakub Kicinski kuba@kernel.org --- Documentation/netlink/specs/ethtool.yaml | 151 ++++++++++++++++++ Documentation/networking/ethtool-netlink.rst | 9 +- .../uapi/linux/ethtool_netlink_generated.h | 34 ++++ net/ethtool/ioctl.c | 6 +- net/ethtool/rss.c | 105 ++++++++++-- 5 files changed, 291 insertions(+), 14 deletions(-)
diff --git a/Documentation/netlink/specs/ethtool.yaml b/Documentation/netlink/specs/ethtool.yaml index 49e782a33eb6..c38c03c624f0 100644 --- a/Documentation/netlink/specs/ethtool.yaml +++ b/Documentation/netlink/specs/ethtool.yaml @@ -158,6 +158,35 @@ c-version-name: ethtool-genl-version - name: pse-event-sw-pw-control-error doc: PSE faced an error managing the power control from software + - + name: rxfh-fields + name-prefix: rxh- + enum-name: + header: linux/ethtool.h + type: flags + entries: + - + name: l2da + value: 1 + - + name: vlan + - + name: l3-proto + - + name: ip-src + - + name: ip-dst + - + name: l4-b-0-1 + doc: src port in case of TCP/UDP/SCTP + - + name: l4-b-2-3 + doc: dst port in case of TCP/UDP/SCTP + - + name: gtp-teid + - + name: discard + value: 31
attribute-sets: - @@ -1447,6 +1476,123 @@ c-version-name: ethtool-genl-version name: pse-prio type: u32 name-prefix: ethtool-a- + - + name: flow + attr-cnt-name: --ethtool-a-flow-cnt + doc: | + Flow types, corresponding to those defined in the old + ethtool header for RXFH and RXNFC as ${PROTO}_FLOW. + The values are not matching the old ones to avoid carrying + into Netlink the IP_USER_FLOW vs IPV4_FLOW vs IPV4_USER_FLOW confusion. + attributes: + - + name: ether + type: uint + enum: rxfh-fields + - + name: ip4 + type: uint + enum: rxfh-fields + - + name: ip6 + type: uint + enum: rxfh-fields + - + name: tcp4 + type: uint + enum: rxfh-fields + - + name: tcp6 + type: uint + enum: rxfh-fields + - + name: udp4 + type: uint + enum: rxfh-fields + - + name: udp6 + type: uint + enum: rxfh-fields + - + name: sctp4 + type: uint + enum: rxfh-fields + - + name: sctp6 + type: uint + enum: rxfh-fields + - + name: ah4 + type: uint + enum: rxfh-fields + - + name: ah6 + type: uint + enum: rxfh-fields + - + name: esp4 + type: uint + enum: rxfh-fields + - + name: esp6 + type: uint + enum: rxfh-fields + - + name: ah-esp4 + type: uint + enum: rxfh-fields + - + name: ah-esp6 + type: uint + enum: rxfh-fields + - + name: gtpu4 + type: uint + enum: rxfh-fields + - + name: gtpu6 + type: uint + enum: rxfh-fields + - + name: gtpc4 + type: uint + enum: rxfh-fields + - + name: gtpc6 + type: uint + enum: rxfh-fields + - + name: gtpc-teid4 + type: uint + enum: rxfh-fields + - + name: gtpc-teid6 + type: uint + enum: rxfh-fields + - + name: gtpu-eh4 + type: uint + enum: rxfh-fields + - + name: gtpu-eh6 + type: uint + enum: rxfh-fields + - + name: gtpu-ul4 + type: uint + enum: rxfh-fields + - + name: gtpu-ul6 + type: uint + enum: rxfh-fields + - + name: gtpu-dl4 + type: uint + enum: rxfh-fields + - + name: gtpu-dl6 + type: uint + enum: rxfh-fields - name: rss attr-cnt-name: __ethtool-a-rss-cnt @@ -1478,6 +1624,10 @@ c-version-name: ethtool-genl-version - name: start-context type: u32 + - + name: flow-hash + type: nest + nested-attributes: flow - name: plca attr-cnt-name: __ethtool-a-plca-cnt @@ -2307,6 +2457,7 @@ c-version-name: ethtool-genl-version - indir - hkey - input-xfrm + - flow-hash dump: request: attributes: diff --git a/Documentation/networking/ethtool-netlink.rst b/Documentation/networking/ethtool-netlink.rst index 07e9808ebd2c..248bc3d93da9 100644 --- a/Documentation/networking/ethtool-netlink.rst +++ b/Documentation/networking/ethtool-netlink.rst @@ -1969,14 +1969,15 @@ used to ignore context 0s and only dump additional contexts).
