linux/drivers/net/ethernet/sfc/mcdi_functions.c
Edward Cree 1679c72cf4 sfc: de-indirect TSO handling
Remove the tx_queue->handle_tso function pointer, and just use
 tx_queue->tso_version to decide which function to call, thus removing
 an indirect call from the fast path.
Instead of passing a tso_v2 flag to efx_mcdi_tx_init(), set the desired
 tx_queue->tso_version before calling it.
In efx_mcdi_tx_init(), report back failure to obtain a TSOv2 context by
 setting tx_queue->tso_version to 0, which will cause the TX path to
 use the GSO-based fallback.

Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-11 17:15:22 -07:00

432 lines
12 KiB
C

// SPDX-License-Identifier: GPL-2.0-only
/****************************************************************************
* Driver for Solarflare network controllers and boards
* Copyright 2019 Solarflare Communications Inc.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 as published
* by the Free Software Foundation, incorporated herein by reference.
*/
#include "net_driver.h"
#include "efx.h"
#include "nic.h"
#include "mcdi_functions.h"
#include "mcdi.h"
#include "mcdi_pcol.h"
int efx_mcdi_free_vis(struct efx_nic *efx)
{
MCDI_DECLARE_BUF_ERR(outbuf);
size_t outlen;
int rc = efx_mcdi_rpc_quiet(efx, MC_CMD_FREE_VIS, NULL, 0,
outbuf, sizeof(outbuf), &outlen);
/* -EALREADY means nothing to free, so ignore */
if (rc == -EALREADY)
rc = 0;
if (rc)
efx_mcdi_display_error(efx, MC_CMD_FREE_VIS, 0, outbuf, outlen,
rc);
return rc;
}
int efx_mcdi_alloc_vis(struct efx_nic *efx, unsigned int min_vis,
unsigned int max_vis, unsigned int *vi_base,
unsigned int *allocated_vis)
{
MCDI_DECLARE_BUF(outbuf, MC_CMD_ALLOC_VIS_OUT_LEN);
MCDI_DECLARE_BUF(inbuf, MC_CMD_ALLOC_VIS_IN_LEN);
size_t outlen;
int rc;
MCDI_SET_DWORD(inbuf, ALLOC_VIS_IN_MIN_VI_COUNT, min_vis);
MCDI_SET_DWORD(inbuf, ALLOC_VIS_IN_MAX_VI_COUNT, max_vis);
rc = efx_mcdi_rpc(efx, MC_CMD_ALLOC_VIS, inbuf, sizeof(inbuf),
outbuf, sizeof(outbuf), &outlen);
if (rc != 0)
return rc;
if (outlen < MC_CMD_ALLOC_VIS_OUT_LEN)
return -EIO;
netif_dbg(efx, drv, efx->net_dev, "base VI is A0x%03x\n",
MCDI_DWORD(outbuf, ALLOC_VIS_OUT_VI_BASE));
if (vi_base)
*vi_base = MCDI_DWORD(outbuf, ALLOC_VIS_OUT_VI_BASE);
if (allocated_vis)
*allocated_vis = MCDI_DWORD(outbuf, ALLOC_VIS_OUT_VI_COUNT);
return 0;
}
int efx_mcdi_ev_probe(struct efx_channel *channel)
{
return efx_nic_alloc_buffer(channel->efx, &channel->eventq.buf,
(channel->eventq_mask + 1) *
sizeof(efx_qword_t),
GFP_KERNEL);
}
int efx_mcdi_ev_init(struct efx_channel *channel, bool v1_cut_thru, bool v2)
{
MCDI_DECLARE_BUF(inbuf,
MC_CMD_INIT_EVQ_V2_IN_LEN(EFX_MAX_EVQ_SIZE * 8 /
EFX_BUF_SIZE));
MCDI_DECLARE_BUF(outbuf, MC_CMD_INIT_EVQ_V2_OUT_LEN);
size_t entries = channel->eventq.buf.len / EFX_BUF_SIZE;
struct efx_nic *efx = channel->efx;
size_t inlen, outlen;
dma_addr_t dma_addr;
int rc, i;
/* Fill event queue with all ones (i.e. empty events) */
memset(channel->eventq.buf.addr, 0xff, channel->eventq.buf.len);
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_SIZE, channel->eventq_mask + 1);
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_INSTANCE, channel->channel);
/* INIT_EVQ expects index in vector table, not absolute */
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_IRQ_NUM, channel->channel);
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_TMR_MODE,
MC_CMD_INIT_EVQ_IN_TMR_MODE_DIS);
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_TMR_LOAD, 0);
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_TMR_RELOAD, 0);
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_COUNT_MODE,
MC_CMD_INIT_EVQ_IN_COUNT_MODE_DIS);
MCDI_SET_DWORD(inbuf, INIT_EVQ_IN_COUNT_THRSHLD, 0);
if (v2) {
/* Use the new generic approach to specifying event queue
* configuration, requesting lower latency or higher throughput.
