d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
4707bb0430
freebsd-dev/sys/dev/bnxt/bnxt_txrx.c
Vincenzo Maffione 35d8a463e8 iflib: leave only 1 receive descriptor unused 
The pidx argument of isc_rxd_flush() indicates which is the last valid
receive descriptor to be used by the NIC. However, current code has
multiple issues:
  - Intel drivers write pidx to their RDT register, which means that
    NICs will only use the descriptors up to pidx-1 (modulo ring size N),
    and won't actually use the one pointed by pidx. This does not break
    reception, but it is anyway confusing and suboptimal (the NIC will
    actually see only N-2 descriptors as available, rather than N-1).
    Other drivers (if_vmx, if_bnxt, if_mgb) adhere to this semantic).
  - The semantic used by Intel (RDT is one descriptor past the last
     valid one) is used by most (if not all) NICs, and it is also used
     on the TX side (also in iflib). Since iflib is not currently
     using this semantic for RX, it must decrement fl->ifl_pidx
     (modulo N) before calling isc_rxd_flush(), and then the
     per-driver callback implementation must increment the index
     again (to match the real semantic). This is confusing and suboptimal.
  -  The iflib refill function is also called at initialization.
     However, in case the ring size is smaller than 128 (e.g. if_mgb),
     the refill function will actually prepare all the receive
     descriptors (N), without leaving one unused, as most of NICs assume
     (e.g. to avoid RDT to overrun RDH). I can speculate that the code
     looks like this right now because this issue showed up during
     testing (e.g. with if_mgb), and it was easy to workaround by
     decrementing pidx before isc_rxd_flush().

The goal of this change is to simplify the code (removing a bunch
of instructions from the RX fast path), and to make the semantic of
isc_rxd_flush() consistent across drivers. To achieve this, we:
  - change the semantics of the pidx argument to the usual one (that
    is the index one past the last valid one), so that both iflib and
    drivers avoid the decrement/increment dance.
  - fix the initialization code to prepare at most N-1 descriptors.

Reviewed by:	markj
MFC after:	2 weeks
Differential Revision:	https://reviews.freebsd.org/D26191
2020-09-01 20:41:47 +00:00

677 lines
19 KiB
C
Raw Blame History

/*-
* Broadcom NetXtreme-C/E network driver.
*
* Copyright (c) 2016 Broadcom, All Rights Reserved.
* The term Broadcom refers to Broadcom Limited and/or its subsidiaries
*
* 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/endian.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/ethernet.h>
#include <net/iflib.h>
#include "opt_inet.h"
#include "opt_inet6.h"
#include "opt_rss.h"
#include "bnxt.h"
/*
* Function prototypes
*/
static int bnxt_isc_txd_encap(void *sc, if_pkt_info_t pi);
static void bnxt_isc_txd_flush(void *sc, uint16_t txqid, qidx_t pidx);
static int bnxt_isc_txd_credits_update(void *sc, uint16_t txqid, bool clear);
static void bnxt_isc_rxd_refill(void *sc, if_rxd_update_t iru);
/* uint16_t rxqid, uint8_t flid,
uint32_t pidx, uint64_t *paddrs, caddr_t *vaddrs, uint16_t count,
uint16_t buf_size);
*/
static void bnxt_isc_rxd_flush(void *sc, uint16_t rxqid, uint8_t flid,
