freebsd-dev/sys/dev/netmap/ixgbe_netmap.h
Luigi Rizzo 9b034c6f08 Properly disable crc stripping when operating in netmap mode.
Contrarily to what i wrote in my previous commit, the 82599
does include the CRC in the length. The operating mode is
reset in ixgbe_init_locked() and so we need to hook into
the places where the two registers (HLREG0 and RDRXCTL) are
modified.
2012-04-13 16:42:54 +00:00

616 lines
20 KiB
C

/*
* Copyright (C) 2011 Matteo Landi, Luigi Rizzo. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*/
/*
* $FreeBSD$
* $Id: ixgbe_netmap.h 10627 2012-02-23 19:37:15Z luigi $
*
* netmap modifications for ixgbe
*
* This file is meant to be a reference on how to implement
* netmap support for a network driver.
* This file contains code but only static or inline functions
* that are used by a single driver. To avoid replication of
* code we just #include it near the beginning of the
* standard driver.
*/
#include <net/netmap.h>
#include <sys/selinfo.h>
/*
* Some drivers may need the following headers. Others
* already include them by default
#include <vm/vm.h>
#include <vm/pmap.h>
*/
#include <dev/netmap/netmap_kern.h>
/*
* ix_crcstrip: 0: keep CRC in rx frames (default), 1: strip it.
* During regular operations the CRC is stripped, but on some
* hardware reception of frames not multiple of 64 is slower,
* so using crcstrip=0 helps in benchmarks.
*
* ix_rx_miss, ix_rx_miss_bufs:
* count packets that might be missed due to lost interrupts.
*
* ix_use_dd
* use the dd bit for completed tx transmissions.
* This is tricky, much better to use TDH for now.
*/
SYSCTL_DECL(_dev_netmap);
static int ix_write_len;
SYSCTL_INT(_dev_netmap, OID_AUTO, ix_write_len,
CTLFLAG_RW, &ix_write_len, 0, "write rx len");
static int ix_rx_miss, ix_rx_miss_bufs, ix_use_dd, ix_crcstrip;
SYSCTL_INT(_dev_netmap, OID_AUTO, ix_crcstrip,
CTLFLAG_RW, &ix_crcstrip, 0, "strip CRC on rx frames");
SYSCTL_INT(_dev_netmap, OID_AUTO, ix_use_dd,
CTLFLAG_RW, &ix_use_dd, 0, "use dd instead of tdh to detect tx frames");
SYSCTL_INT(_dev_netmap, OID_AUTO, ix_rx_miss,
CTLFLAG_RW, &ix_rx_miss, 0, "potentially missed rx intr");
SYSCTL_INT(_dev_netmap, OID_AUTO, ix_rx_miss_bufs,
CTLFLAG_RW, &ix_rx_miss_bufs, 0, "potentially missed rx intr bufs");
/*
* wrapper to export locks to the generic netmap code.
*/
static void
ixgbe_netmap_lock_wrapper(struct ifnet *_a, int what, u_int queueid)
{
struct adapter *adapter = _a->if_softc;
ASSERT(queueid < adapter->num_queues);
switch (what) {
case NETMAP_CORE_LOCK:
IXGBE_CORE_LOCK(adapter);
break;
case NETMAP_CORE_UNLOCK:
IXGBE_CORE_UNLOCK(adapter);
break;
case NETMAP_TX_LOCK:
IXGBE_TX_LOCK(&adapter->tx_rings[queueid]);
break;
case NETMAP_TX_UNLOCK:
IXGBE_TX_UNLOCK(&adapter->tx_rings[queueid]);
break;
case NETMAP_RX_LOCK:
IXGBE_RX_LOCK(&adapter->rx_rings[queueid]);
break;
case NETMAP_RX_UNLOCK:
IXGBE_RX_UNLOCK(&adapter->rx_rings[queueid]);
break;
}
}
static void
set_crcstrip(struct ixgbe_hw *hw, int onoff)
{
/* crc stripping is set in two places:
* IXGBE_HLREG0 (modified on init_locked and hw reset)
* IXGBE_RDRXCTL (set by the original driver in
* ixgbe_setup_hw_rsc() called in init_locked.
* We disable the setting when netmap is compiled in).
* We update the values here, but also in ixgbe.c because
* init_locked sometimes is called outside our control.
