freebsd-dev/sys/dev/ice/ice_iflib_txrx.c
Eric Joyner 71d104536b ice(4): Introduce new driver for Intel E800 Ethernet controllers
The ice(4) driver is the driver for the Intel E8xx series Ethernet
controllers; currently with codenames Columbiaville and
Columbia Park.

These new controllers support 100G speeds, as well as introducing
more queues, better virtualization support, and more offload
capabilities. Future work will enable virtual functions (like
in ixl(4)) and the other functionality outlined above.

For full functionality, the kernel should be compiled with
"device ice_ddp" like in the amd64 NOTES file, and/or
ice_ddp_load="YES" should be added to /boot/loader.conf so that
the DDP package file included in this commit can be downloaded
to the adapter. Otherwise, the adapter will fall back to a single
queue mode with limited functionality.

A man page for this driver will be forthcoming.

MFC after:	1 month
Relnotes:	yes
Sponsored by:	Intel Corporation
Differential Revision:	https://reviews.freebsd.org/D21959
2020-05-26 23:35:10 +00:00

402 lines
12 KiB
C

/* SPDX-License-Identifier: BSD-3-Clause */
/* Copyright (c) 2020, 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.
*/
/*$FreeBSD$*/
/**
* @file ice_iflib_txrx.c
* @brief iflib Tx/Rx hotpath
*
* Main location for the iflib Tx/Rx hotpath implementation.
*
* Contains the implementation for the iflib function callbacks and the
* if_txrx ops structure.
*/
#include "ice_iflib.h"
/* Tx/Rx hotpath utility functions */
#include "ice_common_txrx.h"
/*
* iflib txrx method declarations
*/
static int ice_ift_txd_encap(void *arg, if_pkt_info_t pi);
static int ice_ift_rxd_pkt_get(void *arg, if_rxd_info_t ri);
static void ice_ift_txd_flush(void *arg, uint16_t txqid, qidx_t pidx);
static int ice_ift_txd_credits_update(void *arg, uint16_t txqid, bool clear);
static int ice_ift_rxd_available(void *arg, uint16_t rxqid, qidx_t pidx, qidx_t budget);
static void ice_ift_rxd_flush(void *arg, uint16_t rxqid, uint8_t flidx, qidx_t pidx);
static void ice_ift_rxd_refill(void *arg, if_rxd_update_t iru);
/* Macro to help extract the NIC mode flexible Rx descriptor fields from the
* advanced 32byte Rx descriptors.
*/
#define RX_FLEX_NIC(desc, field) \
(((struct ice_32b_rx_flex_desc_nic *)desc)->field)
/**
* @var ice_txrx
* @brief Tx/Rx operations for the iflib stack
*
* Structure defining the Tx and Rx related operations that iflib can request
* the driver to perform. These are the main entry points for the hot path of
* the transmit and receive paths in the iflib driver.
*/
struct if_txrx ice_txrx = {
.ift_txd_encap = ice_ift_txd_encap,
.ift_txd_flush = ice_ift_txd_flush,
.ift_txd_credits_update = ice_ift_txd_credits_update,
.ift_rxd_available = ice_ift_rxd_available,
.ift_rxd_pkt_get = ice_ift_rxd_pkt_get,
.ift_rxd_refill = ice_ift_rxd_refill,
.ift_rxd_flush = ice_ift_rxd_flush,
};
/**
* ice_ift_txd_encap - prepare Tx descriptors for a packet
* @arg: the iflib softc structure pointer
* @pi: packet info
*
* Prepares and encapsulates the given packet into into Tx descriptors, in
* preparation for sending to the transmit engine. Sets the necessary context
* descriptors for TSO and other offloads, and prepares the last descriptor
* for the writeback status.
*
* Return 0 on success, non-zero error code on failure.
*/
static int
ice_ift_txd_encap(void *arg, if_pkt_info_t pi)
{
struct ice_softc *sc = (struct ice_softc *)arg;
struct ice_tx_queue *txq = &sc->pf_vsi.tx_queues[pi->ipi_qsidx];
int nsegs = pi->ipi_nsegs;
bus_dma_segment_t *segs = pi->ipi_segs;
struct ice_tx_desc *txd = NULL;
int i, j, mask, pidx_last;
u32 cmd, off;
cmd = off = 0;
i = pi->ipi_pidx;
/* Set up the TSO/CSUM offload */
