@@ -210,6 +210,11 @@ Linux AF_PACKET
M: John W. Linville <linville@tuxdriver.com>
F: drivers/net/af_packet/
+Chelsio cxgbe
+M: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
+F: drivers/net/cxgbe/
+F: doc/guides/nics/cxgbe.rst
+
Cisco enic
F: drivers/net/enic/
@@ -208,6 +208,16 @@ CONFIG_RTE_LIBRTE_MLX4_TX_MP_CACHE=8
CONFIG_RTE_LIBRTE_MLX4_SOFT_COUNTERS=1
#
+# Compile burst-oriented Chelsio Terminator 10GbE/40GbE (CXGBE) PMD
+#
+CONFIG_RTE_LIBRTE_CXGBE_PMD=y
+CONFIG_RTE_LIBRTE_CXGBE_DEBUG=n
+CONFIG_RTE_LIBRTE_CXGBE_DEBUG_REG=n
+CONFIG_RTE_LIBRTE_CXGBE_DEBUG_MBOX=n
+CONFIG_RTE_LIBRTE_CXGBE_DEBUG_TX=n
+CONFIG_RTE_LIBRTE_CXGBE_DEBUG_RX=n
+
+#
# Compile burst-oriented Cisco ENIC PMD driver
#
CONFIG_RTE_LIBRTE_ENIC_PMD=y
@@ -33,6 +33,7 @@ include $(RTE_SDK)/mk/rte.vars.mk
DIRS-$(CONFIG_RTE_LIBRTE_PMD_AF_PACKET) += af_packet
DIRS-$(CONFIG_RTE_LIBRTE_PMD_BOND) += bonding
+DIRS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += cxgbe
DIRS-$(CONFIG_RTE_LIBRTE_E1000_PMD) += e1000
DIRS-$(CONFIG_RTE_LIBRTE_ENIC_PMD) += enic
DIRS-$(CONFIG_RTE_LIBRTE_FM10K_PMD) += fm10k
new file mode 100644
@@ -0,0 +1,78 @@
+# BSD LICENSE
+#
+# Copyright(c) 2014-2015 Chelsio Communications.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * 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.
+# * Neither the name of Chelsio Communications nor the names of its
+# contributors may be used to endorse or promote products derived
+# from this software without specific prior written permission.
+#
+# 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.
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# library name
+#
+LIB = librte_pmd_cxgbe.a
+
+CFLAGS += -I$(SRCDIR)/base/
+CFLAGS += -I$(SRCDIR)
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+EXPORT_MAP := rte_pmd_cxgbe_version.map
+
+LIBABIVER := 1
+
+ifeq ($(CC), icc)
+#
+# CFLAGS for icc
+#
+CFLAGS_BASE_DRIVER = -wd174 -wd593 -wd869 -wd981 -wd2259
+else
+#
+# CFLAGS for gcc
+#
+ifeq ($(shell test $(GCC_VERSION) -ge 44 && echo 1), 1)
+CFLAGS += -Wno-deprecated
+endif
+CFLAGS_BASE_DRIVER = -Wno-unused-parameter -Wno-unused-value
+CFLAGS_BASE_DRIVER += -Wno-strict-aliasing -Wno-format-extra-args
+
+endif
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += cxgbe_ethdev.c
+SRCS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += cxgbe_main.c
+SRCS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += sge.c
+SRCS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += base/t4_hw.c
+
+# this lib depends upon:
+DEPDIRS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += lib/librte_eal lib/librte_ether
+DEPDIRS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += lib/librte_mempool lib/librte_mbuf
+DEPDIRS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += lib/librte_net lib/librte_malloc
+
+include $(RTE_SDK)/mk/rte.lib.mk
new file mode 100644
@@ -0,0 +1,48 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2014-2015 Chelsio Communications.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * 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.
+ * * Neither the name of Chelsio Communications nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * 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.
+ */
+
+#ifndef _CXGBE_H_
+#define _CXGBE_H_
+
+#include "common.h"
+#include "t4_regs.h"
+
+#define CXGBE_MIN_RING_DESC_SIZE 1024 /* Min TX/RX descriptor ring size */
+#define CXGBE_MAX_RING_DESC_SIZE 4096 /* Max TX/RX descriptor ring size */
+
+#define CXGBE_DEFAULT_TX_DESC_SIZE 1024 /* Default TX ring size */
+#define CXGBE_DEFAULT_RX_DESC_SIZE 1024 /* Default RX ring size */
+
+int cxgbe_probe(struct adapter *adapter);
+
+#endif /* _CXGBE_H_ */
new file mode 100644
@@ -0,0 +1,266 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2014-2015 Chelsio Communications.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * 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.
+ * * Neither the name of Chelsio Communications nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * 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.
+ */
+
+#ifndef _CXGBE_COMPAT_H_
+#define _CXGBE_COMPAT_H_
+
+#include <string.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdarg.h>
+
+#include <rte_common.h>
+#include <rte_memcpy.h>
+#include <rte_byteorder.h>
+#include <rte_cycles.h>
+#include <rte_spinlock.h>
+#include <rte_log.h>
+
+#define dev_printf(level, fmt, args...) \
+ RTE_LOG(level, PMD, "rte_cxgbe_pmd: " fmt, ## args)
+
+#define dev_err(x, args...) dev_printf(ERR, args)
+#define dev_info(x, args...) dev_printf(INFO, args)
+#define dev_warn(x, args...) dev_printf(WARNING, args)
+
+#ifdef RTE_LIBRTE_CXGBE_DEBUG
+#define dev_debug(x, args...) dev_printf(DEBUG, args)
+#else
+#define dev_debug(x, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_CXGBE_DEBUG_REG
+#define CXGBE_DEBUG_REG(x, args...) dev_printf(DEBUG, "REG:" args)
+#else
+#define CXGBE_DEBUG_REG(x, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_CXGBE_DEBUG_MBOX
+#define CXGBE_DEBUG_MBOX(x, args...) dev_printf(DEBUG, "MBOX:" args)
+#else
+#define CXGBE_DEBUG_MBOX(x, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_CXGBE_DEBUG_TX
+#define CXGBE_DEBUG_TX(x, args...) dev_printf(DEBUG, "TX:" args)
+#else
+#define CXGBE_DEBUG_TX(x, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_CXGBE_DEBUG_RX
+#define CXGBE_DEBUG_RX(x, args...) dev_printf(DEBUG, "RX:" args)
+#else
+#define CXGBE_DEBUG_RX(x, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_CXGBE_DEBUG
+#define CXGBE_FUNC_TRACE() \