*/
uint32_t hl, rxc;
hl = IXGBE_READ_REG(hw, IXGBE_HLREG0);
rxc = IXGBE_READ_REG(hw, IXGBE_RDRXCTL);
if (netmap_verbose)
D("%s read HLREG 0x%x rxc 0x%x",
onoff ? "enter" : "exit", hl, rxc);
/* hw requirements ... */
rxc &= ~IXGBE_RDRXCTL_RSCFRSTSIZE;
rxc |= IXGBE_RDRXCTL_RSCACKC;
if (onoff && !ix_crcstrip) {
/* keep the crc. Fast rx */
hl &= ~IXGBE_HLREG0_RXCRCSTRP;
rxc &= ~IXGBE_RDRXCTL_CRCSTRIP;
} else {
/* reset default mode */
hl |= IXGBE_HLREG0_RXCRCSTRP;
rxc |= IXGBE_RDRXCTL_CRCSTRIP;
}
if (netmap_verbose)
D("%s write HLREG 0x%x rxc 0x%x",
onoff ? "enter" : "exit", hl, rxc);
IXGBE_WRITE_REG(hw, IXGBE_HLREG0, hl);
IXGBE_WRITE_REG(hw, IXGBE_RDRXCTL, rxc);
}
/*
* Register/unregister. We are already under core lock.
* Only called on the first register or the last unregister.
*/
static int
ixgbe_netmap_reg(struct ifnet *ifp, int onoff)
{
struct adapter *adapter = ifp->if_softc;
struct netmap_adapter *na = NA(ifp);
int error = 0;
if (na == NULL)
return EINVAL; /* no netmap support here */
ixgbe_disable_intr(adapter);
/* Tell the stack that the interface is no longer active */
ifp->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE);
set_crcstrip(&adapter->hw, onoff);
if (onoff) { /* enable netmap mode */
ifp->if_capenable |= IFCAP_NETMAP;
/* save if_transmit and replace with our routine */
na->if_transmit = ifp->if_transmit;
ifp->if_transmit = netmap_start;
/*
* reinitialize the adapter, now with netmap flag set,
* so the rings will be set accordingly.
*/
ixgbe_init_locked(adapter);
if ((ifp->if_drv_flags & (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)) == 0) {
error = ENOMEM;
goto fail;
}
} else { /* reset normal mode (explicit request or netmap failed) */
fail:
/* restore if_transmit */
ifp->if_transmit = na->if_transmit;
ifp->if_capenable &= ~IFCAP_NETMAP;
/* initialize the card, this time in standard mode */
ixgbe_init_locked(adapter); /* also enables intr */
}
set_crcstrip(&adapter->hw, onoff);
return (error);
}
/*
* Reconcile kernel and user view of the transmit ring.
* This routine might be called frequently so it must be efficient.
*
* Userspace has filled tx slots up to ring->cur (excluded).
* The last unused slot previously known to the kernel was kring->nkr_hwcur,
* and the last interrupt reported kring->nr_hwavail slots available.
*
* This function runs under lock (acquired from the caller or internally).
* It must first update ring->avail to what the kernel knows,
* subtract the newly used slots (ring->cur - kring->nkr_hwcur)
* from both avail and nr_hwavail, and set ring->nkr_hwcur = ring->cur
* issuing a dmamap_sync on all slots.
*
* Since ring comes from userspace, its content must be read only once,
* and validated before being used to update the kernel's structures.
* (this is also true for every use of ring in the kernel).
*
* ring->avail is never used, only checked for bogus values.
*
* do_lock is set iff the function is called from the ioctl handler.
* In this case, grab a lock around the body, and also reclaim transmitted
* buffers irrespective of interrupt mitigation.
*/
static int
ixgbe_netmap_txsync(struct ifnet *ifp, u_int ring_nr, int do_lock)
{
struct adapter *adapter = ifp->if_softc;
struct tx_ring *txr = &adapter->tx_rings[ring_nr];
struct netmap_adapter *na = NA(adapter->ifp);
struct netmap_kring *kring = &na->tx_rings[ring_nr];
struct netmap_ring *ring = kring->ring;
u_int j, k = ring->cur, l, n = 0, lim = kring->nkr_num_slots - 1;
/*
* ixgbe can generate an interrupt on every tx packet, but it
* seems very expensive, so we interrupt once every half ring,
* or when requested with NS_REPORT
*/
int report_frequency = kring->nkr_num_slots >> 1;
if (k > lim)
return netmap_ring_reinit(kring);
if (do_lock)
IXGBE_TX_LOCK(txr);
bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map,
BUS_DMASYNC_POSTREAD);
/*
* Process new packets to send. j is the current index in the
* netmap ring, l is the corresponding index in the NIC ring.
