1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
|
// SPDX-License-Identifier: GPL-2.0-only
/****************************************************************************
* Driver for Solarflare network controllers and boards
* Copyright 2018 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_common.h"
#include "tx_common.h"
static unsigned int efx_tx_cb_page_count(struct efx_tx_queue *tx_queue)
{
return DIV_ROUND_UP(tx_queue->ptr_mask + 1,
PAGE_SIZE >> EFX_TX_CB_ORDER);
}
int efx_probe_tx_queue(struct efx_tx_queue *tx_queue)
{
struct efx_nic *efx = tx_queue->efx;
unsigned int entries;
int rc;
/* Create the smallest power-of-two aligned ring */
entries = max(roundup_pow_of_two(efx->txq_entries), EFX_MIN_DMAQ_SIZE);
EFX_WARN_ON_PARANOID(entries > EFX_MAX_DMAQ_SIZE);
tx_queue->ptr_mask = entries - 1;
netif_dbg(efx, probe, efx->net_dev,
"creating TX queue %d size %#x mask %#x\n",
tx_queue->queue, efx->txq_entries, tx_queue->ptr_mask);
/* Allocate software ring */
tx_queue->buffer = kcalloc(entries, sizeof(*tx_queue->buffer),
GFP_KERNEL);
if (!tx_queue->buffer)
return -ENOMEM;
tx_queue->cb_page = kcalloc(efx_tx_cb_page_count(tx_queue),
sizeof(tx_queue->cb_page[0]), GFP_KERNEL);
if (!tx_queue->cb_page) {
rc = -ENOMEM;
goto fail1;
}
/* Allocate hardware ring */
rc = efx_nic_probe_tx(tx_queue);
if (rc)
goto fail2;
return 0;
fail2:
kfree(tx_queue->cb_page);
tx_queue->cb_page = NULL;
fail1:
kfree(tx_queue->buffer);
tx_queue->buffer = NULL;
return rc;
}
void efx_init_tx_queue(struct efx_tx_queue *tx_queue)
{
struct efx_nic *efx = tx_queue->efx;
netif_dbg(efx, drv, efx->net_dev,
"initialising TX queue %d\n", tx_queue->queue);
tx_queue->insert_count = 0;
tx_queue->write_count = 0;
tx_queue->packet_write_count = 0;
tx_queue->old_write_count = 0;
tx_queue->read_count = 0;
tx_queue->old_read_count = 0;
tx_queue->empty_read_count = 0 | EFX_EMPTY_COUNT_VALID;
tx_queue->xmit_more_available = false;
tx_queue->timestamping = (efx_ptp_use_mac_tx_timestamps(efx) &&
tx_queue->channel == efx_ptp_channel(efx));
tx_queue->completed_timestamp_major = 0;
tx_queue->completed_timestamp_minor = 0;
tx_queue->xdp_tx = efx_channel_is_xdp_tx(tx_queue->channel);
/* Set up default function pointers. These may get replaced by
* efx_nic_init_tx() based off NIC/queue capabilities.
