Copyright (c) 2001-2017, Intel Corporation
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.
3. Neither the name of the Intel Corporation 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.
/*
* Copyright (c) 2011 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Coyote Point Systems, Inc.
*
* 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 NETBSD FOUNDATION, INC. 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 FOUNDATION 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.
*/
/*
* TxDescriptors Valid Range: 64-4096 Default Value: 2048 This value is the
* number of transmit descriptors allocated by the driver. Increasing this
* value allows the driver to queue more transmits. Each descriptor is 16
* bytes. Performance tests have show the 2K value to be optimal for top
* performance.
*/
#define DEFAULT_TXD 2048
#define MAX_TXD 4096
#define MIN_TXD 64
/*
* RxDescriptors Valid Range: 64-4096 Default Value: 2048 This value is the
* number of receive descriptors allocated for each RX queue. Increasing this
* value allows the driver to buffer more incoming packets. Each descriptor
* is 16 bytes. A receive buffer is also allocated for each descriptor.
*
* Note: with 8 rings and a dual port card, it is possible to bump up
* against the system mbuf pool limit, you can tune nmbclusters
* to adjust for this.
*/
#define DEFAULT_RXD 2048
#define MAX_RXD 4096
#define MIN_RXD 64
/* Alignment for rings */
#define DBA_ALIGN 128
/*
* This is the max watchdog interval, ie. the time that can
* pass between any two TX clean operations, such only happening
* when the TX hardware is functioning.
*/
#define IXGBE_WATCHDOG (10 * hz)
/*
* This parameters control when the driver calls the routine to reclaim
* transmit descriptors.
*/
#define IXGBE_TX_CLEANUP_THRESHOLD(_a) ((_a)->num_tx_desc / 8)
#define IXGBE_TX_OP_THRESHOLD(_a) ((_a)->num_tx_desc / 32)
/* These defines are used in MTU calculations */
#define IXGBE_MAX_FRAME_SIZE 9728
#define IXGBE_MTU_HDR (ETHER_HDR_LEN + ETHER_CRC_LEN)
#define IXGBE_MTU_HDR_VLAN (ETHER_HDR_LEN + ETHER_CRC_LEN + \
ETHER_VLAN_ENCAP_LEN)
#define IXGBE_MAX_MTU (IXGBE_MAX_FRAME_SIZE - IXGBE_MTU_HDR)
#define IXGBE_MAX_MTU_VLAN (IXGBE_MAX_FRAME_SIZE - IXGBE_MTU_HDR_VLAN)
/* Flow control constants */
#define IXGBE_FC_PAUSE 0xFFFF
/*
* Used for optimizing small rx mbufs. Effort is made to keep the copy
* small and aligned for the CPU L1 cache.
*
* MHLEN is typically 168 bytes, giving us 8-byte alignment. Getting
* 32 byte alignment needed for the fast bcopy results in 8 bytes being
* wasted. Getting 64 byte alignment, which _should_ be ideal for
* modern Intel CPUs, results in 40 bytes wasted and a significant drop
* in observed efficiency of the optimization, 97.9% -> 81.8%.
*/
#define IXGBE_RX_COPY_LEN_MAX (MHLEN - ETHER_ALIGN)
/*
* Default TX WTHRESH value.
* Currently, we don't use the Tx Head Pointer Write Back function.
*/
#define IXGBE_TX_WTHRESH 5
/*
* The max number of descriptors that one packet can use is 40 - WTHRESH - 2.
* Though 82598 does not have this limit, we don't want long TX chain.
* 33 should be large enough even for 64K TSO
* (32 * 2K mbuf cluster and 1 x mbuf header).
*
* Reference: 82599-X550 datasheet 7.2.1.1 "Transmit Storage in System Memory".