+ RTE_LOG(DEBUG, PMD, "CXGBE trace: %s\n", __func__)
+#else
+#define CXGBE_FUNC_TRACE() do { } while (0)
+#endif
+
+#define pr_err(y, args...) dev_err(0, y, ##args)
+#define pr_warn(y, args...) dev_warn(0, y, ##args)
+#define pr_info(y, args...) dev_info(0, y, ##args)
+#define BUG() pr_err("BUG at %s:%d", __func__, __LINE__)
+
+#define ASSERT(x) do {\
+ if (!(x)) \
+ rte_panic("CXGBE: x"); \
+} while (0)
+#define BUG_ON(x) ASSERT(!(x))
+
+#ifndef WARN_ON
+#define WARN_ON(x) do { \
+ int ret = !!(x); \
+ if (unlikely(ret)) \
+ pr_warn("WARN_ON: \"" #x "\" at %s:%d\n", __func__, __LINE__); \
+} while (0)
+#endif
+
+#define __iomem
+
+#ifndef BIT
+#define BIT(n) (1 << (n))
+#endif
+
+#define L1_CACHE_SHIFT 6
+#define L1_CACHE_BYTES BIT(L1_CACHE_SHIFT)
+
+#define PAGE_SHIFT 12
+#define ALIGN(x, a) (((x) + (a) - 1) & ~((a) - 1))
+#define PTR_ALIGN(p, a) ((typeof(p))ALIGN((unsigned long)(p), (a)))
+
+#define VLAN_HLEN 4
+
+#define rmb() rte_rmb() /* dpdk rte provided rmb */
+#define wmb() rte_wmb() /* dpdk rte provided wmb */
+
+typedef uint8_t u8;
+typedef int8_t s8;
+typedef uint16_t u16;
+typedef uint32_t u32;
+typedef int32_t s32;
+typedef uint64_t u64;
+typedef int bool;
+typedef uint64_t dma_addr_t;
+
+#ifndef __le16
+#define __le16 uint16_t
+#endif
+#ifndef __le32
+#define __le32 uint32_t
+#endif
+#ifndef __le64
+#define __le64 uint64_t
+#endif
+#ifndef __be16
+#define __be16 uint16_t
+#endif
+#ifndef __be32
+#define __be32 uint32_t
+#endif
+#ifndef __be64
+#define __be64 uint64_t
+#endif
+#ifndef __u8
+#define __u8 uint8_t
+#endif
+#ifndef __u16
+#define __u16 uint16_t
+#endif
+#ifndef __u32
+#define __u32 uint32_t
+#endif
+#ifndef __u64
+#define __u64 uint64_t
+#endif
+
+#define FALSE 0
+#define TRUE 1
+#define false 0
+#define true 1
+
+#define min(a, b) RTE_MIN(a, b)
+#define max(a, b) RTE_MAX(a, b)
+
+/*
+ * round up val _p to a power of 2 size _s
+ */
+#define roundup(_p, _s) (((unsigned long)(_p) + (_s - 1)) & ~(_s - 1))
+
+#undef container_of
+#define container_of(ptr, type, member) ({ \
+ typeof(((type *)0)->member)(*__mptr) = (ptr); \
+ (type *)((char *)__mptr - offsetof(type, member)); })
+
+#define ARRAY_SIZE(arr) RTE_DIM(arr)
+
+#define cpu_to_be16(o) rte_cpu_to_be_16(o)
+#define cpu_to_be32(o) rte_cpu_to_be_32(o)
+#define cpu_to_be64(o) rte_cpu_to_be_64(o)
+#define cpu_to_le32(o) rte_cpu_to_le_32(o)
+#define be16_to_cpu(o) rte_be_to_cpu_16(o)
+#define be32_to_cpu(o) rte_be_to_cpu_32(o)
+#define be64_to_cpu(o) rte_be_to_cpu_64(o)
+#define le32_to_cpu(o) rte_le_to_cpu_32(o)
+
+#define DIV_ROUND_UP(n, d) (((n) + (d) - 1) / (d))
+#define DELAY(x) rte_delay_us(x)
+#define udelay(x) DELAY(x)
+#define msleep(x) DELAY(1000 * (x))
+#define usleep_range(min, max) msleep(DIV_ROUND_UP(min, 1000))
+
+static inline uint8_t hweight32(uint32_t word32)
+{
+ uint32_t res = word32 - ((word32 >> 1) & 0x55555555);
+
+ res = (res & 0x33333333) + ((res >> 2) & 0x33333333);
+ res = (res + (res >> 4)) & 0x0F0F0F0F;
+ res = res + (res >> 8);
+ return (res + (res >> 16)) & 0x000000FF;
+
+} /* weight32 */
+
+/**
+ * fls - find last (most-significant) bit set
+ * @x: the word to search
+ *
+ * This is defined the same way as ffs.
+ * Note fls(0) = 0, fls(1) = 1, fls(0x80000000) = 32.
+ */
+static inline int fls(int x)
+{
+ return x ? sizeof(x) * 8 - __builtin_clz(x) : 0;
+}
+
+static inline unsigned long ilog2(unsigned long n)
+{
+ unsigned int e = 0;
+
+ while (n) {
+ if (n & ~((1 << 8) - 1)) {
+ e += 8;
+ n >>= 8;
+ continue;
+ }
+
+ if (n & ~((1 << 4) - 1)) {
+ e += 4;
+ n >>= 4;
+ }
+
+ for (;;) {
+ n >>= 1;
+ if (n == 0)
+ break;
+ e++;
+ }
+ }
+
+ return e;
+}
+
+static inline void writel(unsigned int val, volatile void __iomem *addr)
+{
+ *(volatile unsigned int *)addr = val;
+}
+
+static inline void writeq(u64 val, volatile void __iomem *addr)
+{
+ writel(val, addr);
+ writel(val >> 32, (void *)((uintptr_t)addr + 4));
+}
+
+#endif /* _CXGBE_COMPAT_H_ */
new file mode 100644
@@ -0,0 +1,169 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2014-2015 Chelsio Communications.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * 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.
+ * * Neither the name of Chelsio Communications nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * 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.
+ */
+
+#include <sys/queue.h>
+#include <stdio.h>
+#include <errno.h>
+#include <stdint.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdarg.h>
+#include <inttypes.h>
+#include <netinet/in.h>
+
+#include <rte_byteorder.h>
+#include <rte_common.h>
+#include <rte_cycles.h>
+#include <rte_interrupts.h>
+#include <rte_log.h>
+#include <rte_debug.h>
+#include <rte_pci.h>
+#include <rte_atomic.h>
+#include <rte_branch_prediction.h>
+#include <rte_memory.h>
+#include <rte_memzone.h>
+#include <rte_tailq.h>
+#include <rte_eal.h>
+#include <rte_alarm.h>
+#include <rte_ether.h>
+#include <rte_ethdev.h>
+#include <rte_atomic.h>
+#include <rte_malloc.h>
+#include <rte_random.h>
+#include <rte_dev.h>
+
+#include "cxgbe.h"
+
+/*
+ * Macros needed to support the PCI Device ID Table ...
+ */
+#define CH_PCI_DEVICE_ID_TABLE_DEFINE_BEGIN \
+ static struct rte_pci_id cxgb4_pci_tbl[] = {
+#define CH_PCI_DEVICE_ID_FUNCTION 0x4
+
+#define PCI_VENDOR_ID_CHELSIO 0x1425
+
+#define CH_PCI_ID_TABLE_ENTRY(devid) \
+ { RTE_PCI_DEVICE(PCI_VENDOR_ID_CHELSIO, (devid)) }
+
+#define CH_PCI_DEVICE_ID_TABLE_DEFINE_END \
+ { .vendor_id = 0, } \
+ }
+
+/*
+ *... and the PCI ID Table itself ...