*/
tx_queue->handle_tso = efx_enqueue_skb_tso;
/* Set up TX descriptor ring */
efx_nic_init_tx(tx_queue);
tx_queue->initialised = true;
}
void efx_fini_tx_queue(struct efx_tx_queue *tx_queue)
{
struct efx_tx_buffer *buffer;
netif_dbg(tx_queue->efx, drv, tx_queue->efx->net_dev,
"shutting down TX queue %d\n", tx_queue->queue);
if (!tx_queue->buffer)
return;
/* Free any buffers left in the ring */
while (tx_queue->read_count != tx_queue->write_count) {
unsigned int pkts_compl = 0, bytes_compl = 0;
buffer = &tx_queue->buffer[tx_queue->read_count & tx_queue->ptr_mask];
efx_dequeue_buffer(tx_queue, buffer, &pkts_compl, &bytes_compl);
++tx_queue->read_count;
}
tx_queue->xmit_more_available = false;
netdev_tx_reset_queue(tx_queue->core_txq);
}
void efx_remove_tx_queue(struct efx_tx_queue *tx_queue)
{
int i;
if (!tx_queue->buffer)
return;
netif_dbg(tx_queue->efx, drv, tx_queue->efx->net_dev,
"destroying TX queue %d\n", tx_queue->queue);
efx_nic_remove_tx(tx_queue);
if (tx_queue->cb_page) {
for (i = 0; i < efx_tx_cb_page_count(tx_queue); i++)
efx_nic_free_buffer(tx_queue->efx,
&tx_queue->cb_page[i]);
kfree(tx_queue->cb_page);
tx_queue->cb_page = NULL;
}
kfree(tx_queue->buffer);
tx_queue->buffer = NULL;
}
void efx_dequeue_buffer(struct efx_tx_queue *tx_queue,
struct efx_tx_buffer *buffer,
unsigned int *pkts_compl,
unsigned int *bytes_compl)
{
if (buffer->unmap_len) {
struct device *dma_dev = &tx_queue->efx->pci_dev->dev;
dma_addr_t unmap_addr = buffer->dma_addr - buffer->dma_offset;
if (buffer->flags & EFX_TX_BUF_MAP_SINGLE)
dma_unmap_single(dma_dev, unmap_addr, buffer->unmap_len,
DMA_TO_DEVICE);
else
dma_unmap_page(dma_dev, unmap_addr, buffer->unmap_len,
DMA_TO_DEVICE);
buffer->unmap_len = 0;
}
if (buffer->flags & EFX_TX_BUF_SKB) {
struct sk_buff *skb = (struct sk_buff *)buffer->skb;
EFX_WARN_ON_PARANOID(!pkts_compl || !bytes_compl);
(*pkts_compl)++;
(*bytes_compl) += skb->len;
if (tx_queue->timestamping &&
(tx_queue->completed_timestamp_major ||
tx_queue->completed_timestamp_minor)) {
struct skb_shared_hwtstamps hwtstamp;
hwtstamp.hwtstamp =
efx_ptp_nic_to_kernel_time(tx_queue);
skb_tstamp_tx(skb, &hwtstamp);
tx_queue->completed_timestamp_major = 0;
tx_queue->completed_timestamp_minor = 0;
}
dev_consume_skb_any((struct sk_buff *)buffer->skb);
netif_vdbg(tx_queue->efx, tx_done, tx_queue->efx->net_dev,
"TX queue %d transmission id %x complete\n",
tx_queue->queue, tx_queue->read_count);
} else if (buffer->flags & EFX_TX_BUF_XDP) {
xdp_return_frame_rx_napi(buffer->xdpf);
}
buffer->len = 0;
buffer->flags = 0;
}
/* Remove packets from the TX queue
*
* This removes packets from the TX queue, up to and including the
* specified index.