* The two numbers differ because upon a *_init() we reset
* the NIC ring but leave the netmap ring unchanged.
* For the transmit ring, we have
*
* j = kring->nr_hwcur
* l = IXGBE_TDT (not tracked in the driver)
* and
* j == (l + kring->nkr_hwofs) % ring_size
*
* In this driver kring->nkr_hwofs >= 0, but for other
* drivers it might be negative as well.
*/
j = kring->nr_hwcur;
if (j != k) { /* we have new packets to send */
prefetch(&ring->slot[j]);
l = netmap_idx_k2n(kring, j); /* NIC index */
prefetch(&txr->tx_buffers[l]);
for (n = 0; j != k; n++) {
/*
* Collect per-slot info.
* Note that txbuf and curr are indexed by l.
*
* In this driver we collect the buffer address
* (using the PNMB() macro) because we always
* need to rewrite it into the NIC ring.
* Many other drivers preserve the address, so
* we only need to access it if NS_BUF_CHANGED
* is set.
* XXX note, on this device the dmamap* calls are
* not necessary because tag is 0, however just accessing
* the per-packet tag kills 1Mpps at 900 MHz.
*/
struct netmap_slot *slot = &ring->slot[j];
union ixgbe_adv_tx_desc *curr = &txr->tx_base[l];
struct ixgbe_tx_buf *txbuf = &txr->tx_buffers[l];
uint64_t paddr;
// XXX type for flags and len ?
int flags = ((slot->flags & NS_REPORT) ||
j == 0 || j == report_frequency) ?
IXGBE_TXD_CMD_RS : 0;
u_int len = slot->len;
void *addr = PNMB(slot, &paddr);
j = (j == lim) ? 0 : j + 1;
l = (l == lim) ? 0 : l + 1;
prefetch(&ring->slot[j]);
prefetch(&txr->tx_buffers[l]);
/*
* Quick check for valid addr and len.
* NMB() returns netmap_buffer_base for invalid
* buffer indexes (but the address is still a
* valid one to be used in a ring). slot->len is
* unsigned so no need to check for negative values.
*/
if (addr == netmap_buffer_base || len > NETMAP_BUF_SIZE) {
ring_reset:
if (do_lock)
IXGBE_TX_UNLOCK(txr);
return netmap_ring_reinit(kring);
}
if (slot->flags & NS_BUF_CHANGED) {
/* buffer has changed, unload and reload map */
netmap_reload_map(txr->txtag, txbuf->map, addr);
slot->flags &= ~NS_BUF_CHANGED;
}
slot->flags &= ~NS_REPORT;
/*
* Fill the slot in the NIC ring.
* In this driver we need to rewrite the buffer
* address in the NIC ring. Other drivers do not
* need this.
* Use legacy descriptor, it is faster.
*/
curr->read.buffer_addr = htole64(paddr);
curr->read.olinfo_status = 0;
curr->read.cmd_type_len = htole32(len | flags |
IXGBE_ADVTXD_DCMD_IFCS | IXGBE_TXD_CMD_EOP);
/* make sure changes to the buffer are synced */
bus_dmamap_sync(txr->txtag, txbuf->map, BUS_DMASYNC_PREWRITE);
}
kring->nr_hwcur = k; /* the saved ring->cur */
/* decrease avail by number of packets sent */
kring->nr_hwavail -= n;
/* synchronize the NIC ring */
bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
/* (re)start the transmitter up to slot l (excluded) */
IXGBE_WRITE_REG(&adapter->hw, IXGBE_TDT(txr->me), l);
}
/*
* Reclaim buffers for completed transmissions.
* Because this is expensive (we read a NIC register etc.)
* we only do it in specific cases (see below).
* In all cases kring->nr_kflags indicates which slot will be
* checked upon a tx interrupt (nkr_num_slots means none).
*/
if (do_lock) {
j = 1; /* forced reclaim, ignore interrupts */
kring->nr_kflags = kring->nkr_num_slots;
} else if (kring->nr_hwavail > 0) {
j = 0; /* buffers still available: no reclaim, ignore intr. */
kring->nr_kflags = kring->nkr_num_slots;
} else {
/*
* no buffers available, locate a slot for which we request
* ReportStatus (approximately half ring after next_to_clean)
* and record it in kring->nr_kflags.
* If the slot has DD set, do the reclaim looking at TDH,
* otherwise we go to sleep (in netmap_poll()) and will be
* woken up when slot nr_kflags will be ready.
