net/sfc/base: add API to control UDP tunnel ports
authorIvan Malov <ivan.malov@oktetlabs.ru>
Sun, 24 Dec 2017 10:46:38 +0000 (10:46 +0000)
committerFerruh Yigit <ferruh.yigit@intel.com>
Tue, 16 Jan 2018 17:47:49 +0000 (18:47 +0100)
HW needs to know which UDP packets should be treated as tunnel
encapsulation to do inner packet recognition, classification and
offloads.

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
drivers/net/sfc/Makefile
drivers/net/sfc/base/ef10_nic.c
drivers/net/sfc/base/efx.h
drivers/net/sfc/base/efx_check.h
drivers/net/sfc/base/efx_impl.h
drivers/net/sfc/base/efx_mcdi.h
drivers/net/sfc/base/efx_tunnel.c [new file with mode: 0644]
drivers/net/sfc/efsys.h

index 2cfd62a..9541815 100644 (file)
@@ -122,6 +122,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += efx_phy.c
 SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += efx_port.c
 SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += efx_rx.c
 SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += efx_sram.c
+SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += efx_tunnel.c
 SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += efx_tx.c
 SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += efx_vpd.c
 SRCS-$(CONFIG_RTE_LIBRTE_SFC_EFX_PMD) += mcdi_mon.c
index 53ec227..dc6c22b 100644 (file)
@@ -1076,12 +1076,20 @@ ef10_get_datapath_caps(
         * Check if firmware supports VXLAN and NVGRE tunnels.
         * The capability indicates Geneve protocol support as well.
         */
-       if (CAP_FLAG(flags, VXLAN_NVGRE))
+       if (CAP_FLAG(flags, VXLAN_NVGRE)) {
                encp->enc_tunnel_encapsulations_supported =
                    (1u << EFX_TUNNEL_PROTOCOL_VXLAN) |
                    (1u << EFX_TUNNEL_PROTOCOL_GENEVE) |
                    (1u << EFX_TUNNEL_PROTOCOL_NVGRE);
 
+               EFX_STATIC_ASSERT(EFX_TUNNEL_MAXNENTRIES ==
+                   MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_IN_ENTRIES_MAXNUM);
+               encp->enc_tunnel_config_udp_entries_max =
+                   EFX_TUNNEL_MAXNENTRIES;
+       } else {
+               encp->enc_tunnel_config_udp_entries_max = 0;
+       }
+
 #undef CAP_FLAG
 #undef CAP_FLAG2
 
index ec65dcf..8c59a99 100644 (file)
@@ -1205,6 +1205,11 @@ typedef struct efx_nic_cfg_s {
        boolean_t               enc_pm_and_rxdp_counters;
        boolean_t               enc_mac_stats_40g_tx_size_bins;
        uint32_t                enc_tunnel_encapsulations_supported;
+       /*
+        * NIC global maximum for unique UDP tunnel ports shared by all
+        * functions.
+        */
+       uint32_t                enc_tunnel_config_udp_entries_max;
        /* External port identifier */
        uint8_t                 enc_external_port;
        uint32_t                enc_mcdi_max_payload_length;
@@ -2631,6 +2636,52 @@ efx_lic_finish_partition(
 
 #endif /* EFSYS_OPT_LICENSING */
 
+/* TUNNEL */
+
+#if EFSYS_OPT_TUNNEL
+
+extern __checkReturn   efx_rc_t
+efx_tunnel_init(
+       __in            efx_nic_t *enp);
+
+extern                 void
+efx_tunnel_fini(
+       __in            efx_nic_t *enp);
+
+/*
+ * For overlay network encapsulation using UDP, the firmware needs to know
+ * the configured UDP port for the overlay so it can decode encapsulated
+ * frames correctly.
+ * The UDP port/protocol list is global.
+ */
+
+extern __checkReturn   efx_rc_t
+efx_tunnel_config_udp_add(
+       __in            efx_nic_t *enp,
+       __in            uint16_t port /* host/cpu-endian */,
+       __in            efx_tunnel_protocol_t protocol);
+
+extern __checkReturn   efx_rc_t
+efx_tunnel_config_udp_remove(
+       __in            efx_nic_t *enp,
+       __in            uint16_t port /* host/cpu-endian */,
+       __in            efx_tunnel_protocol_t protocol);
+
+extern                 void
+efx_tunnel_config_clear(
+       __in            efx_nic_t *enp);
+
+/**
+ * Apply tunnel UDP ports configuration to hardware.
+ *
+ * EAGAIN is returned if hardware will be reset (datapath and managment CPU
+ * reboot).
+ */
+extern __checkReturn   efx_rc_t
+efx_tunnel_reconfigure(
+       __in            efx_nic_t *enp);
+
+#endif /* EFSYS_OPT_TUNNEL */
 