qidx_t pidx);
static int bnxt_isc_rxd_available(void *sc, uint16_t rxqid, qidx_t idx,
qidx_t budget);
static int bnxt_isc_rxd_pkt_get(void *sc, if_rxd_info_t ri);
static int bnxt_intr(void *sc);
struct if_txrx bnxt_txrx = {
.ift_txd_encap = bnxt_isc_txd_encap,
.ift_txd_flush = bnxt_isc_txd_flush,
.ift_txd_credits_update = bnxt_isc_txd_credits_update,
.ift_rxd_available = bnxt_isc_rxd_available,
.ift_rxd_pkt_get = bnxt_isc_rxd_pkt_get,
.ift_rxd_refill = bnxt_isc_rxd_refill,
.ift_rxd_flush = bnxt_isc_rxd_flush,
.ift_legacy_intr = bnxt_intr
};
/*
* Device Dependent Packet Transmit and Receive Functions
*/
static const uint16_t bnxt_tx_lhint[] = {
TX_BD_SHORT_FLAGS_LHINT_LT512,
TX_BD_SHORT_FLAGS_LHINT_LT1K,
TX_BD_SHORT_FLAGS_LHINT_LT2K,
TX_BD_SHORT_FLAGS_LHINT_LT2K,
TX_BD_SHORT_FLAGS_LHINT_GTE2K,
};
static int
bnxt_isc_txd_encap(void *sc, if_pkt_info_t pi)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
struct bnxt_ring *txr = &softc->tx_rings[pi->ipi_qsidx];
struct tx_bd_long *tbd;
struct tx_bd_long_hi *tbdh;
bool need_hi = false;
uint16_t flags_type;
uint16_t lflags;
uint32_t cfa_meta;
int seg = 0;
/* If we have offloads enabled, we need to use two BDs. */
if ((pi->ipi_csum_flags & (CSUM_OFFLOAD | CSUM_TSO | CSUM_IP)) ||
pi->ipi_mflags & M_VLANTAG)
need_hi = true;
/* TODO: Devices before Cu+B1 need to not mix long and short BDs */
need_hi = true;
pi->ipi_new_pidx = pi->ipi_pidx;
tbd = &((struct tx_bd_long *)txr->vaddr)[pi->ipi_new_pidx];
pi->ipi_ndescs = 0;
/* No need to byte-swap the opaque value */
tbd->opaque = ((pi->ipi_nsegs + need_hi) << 24) | pi->ipi_new_pidx;
tbd->len = htole16(pi->ipi_segs[seg].ds_len);
tbd->addr = htole64(pi->ipi_segs[seg++].ds_addr);
flags_type = ((pi->ipi_nsegs + need_hi) <<
TX_BD_SHORT_FLAGS_BD_CNT_SFT) & TX_BD_SHORT_FLAGS_BD_CNT_MASK;
if (pi->ipi_len >= 2048)
flags_type |= TX_BD_SHORT_FLAGS_LHINT_GTE2K;
else
flags_type |= bnxt_tx_lhint[pi->ipi_len >> 9];
if (need_hi) {
flags_type |= TX_BD_LONG_TYPE_TX_BD_LONG;
pi->ipi_new_pidx = RING_NEXT(txr, pi->ipi_new_pidx);
tbdh = &((struct tx_bd_long_hi *)txr->vaddr)[pi->ipi_new_pidx];
tbdh->mss = htole16(pi->ipi_tso_segsz);
tbdh->hdr_size = htole16((pi->ipi_ehdrlen + pi->ipi_ip_hlen +
pi->ipi_tcp_hlen) >> 1);
tbdh->cfa_action = 0;
lflags = 0;
cfa_meta = 0;
if (pi->ipi_mflags & M_VLANTAG) {
/* TODO: Do we need to byte-swap the vtag here? */
cfa_meta = TX_BD_LONG_CFA_META_KEY_VLAN_TAG |
pi->ipi_vtag;
cfa_meta |= TX_BD_LONG_CFA_META_VLAN_TPID_TPID8100;
}
tbdh->cfa_meta = htole32(cfa_meta);
if (pi->ipi_csum_flags & CSUM_TSO) {
lflags |= TX_BD_LONG_LFLAGS_LSO |
TX_BD_LONG_LFLAGS_T_IPID;
}
else if(pi->ipi_csum_flags & CSUM_OFFLOAD) {
lflags |= TX_BD_LONG_LFLAGS_TCP_UDP_CHKSUM |
TX_BD_LONG_LFLAGS_IP_CHKSUM;
}
else if(pi->ipi_csum_flags & CSUM_IP) {
lflags |= TX_BD_LONG_LFLAGS_IP_CHKSUM;
}
tbdh->lflags = htole16(lflags);
}
else {
flags_type |= TX_BD_SHORT_TYPE_TX_BD_SHORT;
}
for (; seg < pi->ipi_nsegs; seg++) {
tbd->flags_type = htole16(flags_type);
pi->ipi_new_pidx = RING_NEXT(txr, pi->ipi_new_pidx);
tbd = &((struct tx_bd_long *)txr->vaddr)[pi->ipi_new_pidx];
tbd->len = htole16(pi->ipi_segs[seg].ds_len);
tbd->addr = htole64(pi->ipi_segs[seg].ds_addr);
flags_type = TX_BD_SHORT_TYPE_TX_BD_SHORT;