Kernel response contents:
-===================================== ====== ========================== +===================================== ====== =============================== ``ETHTOOL_A_RSS_HEADER`` nested reply header ``ETHTOOL_A_RSS_CONTEXT`` u32 context number ``ETHTOOL_A_RSS_HFUNC`` u32 RSS hash func ``ETHTOOL_A_RSS_INDIR`` binary Indir table bytes ``ETHTOOL_A_RSS_HKEY`` binary Hash key bytes ``ETHTOOL_A_RSS_INPUT_XFRM`` u32 RSS input data transformation -===================================== ====== ========================== + ``ETHTOOL_A_RSS_FLOW_HASH`` nested Header fields included in hash +===================================== ====== ===============================
ETHTOOL_A_RSS_HFUNC attribute is bitmap indicating the hash function being used. Current supported options are toeplitz, xor or crc32. @@ -1985,6 +1986,8 @@ indicates queue number. ETHTOOL_A_RSS_INPUT_XFRM attribute is a bitmap indicating the type of transformation applied to the input protocol fields before given to the RSS hfunc. Current supported options are symmetric-xor and symmetric-or-xor. +ETHTOOL_A_RSS_FLOW_HASH carries per-flow type bitmask of which header +fields are included in the hash calculation.
PLCA_GET_CFG ============ @@ -2436,7 +2439,7 @@ are netlink only. ``ETHTOOL_SFLAGS`` ``ETHTOOL_MSG_FEATURES_SET`` ``ETHTOOL_GPFLAGS`` ``ETHTOOL_MSG_PRIVFLAGS_GET`` ``ETHTOOL_SPFLAGS`` ``ETHTOOL_MSG_PRIVFLAGS_SET`` - ``ETHTOOL_GRXFH`` n/a + ``ETHTOOL_GRXFH`` ``ETHTOOL_MSG_RSS_GET`` ``ETHTOOL_SRXFH`` n/a ``ETHTOOL_GGRO`` ``ETHTOOL_MSG_FEATURES_GET`` ``ETHTOOL_SGRO`` ``ETHTOOL_MSG_FEATURES_SET`` diff --git a/include/uapi/linux/ethtool_netlink_generated.h b/include/uapi/linux/ethtool_netlink_generated.h index 8f30ffa1cd14..96027e26ffba 100644 --- a/include/uapi/linux/ethtool_netlink_generated.h +++ b/include/uapi/linux/ethtool_netlink_generated.h @@ -678,6 +678,39 @@ enum { ETHTOOL_A_PSE_MAX = (__ETHTOOL_A_PSE_CNT - 1) };
+enum { + ETHTOOL_A_FLOW_ETHER = 1, + ETHTOOL_A_FLOW_IP4, + ETHTOOL_A_FLOW_IP6, + ETHTOOL_A_FLOW_TCP4, + ETHTOOL_A_FLOW_TCP6, + ETHTOOL_A_FLOW_UDP4, + ETHTOOL_A_FLOW_UDP6, + ETHTOOL_A_FLOW_SCTP4, + ETHTOOL_A_FLOW_SCTP6, + ETHTOOL_A_FLOW_AH4, + ETHTOOL_A_FLOW_AH6, + ETHTOOL_A_FLOW_ESP4, + ETHTOOL_A_FLOW_ESP6, + ETHTOOL_A_FLOW_AH_ESP4, + ETHTOOL_A_FLOW_AH_ESP6, + ETHTOOL_A_FLOW_GTPU4, + ETHTOOL_A_FLOW_GTPU6, + ETHTOOL_A_FLOW_GTPC4, + ETHTOOL_A_FLOW_GTPC6, + ETHTOOL_A_FLOW_GTPC_TEID4, + ETHTOOL_A_FLOW_GTPC_TEID6, + ETHTOOL_A_FLOW_GTPU_EH4, + ETHTOOL_A_FLOW_GTPU_EH6, + ETHTOOL_A_FLOW_GTPU_UL4, + ETHTOOL_A_FLOW_GTPU_UL6, + ETHTOOL_A_FLOW_GTPU_DL4, + ETHTOOL_A_FLOW_GTPU_DL6, + + __ETHTOOL_A_FLOW_CNT, + ETHTOOL_A_FLOW_MAX = (__ETHTOOL_A_FLOW_CNT - 1) +}; + enum { ETHTOOL_A_RSS_UNSPEC, ETHTOOL_A_RSS_HEADER, @@ -687,6 +720,7 @@ enum { ETHTOOL_A_RSS_HKEY, ETHTOOL_A_RSS_INPUT_XFRM, ETHTOOL_A_RSS_START_CONTEXT, + ETHTOOL_A_RSS_FLOW_HASH,