* The options that actually get used appear in the output.
*/
MCDI_POPULATE_DWORD_2(inbuf, INIT_EVQ_V2_IN_FLAGS,
INIT_EVQ_V2_IN_FLAG_INTERRUPTING, 1,
INIT_EVQ_V2_IN_FLAG_TYPE,
MC_CMD_INIT_EVQ_V2_IN_FLAG_TYPE_AUTO);
} else {
MCDI_POPULATE_DWORD_4(inbuf, INIT_EVQ_IN_FLAGS,
INIT_EVQ_IN_FLAG_INTERRUPTING, 1,
INIT_EVQ_IN_FLAG_RX_MERGE, 1,
INIT_EVQ_IN_FLAG_TX_MERGE, 1,
INIT_EVQ_IN_FLAG_CUT_THRU, v1_cut_thru);
}
dma_addr = channel->eventq.buf.dma_addr;
for (i = 0; i < entries; ++i) {
MCDI_SET_ARRAY_QWORD(inbuf, INIT_EVQ_IN_DMA_ADDR, i, dma_addr);
dma_addr += EFX_BUF_SIZE;
}
inlen = MC_CMD_INIT_EVQ_IN_LEN(entries);
rc = efx_mcdi_rpc(efx, MC_CMD_INIT_EVQ, inbuf, inlen,
outbuf, sizeof(outbuf), &outlen);
if (outlen >= MC_CMD_INIT_EVQ_V2_OUT_LEN)
netif_dbg(efx, drv, efx->net_dev,
"Channel %d using event queue flags %08x\n",
channel->channel,
MCDI_DWORD(outbuf, INIT_EVQ_V2_OUT_FLAGS));
return rc;
}
void efx_mcdi_ev_remove(struct efx_channel *channel)
{
efx_nic_free_buffer(channel->efx, &channel->eventq.buf);
}
void efx_mcdi_ev_fini(struct efx_channel *channel)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_FINI_EVQ_IN_LEN);
MCDI_DECLARE_BUF_ERR(outbuf);
struct efx_nic *efx = channel->efx;
size_t outlen;
int rc;
MCDI_SET_DWORD(inbuf, FINI_EVQ_IN_INSTANCE, channel->channel);
rc = efx_mcdi_rpc_quiet(efx, MC_CMD_FINI_EVQ, inbuf, sizeof(inbuf),
outbuf, sizeof(outbuf), &outlen);
if (rc && rc != -EALREADY)
goto fail;
return;
fail:
efx_mcdi_display_error(efx, MC_CMD_FINI_EVQ, MC_CMD_FINI_EVQ_IN_LEN,
outbuf, outlen, rc);
}
int efx_mcdi_tx_init(struct efx_tx_queue *tx_queue)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_INIT_TXQ_IN_LEN(EFX_MAX_DMAQ_SIZE * 8 /
EFX_BUF_SIZE));
bool csum_offload = tx_queue->type & EFX_TXQ_TYPE_OUTER_CSUM;
bool inner_csum = tx_queue->type & EFX_TXQ_TYPE_INNER_CSUM;
size_t entries = tx_queue->txd.buf.len / EFX_BUF_SIZE;
struct efx_channel *channel = tx_queue->channel;
struct efx_nic *efx = tx_queue->efx;
dma_addr_t dma_addr;
size_t inlen;
int rc, i;
BUILD_BUG_ON(MC_CMD_INIT_TXQ_OUT_LEN != 0);
MCDI_SET_DWORD(inbuf, INIT_TXQ_IN_SIZE, tx_queue->ptr_mask + 1);
MCDI_SET_DWORD(inbuf, INIT_TXQ_IN_TARGET_EVQ, channel->channel);
MCDI_SET_DWORD(inbuf, INIT_TXQ_IN_LABEL, tx_queue->label);