*/
#define IXGBE_82599_SCATTER_MAX (40 - IXGBE_TX_WTHRESH - 2)
#define IXGBE_SCATTER_DEFAULT 33
/* Defines for printing debug information */
#define DEBUG_INIT 0
#define DEBUG_IOCTL 0
#define DEBUG_HW 0
#define INIT_DEBUGOUT(S) if (DEBUG_INIT) printf(S "\n")
#define INIT_DEBUGOUT1(S, A) if (DEBUG_INIT) printf(S "\n", A)
#define INIT_DEBUGOUT2(S, A, B) if (DEBUG_INIT) printf(S "\n", A, B)
#define IOCTL_DEBUGOUT(S) if (DEBUG_IOCTL) printf(S "\n")
#define IOCTL_DEBUGOUT1(S, A) if (DEBUG_IOCTL) printf(S "\n", A)
#define IOCTL_DEBUGOUT2(S, A, B) if (DEBUG_IOCTL) printf(S "\n", A, B)
#define HW_DEBUGOUT(S) if (DEBUG_HW) printf(S "\n")
#define HW_DEBUGOUT1(S, A) if (DEBUG_HW) printf(S "\n", A)
#define HW_DEBUGOUT2(S, A, B) if (DEBUG_HW) printf(S "\n", A, B)
/* Using 1FF (the max value), the interval is ~1.05ms */
#define IXGBE_LINK_ITR_QUANTA 0x1FF
#define IXGBE_LINK_ITR ((IXGBE_LINK_ITR_QUANTA << 3) & \
IXGBE_EITR_ITR_INT_MASK)
/************************************************************************
* vendor_info_array
*
* Contains the list of Subvendor/Subdevice IDs on
* which the driver should load.
************************************************************************/
typedef struct _ixgbe_vendor_info_t {
unsigned int vendor_id;
unsigned int device_id;
unsigned int subvendor_id;
unsigned int subdevice_id;
unsigned int index;
} ixgbe_vendor_info_t;
/* This is used to get SFP+ module data */
struct ixgbe_i2c_req {
u8 dev_addr;
u8 offset;
u8 len;
u8 data[8];
};
/*
* Driver queue struct: this is the interrupt container
* for the associated tx and rx ring.
*/
struct ix_queue {
struct ixgbe_softc *sc;
u32 msix; /* This queue's MSI-X vector */
u32 eitr_setting;
u8 me;
struct resource *res;
int busy;
struct tx_ring *txr;
struct rx_ring *rxr;
struct work wq_cookie;
void *que_si;
/* Per queue event counters */
struct evcnt irqs; /* Hardware interrupt */
struct evcnt handleq; /* software_interrupt */
struct evcnt req; /* deferred */
char namebuf[32]; /* Name for sysctl */
char evnamebuf[32]; /* Name for evcnt */
/* Lock for disabled_count and this queue's EIMS/EIMC bit */
kmutex_t dc_mtx;
/*
* disabled_count means:
* 0 : this queue is enabled
* > 0 : this queue is disabled
* the value is ixgbe_disable_queue() called count
*/
int disabled_count;
bool txrx_use_workqueue;
};
/*
* The transmit ring, one per queue
*/
struct tx_ring {
struct ixgbe_softc *sc;
kmutex_t tx_mtx;
u8 me;
u32 tail;
int busy;
union ixgbe_adv_tx_desc *tx_base;
struct ixgbe_tx_buf *tx_buffers;
struct ixgbe_dma_alloc txdma;
volatile u16 tx_avail;
u16 next_avail_desc;
u16 next_to_clean;
u16 num_desc;
ixgbe_dma_tag_t *txtag;
#if 0
char mtx_name[16]; /* NetBSD has no mutex name */
#endif
pcq_t *txr_interq;
struct work wq_cookie;
void *txr_si;
bool txr_no_space; /* Like IFF_OACTIVE */
/*
* Shadow VFTA table, this is needed because
* the real vlan filter table gets cleared during
* a soft reset and the driver needs to be able
* to repopulate it.