__ETHTOOL_A_RSS_CNT, ETHTOOL_A_RSS_MAX = (__ETHTOOL_A_RSS_CNT - 1) diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c index 67f6d900a4ee..cccb4694f5e1 100644 --- a/net/ethtool/ioctl.c +++ b/net/ethtool/ioctl.c @@ -1101,7 +1101,11 @@ ethtool_set_rxfh_fields(struct net_device *dev, u32 cmd, void __user *useraddr) rc = ops->set_rxfh_fields(dev, &fields, NULL); exit_unlock: mutex_unlock(&dev->ethtool->rss_lock); - return rc; + if (rc) + return rc; + + ethtool_rss_notify(dev, fields.rss_context); + return 0; }
static noinline_for_stack int diff --git a/net/ethtool/rss.c b/net/ethtool/rss.c index 37a7b20fcd07..41ab9fc67652 100644 --- a/net/ethtool/rss.c +++ b/net/ethtool/rss.c @@ -12,6 +12,7 @@ struct rss_req_info {
struct rss_reply_data { struct ethnl_reply_data base; + bool has_flow_hash; bool no_key_fields; u32 indir_size; u32 hkey_size; @@ -19,6 +20,37 @@ struct rss_reply_data { u32 input_xfrm; u32 *indir_table; u8 *hkey; + int flow_hash[__ETHTOOL_A_FLOW_CNT]; +}; + +static const u8 ethtool_rxfh_ft_nl2ioctl[] = { + [ETHTOOL_A_FLOW_ETHER] = ETHER_FLOW, + [ETHTOOL_A_FLOW_IP4] = IPV4_FLOW, + [ETHTOOL_A_FLOW_IP6] = IPV6_FLOW, + [ETHTOOL_A_FLOW_TCP4] = TCP_V4_FLOW, + [ETHTOOL_A_FLOW_UDP4] = UDP_V4_FLOW, + [ETHTOOL_A_FLOW_SCTP4] = SCTP_V4_FLOW, + [ETHTOOL_A_FLOW_AH_ESP4] = AH_ESP_V4_FLOW, + [ETHTOOL_A_FLOW_TCP6] = TCP_V6_FLOW, + [ETHTOOL_A_FLOW_UDP6] = UDP_V6_FLOW, + [ETHTOOL_A_FLOW_SCTP6] = SCTP_V6_FLOW, + [ETHTOOL_A_FLOW_AH_ESP6] = AH_ESP_V6_FLOW, + [ETHTOOL_A_FLOW_AH4] = AH_V4_FLOW, + [ETHTOOL_A_FLOW_ESP4] = ESP_V4_FLOW, + [ETHTOOL_A_FLOW_AH6] = AH_V6_FLOW, + [ETHTOOL_A_FLOW_ESP6] = ESP_V6_FLOW, + [ETHTOOL_A_FLOW_GTPU4] = GTPU_V4_FLOW, + [ETHTOOL_A_FLOW_GTPU6] = GTPU_V6_FLOW, + [ETHTOOL_A_FLOW_GTPC4] = GTPC_V4_FLOW, + [ETHTOOL_A_FLOW_GTPC6] = GTPC_V6_FLOW, + [ETHTOOL_A_FLOW_GTPC_TEID4] = GTPC_TEID_V4_FLOW, + [ETHTOOL_A_FLOW_GTPC_TEID6] = GTPC_TEID_V6_FLOW, + [ETHTOOL_A_FLOW_GTPU_EH4] = GTPU_EH_V4_FLOW, + [ETHTOOL_A_FLOW_GTPU_EH6] = GTPU_EH_V6_FLOW, + [ETHTOOL_A_FLOW_GTPU_UL4] = GTPU_UL_V4_FLOW, + [ETHTOOL_A_FLOW_GTPU_UL6] = GTPU_UL_V6_FLOW, + [ETHTOOL_A_FLOW_GTPU_DL4] = GTPU_DL_V4_FLOW, + [ETHTOOL_A_FLOW_GTPU_DL6] = GTPU_DL_V6_FLOW, };
#define RSS_REQINFO(__req_base) \ @@ -49,6 +81,37 @@ rss_parse_request(struct ethnl_req_info *req_info, struct nlattr **tb, return 0; }