if (pi->ipi_csum_flags & ICE_CSUM_OFFLOAD) {
/* Set up the TSO context descriptor if required */
if (pi->ipi_csum_flags & CSUM_TSO) {
if (ice_tso_detect_sparse(pi))
return (EFBIG);
i = ice_tso_setup(txq, pi);
}
ice_tx_setup_offload(txq, pi, &cmd, &off);
}
if (pi->ipi_mflags & M_VLANTAG)
cmd |= ICE_TX_DESC_CMD_IL2TAG1;
mask = txq->desc_count - 1;
for (j = 0; j < nsegs; j++) {
bus_size_t seglen;
txd = &txq->tx_base[i];
seglen = segs[j].ds_len;
txd->buf_addr = htole64(segs[j].ds_addr);
txd->cmd_type_offset_bsz =
htole64(ICE_TX_DESC_DTYPE_DATA
| ((u64)cmd << ICE_TXD_QW1_CMD_S)
| ((u64)off << ICE_TXD_QW1_OFFSET_S)
| ((u64)seglen << ICE_TXD_QW1_TX_BUF_SZ_S)
| ((u64)htole16(pi->ipi_vtag) << ICE_TXD_QW1_L2TAG1_S));
txq->stats.tx_bytes += seglen;
pidx_last = i;
i = (i+1) & mask;
}
/* Set the last descriptor for report */
#define ICE_TXD_CMD (ICE_TX_DESC_CMD_EOP | ICE_TX_DESC_CMD_RS)
txd->cmd_type_offset_bsz |=
htole64(((u64)ICE_TXD_CMD << ICE_TXD_QW1_CMD_S));
/* Add to report status array */
txq->tx_rsq[txq->tx_rs_pidx] = pidx_last;
txq->tx_rs_pidx = (txq->tx_rs_pidx+1) & mask;
MPASS(txq->tx_rs_pidx != txq->tx_rs_cidx);
pi->ipi_new_pidx = i;
++txq->stats.tx_packets;
return (0);
}
/**
* ice_ift_txd_flush - Flush Tx descriptors to hardware
* @arg: device specific softc pointer
* @txqid: the Tx queue to flush
* @pidx: descriptor index to advance tail to
*
* Advance the Transmit Descriptor Tail (TDT). This indicates to hardware that
* frames are available for transmit.
*/
static void
ice_ift_txd_flush(void *arg, uint16_t txqid, qidx_t pidx)
{
struct ice_softc *sc = (struct ice_softc *)arg;
struct ice_tx_queue *txq = &sc->pf_vsi.tx_queues[txqid];
struct ice_hw *hw = &sc->hw;
wr32(hw, txq->tail, pidx);
}
/**
* ice_ift_txd_credits_update - cleanup Tx descriptors
* @arg: device private softc
* @txqid: the Tx queue to update
* @clear: if false, only report, do not actually clean
*
* If clear is false, iflib is asking if we *could* clean up any Tx
* descriptors.
*
* If clear is true, iflib is requesting to cleanup and reclaim used Tx
* descriptors.
*/
static int
ice_ift_txd_credits_update(void *arg, uint16_t txqid, bool clear)
{
struct ice_softc *sc = (struct ice_softc *)arg;
struct ice_tx_queue *txq = &sc->pf_vsi.tx_queues[txqid];
qidx_t processed = 0;
qidx_t cur, prev, ntxd, rs_cidx;
int32_t delta;
bool is_done;
rs_cidx = txq->tx_rs_cidx;
if (rs_cidx == txq->tx_rs_pidx)
return (0);
cur = txq->tx_rsq[rs_cidx];
MPASS(cur != QIDX_INVALID);
is_done = ice_is_tx_desc_done(&txq->tx_base[cur]);
if (!is_done)
return (0);
else if (clear == false)
return (1);
prev = txq->tx_cidx_processed;
ntxd = txq->desc_count;
do {
MPASS(prev != cur);
delta = (int32_t)cur - (int32_t)prev;
if (delta < 0)
delta += ntxd;
MPASS(delta > 0);
processed += delta;
prev = cur;
rs_cidx = (rs_cidx + 1) & (ntxd-1);
if (rs_cidx == txq->tx_rs_pidx)
break;
cur = txq->tx_rsq[rs_cidx];
MPASS(cur != QIDX_INVALID);
is_done = ice_is_tx_desc_done(&txq->tx_base[cur]);
} while (is_done);
txq->tx_rs_cidx = rs_cidx;
txq->tx_cidx_processed = prev;
return (processed);
}
/**
* ice_ift_rxd_available - Return number of available Rx packets
* @arg: device private softc
* @rxqid: the Rx queue id
* @pidx: descriptor start point
* @budget: maximum Rx budget
*
* Determines how many Rx packets are available on the queue, up to a maximum
* of the given budget.
*/
static int
ice_ift_rxd_available(void *arg, uint16_t rxqid, qidx_t pidx, qidx_t budget)
{
struct ice_softc *sc = (struct ice_softc *)arg;
struct ice_rx_queue *rxq = &sc->pf_vsi.rx_queues[rxqid];
union ice_32b_rx_flex_desc *rxd;
uint16_t status0;
int cnt, i, nrxd;
nrxd = rxq->desc_count;
for (cnt = 0, i = pidx; cnt < nrxd - 1 && cnt < budget;) {
rxd = &rxq->rx_base[i];