MCDI_SET_DWORD(inbuf, INIT_TXQ_IN_INSTANCE, tx_queue->queue);
MCDI_SET_DWORD(inbuf, INIT_TXQ_IN_OWNER_ID, 0);
MCDI_SET_DWORD(inbuf, INIT_TXQ_IN_PORT_ID, efx->vport_id);
dma_addr = tx_queue->txd.buf.dma_addr;
netif_dbg(efx, hw, efx->net_dev, "pushing TXQ %d. %zu entries (%llx)\n",
tx_queue->queue, entries, (u64)dma_addr);
for (i = 0; i < entries; ++i) {
MCDI_SET_ARRAY_QWORD(inbuf, INIT_TXQ_IN_DMA_ADDR, i, dma_addr);
dma_addr += EFX_BUF_SIZE;
}
inlen = MC_CMD_INIT_TXQ_IN_LEN(entries);
do {
bool tso_v2 = tx_queue->tso_version == 2;
/* TSOv2 implies IP header checksum offload for TSO frames,
* so we can safely disable IP header checksum offload for
* everything else. If we don't have TSOv2, then we have to
* enable IP header checksum offload, which is strictly
* incorrect but better than breaking TSO.
*/
MCDI_POPULATE_DWORD_6(inbuf, INIT_TXQ_IN_FLAGS,
/* This flag was removed from mcdi_pcol.h for
* the non-_EXT version of INIT_TXQ. However,
* firmware still honours it.
*/
INIT_TXQ_EXT_IN_FLAG_TSOV2_EN, tso_v2,
INIT_TXQ_IN_FLAG_IP_CSUM_DIS, !(csum_offload && tso_v2),
INIT_TXQ_IN_FLAG_TCP_CSUM_DIS, !csum_offload,
INIT_TXQ_EXT_IN_FLAG_TIMESTAMP, tx_queue->timestamping,
INIT_TXQ_IN_FLAG_INNER_IP_CSUM_EN, inner_csum && !tso_v2,
INIT_TXQ_IN_FLAG_INNER_TCP_CSUM_EN, inner_csum);
rc = efx_mcdi_rpc_quiet(efx, MC_CMD_INIT_TXQ, inbuf, inlen,
NULL, 0, NULL);
if (rc == -ENOSPC && tso_v2) {
/* Retry without TSOv2 if we're short on contexts. */
tx_queue->tso_version = 0;
netif_warn(efx, probe, efx->net_dev,
"TSOv2 context not available to segment in "
"hardware. TCP performance may be reduced.\n"
);
} else if (rc) {
efx_mcdi_display_error(efx, MC_CMD_INIT_TXQ,
MC_CMD_INIT_TXQ_EXT_IN_LEN,
NULL, 0, rc);
goto fail;
}
} while (rc);
return 0;
fail:
return rc;
}
void efx_mcdi_tx_remove(struct efx_tx_queue *tx_queue)
{
efx_nic_free_buffer(tx_queue->efx, &tx_queue->txd.buf);
}
void efx_mcdi_tx_fini(struct efx_tx_queue *tx_queue)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_FINI_TXQ_IN_LEN);
MCDI_DECLARE_BUF_ERR(outbuf);
struct efx_nic *efx = tx_queue->efx;
size_t outlen;
int rc;
MCDI_SET_DWORD(inbuf, FINI_TXQ_IN_INSTANCE,
tx_queue->queue);
rc = efx_mcdi_rpc_quiet(efx, MC_CMD_FINI_TXQ, inbuf, sizeof(inbuf),
