ligang/vpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
ee577115ffad0147c75cb9a7dce2570afeed9a8c
vpp/src/plugins/avf/output.c
Leyi Rong 0638619e05 avf: 512-bit SIMD version of avf_tx_prepare 
Exploiting AVX-512 operations on avf_tx_prepare().

Type: improvement

Signed-off-by: Leyi Rong <leyi.rong@intel.com>
Change-Id: I01e0b4a2e2d440659b4298668a868d983f5091c3
2023-03-13 10:40:51 +08:00

540 lines
14 KiB
C
Raw Blame History

/*
*------------------------------------------------------------------
* Copyright (c) 2018 Cisco and/or its affiliates.
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at:
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*------------------------------------------------------------------
*/
#include <vlib/vlib.h>
#include <vlib/unix/unix.h>
#include <vlib/pci/pci.h>
#include <vppinfra/ring.h>
#include <vppinfra/vector/ip_csum.h>
#include <vnet/ethernet/ethernet.h>
#include <vnet/ip/ip4_packet.h>
#include <vnet/ip/ip6_packet.h>
#include <vnet/udp/udp_packet.h>
#include <vnet/tcp/tcp_packet.h>
#include <vnet/devices/devices.h>
#include <avf/avf.h>
static_always_inline u8
avf_tx_desc_get_dtyp (avf_tx_desc_t * d)
{
return d->qword[1] & 0x0f;
}
struct avf_ip4_psh
{
u32 src;
u32 dst;
u8 zero;
u8 proto;
u16 l4len;
};
struct avf_ip6_psh
{
ip6_address_t src;
ip6_address_t dst;
u32 l4len;
u32 proto;
};
static_always_inline u64
avf_tx_prepare_cksum (vlib_buffer_t * b, u8 is_tso)
{
u64 flags = 0;
if (!is_tso && !(b->flags & VNET_BUFFER_F_OFFLOAD))
return 0;
vnet_buffer_oflags_t oflags = vnet_buffer (b)->oflags;
u32 is_tcp = is_tso || oflags & VNET_BUFFER_OFFLOAD_F_TCP_CKSUM;
u32 is_udp = !is_tso && oflags & VNET_BUFFER_OFFLOAD_F_UDP_CKSUM;
if (!is_tcp && !is_udp)
return 0;
u32 is_ip4 = b->flags & VNET_BUFFER_F_IS_IP4;
u32 is_ip6 = b->flags & VNET_BUFFER_F_IS_IP6;
ASSERT (!(is_tcp && is_udp));
ASSERT (is_ip4 || is_ip6);
i16 l2_hdr_offset = b->current_data;
i16 l3_hdr_offset = vnet_buffer (b)->l3_hdr_offset;
i16 l4_hdr_offset = vnet_buffer (b)->l4_hdr_offset;
u16 l2_len = l3_hdr_offset - l2_hdr_offset;
u16 l3_len = l4_hdr_offset - l3_hdr_offset;
ip4_header_t *ip4 = (void *) (b->data + l3_hdr_offset);
ip6_header_t *ip6 = (void *) (b->data + l3_hdr_offset);
tcp_header_t *tcp = (void *) (b->data + l4_hdr_offset);
udp_header_t *udp = (void *) (b->data + l4_hdr_offset);
u16 l4_len = is_tcp ? tcp_header_bytes (tcp) : sizeof (udp_header_t);
u16 sum = 0;
flags |= AVF_TXD_OFFSET_MACLEN (l2_len) |
AVF_TXD_OFFSET_IPLEN (l3_len) | AVF_TXD_OFFSET_L4LEN (l4_len);
flags |= is_ip4 ? AVF_TXD_CMD_IIPT_IPV4 : AVF_TXD_CMD_IIPT_IPV6;
flags |= is_tcp ? AVF_TXD_CMD_L4T_TCP : AVF_TXD_CMD_L4T_UDP;
if (is_ip4)
ip4->checksum = 0;
if (is_tso)
{
if (is_ip4)
ip4->length = 0;
else
ip6->payload_length = 0;
}
if (is_ip4)
{
struct avf_ip4_psh psh = { 0 };
psh.src = ip4->src_address.as_u32;
psh.dst = ip4->dst_address.as_u32;
psh.proto = ip4->protocol;
psh.l4len =
is_tso ? 0 :
clib_host_to_net_u16 (clib_net_to_host_u16 (ip4->length) -
(l4_hdr_offset - l3_hdr_offset));
sum = ~clib_ip_csum ((u8 *) &psh, sizeof (psh));
}
else
{
struct avf_ip6_psh psh = { 0 };
psh.src = ip6->src_address;
psh.dst = ip6->dst_address;
psh.proto = clib_host_to_net_u32 ((u32) ip6->protocol);
psh.l4len = is_tso ? 0 : ip6->payload_length;
sum = ~clib_ip_csum ((u8 *) &psh, sizeof (psh));
}
if (is_tcp)
tcp->checksum = sum;
else
udp->checksum = sum;
return flags;