*/
static void efx_dequeue_buffers(struct efx_tx_queue *tx_queue,
unsigned int index,
unsigned int *pkts_compl,
unsigned int *bytes_compl)
{
struct efx_nic *efx = tx_queue->efx;
unsigned int stop_index, read_ptr;
stop_index = (index + 1) & tx_queue->ptr_mask;
read_ptr = tx_queue->read_count & tx_queue->ptr_mask;
while (read_ptr != stop_index) {
struct efx_tx_buffer *buffer = &tx_queue->buffer[read_ptr];
if (!efx_tx_buffer_in_use(buffer)) {
netif_err(efx, tx_err, efx->net_dev,
"TX queue %d spurious TX completion id %d\n",
tx_queue->queue, read_ptr);
efx_schedule_reset(efx, RESET_TYPE_TX_SKIP);
return;
}
efx_dequeue_buffer(tx_queue, buffer, pkts_compl, bytes_compl);
++tx_queue->read_count;
read_ptr = tx_queue->read_count & tx_queue->ptr_mask;
}
}
void efx_xmit_done_check_empty(struct efx_tx_queue *tx_queue)
{
if ((int)(tx_queue->read_count - tx_queue->old_write_count) >= 0) {
tx_queue->old_write_count = READ_ONCE(tx_queue->write_count);
if (tx_queue->read_count == tx_queue->old_write_count) {
/* Ensure that read_count is flushed. */
smp_mb();
tx_queue->empty_read_count =
tx_queue->read_count | EFX_EMPTY_COUNT_VALID;
}
}
}
void efx_xmit_done(struct efx_tx_queue *tx_queue, unsigned int index)
{
unsigned int fill_level, pkts_compl = 0, bytes_compl = 0;
struct efx_nic *efx = tx_queue->efx;
struct efx_tx_queue *txq2;
EFX_WARN_ON_ONCE_PARANOID(index > tx_queue->ptr_mask);
efx_dequeue_buffers(tx_queue, index, &pkts_compl, &bytes_compl);
tx_queue->pkts_compl += pkts_compl;
tx_queue->bytes_compl += bytes_compl;
if (pkts_compl > 1)
++tx_queue->merge_events;
/* See if we need to restart the netif queue. This memory
* barrier ensures that we write read_count (inside
* efx_dequeue_buffers()) before reading the queue status.
*/
smp_mb();
if (unlikely(netif_tx_queue_stopped(tx_queue->core_txq)) &&
likely(efx->port_enabled) &&
likely(netif_device_present(efx->net_dev))) {
txq2 = efx_tx_queue_partner(tx_queue);
fill_level = max(tx_queue->insert_count - tx_queue->read_count,
txq2->insert_count - txq2->read_count);
if (fill_level <= efx->txq_wake_thresh)
netif_tx_wake_queue(tx_queue->core_txq);
}
efx_xmit_done_check_empty(tx_queue);
}
/* Remove buffers put into a tx_queue for the current packet.
* None of the buffers must have an skb attached.
*/
void efx_enqueue_unwind(struct efx_tx_queue *tx_queue,
unsigned int insert_count)
{
struct efx_tx_buffer *buffer;
unsigned int bytes_compl = 0;
unsigned int pkts_compl = 0;
/* Work backwards until we hit the original insert pointer value */
while (tx_queue->insert_count != insert_count) {
--tx_queue->insert_count;
buffer = __efx_tx_queue_get_insert_buffer(tx_queue);
efx_dequeue_buffer(tx_queue, buffer, &pkts_compl, &bytes_compl);
}
}
struct efx_tx_buffer *efx_tx_map_chunk(struct efx_tx_queue *tx_queue,
dma_addr_t dma_addr, size_t len)
{
const struct efx_nic_type *nic_type = tx_queue->efx->type;
struct efx_tx_buffer *buffer;
unsigned int dma_len;
/* Map the fragment taking account of NIC-dependent DMA limits. */
do {
buffer = efx_tx_queue_get_insert_buffer(tx_queue);
dma_len = nic_type->tx_limit_len(tx_queue, dma_addr, len);
buffer->len = dma_len;
buffer->dma_addr = dma_addr;
buffer->flags = EFX_TX_BUF_CONT;
len -= dma_len;
dma_addr += dma_len;
++tx_queue->insert_count;
} while (len);
return buffer;
}
int efx_tx_tso_header_length(struct sk_buff *skb)
{
size_t header_len;
if (skb->encapsulation)
header_len = skb_inner_transport_header(skb) -
skb->data +
(inner_tcp_hdr(skb)->doff << 2u);
else
header_len = skb_transport_header(skb) - skb->data +
(tcp_hdr(skb)->doff << 2u);
return header_len;
}
/* Map all data from an SKB for DMA and create descriptors on the queue. */
int efx_tx_map_data(struct efx_tx_queue *tx_queue, struct sk_buff *skb,
unsigned int segment_count)
{
struct efx_nic *efx = tx_queue->efx;
struct device *dma_dev = &efx->pci_dev->dev;
unsigned int frag_index, nr_frags;
dma_addr_t dma_addr, unmap_addr;
unsigned short dma_flags;
size_t len, unmap_len;
nr_frags = skb_shinfo(skb)->nr_frags;
frag_index = 0;
/* Map header data. */
len = skb_headlen(skb);
dma_addr = dma_map_single(dma_dev, skb->data, len, DMA_TO_DEVICE);
dma_flags = EFX_TX_BUF_MAP_SINGLE;
unmap_len = len;
unmap_addr = dma_addr;
if (unlikely(dma_mapping_error(dma_dev, dma_addr)))
return -EIO;
if (segment_count) {
/* For TSO we need to put the header in to a separate
* descriptor. Map this separately if necessary.