status0 = le16toh(rxd->wb.status_error0);
if ((status0 & BIT(ICE_RX_FLEX_DESC_STATUS0_DD_S)) == 0)
break;
if (++i == nrxd)
i = 0;
if (status0 & BIT(ICE_RX_FLEX_DESC_STATUS0_EOF_S))
cnt++;
}
return (cnt);
}
/**
* ice_ift_rxd_pkt_get - Called by iflib to send data to upper layer
* @arg: device specific softc
* @ri: receive packet info
*
* This function is called by iflib, and executes in ithread context. It is
* called by iflib to obtain data which has been DMA'ed into host memory.
* Returns zero on success, and an error code on failure.
*/
static int
ice_ift_rxd_pkt_get(void *arg, if_rxd_info_t ri)
{
struct ice_softc *sc = (struct ice_softc *)arg;
struct ice_rx_queue *rxq = &sc->pf_vsi.rx_queues[ri->iri_qsidx];
union ice_32b_rx_flex_desc *cur;
u16 status0, plen, vtag, ptype;
bool eop;
size_t cidx;
int i;
cidx = ri->iri_cidx;
i = 0;
do {
/* 5 descriptor receive limit */
MPASS(i < ICE_MAX_RX_SEGS);
cur = &rxq->rx_base[cidx];
status0 = le16toh(cur->wb.status_error0);
plen = le16toh(cur->wb.pkt_len) &
ICE_RX_FLX_DESC_PKT_LEN_M;
ptype = le16toh(cur->wb.ptype_flex_flags0) &
ICE_RX_FLEX_DESC_PTYPE_M;
/* we should never be called without a valid descriptor */
MPASS((status0 & BIT(ICE_RX_FLEX_DESC_STATUS0_DD_S)) != 0);
ri->iri_len += plen;
cur->wb.status_error0 = 0;
eop = (status0 & BIT(ICE_RX_FLEX_DESC_STATUS0_EOF_S));
if (status0 & BIT(ICE_RX_FLEX_DESC_STATUS0_L2TAG1P_S))
vtag = le16toh(cur->wb.l2tag1);
else
vtag = 0;
/*
* Make sure packets with bad L2 values are discarded.
* NOTE: Only the EOP descriptor has valid error results.
*/
if (eop && (status0 & BIT(ICE_RX_FLEX_DESC_STATUS0_RXE_S))) {
rxq->stats.desc_errs++;
return (EBADMSG);
}
ri->iri_frags[i].irf_flid = 0;
ri->iri_frags[i].irf_idx = cidx;
ri->iri_frags[i].irf_len = plen;
if (++cidx == rxq->desc_count)
cidx = 0;
i++;
} while (!eop);
/* capture soft statistics for this Rx queue */
rxq->stats.rx_packets++;
rxq->stats.rx_bytes += ri->iri_len;
if ((iflib_get_ifp(sc->ctx)->if_capenable & IFCAP_RXCSUM) != 0)
ice_rx_checksum(rxq, &ri->iri_csum_flags,
&ri->iri_csum_data, status0, ptype);
ri->iri_flowid = le32toh(RX_FLEX_NIC(&cur->wb, rss_hash));
ri->iri_rsstype = ice_ptype_to_hash(ptype);
ri->iri_vtag = vtag;
ri->iri_nfrags = i;
if (vtag)
ri->iri_flags |= M_VLANTAG;
return (0);
}
/**
* ice_ift_rxd_refill - Prepare Rx descriptors for re-use by hardware
* @arg: device specific softc structure
* @iru: the Rx descriptor update structure
*
* Update the Rx descriptor indices for a given queue, assigning new physical
* addresses to the descriptors, preparing them for re-use by the hardware.
*/
static void
ice_ift_rxd_refill(void *arg, if_rxd_update_t iru)
{
struct ice_softc *sc = (struct ice_softc *)arg;
struct ice_rx_queue *rxq;
uint32_t next_pidx;
int i;
uint64_t *paddrs;
uint32_t pidx;
uint16_t qsidx, count;
paddrs = iru->iru_paddrs;
pidx = iru->iru_pidx;
qsidx = iru->iru_qsidx;
count = iru->iru_count;
rxq = &(sc->pf_vsi.rx_queues[qsidx]);
for (i = 0, next_pidx = pidx; i < count; i++) {
rxq->rx_base[next_pidx].read.pkt_addr = htole64(paddrs[i]);
if (++next_pidx == (uint32_t)rxq->desc_count)
next_pidx = 0;
}
}
/**
* ice_ift_rxd_flush - Flush Rx descriptors to hardware
* @arg: device specific softc pointer
* @rxqid: the Rx queue to flush
* @flidx: unused parameter
* @pidx: descriptor index to advance tail to
*
* Advance the Receive Descriptor Tail (RDT). This indicates to hardware that
* software is done with the descriptor and it can be recycled.
*/
static void
ice_ift_rxd_flush(void *arg, uint16_t rxqid, uint8_t flidx __unused,
qidx_t pidx)
{
struct ice_softc *sc = (struct ice_softc *)arg;
struct ice_rx_queue *rxq = &sc->pf_vsi.rx_queues[rxqid];
struct ice_hw *hw = &sc->hw;
wr32(hw, rxq->tail, pidx);
}