+static void +rss_prepare_flow_hash(const struct rss_req_info *req, struct net_device *dev, + struct rss_reply_data *data, const struct genl_info *info) +{ + int i; + + data->has_flow_hash = false; + + if (!dev->ethtool_ops->get_rxfh_fields) + return; + if (req->rss_context && !dev->ethtool_ops->rxfh_per_ctx_fields) + return; + + mutex_lock(&dev->ethtool->rss_lock); + for (i = 1; i < __ETHTOOL_A_FLOW_CNT; i++) { + struct ethtool_rxfh_fields fields = { + .flow_type = ethtool_rxfh_ft_nl2ioctl[i], + .rss_context = req->rss_context, + }; + + if (dev->ethtool_ops->get_rxfh_fields(dev, &fields)) { + data->flow_hash[i] = -1; /* Unsupported */ + continue; + } + + data->flow_hash[i] = fields.data; + data->has_flow_hash = true; + } + mutex_unlock(&dev->ethtool->rss_lock); +} + static int rss_prepare_get(const struct rss_req_info *request, struct net_device *dev, struct rss_reply_data *data, const struct genl_info *info) @@ -153,6 +216,8 @@ static int rss_prepare(const struct rss_req_info *request, struct net_device *dev, struct rss_reply_data *data, const struct genl_info *info) { + rss_prepare_flow_hash(request, dev, data, info); + if (request->rss_context) return rss_prepare_ctx(request, dev, data, info); return rss_prepare_get(request, dev, data, info); @@ -190,7 +255,10 @@ rss_reply_size(const struct ethnl_req_info *req_base, nla_total_size(sizeof(u32)) + /* _RSS_HFUNC */ nla_total_size(sizeof(u32)) + /* _RSS_INPUT_XFRM */ nla_total_size(sizeof(u32) * data->indir_size) + /* _RSS_INDIR */ - nla_total_size(data->hkey_size); /* _RSS_HKEY */ + nla_total_size(data->hkey_size) + /* _RSS_HKEY */ + nla_total_size(0) + /* _RSS_FLOW_HASH */ + nla_total_size(sizeof(u32)) * ETHTOOL_A_FLOW_MAX + + 0;
return len; } @@ -211,17 +279,34 @@ rss_fill_reply(struct sk_buff *skb, const struct ethnl_req_info *req_base, sizeof(u32) * data->indir_size, data->indir_table))) return -EMSGSIZE;
- if (data->no_key_fields) - return 0; - - if ((data->hfunc && - nla_put_u32(skb, ETHTOOL_A_RSS_HFUNC, data->hfunc)) || - (data->input_xfrm && - nla_put_u32(skb, ETHTOOL_A_RSS_INPUT_XFRM, data->input_xfrm)) || - (data->hkey_size && - nla_put(skb, ETHTOOL_A_RSS_HKEY, data->hkey_size, data->hkey))) + if (!data->no_key_fields && + ((data->hfunc && + nla_put_u32(skb, ETHTOOL_A_RSS_HFUNC, data->hfunc)) || + (data->input_xfrm && + nla_put_u32(skb, ETHTOOL_A_RSS_INPUT_XFRM, data->input_xfrm)) || + (data->hkey_size && + nla_put(skb, ETHTOOL_A_RSS_HKEY, data->hkey_size, data->hkey)))) return -EMSGSIZE;
+ if (data->has_flow_hash) { + struct nlattr *nest; + int i; + + nest = nla_nest_start(skb, ETHTOOL_A_RSS_FLOW_HASH); + if (!nest) + return -EMSGSIZE; + + for (i = 1; i < __ETHTOOL_A_FLOW_CNT; i++) { + if (data->flow_hash[i] >= 0 && + nla_put_uint(skb, i, data->flow_hash[i])) { + nla_nest_cancel(skb, nest); + return -EMSGSIZE; + } + } + + nla_nest_end(skb, nest); + } + return 0; }
Test reading RXFH fields over IOCTL and netlink.