*/
u32 shadow_vfta[IXGBE_VFTA_SIZE];
/* Info about the interface */
int advertise; /* link speeds */
bool enable_aim; /* adaptive interrupt moderation */
int max_interrupt_rate;
int link_active; /* Use LINK_STATE_* value */
u16 max_frame_size;
u16 num_segs;
u32 link_speed;
bool link_up;
bool link_enabled;
u32 vector;
u16 dmac;
u32 phy_layer;
/* Power management-related */
bool wol_support;
u32 wufc;
/* Mbuf cluster size */
u32 rx_mbuf_sz;
bool schedule_wqs_ok;
/* Flow Director */
int fdir_reinit;
/* Admin task */
struct workqueue *admin_wq; /* Link, SFP, PHY and FDIR */
struct work admin_wc;
u_int admin_pending;
volatile u32 task_requests;
kmutex_t admin_mtx; /* lock for admin_pending, task_request */
/*
* Don't acquire this mutex while
* holding rx_mtx or tx_mtx, and
* vice versa.
*/
bool txrx_use_workqueue;
/*
* Workqueue for ixgbe_handle_que_work().
*
* que_wq's "enqueued flag" is not required, because twice
* workqueue_enqueue() for ixgbe_handle_que_work() is avoided by
* masking the queue's interrupt by EIMC. See also ixgbe_msix_que().
*/
struct workqueue *que_wq;
/* Workqueue for ixgbe_deferred_mq_start_work() */
struct workqueue *txr_wq;
percpu_t *txr_wq_enqueued;
/*
* Queues:
* This is the irq holder, it has
* and RX/TX pair or rings associated
* with it.
*/
struct ix_queue *queues;
/*
* Transmit rings
* Allocated at run time, an array of rings
*/
struct tx_ring *tx_rings;
u32 num_tx_desc;
u32 tx_process_limit;
/*
* Receive rings
* Allocated at run time, an array of rings
*/
struct rx_ring *rx_rings;
u64 active_queues;
u32 num_rx_desc;
u32 rx_process_limit;
u32 rx_copy_len;
/* Sysctl help messages; displayed with sysctl -d */
#define IXGBE_SYSCTL_DESC_ADV_SPEED \
"\nControl advertised link speed using these flags:\n" \
"\t0x1 - advertise 100M\n" \
"\t0x2 - advertise 1G\n" \
"\t0x4 - advertise 10G\n" \
"\t0x8 - advertise 10M\n" \
"\t0x10 - advertise 2.5G\n" \
"\t0x20 - advertise 5G\n\n" \
"\t5G, 2.5G, 100M and 10M are only supported on certain adapters."
#define IXGBE_SYSCTL_DESC_SET_FC \
"\nSet flow control mode using these values:\n" \
"\t0 - off\n" \
"\t1 - rx pause\n" \
"\t2 - tx pause\n" \
"\t3 - tx and rx pause"
/*
* Find the number of unrefreshed RX descriptors
*/
static __inline u16
ixgbe_rx_unrefreshed(struct rx_ring *rxr)
{
if (rxr->next_to_check > rxr->next_to_refresh)
return (rxr->next_to_check - rxr->next_to_refresh - 1);
else
return ((rxr->num_desc + rxr->next_to_check) -
rxr->next_to_refresh - 1);
}
static __inline int
ixgbe_legacy_ring_empty(struct ifnet *ifp, pcq_t *dummy)
{
UNREFERENCED_1PARAMETER(dummy);
return IFQ_IS_EMPTY(&ifp->if_snd);
}
static __inline int
ixgbe_mq_ring_empty(struct ifnet *dummy, pcq_t *interq)
{
UNREFERENCED_1PARAMETER(dummy);
return (pcq_peek(interq) == NULL);
}
/*
* This checks for a zero mac addr, something that will be likely
* unless the Admin on the Host has created one.
*/
static __inline bool
ixv_check_ether_addr(u8 *addr)
{
bool status = TRUE;
/*
* This checks the sc->recovery_mode software flag which is
* set by ixgbe_fw_recovery_mode().
*
*/
static inline bool
ixgbe_fw_recovery_mode_swflag(struct ixgbe_softc *sc)
{
return (sc->feat_en & IXGBE_FEATURE_RECOVERY_MODE) &&
atomic_load_acq_uint(&sc->recovery_mode);
}