}
flags_type |= TX_BD_SHORT_FLAGS_PACKET_END;
tbd->flags_type = htole16(flags_type);
pi->ipi_new_pidx = RING_NEXT(txr, pi->ipi_new_pidx);
return 0;
}
static void
bnxt_isc_txd_flush(void *sc, uint16_t txqid, qidx_t pidx)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
struct bnxt_ring *tx_ring = &softc->tx_rings[txqid];
/* pidx is what we last set ipi_new_pidx to */
BNXT_TX_DB(tx_ring, pidx);
/* TODO: Cumulus+ doesn't need the double doorbell */
BNXT_TX_DB(tx_ring, pidx);
return;
}
static int
bnxt_isc_txd_credits_update(void *sc, uint16_t txqid, bool clear)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
struct bnxt_cp_ring *cpr = &softc->tx_cp_rings[txqid];
struct tx_cmpl *cmpl = (struct tx_cmpl *)cpr->ring.vaddr;
int avail = 0;
uint32_t cons = cpr->cons;
bool v_bit = cpr->v_bit;
bool last_v_bit;
uint32_t last_cons;
uint16_t type;
uint16_t err;
for (;;) {
last_cons = cons;
last_v_bit = v_bit;
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmpl[cons], v_bit))
goto done;
type = cmpl[cons].flags_type & TX_CMPL_TYPE_MASK;
switch (type) {
case TX_CMPL_TYPE_TX_L2:
err = (le16toh(cmpl[cons].errors_v) &
TX_CMPL_ERRORS_BUFFER_ERROR_MASK) >>
TX_CMPL_ERRORS_BUFFER_ERROR_SFT;
if (err)
device_printf(softc->dev,
"TX completion error %u\n", err);
/* No need to byte-swap the opaque value */
avail += cmpl[cons].opaque >> 24;
/*
* If we're not clearing, iflib only cares if there's
* at least one buffer. Don't scan the whole ring in
* this case.
*/
if (!clear)
goto done;
break;
default:
if (type & 1) {
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
if (!CMP_VALID(&cmpl[cons], v_bit))
goto done;
}
device_printf(softc->dev,
"Unhandled TX completion type %u\n", type);
break;
}
}
done:
if (clear && avail) {
cpr->cons = last_cons;
cpr->v_bit = last_v_bit;
BNXT_CP_IDX_DISABLE_DB(&cpr->ring, cpr->cons);
}
return avail;
}
static void
bnxt_isc_rxd_refill(void *sc, if_rxd_update_t iru)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
struct bnxt_ring *rx_ring;
struct rx_prod_pkt_bd *rxbd;
uint16_t type;
uint16_t i;
uint16_t rxqid;
uint16_t count, len;
uint32_t pidx;
uint8_t flid;
uint64_t *paddrs;
qidx_t *frag_idxs;
rxqid = iru->iru_qsidx;
count = iru->iru_count;
len = iru->iru_buf_size;
pidx = iru->iru_pidx;
flid = iru->iru_flidx;
paddrs = iru->iru_paddrs;
frag_idxs = iru->iru_idxs;
if (flid == 0) {
rx_ring = &softc->rx_rings[rxqid];
type = RX_PROD_PKT_BD_TYPE_RX_PROD_PKT;
}
else {
rx_ring = &softc->ag_rings[rxqid];
type = RX_PROD_AGG_BD_TYPE_RX_PROD_AGG;
}
rxbd = (void *)rx_ring->vaddr;
for (i=0; i<count; i++) {
rxbd[pidx].flags_type = htole16(type);
rxbd[pidx].len = htole16(len);
/* No need to byte-swap the opaque value */
rxbd[pidx].opaque = (((rxqid & 0xff) << 24) | (flid << 16)
| (frag_idxs[i]));
rxbd[pidx].addr = htole64(paddrs[i]);
if (++pidx == rx_ring->ring_size)
pidx = 0;
}
return;
}
static void
bnxt_isc_rxd_flush(void *sc, uint16_t rxqid, uint8_t flid,
qidx_t pidx)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
struct bnxt_ring *rx_ring;
if (flid == 0)
rx_ring = &softc->rx_rings[rxqid];
else
rx_ring = &softc->ag_rings[rxqid];
/*
* We *must* update the completion ring before updating the RX ring
* or we will overrun the completion ring and the device will wedge for
* RX.