outbuf, sizeof(outbuf), &outlen);
if (rc && rc != -EALREADY)
goto fail;
return;
fail:
efx_mcdi_display_error(efx, MC_CMD_FINI_TXQ, MC_CMD_FINI_TXQ_IN_LEN,
outbuf, outlen, rc);
}
int efx_mcdi_rx_probe(struct efx_rx_queue *rx_queue)
{
return efx_nic_alloc_buffer(rx_queue->efx, &rx_queue->rxd.buf,
(rx_queue->ptr_mask + 1) *
sizeof(efx_qword_t),
GFP_KERNEL);
}
void efx_mcdi_rx_init(struct efx_rx_queue *rx_queue)
{
struct efx_channel *channel = efx_rx_queue_channel(rx_queue);
size_t entries = rx_queue->rxd.buf.len / EFX_BUF_SIZE;
MCDI_DECLARE_BUF(inbuf, MC_CMD_INIT_RXQ_V4_IN_LEN);
struct efx_nic *efx = rx_queue->efx;
unsigned int buffer_size;
dma_addr_t dma_addr;
int rc;
int i;
BUILD_BUG_ON(MC_CMD_INIT_RXQ_OUT_LEN != 0);
rx_queue->scatter_n = 0;
rx_queue->scatter_len = 0;
if (efx->type->revision == EFX_REV_EF100)
buffer_size = efx->rx_page_buf_step;
else
buffer_size = 0;
MCDI_SET_DWORD(inbuf, INIT_RXQ_IN_SIZE, rx_queue->ptr_mask + 1);
MCDI_SET_DWORD(inbuf, INIT_RXQ_IN_TARGET_EVQ, channel->channel);
MCDI_SET_DWORD(inbuf, INIT_RXQ_IN_LABEL, efx_rx_queue_index(rx_queue));
MCDI_SET_DWORD(inbuf, INIT_RXQ_IN_INSTANCE,
efx_rx_queue_index(rx_queue));
MCDI_POPULATE_DWORD_2(inbuf, INIT_RXQ_IN_FLAGS,
INIT_RXQ_IN_FLAG_PREFIX, 1,
INIT_RXQ_IN_FLAG_TIMESTAMP, 1);
MCDI_SET_DWORD(inbuf, INIT_RXQ_IN_OWNER_ID, 0);
MCDI_SET_DWORD(inbuf, INIT_RXQ_IN_PORT_ID, efx->vport_id);
MCDI_SET_DWORD(inbuf, INIT_RXQ_V4_IN_BUFFER_SIZE_BYTES, buffer_size);
dma_addr = rx_queue->rxd.buf.dma_addr;
netif_dbg(efx, hw, efx->net_dev, "pushing RXQ %d. %zu entries (%llx)\n",
efx_rx_queue_index(rx_queue), entries, (u64)dma_addr);
for (i = 0; i < entries; ++i) {
MCDI_SET_ARRAY_QWORD(inbuf, INIT_RXQ_IN_DMA_ADDR, i, dma_addr);
dma_addr += EFX_BUF_SIZE;
}
rc = efx_mcdi_rpc(efx, MC_CMD_INIT_RXQ, inbuf, sizeof(inbuf),
NULL, 0, NULL);
if (rc)
netdev_WARN(efx->net_dev, "failed to initialise RXQ %d\n",
efx_rx_queue_index(rx_queue));
}
void efx_mcdi_rx_remove(struct efx_rx_queue *rx_queue)
{
efx_nic_free_buffer(rx_queue->efx, &rx_queue->rxd.buf);
}
void efx_mcdi_rx_fini(struct efx_rx_queue *rx_queue)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_FINI_RXQ_IN_LEN);
MCDI_DECLARE_BUF_ERR(outbuf);
struct efx_nic *efx = rx_queue->efx;