+ */
+#include "t4_pci_id_tbl.h"
+
+static struct eth_dev_ops cxgbe_eth_dev_ops = {
+};
+
+/*
+ * Initialize driver
+ * It returns 0 on success.
+ */
+static int eth_cxgbe_dev_init(struct rte_eth_dev *eth_dev)
+{
+ struct rte_pci_device *pci_dev;
+ struct port_info *pi = (struct port_info *)(eth_dev->data->dev_private);
+ struct adapter *adapter = NULL;
+ char name[RTE_ETH_NAME_MAX_LEN];
+ int err = 0;
+
+ CXGBE_FUNC_TRACE();
+
+ eth_dev->dev_ops = &cxgbe_eth_dev_ops;
+
+ /* for secondary processes, we don't initialise any further as primary
+ * has already done this work.
+ */
+ if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+ return 0;
+
+ pci_dev = eth_dev->pci_dev;
+ snprintf(name, sizeof(name), "cxgbeadapter%d", eth_dev->data->port_id);
+ adapter = rte_zmalloc(name, sizeof(*adapter), 0);
+ if (!adapter)
+ return -1;
+
+ adapter->use_unpacked_mode = 1;
+ adapter->regs = (void *)pci_dev->mem_resource[0].addr;
+ if (!adapter->regs) {
+ dev_err(adapter, "%s: cannot map device registers\n", __func__);
+ err = -ENOMEM;
+ goto out_free_adapter;
+ }
+ adapter->pdev = pci_dev;
+ adapter->eth_dev = eth_dev;
+ pi->adapter = adapter;
+
+ err = cxgbe_probe(adapter);
+ if (err)
+ dev_err(adapter, "%s: cxgbe probe failed with err %d\n",
+ __func__, err);
+
+out_free_adapter:
+ return err;
+}
+
+static struct eth_driver rte_cxgbe_pmd = {
+ {
+ .name = "rte_cxgbe_pmd",
+ .id_table = cxgb4_pci_tbl,
+ .drv_flags = RTE_PCI_DRV_NEED_MAPPING | RTE_PCI_DRV_INTR_LSC,
+ },
+ .eth_dev_init = eth_cxgbe_dev_init,
+ .dev_private_size = sizeof(struct port_info),
+};
+
+/*
+ * Driver initialization routine.
+ * Invoked once at EAL init time.
+ * Register itself as the [Poll Mode] Driver of PCI CXGBE devices.
+ */
+static int rte_cxgbe_pmd_init(const char *name __rte_unused,
+ const char *params __rte_unused)
+{
+ CXGBE_FUNC_TRACE();
+
+ rte_eth_driver_register(&rte_cxgbe_pmd);
+ return 0;
+}
+
+static struct rte_driver rte_cxgbe_driver = {
+ .name = "cxgbe_driver",
+ .type = PMD_PDEV,
+ .init = rte_cxgbe_pmd_init,
+};
+
+PMD_REGISTER_DRIVER(rte_cxgbe_driver);
new file mode 100644
@@ -0,0 +1,706 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2014-2015 Chelsio Communications.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * 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.
+ * * Neither the name of Chelsio Communications nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * 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.
+ */
+
+#include <sys/queue.h>
+#include <stdio.h>
+#include <errno.h>
+#include <stdint.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdarg.h>
+#include <inttypes.h>
+#include <netinet/in.h>
+
+#include <rte_byteorder.h>
+#include <rte_common.h>
+#include <rte_cycles.h>
+#include <rte_interrupts.h>
+#include <rte_log.h>
+#include <rte_debug.h>
+#include <rte_pci.h>
+#include <rte_atomic.h>
+#include <rte_branch_prediction.h>
+#include <rte_memory.h>
+#include <rte_memzone.h>
+#include <rte_tailq.h>
+#include <rte_eal.h>
+#include <rte_alarm.h>
+#include <rte_ether.h>
+#include <rte_ethdev.h>
+#include <rte_atomic.h>
+#include <rte_malloc.h>
+#include <rte_random.h>
+#include <rte_dev.h>
+
+#include "common.h"
+#include "t4_regs.h"
+#include "t4_msg.h"
+#include "cxgbe.h"
+
+static void setup_memwin(struct adapter *adap)
+{
+ u32 mem_win0_base;
+
+ /* For T5, only relative offset inside the PCIe BAR is passed */
+ mem_win0_base = MEMWIN0_BASE;
+
+ /*
+ * Set up memory window for accessing adapter memory ranges. (Read
+ * back MA register to ensure that changes propagate before we attempt
+ * to use the new values.)
+ */
+ t4_write_reg(adap,
+ PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN,
+ MEMWIN_NIC),
+ mem_win0_base | V_BIR(0) |
+ V_WINDOW(ilog2(MEMWIN0_APERTURE) - X_WINDOW_SHIFT));
+ t4_read_reg(adap,
+ PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN,
+ MEMWIN_NIC));
+}
+
+static void print_port_info(struct adapter *adap)
+{
+ int i;
+ char buf[80];
+ struct rte_pci_addr *loc = &adap->pdev->addr;
+
+ for_each_port(adap, i) {
+ const struct port_info *pi = &adap->port[i];
+ char *bufp = buf;
+
+ if (pi->link_cfg.supported & FW_PORT_CAP_SPEED_100M)
+ bufp += sprintf(bufp, "100/");
+ if (pi->link_cfg.supported & FW_PORT_CAP_SPEED_1G)
+ bufp += sprintf(bufp, "1000/");
+ if (pi->link_cfg.supported & FW_PORT_CAP_SPEED_10G)
+ bufp += sprintf(bufp, "10G/");
+ if (pi->link_cfg.supported & FW_PORT_CAP_SPEED_40G)
+ bufp += sprintf(bufp, "40G/");
+ if (bufp != buf)
+ --bufp;
+ sprintf(bufp, "BASE-%s",
+ t4_get_port_type_description(pi->port_type));
+
+ dev_info(adap,
+ " " PCI_PRI_FMT " Chelsio rev %d %s %s\n",
+ loc->domain, loc->bus, loc->devid, loc->function,
+ CHELSIO_CHIP_RELEASE(adap->params.chip), buf,
+ (adap->flags & USING_MSIX) ? " MSI-X" :
+ (adap->flags & USING_MSI) ? " MSI" : "");
+ }
+}
+
+/*
+ * Tweak configuration based on system architecture, etc. Most of these have
+ * defaults assigned to them by Firmware Configuration Files (if we're using
+ * them) but need to be explicitly set if we're using hard-coded
+ * initialization. So these are essentially common tweaks/settings for
+ * Configuration Files and hard-coded initialization ...
+ */
+static int adap_init0_tweaks(struct adapter *adapter)
+{
+ u8 rx_dma_offset;
+
+ /*
+ * Fix up various Host-Dependent Parameters like Page Size, Cache
+ * Line Size, etc. The firmware default is for a 4KB Page Size and
+ * 64B Cache Line Size ...
+ */
+ t4_fixup_host_params_compat(adapter, PAGE_SIZE, L1_CACHE_BYTES,
+ T5_LAST_REV);
+
+ /*
+ * Keep the chip default offset to deliver Ingress packets into our
+ * DMA buffers to zero
+ */
+ rx_dma_offset = 0;
+ t4_set_reg_field(adapter, A_SGE_CONTROL, V_PKTSHIFT(M_PKTSHIFT),
+ V_PKTSHIFT(rx_dma_offset));
+
+ /*
+ * Don't include the "IP Pseudo Header" in CPL_RX_PKT checksums: Linux
+ * adds the pseudo header itself.