*/
if (softc->rx_cp_rings[rxqid].cons != UINT32_MAX)
BNXT_CP_IDX_DISABLE_DB(&softc->rx_cp_rings[rxqid].ring,
softc->rx_cp_rings[rxqid].cons);
BNXT_RX_DB(rx_ring, pidx);
/* TODO: Cumulus+ doesn't need the double doorbell */
BNXT_RX_DB(rx_ring, pidx);
return;
}
static int
bnxt_isc_rxd_available(void *sc, uint16_t rxqid, qidx_t idx, qidx_t budget)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
struct bnxt_cp_ring *cpr = &softc->rx_cp_rings[rxqid];
struct rx_pkt_cmpl *rcp;
struct rx_tpa_end_cmpl *rtpae;
struct cmpl_base *cmp = (struct cmpl_base *)cpr->ring.vaddr;
int avail = 0;
uint32_t cons = cpr->cons;
bool v_bit = cpr->v_bit;
uint8_t ags;
int i;
uint16_t type;
for (;;) {
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmp[cons], v_bit))
goto cmpl_invalid;
type = le16toh(cmp[cons].type) & CMPL_BASE_TYPE_MASK;
switch (type) {
case CMPL_BASE_TYPE_RX_L2:
rcp = (void *)&cmp[cons];
ags = (rcp->agg_bufs_v1 & RX_PKT_CMPL_AGG_BUFS_MASK) >>
RX_PKT_CMPL_AGG_BUFS_SFT;
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmp[cons], v_bit))
goto cmpl_invalid;
/* Now account for all the AG completions */
for (i=0; i<ags; i++) {
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmp[cons], v_bit))
goto cmpl_invalid;
}
avail++;
break;
case CMPL_BASE_TYPE_RX_TPA_END:
rtpae = (void *)&cmp[cons];
ags = (rtpae->agg_bufs_v1 &
RX_TPA_END_CMPL_AGG_BUFS_MASK) >>
RX_TPA_END_CMPL_AGG_BUFS_SFT;
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmp[cons], v_bit))
goto cmpl_invalid;
/* Now account for all the AG completions */
for (i=0; i<ags; i++) {
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmp[cons], v_bit))
goto cmpl_invalid;
}
avail++;
break;
case CMPL_BASE_TYPE_RX_TPA_START:
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmp[cons], v_bit))
goto cmpl_invalid;
break;
case CMPL_BASE_TYPE_RX_AGG:
break;
default:
device_printf(softc->dev,
"Unhandled completion type %d on RXQ %d\n",
type, rxqid);
/* Odd completion types use two completions */
if (type & 1) {
NEXT_CP_CONS_V(&cpr->ring, cons, v_bit);
CMPL_PREFETCH_NEXT(cpr, cons);
if (!CMP_VALID(&cmp[cons], v_bit))
goto cmpl_invalid;
}
break;
}
if (avail > budget)
break;
}
cmpl_invalid:
return avail;
}
static void
bnxt_set_rsstype(if_rxd_info_t ri, uint8_t rss_hash_type)
{
uint8_t rss_profile_id;
rss_profile_id = BNXT_GET_RSS_PROFILE_ID(rss_hash_type);
switch (rss_profile_id) {
case BNXT_RSS_HASH_TYPE_TCPV4:
ri->iri_rsstype = M_HASHTYPE_RSS_TCP_IPV4;
break;