size_t outlen;
int rc;
MCDI_SET_DWORD(inbuf, FINI_RXQ_IN_INSTANCE,
efx_rx_queue_index(rx_queue));
rc = efx_mcdi_rpc_quiet(efx, MC_CMD_FINI_RXQ, inbuf, sizeof(inbuf),
outbuf, sizeof(outbuf), &outlen);
if (rc && rc != -EALREADY)
goto fail;
return;
fail:
efx_mcdi_display_error(efx, MC_CMD_FINI_RXQ, MC_CMD_FINI_RXQ_IN_LEN,
outbuf, outlen, rc);
}
int efx_fini_dmaq(struct efx_nic *efx)
{
struct efx_tx_queue *tx_queue;
struct efx_rx_queue *rx_queue;
struct efx_channel *channel;
int pending;
/* If the MC has just rebooted, the TX/RX queues will have already been
* torn down, but efx->active_queues needs to be set to zero.
*/
if (efx->must_realloc_vis) {
atomic_set(&efx->active_queues, 0);
return 0;
}
/* Do not attempt to write to the NIC during EEH recovery */
if (efx->state != STATE_RECOVERY) {
efx_for_each_channel(channel, efx) {
efx_for_each_channel_rx_queue(rx_queue, channel)
efx_mcdi_rx_fini(rx_queue);
efx_for_each_channel_tx_queue(tx_queue, channel)
efx_mcdi_tx_fini(tx_queue);
}
wait_event_timeout(efx->flush_wq,
atomic_read(&efx->active_queues) == 0,
msecs_to_jiffies(EFX_MAX_FLUSH_TIME));
pending = atomic_read(&efx->active_queues);
if (pending) {
netif_err(efx, hw, efx->net_dev, "failed to flush %d queues\n",
pending);
return -ETIMEDOUT;
}
}
return 0;
}
int efx_mcdi_window_mode_to_stride(struct efx_nic *efx, u8 vi_window_mode)
{
switch (vi_window_mode) {
case MC_CMD_GET_CAPABILITIES_V3_OUT_VI_WINDOW_MODE_8K:
efx->vi_stride = 8192;
break;
case MC_CMD_GET_CAPABILITIES_V3_OUT_VI_WINDOW_MODE_16K:
efx->vi_stride = 16384;
break;
case MC_CMD_GET_CAPABILITIES_V3_OUT_VI_WINDOW_MODE_64K:
efx->vi_stride = 65536;
break;
default:
netif_err(efx, probe, efx->net_dev,
"Unrecognised VI window mode %d\n",
vi_window_mode);
return -EIO;
}
netif_dbg(efx, probe, efx->net_dev, "vi_stride = %u\n",
efx->vi_stride);
return 0;
}
int efx_get_pf_index(struct efx_nic *efx, unsigned int *pf_index)
{
MCDI_DECLARE_BUF(outbuf, MC_CMD_GET_FUNCTION_INFO_OUT_LEN);
size_t outlen;
int rc;
rc = efx_mcdi_rpc(efx, MC_CMD_GET_FUNCTION_INFO, NULL, 0, outbuf,
sizeof(outbuf), &outlen);
if (rc)
return rc;
if (outlen < sizeof(outbuf))
return -EIO;
*pf_index = MCDI_DWORD(outbuf, GET_FUNCTION_INFO_OUT_PF);
return 0;
}