+ */
+ t4_tp_wr_bits_indirect(adapter, A_TP_INGRESS_CONFIG,
+ F_CSUM_HAS_PSEUDO_HDR, 0);
+
+ return 0;
+}
+
+/*
+ * Attempt to initialize the adapter via a Firmware Configuration File.
+ */
+static int adap_init0_config(struct adapter *adapter, int reset)
+{
+ struct fw_caps_config_cmd caps_cmd;
+ unsigned long mtype = 0, maddr = 0;
+ u32 finiver, finicsum, cfcsum;
+ int ret;
+ int config_issued = 0;
+ int cfg_addr;
+ char config_name[20];
+
+ /*
+ * Reset device if necessary.
+ */
+ if (reset) {
+ ret = t4_fw_reset(adapter, adapter->mbox,
+ F_PIORSTMODE | F_PIORST);
+ if (ret < 0) {
+ dev_warn(adapter, "Firmware reset failed, error %d\n",
+ -ret);
+ goto bye;
+ }
+ }
+
+ cfg_addr = t4_flash_cfg_addr(adapter);
+ if (cfg_addr < 0) {
+ ret = cfg_addr;
+ dev_warn(adapter, "Finding address for firmware config file in flash failed, error %d\n",
+ -ret);
+ goto bye;
+ }
+
+ strcpy(config_name, "On Flash");
+ mtype = FW_MEMTYPE_CF_FLASH;
+ maddr = cfg_addr;
+
+ /*
+ * Issue a Capability Configuration command to the firmware to get it
+ * to parse the Configuration File. We don't use t4_fw_config_file()
+ * because we want the ability to modify various features after we've
+ * processed the configuration file ...
+ */
+ memset(&caps_cmd, 0, sizeof(caps_cmd));
+ caps_cmd.op_to_write = cpu_to_be32(V_FW_CMD_OP(FW_CAPS_CONFIG_CMD) |
+ F_FW_CMD_REQUEST | F_FW_CMD_READ);
+ caps_cmd.cfvalid_to_len16 =
+ cpu_to_be32(F_FW_CAPS_CONFIG_CMD_CFVALID |
+ V_FW_CAPS_CONFIG_CMD_MEMTYPE_CF(mtype) |
+ V_FW_CAPS_CONFIG_CMD_MEMADDR64K_CF(maddr >> 16) |
+ FW_LEN16(caps_cmd));
+ ret = t4_wr_mbox(adapter, adapter->mbox, &caps_cmd, sizeof(caps_cmd),
+ &caps_cmd);
+ /*
+ * If the CAPS_CONFIG failed with an ENOENT (for a Firmware
+ * Configuration File in FLASH), our last gasp effort is to use the
+ * Firmware Configuration File which is embedded in the firmware. A
+ * very few early versions of the firmware didn't have one embedded
+ * but we can ignore those.
+ */
+ if (ret == -ENOENT) {
+ dev_info(adapter, "%s: Going for embedded config in firmware..\n",
+ __func__);
+
+ memset(&caps_cmd, 0, sizeof(caps_cmd));
+ caps_cmd.op_to_write =
+ cpu_to_be32(V_FW_CMD_OP(FW_CAPS_CONFIG_CMD) |
+ F_FW_CMD_REQUEST | F_FW_CMD_READ);
+ caps_cmd.cfvalid_to_len16 = cpu_to_be32(FW_LEN16(caps_cmd));
+ ret = t4_wr_mbox(adapter, adapter->mbox, &caps_cmd,
+ sizeof(caps_cmd), &caps_cmd);
+ strcpy(config_name, "Firmware Default");
+ }
+
+ config_issued = 1;
+ if (ret < 0)
+ goto bye;
+
+ finiver = be32_to_cpu(caps_cmd.finiver);
+ finicsum = be32_to_cpu(caps_cmd.finicsum);
+ cfcsum = be32_to_cpu(caps_cmd.cfcsum);
+ if (finicsum != cfcsum)
+ dev_warn(adapter, "Configuration File checksum mismatch: [fini] csum=%#x, computed csum=%#x\n",
+ finicsum, cfcsum);
+
+ /*
+ * If we're a pure NIC driver then disable all offloading facilities.
+ * This will allow the firmware to optimize aspects of the hardware
+ * configuration which will result in improved performance.
+ */
+ caps_cmd.niccaps &= cpu_to_be16(~(FW_CAPS_CONFIG_NIC_HASHFILTER |
+ FW_CAPS_CONFIG_NIC_ETHOFLD));
+ caps_cmd.toecaps = 0;
+ caps_cmd.iscsicaps = 0;
+ caps_cmd.rdmacaps = 0;
+ caps_cmd.fcoecaps = 0;
+
+ /*
+ * And now tell the firmware to use the configuration we just loaded.
+ */
+ caps_cmd.op_to_write = cpu_to_be32(V_FW_CMD_OP(FW_CAPS_CONFIG_CMD) |
+ F_FW_CMD_REQUEST | F_FW_CMD_WRITE);
+ caps_cmd.cfvalid_to_len16 = htonl(FW_LEN16(caps_cmd));
+ ret = t4_wr_mbox(adapter, adapter->mbox, &caps_cmd, sizeof(caps_cmd),
+ NULL);
+ if (ret < 0) {
+ dev_warn(adapter, "Unable to finalize Firmware Capabilities %d\n",
+ -ret);
+ goto bye;
+ }
+
+ /*
+ * Tweak configuration based on system architecture, etc.
+ */
+ ret = adap_init0_tweaks(adapter);
+ if (ret < 0) {
+ dev_warn(adapter, "Unable to do init0-tweaks %d\n", -ret);
+ goto bye;
+ }
+
+ /*
+ * And finally tell the firmware to initialize itself using the
+ * parameters from the Configuration File.
+ */
+ ret = t4_fw_initialize(adapter, adapter->mbox);
+ if (ret < 0) {
+ dev_warn(adapter, "Initializing Firmware failed, error %d\n",
+ -ret);
+ goto bye;
+ }
+
+ /*
+ * Return successfully and note that we're operating with parameters
+ * not supplied by the driver, rather than from hard-wired
+ * initialization constants burried in the driver.
+ */
+ dev_info(adapter,
+ "Successfully configured using Firmware Configuration File \"%s\", version %#x, computed checksum %#x\n",
+ config_name, finiver, cfcsum);
+
+ return 0;
+
+ /*
+ * Something bad happened. Return the error ... (If the "error"
+ * is that there's no Configuration File on the adapter we don't
+ * want to issue a warning since this is fairly common.)
+ */
+bye:
+ if (config_issued && ret != -ENOENT)
+ dev_warn(adapter, "\"%s\" configuration file error %d\n",
+ config_name, -ret);
+
+ dev_debug(adapter, "%s: returning ret = %d ..\n", __func__, ret);
+ return ret;
+}
+
+static int adap_init0(struct adapter *adap)
+{
+ int ret = 0;
+ u32 v, port_vec;
+ enum dev_state state;
+ u32 params[7], val[7];
+ int reset = 1;
+ int mbox = adap->mbox;
+
+ /*
+ * Contact FW, advertising Master capability.