case BNXT_RSS_HASH_TYPE_UDPV4:
ri->iri_rsstype = M_HASHTYPE_RSS_UDP_IPV4;
break;
case BNXT_RSS_HASH_TYPE_IPV4:
ri->iri_rsstype = M_HASHTYPE_RSS_IPV4;
break;
case BNXT_RSS_HASH_TYPE_TCPV6:
ri->iri_rsstype = M_HASHTYPE_RSS_TCP_IPV6;
break;
case BNXT_RSS_HASH_TYPE_UDPV6:
ri->iri_rsstype = M_HASHTYPE_RSS_UDP_IPV6;
break;
case BNXT_RSS_HASH_TYPE_IPV6:
ri->iri_rsstype = M_HASHTYPE_RSS_IPV6;
break;
default:
ri->iri_rsstype = M_HASHTYPE_OPAQUE_HASH;
break;
}
}
static int
bnxt_pkt_get_l2(struct bnxt_softc *softc, if_rxd_info_t ri,
struct bnxt_cp_ring *cpr, uint16_t flags_type)
{
struct rx_pkt_cmpl *rcp;
struct rx_pkt_cmpl_hi *rcph;
struct rx_abuf_cmpl *acp;
uint32_t flags2;
uint32_t errors;
uint8_t ags;
int i;
rcp = &((struct rx_pkt_cmpl *)cpr->ring.vaddr)[cpr->cons];
/* Extract from the first 16-byte BD */
if (flags_type & RX_PKT_CMPL_FLAGS_RSS_VALID) {
ri->iri_flowid = le32toh(rcp->rss_hash);
bnxt_set_rsstype(ri, rcp->rss_hash_type);
}
else {
ri->iri_rsstype = M_HASHTYPE_NONE;
}
ags = (rcp->agg_bufs_v1 & RX_PKT_CMPL_AGG_BUFS_MASK) >>
RX_PKT_CMPL_AGG_BUFS_SFT;
ri->iri_nfrags = ags + 1;
/* No need to byte-swap the opaque value */
ri->iri_frags[0].irf_flid = (rcp->opaque >> 16) & 0xff;
ri->iri_frags[0].irf_idx = rcp->opaque & 0xffff;
ri->iri_frags[0].irf_len = le16toh(rcp->len);
ri->iri_len = le16toh(rcp->len);
/* Now the second 16-byte BD */
NEXT_CP_CONS_V(&cpr->ring, cpr->cons, cpr->v_bit);
ri->iri_cidx = RING_NEXT(&cpr->ring, ri->iri_cidx);
rcph = &((struct rx_pkt_cmpl_hi *)cpr->ring.vaddr)[cpr->cons];
flags2 = le32toh(rcph->flags2);
errors = le16toh(rcph->errors_v2);
if ((flags2 & RX_PKT_CMPL_FLAGS2_META_FORMAT_MASK) ==
RX_PKT_CMPL_FLAGS2_META_FORMAT_VLAN) {
ri->iri_flags |= M_VLANTAG;
/* TODO: Should this be the entire 16-bits? */
ri->iri_vtag = le32toh(rcph->metadata) &
(RX_PKT_CMPL_METADATA_VID_MASK | RX_PKT_CMPL_METADATA_DE |
RX_PKT_CMPL_METADATA_PRI_MASK);
}
if (flags2 & RX_PKT_CMPL_FLAGS2_IP_CS_CALC) {
ri->iri_csum_flags |= CSUM_IP_CHECKED;
if (!(errors & RX_PKT_CMPL_ERRORS_IP_CS_ERROR))
ri->iri_csum_flags |= CSUM_IP_VALID;
}
if (flags2 & (RX_PKT_CMPL_FLAGS2_L4_CS_CALC |
RX_PKT_CMPL_FLAGS2_T_L4_CS_CALC)) {
ri->iri_csum_flags |= CSUM_L4_CALC;
if (!(errors & (RX_PKT_CMPL_ERRORS_L4_CS_ERROR |
RX_PKT_CMPL_ERRORS_T_L4_CS_ERROR))) {
ri->iri_csum_flags |= CSUM_L4_VALID;
ri->iri_csum_data = 0xffff;
}
}
/* And finally the ag ring stuff. */
for (i=1; i < ri->iri_nfrags; i++) {
NEXT_CP_CONS_V(&cpr->ring, cpr->cons, cpr->v_bit);