 
 #ifdef __cplusplus
index 33c16a6..ef5936b 100644 (file)
 # endif
 #endif
 
+/* Support hardware assistance for tunnels */
+#if EFSYS_OPT_TUNNEL
+# if !EFSYS_OPT_MEDFORD
+#  error "TUNNEL requires MEDFORD"
+# endif
+#endif /* EFSYS_OPT_TUNNEL */
+
 #endif /* _SYS_EFX_CHECK_H */
index 16e5c20..7c971b3 100644 (file)
@@ -74,6 +74,7 @@ extern "C" {
 #define        EFX_MOD_MON             0x00000400
 #define        EFX_MOD_FILTER          0x00001000
 #define        EFX_MOD_LIC             0x00002000
+#define        EFX_MOD_TUNNEL          0x00004000
 
 #define        EFX_RESET_PHY           0x00000001
 #define        EFX_RESET_RXQ_ERR       0x00000002
@@ -262,6 +263,12 @@ efx_filter_reconfigure(
 
 #endif /* EFSYS_OPT_FILTER */
 
+#if EFSYS_OPT_TUNNEL
+typedef struct efx_tunnel_ops_s {
+       boolean_t       (*eto_udp_encap_supported)(efx_nic_t *);
+       efx_rc_t        (*eto_reconfigure)(efx_nic_t *);
+} efx_tunnel_ops_t;
+#endif /* EFSYS_OPT_TUNNEL */
 
 typedef struct efx_port_s {
        efx_mac_type_t          ep_mac_type;
@@ -433,6 +440,22 @@ siena_filter_tbl_clear(
 
 #if EFSYS_OPT_MCDI
 
+#define        EFX_TUNNEL_MAXNENTRIES  (16)
+
+#if EFSYS_OPT_TUNNEL
+
+typedef struct efx_tunnel_udp_entry_s {
+       uint16_t                        etue_port; /* host/cpu-endian */
+       uint16_t                        etue_protocol;
+} efx_tunnel_udp_entry_t;
+
+typedef struct efx_tunnel_cfg_s {
+       efx_tunnel_udp_entry_t  etc_udp_entries[EFX_TUNNEL_MAXNENTRIES];
+       unsigned int            etc_udp_entries_num;
+} efx_tunnel_cfg_t;
+
+#endif /* EFSYS_OPT_TUNNEL */
+
 typedef struct efx_mcdi_ops_s {
        efx_rc_t        (*emco_init)(efx_nic_t *, const efx_mcdi_transport_t *);
        void            (*emco_send_request)(efx_nic_t *, void *, size_t,
@@ -645,6 +668,10 @@ struct efx_nic_s {
        efx_filter_t            en_filter;
        const efx_filter_ops_t  *en_efop;
 #endif /* EFSYS_OPT_FILTER */
+#if EFSYS_OPT_TUNNEL
+       efx_tunnel_cfg_t        en_tunnel_cfg;
+       const efx_tunnel_ops_t  *en_etop;
+#endif /* EFSYS_OPT_TUNNEL */
 #if EFSYS_OPT_MCDI
        efx_mcdi_t              en_mcdi;
 #endif /* EFSYS_OPT_MCDI */
index fba807a..a95cdc0 100644 (file)
@@ -382,6 +382,10 @@ efx_mcdi_phy_module_get_info(
        EFX_WORD_FIELD(*MCDI_OUT2(_emr, efx_word_t, _ofst),             \
                    EFX_WORD_0)
 