*/
size_t header_len = efx_tx_tso_header_length(skb);
if (header_len != len) {
tx_queue->tso_long_headers++;
efx_tx_map_chunk(tx_queue, dma_addr, header_len);
len -= header_len;
dma_addr += header_len;
}
}
/* Add descriptors for each fragment. */
do {
struct efx_tx_buffer *buffer;
skb_frag_t *fragment;
buffer = efx_tx_map_chunk(tx_queue, dma_addr, len);
/* The final descriptor for a fragment is responsible for
* unmapping the whole fragment.
*/
buffer->flags = EFX_TX_BUF_CONT | dma_flags;
buffer->unmap_len = unmap_len;
buffer->dma_offset = buffer->dma_addr - unmap_addr;
if (frag_index >= nr_frags) {
/* Store SKB details with the final buffer for
* the completion.
*/
buffer->skb = skb;
buffer->flags = EFX_TX_BUF_SKB | dma_flags;
return 0;
}
/* Move on to the next fragment. */
fragment = &skb_shinfo(skb)->frags[frag_index++];
len = skb_frag_size(fragment);
dma_addr = skb_frag_dma_map(dma_dev, fragment, 0, len,
DMA_TO_DEVICE);
dma_flags = 0;
unmap_len = len;
unmap_addr = dma_addr;
if (unlikely(dma_mapping_error(dma_dev, dma_addr)))
return -EIO;
} while (1);
}
unsigned int efx_tx_max_skb_descs(struct efx_nic *efx)
{
/* Header and payload descriptor for each output segment, plus
* one for every input fragment boundary within a segment
*/
unsigned int max_descs = EFX_TSO_MAX_SEGS * 2 + MAX_SKB_FRAGS;
/* Possibly one more per segment for option descriptors */
if (efx_nic_rev(efx) >= EFX_REV_HUNT_A0)
max_descs += EFX_TSO_MAX_SEGS;
/* Possibly more for PCIe page boundaries within input fragments */
if (PAGE_SIZE > EFX_PAGE_SIZE)
max_descs += max_t(unsigned int, MAX_SKB_FRAGS,
DIV_ROUND_UP(GSO_MAX_SIZE, EFX_PAGE_SIZE));
return max_descs;
}
/*
* Fallback to software TSO.
*
* This is used if we are unable to send a GSO packet through hardware TSO.
* This should only ever happen due to per-queue restrictions - unsupported
* packets should first be filtered by the feature flags.
*
* Returns 0 on success, error code otherwise.
*/
int efx_tx_tso_fallback(struct efx_tx_queue *tx_queue, struct sk_buff *skb)
{
struct sk_buff *segments, *next;
segments = skb_gso_segment(skb, 0);
if (IS_ERR(segments))
return PTR_ERR(segments);
dev_consume_skb_any(skb);
skb_list_walk_safe(segments, skb, next) {
skb_mark_not_on_list(skb);
efx_enqueue_skb(tx_queue, skb);
}
return 0;
}
|