+ */
+ ret = t4_fw_hello(adap, adap->mbox, adap->mbox, MASTER_MAY, &state);
+ if (ret < 0) {
+ dev_err(adap, "%s: could not connect to FW, error %d\n",
+ __func__, -ret);
+ goto bye;
+ }
+
+ CXGBE_DEBUG_MBOX(adap, "%s: adap->mbox = %d; ret = %d\n", __func__,
+ adap->mbox, ret);
+
+ if (ret == mbox)
+ adap->flags |= MASTER_PF;
+
+ if (state == DEV_STATE_INIT) {
+ /*
+ * Force halt and reset FW because a previous instance may have
+ * exited abnormally without properly shutting down
+ */
+ ret = t4_fw_halt(adap, adap->mbox, reset);
+ if (ret < 0) {
+ dev_err(adap, "Failed to halt. Exit.\n");
+ goto bye;
+ }
+
+ ret = t4_fw_restart(adap, adap->mbox, reset);
+ if (ret < 0) {
+ dev_err(adap, "Failed to restart. Exit.\n");
+ goto bye;
+ }
+ state &= ~DEV_STATE_INIT;
+ }
+
+ t4_get_fw_version(adap, &adap->params.fw_vers);
+ t4_get_tp_version(adap, &adap->params.tp_vers);
+
+ dev_info(adap, "fw: %u.%u.%u.%u, TP: %u.%u.%u.%u\n",
+ G_FW_HDR_FW_VER_MAJOR(adap->params.fw_vers),
+ G_FW_HDR_FW_VER_MINOR(adap->params.fw_vers),
+ G_FW_HDR_FW_VER_MICRO(adap->params.fw_vers),
+ G_FW_HDR_FW_VER_BUILD(adap->params.fw_vers),
+ G_FW_HDR_FW_VER_MAJOR(adap->params.tp_vers),
+ G_FW_HDR_FW_VER_MINOR(adap->params.tp_vers),
+ G_FW_HDR_FW_VER_MICRO(adap->params.tp_vers),
+ G_FW_HDR_FW_VER_BUILD(adap->params.tp_vers));
+
+ ret = t4_get_core_clock(adap, &adap->params.vpd);
+ if (ret < 0) {
+ dev_err(adap, "%s: could not get core clock, error %d\n",
+ __func__, -ret);
+ goto bye;
+ }
+
+ /*
+ * Find out what ports are available to us. Note that we need to do
+ * this before calling adap_init0_no_config() since it needs nports
+ * and portvec ...
+ */
+ v = V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DEV) |
+ V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DEV_PORTVEC);
+ ret = t4_query_params(adap, adap->mbox, adap->pf, 0, 1, &v, &port_vec);
+ if (ret < 0) {
+ dev_err(adap, "%s: failure in t4_queury_params; error = %d\n",
+ __func__, ret);
+ goto bye;
+ }
+
+ adap->params.nports = hweight32(port_vec);
+ adap->params.portvec = port_vec;
+
+ dev_debug(adap, "%s: adap->params.nports = %u\n", __func__,
+ adap->params.nports);
+
+ /*
+ * If the firmware is initialized already (and we're not forcing a
+ * master initialization), note that we're living with existing
+ * adapter parameters. Otherwise, it's time to try initializing the
+ * adapter ...
+ */
+ if (state == DEV_STATE_INIT) {
+ dev_info(adap, "Coming up as %s: Adapter already initialized\n",
+ adap->flags & MASTER_PF ? "MASTER" : "SLAVE");
+ } else {
+ dev_info(adap, "Coming up as MASTER: Initializing adapter\n");
+
+ ret = adap_init0_config(adap, reset);
+ if (ret == -ENOENT) {
+ dev_err(adap,
+ "No Configuration File present on adapter. Using hard-wired configuration parameters.\n");
+ goto bye;
+ }
+ }
+ if (ret < 0) {
+ dev_err(adap, "could not initialize adapter, error %d\n", -ret);
+ goto bye;
+ }
+
+ /*
+ * Give the SGE code a chance to pull in anything that it needs ...
+ * Note that this must be called after we retrieve our VPD parameters
+ * in order to know how to convert core ticks to seconds, etc.
+ */
+ ret = t4_sge_init(adap);
+ if (ret < 0) {
+ dev_err(adap, "t4_sge_init failed with error %d\n",
+ -ret);
+ goto bye;
+ }
+
+ /*
+ * Grab some of our basic fundamental operating parameters.
+ */
+#define FW_PARAM_DEV(param) \
+ (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DEV) | \
+ V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DEV_##param))
+
+#define FW_PARAM_PFVF(param) \
+ (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_PFVF) | \
+ V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_PFVF_##param) | \
+ V_FW_PARAMS_PARAM_Y(0) | \
+ V_FW_PARAMS_PARAM_Z(0))
+
+ /* If we're running on newer firmware, let it know that we're
+ * prepared to deal with encapsulated CPL messages. Older
+ * firmware won't understand this and we'll just get
+ * unencapsulated messages ...
+ */
+ params[0] = FW_PARAM_PFVF(CPLFW4MSG_ENCAP);
+ val[0] = 1;
+ (void)t4_set_params(adap, adap->mbox, adap->pf, 0, 1, params, val);
+
+ /*
+ * Find out whether we're allowed to use the T5+ ULPTX MEMWRITE DSGL
+ * capability. Earlier versions of the firmware didn't have the
+ * ULPTX_MEMWRITE_DSGL so we'll interpret a query failure as no
+ * permission to use ULPTX MEMWRITE DSGL.
+ */
+ if (is_t4(adap->params.chip)) {
+ adap->params.ulptx_memwrite_dsgl = false;
+ } else {
+ params[0] = FW_PARAM_DEV(ULPTX_MEMWRITE_DSGL);
+ ret = t4_query_params(adap, adap->mbox, adap->pf, 0,
+ 1, params, val);
+ adap->params.ulptx_memwrite_dsgl = (ret == 0 && val[0] != 0);
+ }
+
+ /*
+ * The MTU/MSS Table is initialized by now, so load their values. If
+ * we're initializing the adapter, then we'll make any modifications
+ * we want to the MTU/MSS Table and also initialize the congestion
+ * parameters.
+ */
+ t4_read_mtu_tbl(adap, adap->params.mtus, NULL);
+ if (state != DEV_STATE_INIT) {
+ int i;
+
+ /*
+ * The default MTU Table contains values 1492 and 1500.
+ * However, for TCP, it's better to have two values which are
+ * a multiple of 8 +/- 4 bytes apart near this popular MTU.
+ * This allows us to have a TCP Data Payload which is a
+ * multiple of 8 regardless of what combination of TCP Options
+ * are in use (always a multiple of 4 bytes) which is
+ * important for performance reasons. For instance, if no
+ * options are in use, then we have a 20-byte IP header and a
+ * 20-byte TCP header. In this case, a 1500-byte MSS would
+ * result in a TCP Data Payload of 1500 - 40 == 1460 bytes
+ * which is not a multiple of 8. So using an MSS of 1488 in
+ * this case results in a TCP Data Payload of 1448 bytes which
+ * is a multiple of 8. On the other hand, if 12-byte TCP Time
+ * Stamps have been negotiated, then an MTU of 1500 bytes
+ * results in a TCP Data Payload of 1448 bytes which, as
+ * above, is a multiple of 8 bytes ...