+#define        MCDI_OUT_WORD_FIELD(_emr, _ofst, _field)                        \
+       EFX_WORD_FIELD(*MCDI_OUT2(_emr, efx_word_t, _ofst),             \
+                      MC_CMD_ ## _field)
+
 #define        MCDI_OUT_DWORD(_emr, _ofst)                                     \
        EFX_DWORD_FIELD(*MCDI_OUT2(_emr, efx_dword_t, _ofst),           \
                        EFX_DWORD_0)
diff --git a/drivers/net/sfc/base/efx_tunnel.c b/drivers/net/sfc/base/efx_tunnel.c
new file mode 100644 (file)
index 0000000..8a8ac32
--- /dev/null
@@ -0,0 +1,487 @@
+/*
+ * Copyright (c) 2017 Solarflare Communications Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+ * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+ * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+ * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+ * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
+ * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * The views and conclusions contained in the software and documentation are
+ * those of the authors and should not be interpreted as representing official
+ * policies, either expressed or implied, of the FreeBSD Project.
+ */
+
+#include "efx.h"
+#include "efx_impl.h"
+
+
+#if EFSYS_OPT_TUNNEL
+
+#if EFSYS_OPT_SIENA || EFSYS_OPT_HUNTINGTON
+static const efx_tunnel_ops_t  __efx_tunnel_dummy_ops = {
+       NULL,   /* eto_udp_encap_supported */
+       NULL,   /* eto_reconfigure */
+};
+#endif /* EFSYS_OPT_SIENA || EFSYS_OPT_HUNTINGTON */
+
+#if EFSYS_OPT_MEDFORD
+static __checkReturn   boolean_t
+medford_udp_encap_supported(
+       __in            efx_nic_t *enp);
+
+static __checkReturn   efx_rc_t
+medford_tunnel_reconfigure(
+       __in            efx_nic_t *enp);
+
+static const efx_tunnel_ops_t  __efx_tunnel_medford_ops = {
+       medford_udp_encap_supported,    /* eto_udp_encap_supported */
+       medford_tunnel_reconfigure,     /* eto_reconfigure */
+};
+#endif /* EFSYS_OPT_MEDFORD */
+
+static __checkReturn           efx_rc_t
+efx_mcdi_set_tunnel_encap_udp_ports(
+       __in                    efx_nic_t *enp,
+       __in                    efx_tunnel_cfg_t *etcp,
+       __in                    boolean_t unloading,
+       __out                   boolean_t *resetting)
+{
+       efx_mcdi_req_t req;
+       uint8_t payload[MAX(MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_IN_LENMAX,
+                           MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_OUT_LEN)];
+       efx_word_t flags;
+       efx_rc_t rc;
+       unsigned int i;
+       unsigned int entries_num;
+
+       if (etcp == NULL)
+               entries_num = 0;
+       else
+               entries_num = etcp->etc_udp_entries_num;
+
+       (void) memset(payload, 0, sizeof (payload));
+       req.emr_cmd = MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS;
+       req.emr_in_buf = payload;
+       req.emr_in_length =
+           MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_IN_LEN(entries_num);
+       req.emr_out_buf = payload;
+       req.emr_out_length = MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_OUT_LEN;
+
+       EFX_POPULATE_WORD_1(flags,
+           MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_IN_UNLOADING,
+           (unloading == B_TRUE) ? 1 : 0);
+       MCDI_IN_SET_WORD(req, SET_TUNNEL_ENCAP_UDP_PORTS_IN_FLAGS,
+           EFX_WORD_FIELD(flags, EFX_WORD_0));
+
+       MCDI_IN_SET_WORD(req, SET_TUNNEL_ENCAP_UDP_PORTS_IN_NUM_ENTRIES,
+           entries_num);
+
+       for (i = 0; i < entries_num; ++i) {
+               uint16_t mcdi_udp_protocol;
+
+               switch (etcp->etc_udp_entries[i].etue_protocol) {
+               case EFX_TUNNEL_PROTOCOL_VXLAN:
+                       mcdi_udp_protocol = TUNNEL_ENCAP_UDP_PORT_ENTRY_VXLAN;
+                       break;
+               case EFX_TUNNEL_PROTOCOL_GENEVE:
+                       mcdi_udp_protocol = TUNNEL_ENCAP_UDP_PORT_ENTRY_GENEVE;
+                       break;
+               default:
+                       rc = EINVAL;
+                       goto fail1;
+               }
+
+               /*
+                * UDP port is MCDI native little-endian in the request
+                * and EFX_POPULATE_DWORD cares about conversion from
+                * host/CPU byte order to little-endian.
+                */
+               EFX_STATIC_ASSERT(sizeof (efx_dword_t) ==
+                   TUNNEL_ENCAP_UDP_PORT_ENTRY_LEN);
+               EFX_POPULATE_DWORD_2(
+                   MCDI_IN2(req, efx_dword_t,
+                       SET_TUNNEL_ENCAP_UDP_PORTS_IN_ENTRIES)[i],
+                   TUNNEL_ENCAP_UDP_PORT_ENTRY_UDP_PORT,
+                   etcp->etc_udp_entries[i].etue_port,
+                   TUNNEL_ENCAP_UDP_PORT_ENTRY_PROTOCOL,
+                   mcdi_udp_protocol);
+       }
+
+       efx_mcdi_execute(enp, &req);
+
+       if (req.emr_rc != 0) {
+               rc = req.emr_rc;
+               goto fail2;
+       }
+
+       if (req.emr_out_length_used !=
+           MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_OUT_LEN) {
+               rc = EMSGSIZE;
+               goto fail3;
+       }
+
+       *resetting = MCDI_OUT_WORD_FIELD(req,
+           SET_TUNNEL_ENCAP_UDP_PORTS_OUT_FLAGS,
+           SET_TUNNEL_ENCAP_UDP_PORTS_OUT_RESETTING);
+
+       return (0);
+
+fail3:
+       EFSYS_PROBE(fail3);
+
+fail2:
+       EFSYS_PROBE(fail2);
+
+fail1:
+       EFSYS_PROBE1(fail1, efx_rc_t, rc);
+
+       return (rc);
+}
+
+       __checkReturn   efx_rc_t
+efx_tunnel_init(
+       __in            efx_nic_t *enp)
+{
+       efx_tunnel_cfg_t *etcp = &enp->en_tunnel_cfg;
+       const efx_tunnel_ops_t *etop;
+       efx_rc_t rc;
+
+       EFSYS_ASSERT3U(enp->en_magic, ==, EFX_NIC_MAGIC);
+       EFSYS_ASSERT3U(enp->en_mod_flags, &, EFX_MOD_PROBE);
+       EFSYS_ASSERT(!(enp->en_mod_flags & EFX_MOD_TUNNEL));
+
+       EFX_STATIC_ASSERT(EFX_TUNNEL_MAXNENTRIES ==
+           MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS_IN_ENTRIES_MAXNUM);
+
+       switch (enp->en_family) {
+#if EFSYS_OPT_SIENA
+       case EFX_FAMILY_SIENA:
+               etop = &__efx_tunnel_dummy_ops;
+               break;
+#endif /* EFSYS_OPT_SIENA */
+
+#if EFSYS_OPT_HUNTINGTON
+       case EFX_FAMILY_HUNTINGTON:
+               etop = &__efx_tunnel_dummy_ops;
+               break;
+#endif /* EFSYS_OPT_HUNTINGTON */
+
+#if EFSYS_OPT_MEDFORD
+       case EFX_FAMILY_MEDFORD:
+               etop = &__efx_tunnel_medford_ops;
+               break;
+#endif /* EFSYS_OPT_MEDFORD */
+
+       default:
+               EFSYS_ASSERT(0);
+               rc = ENOTSUP;
+               goto fail1;
+       }
+
+       memset(etcp->etc_udp_entries, 0, sizeof (etcp->etc_udp_entries));
+       etcp->etc_udp_entries_num = 0;
+
+       enp->en_etop = etop;
+       enp->en_mod_flags |= EFX_MOD_TUNNEL;
+
+       return (0);
+
+fail1:
+       EFSYS_PROBE1(fail1, efx_rc_t, rc);
+
+       enp->en_etop = NULL;
+       enp->en_mod_flags &= ~EFX_MOD_TUNNEL;
+
+       return (rc);
+}
+
+                       void
+efx_tunnel_fini(
+       __in            efx_nic_t *enp)
+{
+       boolean_t resetting;
+
+       EFSYS_ASSERT3U(enp->en_magic, ==, EFX_NIC_MAGIC);
+       EFSYS_ASSERT3U(enp->en_mod_flags, &, EFX_MOD_PROBE);
+       EFSYS_ASSERT3U(enp->en_mod_flags, &, EFX_MOD_TUNNEL);
+
+       if ((enp->en_etop->eto_udp_encap_supported != NULL) &&
+           enp->en_etop->eto_udp_encap_supported(enp)) {
+               /*