*/
struct ixgbe_legacy_tx_desc *txd =
(struct ixgbe_legacy_tx_desc *)txr->tx_base;
j = txr->next_to_clean + kring->nkr_num_slots/2;
if (j >= kring->nkr_num_slots)
j -= kring->nkr_num_slots;
// round to the closest with dd set
j= (j < kring->nkr_num_slots / 4 || j >= kring->nkr_num_slots*3/4) ?
0 : report_frequency;
kring->nr_kflags = j; /* the slot to check */
j = txd[j].upper.fields.status & IXGBE_TXD_STAT_DD; // XXX cpu_to_le32 ?
}
if (j) {
int delta;
/*
* Record completed transmissions.
* We (re)use the driver's txr->next_to_clean to keep
* track of the most recently completed transmission.
*
* The datasheet discourages the use of TDH to find out the
* number of sent packets. We should rather check the DD
* status bit in a packet descriptor. However, we only set
* the "report status" bit for some descriptors (a kind of
* interrupt mitigation), so we can only check on those.
* For the time being we use TDH, as we do it infrequently
* enough not to pose performance problems.
*/
if (ix_use_dd) {
struct ixgbe_legacy_tx_desc *txd =
(struct ixgbe_legacy_tx_desc *)txr->tx_base;
l = txr->next_to_clean;
k = netmap_idx_k2n(kring, kring->nr_hwcur);
delta = 0;
while (l != k &&
txd[l].upper.fields.status & IXGBE_TXD_STAT_DD) {
delta++;
l = (l == lim) ? 0 : l + 1;
}
} else {
l = IXGBE_READ_REG(&adapter->hw, IXGBE_TDH(ring_nr));
if (l >= kring->nkr_num_slots) { /* XXX can happen */
D("TDH wrap %d", l);
l -= kring->nkr_num_slots;
}
delta = l - txr->next_to_clean;
}
if (delta) {
/* some tx completed, increment avail */
if (delta < 0)
delta += kring->nkr_num_slots;
txr->next_to_clean = l;
kring->nr_hwavail += delta;
if (kring->nr_hwavail > lim)
goto ring_reset;
}
}
/* update avail to what the kernel knows */
ring->avail = kring->nr_hwavail;
if (do_lock)
IXGBE_TX_UNLOCK(txr);
return 0;
}
/*
* Reconcile kernel and user view of the receive ring.
* Same as for the txsync, this routine must be efficient and
* avoid races in accessing the shared regions.
*
* When called, userspace has read data from slots kring->nr_hwcur
* up to ring->cur (excluded).
*
* The last interrupt reported kring->nr_hwavail slots available
* after kring->nr_hwcur.
* We must subtract the newly consumed slots (cur - nr_hwcur)
* from nr_hwavail, make the descriptors available for the next reads,
* and set kring->nr_hwcur = ring->cur and ring->avail = kring->nr_hwavail.
*
* do_lock has a special meaning: please refer to txsync.
*/
static int
ixgbe_netmap_rxsync(struct ifnet *ifp, u_int ring_nr, int do_lock)
{
struct adapter *adapter = ifp->if_softc;
struct rx_ring *rxr = &adapter->rx_rings[ring_nr];
struct netmap_adapter *na = NA(adapter->ifp);
struct netmap_kring *kring = &na->rx_rings[ring_nr];
struct netmap_ring *ring = kring->ring;
u_int j, l, n, lim = kring->nkr_num_slots - 1;
int force_update = do_lock || kring->nr_kflags & NKR_PENDINTR;
u_int k = ring->cur, resvd = ring->reserved;
if (k > lim)
return netmap_ring_reinit(kring);
if (do_lock)
IXGBE_RX_LOCK(rxr);
/* XXX check sync modes */
bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map,
BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE);
/*
* First part, import newly received packets into the netmap ring.
*
* j is the index of the next free slot in the netmap ring,
* and l is the index of the next received packet in the NIC ring,
* and they may differ in case if_init() has been called while
* in netmap mode. For the receive ring we have
*
* j = (kring->nr_hwcur + kring->nr_hwavail) % ring_size
* l = rxr->next_to_check;
* and
* j == (l + kring->nkr_hwofs) % ring_size
*
* rxr->next_to_check is set to 0 on a ring reinit
*/
if (netmap_no_pendintr || force_update) {
/* XXX apparently the length field in advanced descriptors
* does not include the CRC irrespective of the setting
* of CRCSTRIP. The data sheets say differently.
* Very strange.