+ */
+ for (i = 0; i < NMTUS; i++)
+ if (adap->params.mtus[i] == 1492) {
+ adap->params.mtus[i] = 1488;
+ break;
+ }
+
+ t4_load_mtus(adap, adap->params.mtus, adap->params.a_wnd,
+ adap->params.b_wnd);
+ }
+ t4_init_sge_params(adap);
+ t4_init_tp_params(adap);
+
+ adap->params.drv_memwin = MEMWIN_NIC;
+ adap->flags |= FW_OK;
+ dev_debug(adap, "%s: returning zero..\n", __func__);
+ return 0;
+
+ /*
+ * Something bad happened. If a command timed out or failed with EIO
+ * FW does not operate within its spec or something catastrophic
+ * happened to HW/FW, stop issuing commands.
+ */
+bye:
+ if (ret != -ETIMEDOUT && ret != -EIO)
+ t4_fw_bye(adap, adap->mbox);
+ return ret;
+}
+
+/**
+ * t4_os_portmod_changed - handle port module changes
+ * @adap: the adapter associated with the module change
+ * @port_id: the port index whose module status has changed
+ *
+ * This is the OS-dependent handler for port module changes. It is
+ * invoked when a port module is removed or inserted for any OS-specific
+ * processing.
+ */
+void t4_os_portmod_changed(const struct adapter *adap, int port_id)
+{
+ static const char * const mod_str[] = {
+ NULL, "LR", "SR", "ER", "passive DA", "active DA", "LRM"
+ };
+
+ const struct port_info *pi = &adap->port[port_id];
+
+ if (pi->mod_type == FW_PORT_MOD_TYPE_NONE)
+ dev_info(adap, "Port%d: port module unplugged\n", pi->port_id);
+ else if (pi->mod_type < ARRAY_SIZE(mod_str))
+ dev_info(adap, "Port%d: %s port module inserted\n", pi->port_id,
+ mod_str[pi->mod_type]);
+ else if (pi->mod_type == FW_PORT_MOD_TYPE_NOTSUPPORTED)
+ dev_info(adap, "Port%d: unsupported optical port module inserted\n",
+ pi->port_id);
+ else if (pi->mod_type == FW_PORT_MOD_TYPE_UNKNOWN)
+ dev_info(adap, "Port%d: unknown port module inserted, forcing TWINAX\n",
+ pi->port_id);
+ else if (pi->mod_type == FW_PORT_MOD_TYPE_ERROR)
+ dev_info(adap, "Port%d: transceiver module error\n",
+ pi->port_id);
+ else
+ dev_info(adap, "Port%d: unknown module type %d inserted\n",
+ pi->port_id, pi->mod_type);
+}
+
+int cxgbe_probe(struct adapter *adapter)
+{
+ struct port_info *pi;
+ int func, i;
+ int err = 0;
+
+ func = G_SOURCEPF(t4_read_reg(adapter, A_PL_WHOAMI));
+ adapter->mbox = func;
+ adapter->pf = func;
+
+ t4_os_lock_init(&adapter->mbox_lock);
+ TAILQ_INIT(&adapter->mbox_list);
+
+ err = t4_prep_adapter(adapter);
+ if (err)
+ return err;
+
+ setup_memwin(adapter);
+ err = adap_init0(adapter);
+ if (err) {
+ dev_err(adapter, "%s: Adapter initialization failed, error %d\n",
+ __func__, err);
+ goto out_free;
+ }
+
+ if (!is_t4(adapter->params.chip)) {
+ /*
+ * The userspace doorbell BAR is split evenly into doorbell
+ * regions, each associated with an egress queue. If this
+ * per-queue region is large enough (at least UDBS_SEG_SIZE)
+ * then it can be used to submit a tx work request with an
+ * implied doorbell. Enable write combining on the BAR if
+ * there is room for such work requests.
+ */
+ int s_qpp, qpp, num_seg;
+
+ s_qpp = (S_QUEUESPERPAGEPF0 +
+ (S_QUEUESPERPAGEPF1 - S_QUEUESPERPAGEPF0) *
+ adapter->pf);
+ qpp = 1 << ((t4_read_reg(adapter,
+ A_SGE_EGRESS_QUEUES_PER_PAGE_PF) >> s_qpp)
+ & M_QUEUESPERPAGEPF0);
+ num_seg = PAGE_SIZE / UDBS_SEG_SIZE;
+ if (qpp > num_seg)
+ dev_warn(adapter, "Incorrect SGE EGRESS QUEUES_PER_PAGE configuration, continuing in debug mode\n");
+
+ adapter->bar2 = (void *)adapter->pdev->mem_resource[2].addr;
+ if (!adapter->bar2) {
+ dev_err(adapter, "cannot map device bar2 region\n");
+ err = -ENOMEM;
+ goto out_free;
+ }
+ t4_write_reg(adapter, A_SGE_STAT_CFG, V_STATSOURCE_T5(7) |
+ V_STATMODE(0));
+ }
+
+ for_each_port(adapter, i) {
+ char name[RTE_ETH_NAME_MAX_LEN];
+ struct rte_eth_dev_data *data = NULL;
+ const unsigned int numa_node = rte_socket_id();
+
+ pi = &adapter->port[i];
+ pi->adapter = adapter;
+ pi->xact_addr_filt = -1;
+ pi->port_id = i;
+
+ snprintf(name, sizeof(name), "cxgbe%d",
+ adapter->eth_dev->data->port_id + i);
+
+ if (i == 0) {
+ /* First port is already allocated by DPDK */
+ pi->eth_dev = adapter->eth_dev;
+ goto allocate_mac;
+ }
+
+ /*
+ * now do all data allocation - for eth_dev structure,
+ * and internal (private) data for the remaining ports
+ */
+
+ /* reserve an ethdev entry */
+ pi->eth_dev = rte_eth_dev_allocate(name, RTE_ETH_DEV_PCI);
+ if (!pi->eth_dev)
+ goto out_free;
+
+ data = rte_zmalloc_socket(name, sizeof(*data), 0, numa_node);
+ if (!data)
+ goto out_free;
+
+ data->port_id = adapter->eth_dev->data->port_id + i;
+
+ pi->eth_dev->data = data;
+
+allocate_mac:
+ pi->eth_dev->pci_dev = adapter->pdev;
+ pi->eth_dev->data->dev_private = pi;
+ pi->eth_dev->driver = adapter->eth_dev->driver;
+ pi->eth_dev->dev_ops = adapter->eth_dev->dev_ops;
+ TAILQ_INIT(&pi->eth_dev->link_intr_cbs);
+
+ pi->eth_dev->data->mac_addrs = rte_zmalloc(name,
+ ETHER_ADDR_LEN, 0);
+ if (!pi->eth_dev->data->mac_addrs) {
+ dev_err(adapter, "%s: Mem allocation failed for storing mac addr, aborting\n",
+ __func__);
+ err = -1;
+ goto out_free;
+ }
+ }
+
+ if (adapter->flags & FW_OK) {
+ err = t4_port_init(adapter, adapter->mbox, adapter->pf, 0);
+ if (err) {
+ dev_err(adapter, "%s: t4_port_init failed with err %d\n",
+ __func__, err);
+ goto out_free;
+ }
+ }
+
+ print_port_info(adapter);
+
+ return 0;
+
+out_free:
+ for_each_port(adapter, i) {
+ pi = adap2pinfo(adapter, i);
+ if (pi->viid != 0)
+ t4_free_vi(adapter, adapter->mbox, adapter->pf,
+ 0, pi->viid);
+ /* Skip first port since it'll be de-allocated by DPDK */
+ if (i == 0)
+ continue;
+ if (pi->eth_dev->data)
+ rte_free(pi->eth_dev->data);
+ }
+
+ if (adapter->flags & FW_OK)
+ t4_fw_bye(adapter, adapter->mbox);
+ return -err;
+}
new file mode 100644
@@ -0,0 +1,4 @@
+DPDK_2.1 {
+
+ local: *;
+};
new file mode 100644
@@ -0,0 +1,311 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2014-2015 Chelsio Communications.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * 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.