+                * The UNLOADING flag allows the MC to suppress the datapath
+                * reset if it was set on the last call to
+                * MC_CMD_SET_TUNNEL_ENCAP_UDP_PORTS by all functions
+                */
+               (void) efx_mcdi_set_tunnel_encap_udp_ports(enp, NULL, B_TRUE,
+                   &resetting);
+       }
+
+       enp->en_etop = NULL;
+       enp->en_mod_flags &= ~EFX_MOD_TUNNEL;
+}
+
+static __checkReturn   efx_rc_t
+efx_tunnel_config_find_udp_tunnel_entry(
+       __in            efx_tunnel_cfg_t *etcp,
+       __in            uint16_t port,
+       __out           unsigned int *entryp)
+{
+       unsigned int i;
+
+       for (i = 0; i < etcp->etc_udp_entries_num; ++i) {
+               efx_tunnel_udp_entry_t *p = &etcp->etc_udp_entries[i];
+
+               if (p->etue_port == port) {
+                       *entryp = i;
+                       return (0);
+               }
+       }
+
+       return (ENOENT);
+}
+
+       __checkReturn   efx_rc_t
+efx_tunnel_config_udp_add(
+       __in            efx_nic_t *enp,
+       __in            uint16_t port /* host/cpu-endian */,
+       __in            efx_tunnel_protocol_t protocol)
+{
+       const efx_nic_cfg_t *encp = &enp->en_nic_cfg;
+       efx_tunnel_cfg_t *etcp = &enp->en_tunnel_cfg;
+       efsys_lock_state_t state;
+       efx_rc_t rc;
+       unsigned int entry;
+
+       EFSYS_ASSERT3U(enp->en_mod_flags, &, EFX_MOD_TUNNEL);
+
+       if (protocol >= EFX_TUNNEL_NPROTOS) {
+               rc = EINVAL;
+               goto fail1;
+       }
+
+       if ((encp->enc_tunnel_encapsulations_supported &
+           (1u << protocol)) == 0) {
+               rc = ENOTSUP;
+               goto fail2;
+       }
+
+       EFSYS_LOCK(enp->en_eslp, state);
+
+       rc = efx_tunnel_config_find_udp_tunnel_entry(etcp, port, &entry);
+       if (rc == 0) {
+               rc = EEXIST;
+               goto fail3;
+       }
+
+       if (etcp->etc_udp_entries_num ==
+           encp->enc_tunnel_config_udp_entries_max) {
+               rc = ENOSPC;
+               goto fail4;
+       }
+
+       etcp->etc_udp_entries[etcp->etc_udp_entries_num].etue_port = port;
+       etcp->etc_udp_entries[etcp->etc_udp_entries_num].etue_protocol =
+           protocol;
+
+       etcp->etc_udp_entries_num++;
+
+       EFSYS_UNLOCK(enp->en_eslp, state);
+
+       return (0);
+
+fail4:
+       EFSYS_PROBE(fail4);
+
+fail3:
+       EFSYS_PROBE(fail3);
+       EFSYS_UNLOCK(enp->en_eslp, state);
+
+fail2:
+       EFSYS_PROBE(fail2);
+
+fail1:
+       EFSYS_PROBE1(fail1, efx_rc_t, rc);
+
+       return (rc);
+}
+
+       __checkReturn   efx_rc_t
+efx_tunnel_config_udp_remove(
+       __in            efx_nic_t *enp,
+       __in            uint16_t port /* host/cpu-endian */,
+       __in            efx_tunnel_protocol_t protocol)
+{
+       efx_tunnel_cfg_t *etcp = &enp->en_tunnel_cfg;
+       efsys_lock_state_t state;
+       unsigned int entry;
+       efx_rc_t rc;
+
+       EFSYS_ASSERT3U(enp->en_mod_flags, &, EFX_MOD_TUNNEL);
+
+       EFSYS_LOCK(enp->en_eslp, state);
+
+       rc = efx_tunnel_config_find_udp_tunnel_entry(etcp, port, &entry);
+       if (rc != 0)
+               goto fail1;
+
+       if (etcp->etc_udp_entries[entry].etue_protocol != protocol) {
+               rc = EINVAL;
+               goto fail2;
+       }
+
+       EFSYS_ASSERT3U(etcp->etc_udp_entries_num, >, 0);
+       etcp->etc_udp_entries_num--;
+
+       if (entry < etcp->etc_udp_entries_num) {
+               memmove(&etcp->etc_udp_entries[entry],
+                   &etcp->etc_udp_entries[entry + 1],
+                   (etcp->etc_udp_entries_num - entry) *
+                   sizeof (etcp->etc_udp_entries[0]));
+       }
+