}
static_always_inline u32
avf_tx_fill_ctx_desc (vlib_main_t *vm, avf_txq_t *txq, avf_tx_desc_t *d,
vlib_buffer_t *b)
{
vlib_buffer_t *ctx_ph;
u32 *bi = txq->ph_bufs;
next:
ctx_ph = vlib_get_buffer (vm, bi[0]);
if (PREDICT_FALSE (ctx_ph->ref_count == 255))
{
bi++;
goto next;
}
/* Acquire a reference on the placeholder buffer */
ctx_ph->ref_count++;
u16 l234hdr_sz = vnet_buffer (b)->l4_hdr_offset - b->current_data +
vnet_buffer2 (b)->gso_l4_hdr_sz;
u16 tlen = vlib_buffer_length_in_chain (vm, b) - l234hdr_sz;
d[0].qword[0] = 0;
d[0].qword[1] = AVF_TXD_DTYP_CTX | AVF_TXD_CTX_CMD_TSO
| AVF_TXD_CTX_SEG_MSS (vnet_buffer2 (b)->gso_size) |
AVF_TXD_CTX_SEG_TLEN (tlen);
return bi[0];
}
static_always_inline void
avf_tx_copy_desc (avf_tx_desc_t *d, avf_tx_desc_t *s, u32 n_descs)
{
#if defined CLIB_HAVE_VEC512
while (n_descs >= 8)
{
u64x8u *dv = (u64x8u *) d;
u64x8u *sv = (u64x8u *) s;
dv[0] = sv[0];
dv[1] = sv[1];
/* next */
d += 8;
s += 8;
n_descs -= 8;
}
#elif defined CLIB_HAVE_VEC256
while (n_descs >= 4)
{
u64x4u *dv = (u64x4u *) d;
u64x4u *sv = (u64x4u *) s;
dv[0] = sv[0];
dv[1] = sv[1];
/* next */
d += 4;
s += 4;
n_descs -= 4;
}
#elif defined CLIB_HAVE_VEC128
while (n_descs >= 2)
{
u64x2u *dv = (u64x2u *) d;
u64x2u *sv = (u64x2u *) s;
dv[0] = sv[0];
dv[1] = sv[1];
/* next */
d += 2;
s += 2;
n_descs -= 2;
}
#endif
while (n_descs)
{
d[0].qword[0] = s[0].qword[0];
d[0].qword[1] = s[0].qword[1];
d++;
s++;
n_descs--;
}
}
static_always_inline void
avf_tx_fill_data_desc (vlib_main_t *vm, avf_tx_desc_t *d, vlib_buffer_t *b,
u64 cmd, int use_va_dma)
{
if (use_va_dma)
d->qword[0] = vlib_buffer_get_current_va (b);
else
d->qword[0] = vlib_buffer_get_current_pa (vm, b);
d->qword[1] = (((u64) b->current_length) << 34 | cmd | AVF_TXD_CMD_RSV);
}
static_always_inline u16
avf_tx_prepare (vlib_main_t *vm, vlib_node_runtime_t *node, avf_txq_t *txq,
u32 *buffers, u32 n_packets, u16 *n_enq_descs, int use_va_dma)
{
const u64 cmd_eop = AVF_TXD_CMD_EOP;
u16 n_free_desc, n_desc_left, n_packets_left = n_packets;
#if defined CLIB_HAVE_VEC512
vlib_buffer_t *b[8];
#else
vlib_buffer_t *b[4];
#endif
avf_tx_desc_t *d = txq->tmp_descs;
u32 *tb = txq->tmp_bufs;
n_free_desc = n_desc_left = txq->size - txq->n_enqueued - 8;
if (n_desc_left == 0)
return 0;
while (n_packets_left && n_desc_left)
{
#if defined CLIB_HAVE_VEC512
u32 flags;
u64x8 or_flags_vec512;
u64x8 flags_mask_vec512;
#else
u32 flags, or_flags;
#endif
#if defined CLIB_HAVE_VEC512
if (n_packets_left < 8 || n_desc_left < 8)
#else
if (n_packets_left < 8 || n_desc_left < 4)
#endif
goto one_by_one;
#if defined CLIB_HAVE_VEC512
u64x8 base_ptr = u64x8_splat (vm->buffer_main->buffer_mem_start);
u32x8 buf_indices = u32x8_load_unaligned (buffers);
*(u64x8 *) &b = base_ptr + u64x8_from_u32x8 (
buf_indices << CLIB_LOG2_CACHE_LINE_BYTES);
or_flags_vec512 = u64x8_i64gather (u64x8_load_unaligned (b), 0, 1);
#else
vlib_prefetch_buffer_with_index (vm, buffers[4], LOAD);
vlib_prefetch_buffer_with_index (vm, buffers[5], LOAD);
vlib_prefetch_buffer_with_index (vm, buffers[6], LOAD);
vlib_prefetch_buffer_with_index (vm, buffers[7], LOAD);
b[0] = vlib_get_buffer (vm, buffers[0]);
b[1] = vlib_get_buffer (vm, buffers[1]);
b[2] = vlib_get_buffer (vm, buffers[2]);
b[3] = vlib_get_buffer (vm, buffers[3]);
or_flags = b[0]->flags | b[1]->flags | b[2]->flags | b[3]->flags;
#endif
#if defined CLIB_HAVE_VEC512