+ * * Neither the name of Chelsio Communications nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * 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.
+ */
+
+#include <linux/if_ether.h>
+#include <sys/queue.h>
+#include <stdio.h>
+#include <errno.h>
+#include <stdint.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdarg.h>
+#include <inttypes.h>
+#include <netinet/in.h>
+
+#include <rte_byteorder.h>
+#include <rte_common.h>
+#include <rte_cycles.h>
+#include <rte_interrupts.h>
+#include <rte_log.h>
+#include <rte_debug.h>
+#include <rte_pci.h>
+#include <rte_atomic.h>
+#include <rte_branch_prediction.h>
+#include <rte_memory.h>
+#include <rte_memzone.h>
+#include <rte_tailq.h>
+#include <rte_eal.h>
+#include <rte_alarm.h>
+#include <rte_ether.h>
+#include <rte_ethdev.h>
+#include <rte_atomic.h>
+#include <rte_malloc.h>
+#include <rte_random.h>
+#include <rte_dev.h>
+
+#include "common.h"
+#include "t4_regs.h"
+#include "t4_msg.h"
+#include "cxgbe.h"
+
+/*
+ * Rx buffer sizes for "usembufs" Free List buffers (one ingress packet
+ * per mbuf buffer). We currently only support two sizes for 1500- and
+ * 9000-byte MTUs. We could easily support more but there doesn't seem to be
+ * much need for that ...
+ */
+#define FL_MTU_SMALL 1500
+#define FL_MTU_LARGE 9000
+
+static inline unsigned int fl_mtu_bufsize(struct adapter *adapter,
+ unsigned int mtu)
+{
+ struct sge *s = &adapter->sge;
+
+ return ALIGN(s->pktshift + ETH_HLEN + VLAN_HLEN + mtu, s->fl_align);
+}
+
+#define FL_MTU_SMALL_BUFSIZE(adapter) fl_mtu_bufsize(adapter, FL_MTU_SMALL)
+#define FL_MTU_LARGE_BUFSIZE(adapter) fl_mtu_bufsize(adapter, FL_MTU_LARGE)
+
+/*
+ * Bits 0..3 of rx_sw_desc.dma_addr have special meaning. The hardware uses
+ * these to specify the buffer size as an index into the SGE Free List Buffer
+ * Size register array. We also use bit 4, when the buffer has been unmapped
+ * for DMA, but this is of course never sent to the hardware and is only used
+ * to prevent double unmappings. All of the above requires that the Free List
+ * Buffers which we allocate have the bottom 5 bits free (0) -- i.e. are
+ * 32-byte or or a power of 2 greater in alignment. Since the SGE's minimal
+ * Free List Buffer alignment is 32 bytes, this works out for us ...
+ */
+enum {
+ RX_BUF_FLAGS = 0x1f, /* bottom five bits are special */
+ RX_BUF_SIZE = 0x0f, /* bottom three bits are for buf sizes */
+ RX_UNMAPPED_BUF = 0x10, /* buffer is not mapped */
+
+ /*
+ * XXX We shouldn't depend on being able to use these indices.
+ * XXX Especially when some other Master PF has initialized the
+ * XXX adapter or we use the Firmware Configuration File. We
+ * XXX should really search through the Host Buffer Size register
+ * XXX array for the appropriately sized buffer indices.
+ */
+ RX_SMALL_PG_BUF = 0x0, /* small (PAGE_SIZE) page buffer */
+ RX_LARGE_PG_BUF = 0x1, /* buffer large page buffer */
+
+ RX_SMALL_MTU_BUF = 0x2, /* small MTU buffer */
+ RX_LARGE_MTU_BUF = 0x3, /* large MTU buffer */
+};
+
+/**
+ * t4_sge_init - initialize SGE
+ * @adap: the adapter
+ *
+ * Performs SGE initialization needed every time after a chip reset.
+ * We do not initialize any of the queues here, instead the driver
+ * top-level must request those individually.
+ *
+ * Called in two different modes:
+ *
+ * 1. Perform actual hardware initialization and record hard-coded
+ * parameters which were used. This gets used when we're the
+ * Master PF and the Firmware Configuration File support didn't
+ * work for some reason.
+ *
+ * 2. We're not the Master PF or initialization was performed with
+ * a Firmware Configuration File. In this case we need to grab
+ * any of the SGE operating parameters that we need to have in
+ * order to do our job and make sure we can live with them ...
+ */
+static int t4_sge_init_soft(struct adapter *adap)
+{
+ struct sge *s = &adap->sge;
+ u32 fl_small_pg, fl_large_pg, fl_small_mtu, fl_large_mtu;
+ u32 timer_value_0_and_1, timer_value_2_and_3, timer_value_4_and_5;
+ u32 ingress_rx_threshold;
+
+ /*
+ * Verify that CPL messages are going to the Ingress Queue for
+ * process_responses() and that only packet data is going to the
+ * Free Lists.
+ */
+ if ((t4_read_reg(adap, A_SGE_CONTROL) & F_RXPKTCPLMODE) !=
+ V_RXPKTCPLMODE(X_RXPKTCPLMODE_SPLIT)) {
+ dev_err(adap, "bad SGE CPL MODE\n");
+ return -EINVAL;
+ }
+
+ /*
+ * Validate the Host Buffer Register Array indices that we want to
+ * use ...
+ *
+ * XXX Note that we should really read through the Host Buffer Size
+ * XXX register array and find the indices of the Buffer Sizes which
+ * XXX meet our needs!
+ */
+#define READ_FL_BUF(x) \
+ t4_read_reg(adap, A_SGE_FL_BUFFER_SIZE0 + (x) * sizeof(u32))
+
+ fl_small_pg = READ_FL_BUF(RX_SMALL_PG_BUF);
+ fl_large_pg = READ_FL_BUF(RX_LARGE_PG_BUF);
+ fl_small_mtu = READ_FL_BUF(RX_SMALL_MTU_BUF);
+ fl_large_mtu = READ_FL_BUF(RX_LARGE_MTU_BUF);
+
+ /*
+ * We only bother using the Large Page logic if the Large Page Buffer
+ * is larger than our Page Size Buffer.