*/
int crclen = ix_crcstrip ? 0 : 4;
l = rxr->next_to_check;
j = netmap_idx_n2k(kring, l);
for (n = 0; ; n++) {
union ixgbe_adv_rx_desc *curr = &rxr->rx_base[l];
uint32_t staterr = le32toh(curr->wb.upper.status_error);
if ((staterr & IXGBE_RXD_STAT_DD) == 0)
break;
ring->slot[j].len = le16toh(curr->wb.upper.length) - crclen;
if (ix_write_len)
D("rx[%d] len %d", j, ring->slot[j].len);
bus_dmamap_sync(rxr->ptag,
rxr->rx_buffers[l].pmap, BUS_DMASYNC_POSTREAD);
j = (j == lim) ? 0 : j + 1;
l = (l == lim) ? 0 : l + 1;
}
if (n) { /* update the state variables */
if (netmap_no_pendintr && !force_update) {
/* diagnostics */
ix_rx_miss ++;
ix_rx_miss_bufs += n;
}
rxr->next_to_check = l;
kring->nr_hwavail += n;
}
kring->nr_kflags &= ~NKR_PENDINTR;
}
/*
* Skip past packets that userspace has released
* (from kring->nr_hwcur to ring->cur - ring->reserved excluded),
* and make the buffers available for reception.
* As usual j is the index in the netmap ring, l is the index
* in the NIC ring, and j == (l + kring->nkr_hwofs) % ring_size
*/
j = kring->nr_hwcur;
if (resvd > 0) {
if (resvd + ring->avail >= lim + 1) {
D("XXX invalid reserve/avail %d %d", resvd, ring->avail);
ring->reserved = resvd = 0; // XXX panic...
}
k = (k >= resvd) ? k - resvd : k + lim + 1 - resvd;
}
if (j != k) { /* userspace has released some packets. */
l = netmap_idx_k2n(kring, j);
for (n = 0; j != k; n++) {
/* collect per-slot info, with similar validations
* and flag handling as in the txsync code.
*
* NOTE curr and rxbuf are indexed by l.
* Also, this driver needs to update the physical
* address in the NIC ring, but other drivers
* may not have this requirement.
*/
struct netmap_slot *slot = &ring->slot[j];
union ixgbe_adv_rx_desc *curr = &rxr->rx_base[l];
struct ixgbe_rx_buf *rxbuf = &rxr->rx_buffers[l];
uint64_t paddr;
void *addr = PNMB(slot, &paddr);
if (addr == netmap_buffer_base) /* bad buf */
goto ring_reset;
if (slot->flags & NS_BUF_CHANGED) {
netmap_reload_map(rxr->ptag, rxbuf->pmap, addr);
slot->flags &= ~NS_BUF_CHANGED;
}
curr->wb.upper.status_error = 0;
curr->read.pkt_addr = htole64(paddr);
bus_dmamap_sync(rxr->ptag, rxbuf->pmap,
BUS_DMASYNC_PREREAD);
j = (j == lim) ? 0 : j + 1;
l = (l == lim) ? 0 : l + 1;
}
kring->nr_hwavail -= n;
kring->nr_hwcur = k;
bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
/* IMPORTANT: we must leave one free slot in the ring,
* so move l back by one unit
*/
l = (l == 0) ? lim : l - 1;
IXGBE_WRITE_REG(&adapter->hw, IXGBE_RDT(rxr->me), l);
}
/* tell userspace that there are new packets */
ring->avail = kring->nr_hwavail - resvd;
if (do_lock)
IXGBE_RX_UNLOCK(rxr);
return 0;
ring_reset:
if (do_lock)
IXGBE_RX_UNLOCK(rxr);
return netmap_ring_reinit(kring);
}
/*
* The attach routine, called near the end of ixgbe_attach(),
* fills the parameters for netmap_attach() and calls it.
* It cannot fail, in the worst case (such as no memory)
* netmap mode will be disabled and the driver will only
* operate in standard mode.
*/
static void
ixgbe_netmap_attach(struct adapter *adapter)
{
struct netmap_adapter na;
bzero(&na, sizeof(na));
na.ifp = adapter->ifp;
na.separate_locks = 1; /* this card has separate rx/tx locks */
na.num_tx_desc = adapter->num_tx_desc;
na.num_rx_desc = adapter->num_rx_desc;
na.nm_txsync = ixgbe_netmap_txsync;
na.nm_rxsync = ixgbe_netmap_rxsync;
na.nm_lock = ixgbe_netmap_lock_wrapper;
na.nm_register = ixgbe_netmap_reg;
netmap_attach(&na, adapter->num_queues);
}
/* end of file */