+       memset(&etcp->etc_udp_entries[etcp->etc_udp_entries_num], 0,
+           sizeof (etcp->etc_udp_entries[0]));
+
+       EFSYS_UNLOCK(enp->en_eslp, state);
+
+       return (0);
+
+fail2:
+       EFSYS_PROBE(fail2);
+
+fail1:
+       EFSYS_PROBE1(fail1, efx_rc_t, rc);
+       EFSYS_UNLOCK(enp->en_eslp, state);
+
+       return (rc);
+}
+
+                       void
+efx_tunnel_config_clear(
+       __in                    efx_nic_t *enp)
+{
+       efx_tunnel_cfg_t *etcp = &enp->en_tunnel_cfg;
+       efsys_lock_state_t state;
+
+       EFSYS_ASSERT3U(enp->en_mod_flags, &, EFX_MOD_TUNNEL);
+
+       EFSYS_LOCK(enp->en_eslp, state);
+
+       etcp->etc_udp_entries_num = 0;
+       memset(etcp->etc_udp_entries, 0, sizeof (etcp->etc_udp_entries));
+
+       EFSYS_UNLOCK(enp->en_eslp, state);
+}
+
+       __checkReturn   efx_rc_t
+efx_tunnel_reconfigure(
+       __in            efx_nic_t *enp)
+{
+       const efx_tunnel_ops_t *etop = enp->en_etop;
+       efx_rc_t rc;
+
+       EFSYS_ASSERT3U(enp->en_mod_flags, &, EFX_MOD_TUNNEL);
+
+       if (etop->eto_reconfigure == NULL) {
+               rc = ENOTSUP;
+               goto fail1;
+       }
+
+       if ((rc = enp->en_etop->eto_reconfigure(enp)) != 0)
+               goto fail2;
+
+       return (0);
+
+fail2:
+       EFSYS_PROBE(fail2);
+
+fail1:
+       EFSYS_PROBE1(fail1, efx_rc_t, rc);
+
+       return (rc);
+}
+
+#if EFSYS_OPT_MEDFORD
+static __checkReturn           boolean_t
+medford_udp_encap_supported(
+       __in            efx_nic_t *enp)
+{
+       const efx_nic_cfg_t *encp = &enp->en_nic_cfg;
+       uint32_t udp_tunnels_mask = 0;
+
+       udp_tunnels_mask |= (1u << EFX_TUNNEL_PROTOCOL_VXLAN);
+       udp_tunnels_mask |= (1u << EFX_TUNNEL_PROTOCOL_GENEVE);
+
+       return ((encp->enc_tunnel_encapsulations_supported &
+           udp_tunnels_mask) == 0 ? B_FALSE : B_TRUE);
+}
+
+static __checkReturn   efx_rc_t
+medford_tunnel_reconfigure(
+       __in            efx_nic_t *enp)
+{
+       efx_tunnel_cfg_t *etcp = &enp->en_tunnel_cfg;
+       efx_rc_t rc;
+       boolean_t resetting;
+       efsys_lock_state_t state;
+       efx_tunnel_cfg_t etc;
+
+       EFSYS_LOCK(enp->en_eslp, state);
+       memcpy(&etc, etcp, sizeof (etc));
+       EFSYS_UNLOCK(enp->en_eslp, state);
+
+       if (medford_udp_encap_supported(enp) == B_FALSE) {
+               /*
+                * It is OK to apply empty UDP tunnel ports when UDP
+                * tunnel encapsulations are not supported - just nothing
+                * should be done.
+                */
+               if (etc.etc_udp_entries_num == 0)
+                       return (0);
+               rc = ENOTSUP;
+               goto fail1;
+       } else {
+               /*
+                * All PCI functions can see a reset upon the
+                * MCDI request completion
+                */
+               rc = efx_mcdi_set_tunnel_encap_udp_ports(enp, &etc, B_FALSE,
+                   &resetting);
+               if (rc != 0)
+                       goto fail2;
+
+               /*
+                * Although the caller should be able to handle MC reboot,
+                * it might come in handy to report the impending reboot
+                * by returning EAGAIN
+                */
+               return ((resetting) ? EAGAIN : 0);
+       }
+fail2:
+       EFSYS_PROBE(fail2);
+
+fail1:
+       EFSYS_PROBE1(fail1, efx_rc_t, rc);
+
+       return (rc);
+}
+#endif /* EFSYS_OPT_MEDFORD */
+
+#endif /* EFSYS_OPT_TUNNEL */
index f428b62..ba2ee9a 100644 (file)
@@ -214,6 +214,8 @@ prefetch_read_once(const volatile void *addr)
 
 #define EFSYS_OPT_RX_PACKED_STREAM 0
 
+#define EFSYS_OPT_TUNNEL 0
+
 /* ID */
 
 typedef struct __efsys_identifier_s efsys_identifier_t;