+ */
+ if (fl_large_pg <= fl_small_pg)
+ fl_large_pg = 0;
+
+#undef READ_FL_BUF
+
+ /*
+ * The Page Size Buffer must be exactly equal to our Page Size and the
+ * Large Page Size Buffer should be 0 (per above) or a power of 2.
+ */
+ if (fl_small_pg != PAGE_SIZE ||
+ (fl_large_pg & (fl_large_pg - 1)) != 0) {
+ dev_err(adap, "bad SGE FL page buffer sizes [%d, %d]\n",
+ fl_small_pg, fl_large_pg);
+ return -EINVAL;
+ }
+ if (fl_large_pg)
+ s->fl_pg_order = ilog2(fl_large_pg) - PAGE_SHIFT;
+
+ if (adap->use_unpacked_mode) {
+ int err = 0;
+
+ if (fl_small_mtu < FL_MTU_SMALL_BUFSIZE(adap)) {
+ dev_err(adap, "bad SGE FL small MTU %d\n",
+ fl_small_mtu);
+ err = -EINVAL;
+ }
+ if (fl_large_mtu < FL_MTU_LARGE_BUFSIZE(adap)) {
+ dev_err(adap, "bad SGE FL large MTU %d\n",
+ fl_large_mtu);
+ err = -EINVAL;
+ }
+ if (err)
+ return err;
+ }
+
+ /*
+ * Retrieve our RX interrupt holdoff timer values and counter
+ * threshold values from the SGE parameters.
+ */
+ timer_value_0_and_1 = t4_read_reg(adap, A_SGE_TIMER_VALUE_0_AND_1);
+ timer_value_2_and_3 = t4_read_reg(adap, A_SGE_TIMER_VALUE_2_AND_3);
+ timer_value_4_and_5 = t4_read_reg(adap, A_SGE_TIMER_VALUE_4_AND_5);
+ s->timer_val[0] = core_ticks_to_us(adap,
+ G_TIMERVALUE0(timer_value_0_and_1));
+ s->timer_val[1] = core_ticks_to_us(adap,
+ G_TIMERVALUE1(timer_value_0_and_1));
+ s->timer_val[2] = core_ticks_to_us(adap,
+ G_TIMERVALUE2(timer_value_2_and_3));
+ s->timer_val[3] = core_ticks_to_us(adap,
+ G_TIMERVALUE3(timer_value_2_and_3));
+ s->timer_val[4] = core_ticks_to_us(adap,
+ G_TIMERVALUE4(timer_value_4_and_5));
+ s->timer_val[5] = core_ticks_to_us(adap,
+ G_TIMERVALUE5(timer_value_4_and_5));
+
+ ingress_rx_threshold = t4_read_reg(adap, A_SGE_INGRESS_RX_THRESHOLD);
+ s->counter_val[0] = G_THRESHOLD_0(ingress_rx_threshold);
+ s->counter_val[1] = G_THRESHOLD_1(ingress_rx_threshold);
+ s->counter_val[2] = G_THRESHOLD_2(ingress_rx_threshold);
+ s->counter_val[3] = G_THRESHOLD_3(ingress_rx_threshold);
+
+ return 0;
+}
+
+int t4_sge_init(struct adapter *adap)
+{
+ struct sge *s = &adap->sge;
+ u32 sge_control, sge_control2, sge_conm_ctrl;
+ unsigned int ingpadboundary, ingpackboundary;
+ int ret, egress_threshold;
+
+ /*
+ * Ingress Padding Boundary and Egress Status Page Size are set up by
+ * t4_fixup_host_params().
+ */
+ sge_control = t4_read_reg(adap, A_SGE_CONTROL);
+ s->pktshift = G_PKTSHIFT(sge_control);
+ s->stat_len = (sge_control & F_EGRSTATUSPAGESIZE) ? 128 : 64;
+
+ /*
+ * T4 uses a single control field to specify both the PCIe Padding and
+ * Packing Boundary. T5 introduced the ability to specify these
+ * separately. The actual Ingress Packet Data alignment boundary
+ * within Packed Buffer Mode is the maximum of these two
+ * specifications.
+ */
+ ingpadboundary = 1 << (G_INGPADBOUNDARY(sge_control) +
+ X_INGPADBOUNDARY_SHIFT);
+ s->fl_align = ingpadboundary;
+
+ if (!is_t4(adap->params.chip) && !adap->use_unpacked_mode) {
+ /*
+ * T5 has a weird interpretation of one of the PCIe Packing
+ * Boundary values. No idea why ...
+ */
+ sge_control2 = t4_read_reg(adap, A_SGE_CONTROL2);
+ ingpackboundary = G_INGPACKBOUNDARY(sge_control2);
+ if (ingpackboundary == X_INGPACKBOUNDARY_16B)
+ ingpackboundary = 16;
+ else
+ ingpackboundary = 1 << (ingpackboundary +
+ X_INGPACKBOUNDARY_SHIFT);
+
+ s->fl_align = max(ingpadboundary, ingpackboundary);
+ }
+
+ ret = t4_sge_init_soft(adap);
+ if (ret < 0) {
+ dev_err(adap, "%s: t4_sge_init_soft failed, error %d\n",
+ __func__, -ret);
+ return ret;
+ }
+
+ /*
+ * A FL with <= fl_starve_thres buffers is starving and a periodic
+ * timer will attempt to refill it. This needs to be larger than the
+ * SGE's Egress Congestion Threshold. If it isn't, then we can get
+ * stuck waiting for new packets while the SGE is waiting for us to
+ * give it more Free List entries. (Note that the SGE's Egress
+ * Congestion Threshold is in units of 2 Free List pointers.) For T4,
+ * there was only a single field to control this. For T5 there's the
+ * original field which now only applies to Unpacked Mode Free List
+ * buffers and a new field which only applies to Packed Mode Free List
+ * buffers.
+ */
+ sge_conm_ctrl = t4_read_reg(adap, A_SGE_CONM_CTRL);
+ if (is_t4(adap->params.chip) || adap->use_unpacked_mode)
+ egress_threshold = G_EGRTHRESHOLD(sge_conm_ctrl);
+ else
+ egress_threshold = G_EGRTHRESHOLDPACKING(sge_conm_ctrl);
+ s->fl_starve_thres = 2 * egress_threshold + 1;
+
+ return 0;
+}
@@ -122,6 +122,7 @@ ifeq ($(CONFIG_RTE_BUILD_SHARED_LIB),n)
_LDLIBS-$(CONFIG_RTE_LIBRTE_VMXNET3_PMD) += -lrte_pmd_vmxnet3_uio
_LDLIBS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += -lrte_pmd_virtio
+_LDLIBS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += -lrte_pmd_cxgbe
_LDLIBS-$(CONFIG_RTE_LIBRTE_ENIC_PMD) += -lrte_pmd_enic
_LDLIBS-$(CONFIG_RTE_LIBRTE_I40E_PMD) += -lrte_pmd_i40e
_LDLIBS-$(CONFIG_RTE_LIBRTE_FM10K_PMD) += -lrte_pmd_fm10k