ri->iri_cidx = RING_NEXT(&cpr->ring, ri->iri_cidx);
acp = &((struct rx_abuf_cmpl *)cpr->ring.vaddr)[cpr->cons];
/* No need to byte-swap the opaque value */
ri->iri_frags[i].irf_flid = (acp->opaque >> 16 & 0xff);
ri->iri_frags[i].irf_idx = acp->opaque & 0xffff;
ri->iri_frags[i].irf_len = le16toh(acp->len);
ri->iri_len += le16toh(acp->len);
}
return 0;
}
static int
bnxt_pkt_get_tpa(struct bnxt_softc *softc, if_rxd_info_t ri,
struct bnxt_cp_ring *cpr, uint16_t flags_type)
{
struct rx_tpa_end_cmpl *agend =
&((struct rx_tpa_end_cmpl *)cpr->ring.vaddr)[cpr->cons];
struct rx_abuf_cmpl *acp;
struct bnxt_full_tpa_start *tpas;
uint32_t flags2;
uint8_t ags;
uint8_t agg_id;
int i;
/* Get the agg_id */
agg_id = (agend->agg_id & RX_TPA_END_CMPL_AGG_ID_MASK) >>
RX_TPA_END_CMPL_AGG_ID_SFT;
tpas = &(softc->rx_rings[ri->iri_qsidx].tpa_start[agg_id]);
/* Extract from the first 16-byte BD */
if (le16toh(tpas->low.flags_type) & RX_TPA_START_CMPL_FLAGS_RSS_VALID) {
ri->iri_flowid = le32toh(tpas->low.rss_hash);
bnxt_set_rsstype(ri, tpas->low.rss_hash_type);
}
else {
ri->iri_rsstype = M_HASHTYPE_NONE;
}
ags = (agend->agg_bufs_v1 & RX_TPA_END_CMPL_AGG_BUFS_MASK) >>
RX_TPA_END_CMPL_AGG_BUFS_SFT;
ri->iri_nfrags = ags + 1;
/* No need to byte-swap the opaque value */
ri->iri_frags[0].irf_flid = ((tpas->low.opaque >> 16) & 0xff);
ri->iri_frags[0].irf_idx = (tpas->low.opaque & 0xffff);
ri->iri_frags[0].irf_len = le16toh(tpas->low.len);
ri->iri_len = le16toh(tpas->low.len);
/* Now the second 16-byte BD */
NEXT_CP_CONS_V(&cpr->ring, cpr->cons, cpr->v_bit);
ri->iri_cidx = RING_NEXT(&cpr->ring, ri->iri_cidx);
flags2 = le32toh(tpas->high.flags2);
if ((flags2 & RX_TPA_START_CMPL_FLAGS2_META_FORMAT_MASK) ==
RX_TPA_START_CMPL_FLAGS2_META_FORMAT_VLAN) {
ri->iri_flags |= M_VLANTAG;
/* TODO: Should this be the entire 16-bits? */
ri->iri_vtag = le32toh(tpas->high.metadata) &
(RX_TPA_START_CMPL_METADATA_VID_MASK |
RX_TPA_START_CMPL_METADATA_DE |
RX_TPA_START_CMPL_METADATA_PRI_MASK);
}
if (flags2 & RX_TPA_START_CMPL_FLAGS2_IP_CS_CALC) {
ri->iri_csum_flags |= CSUM_IP_CHECKED;
ri->iri_csum_flags |= CSUM_IP_VALID;
}
if (flags2 & RX_TPA_START_CMPL_FLAGS2_L4_CS_CALC) {
ri->iri_csum_flags |= CSUM_L4_CALC;
ri->iri_csum_flags |= CSUM_L4_VALID;
ri->iri_csum_data = 0xffff;
}
/* Now the ag ring stuff. */
for (i=1; i < ri->iri_nfrags; i++) {
NEXT_CP_CONS_V(&cpr->ring, cpr->cons, cpr->v_bit);
ri->iri_cidx = RING_NEXT(&cpr->ring, ri->iri_cidx);