# ./tools/testing/selftests/drivers/net/hw/rss_api.py TAP version 13 1..3 ok 1 rss_api.test_rxfh_indir_ntf ok 2 rss_api.test_rxfh_indir_ctx_ntf ok 3 rss_api.test_rxfh_fields # Totals: pass:3 fail:0 xfail:0 xpass:0 skip:0 error:0
Signed-off-by: Jakub Kicinski kuba@kernel.org --- .../selftests/drivers/net/hw/rss_api.py | 47 +++++++++++++++++++ 1 file changed, 47 insertions(+)
diff --git a/tools/testing/selftests/drivers/net/hw/rss_api.py b/tools/testing/selftests/drivers/net/hw/rss_api.py index db0f723a674b..6ae908bed1a4 100755 --- a/tools/testing/selftests/drivers/net/hw/rss_api.py +++ b/tools/testing/selftests/drivers/net/hw/rss_api.py @@ -20,6 +20,38 @@ from lib.py import NetDrvEnv return int(output.split()[-1])
+def _ethtool_get_cfg(cfg, fl_type, to_nl=False): + descr = ethtool(f"-n {cfg.ifname} rx-flow-hash {fl_type}").stdout + + if to_nl: + converter = { + "IP SA": "ip-src", + "IP DA": "ip-dst", + "L4 bytes 0 & 1 [TCP/UDP src port]": "l4-b-0-1", + "L4 bytes 2 & 3 [TCP/UDP dst port]": "l4-b-2-3", + } + + ret = set() + else: + converter = { + "IP SA": "s", + "IP DA": "d", + "L3 proto": "t", + "L4 bytes 0 & 1 [TCP/UDP src port]": "f", + "L4 bytes 2 & 3 [TCP/UDP dst port]": "n", + } + + ret = "" + + for line in descr.split("\n")[1:-2]: + # if this raises we probably need to add more keys to converter above + if to_nl: + ret.add(converter[line]) + else: + ret += converter[line] + return ret + + def test_rxfh_indir_ntf(cfg): """ Check that Netlink notifications are generated when RSS indirection @@ -77,6 +109,21 @@ from lib.py import NetDrvEnv ksft_eq(set(ntf["msg"]["indir"]), {1})
+def test_rxfh_fields(cfg): + """ + Test reading Rx Flow Hash over Netlink. + """ + + flow_types = ["tcp4", "tcp6", "udp4", "udp6"] + ethnl = EthtoolFamily() + + cfg_nl = ethnl.rss_get({"header": {"dev-index": cfg.ifindex}}) + for fl_type in flow_types: + one = _ethtool_get_cfg(cfg, fl_type, to_nl=True) + ksft_eq(one, cfg_nl["flow-hash"][fl_type], + comment="Config for " + fl_type) + + def main() -> None: """ Ksft boiler plate main """
Hello:
This series was applied to netdev/net-next.git (main) by Jakub Kicinski kuba@kernel.org:
On Tue, 8 Jul 2025 15:06:35 -0700 you wrote:
Add support for reading flow hash configuration via Netlink ethtool.
$ ynl --family ethtool --dump rss-get [{ "header": { "dev-index": 1, "dev-name": "enp1s0" }, "hfunc": 1, "hkey": b"...", "indir": [0, 1, ...], "flow-hash": { "ether": {"l2da"}, "ah-esp4": {"ip-src", "ip-dst"}, "ah-esp6": {"ip-src", "ip-dst"}, "ah4": {"ip-src", "ip-dst"}, "ah6": {"ip-src", "ip-dst"}, "esp4": {"ip-src", "ip-dst"}, "esp6": {"ip-src", "ip-dst"}, "ip4": {"ip-src", "ip-dst"}, "ip6": {"ip-src", "ip-dst"}, "sctp4": {"ip-src", "ip-dst"}, "sctp6": {"ip-src", "ip-dst"}, "udp4": {"ip-src", "ip-dst"}, "udp6": {"ip-src", "ip-dst"} "tcp4": {"l4-b-0-1", "l4-b-2-3", "ip-src", "ip-dst"}, "tcp6": {"l4-b-0-1", "l4-b-2-3", "ip-src", "ip-dst"}, }, }]
[...]
Here is the summary with links: - [net-next,1/5] ethtool: rss: make sure dump takes the rss lock https://git.kernel.org/netdev/net-next/c/400244eaa2c9 - [net-next,2/5] tools: ynl: decode enums in auto-ints https://git.kernel.org/netdev/net-next/c/f7c595c9d9f4 - [net-next,3/5] ethtool: mark ETHER_FLOW as usable for Rx hash https://git.kernel.org/netdev/net-next/c/d7974697de4d - [net-next,4/5] ethtool: rss: report which fields are configured for hashing https://git.kernel.org/netdev/net-next/c/178331743ca8 - [net-next,5/5] selftests: drv-net: test RSS header field configuration https://git.kernel.org/netdev/net-next/c/0c8754b75e69
You are awesome, thank you!
linux-kselftest-mirror@lists.linaro.org