acp = &((struct rx_abuf_cmpl *)cpr->ring.vaddr)[cpr->cons];
/* No need to byte-swap the opaque value */
ri->iri_frags[i].irf_flid = ((acp->opaque >> 16) & 0xff);
ri->iri_frags[i].irf_idx = (acp->opaque & 0xffff);
ri->iri_frags[i].irf_len = le16toh(acp->len);
ri->iri_len += le16toh(acp->len);
}
/* And finally, the empty BD at the end... */
ri->iri_nfrags++;
/* No need to byte-swap the opaque value */
ri->iri_frags[i].irf_flid = ((agend->opaque >> 16) & 0xff);
ri->iri_frags[i].irf_idx = (agend->opaque & 0xffff);
ri->iri_frags[i].irf_len = le16toh(agend->len);
ri->iri_len += le16toh(agend->len);
return 0;
}
/* If we return anything but zero, iflib will assert... */
static int
bnxt_isc_rxd_pkt_get(void *sc, if_rxd_info_t ri)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
struct bnxt_cp_ring *cpr = &softc->rx_cp_rings[ri->iri_qsidx];
struct cmpl_base *cmp_q = (struct cmpl_base *)cpr->ring.vaddr;
struct cmpl_base *cmp;
struct rx_tpa_start_cmpl *rtpa;
uint16_t flags_type;
uint16_t type;
uint8_t agg_id;
for (;;) {
NEXT_CP_CONS_V(&cpr->ring, cpr->cons, cpr->v_bit);
ri->iri_cidx = RING_NEXT(&cpr->ring, ri->iri_cidx);
CMPL_PREFETCH_NEXT(cpr, cpr->cons);
cmp = &((struct cmpl_base *)cpr->ring.vaddr)[cpr->cons];
flags_type = le16toh(cmp->type);
type = flags_type & CMPL_BASE_TYPE_MASK;
switch (type) {
case CMPL_BASE_TYPE_RX_L2:
return bnxt_pkt_get_l2(softc, ri, cpr, flags_type);
case CMPL_BASE_TYPE_RX_TPA_END:
return bnxt_pkt_get_tpa(softc, ri, cpr, flags_type);
case CMPL_BASE_TYPE_RX_TPA_START:
rtpa = (void *)&cmp_q[cpr->cons];
agg_id = (rtpa->agg_id &
RX_TPA_START_CMPL_AGG_ID_MASK) >>
RX_TPA_START_CMPL_AGG_ID_SFT;
softc->rx_rings[ri->iri_qsidx].tpa_start[agg_id].low = *rtpa;
NEXT_CP_CONS_V(&cpr->ring, cpr->cons, cpr->v_bit);
ri->iri_cidx = RING_NEXT(&cpr->ring, ri->iri_cidx);
CMPL_PREFETCH_NEXT(cpr, cpr->cons);
softc->rx_rings[ri->iri_qsidx].tpa_start[agg_id].high =
((struct rx_tpa_start_cmpl_hi *)cmp_q)[cpr->cons];
break;
default:
device_printf(softc->dev,
"Unhandled completion type %d on RXQ %d get\n",
type, ri->iri_qsidx);
if (type & 1) {
NEXT_CP_CONS_V(&cpr->ring, cpr->cons,
cpr->v_bit);
ri->iri_cidx = RING_NEXT(&cpr->ring,
ri->iri_cidx);
CMPL_PREFETCH_NEXT(cpr, cpr->cons);
}
break;
}
}
return 0;
}
static int
bnxt_intr(void *sc)
{
struct bnxt_softc *softc = (struct bnxt_softc *)sc;
device_printf(softc->dev, "STUB: %s @ %s:%d\n", __func__, __FILE__, __LINE__);
return ENOSYS;
}
Reference in New Issue View Git Blame Copy Permalink