flags_mask_vec512 = u64x8_splat (
VLIB_BUFFER_NEXT_PRESENT | VNET_BUFFER_F_OFFLOAD | VNET_BUFFER_F_GSO);
if (PREDICT_FALSE (
!u64x8_is_all_zero (or_flags_vec512 & flags_mask_vec512)))
#else
if (PREDICT_FALSE (or_flags &
(VLIB_BUFFER_NEXT_PRESENT | VNET_BUFFER_F_OFFLOAD |
VNET_BUFFER_F_GSO)))
#endif
goto one_by_one;
#if defined CLIB_HAVE_VEC512
vlib_buffer_copy_indices (tb, buffers, 8);
avf_tx_fill_data_desc (vm, d + 0, b[0], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 1, b[1], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 2, b[2], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 3, b[3], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 4, b[4], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 5, b[5], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 6, b[6], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 7, b[7], cmd_eop, use_va_dma);
buffers += 8;
n_packets_left -= 8;
n_desc_left -= 8;
d += 8;
tb += 8;
#else
vlib_buffer_copy_indices (tb, buffers, 4);
avf_tx_fill_data_desc (vm, d + 0, b[0], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 1, b[1], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 2, b[2], cmd_eop, use_va_dma);
avf_tx_fill_data_desc (vm, d + 3, b[3], cmd_eop, use_va_dma);
buffers += 4;
n_packets_left -= 4;
n_desc_left -= 4;
d += 4;
tb += 4;
#endif
continue;
one_by_one:
tb[0] = buffers[0];
b[0] = vlib_get_buffer (vm, buffers[0]);
flags = b[0]->flags;
/* No chained buffers or TSO case */
if (PREDICT_TRUE (
(flags & (VLIB_BUFFER_NEXT_PRESENT | VNET_BUFFER_F_GSO)) == 0))
{
u64 cmd = cmd_eop;
if (PREDICT_FALSE (flags & VNET_BUFFER_F_OFFLOAD))
cmd |= avf_tx_prepare_cksum (b[0], 0 /* is_tso */);
avf_tx_fill_data_desc (vm, d, b[0], cmd, use_va_dma);
}
else
{
u16 n_desc_needed = 1;
u64 cmd = 0;
if (flags & VLIB_BUFFER_NEXT_PRESENT)
{
vlib_buffer_t *next = vlib_get_buffer (vm, b[0]->next_buffer);
n_desc_needed = 2;
while (next->flags & VLIB_BUFFER_NEXT_PRESENT)
{
next = vlib_get_buffer (vm, next->next_buffer);
n_desc_needed++;
}
}
if (flags & VNET_BUFFER_F_GSO)
{
n_desc_needed++;
}
else if (PREDICT_FALSE (n_desc_needed > 8))
{
vlib_buffer_free_one (vm, buffers[0]);
vlib_error_count (vm, node->node_index,
AVF_TX_ERROR_SEGMENT_SIZE_EXCEEDED, 1);
n_packets_left -= 1;
buffers += 1;
continue;
}
if (PREDICT_FALSE (n_desc_left < n_desc_needed))
break;
if (flags & VNET_BUFFER_F_GSO)
{
/* Enqueue a context descriptor */
tb[1] = tb[0];
tb[0] = avf_tx_fill_ctx_desc (vm, txq, d, b[0]);
n_desc_left -= 1;
d += 1;
tb += 1;
cmd = avf_tx_prepare_cksum (b[0], 1 /* is_tso */);
}
else if (flags & VNET_BUFFER_F_OFFLOAD)
{
cmd = avf_tx_prepare_cksum (b[0], 0 /* is_tso */);
}
/* Deal with chain buffer if present */
while (b[0]->flags & VLIB_BUFFER_NEXT_PRESENT)
{
avf_tx_fill_data_desc (vm, d, b[0], cmd, use_va_dma);
n_desc_left -= 1;
d += 1;
tb += 1;
tb[0] = b[0]->next_buffer;
b[0] = vlib_get_buffer (vm, b[0]->next_buffer);
}
avf_tx_fill_data_desc (vm, d, b[0], cmd_eop | cmd, use_va_dma);
}
buffers += 1;
n_packets_left -= 1;
n_desc_left -= 1;
d += 1;
tb += 1;
}
*n_enq_descs = n_free_desc - n_desc_left;
return n_packets - n_packets_left;
}
VNET_DEVICE_CLASS_TX_FN (avf_device_class) (vlib_main_t * vm,
vlib_node_runtime_t * node,
vlib_frame_t * frame)
{
vnet_interface_output_runtime_t *rd = (void *) node->runtime_data;
avf_device_t *ad = avf_get_device (rd->dev_instance);
vnet_hw_if_tx_frame_t *tf = vlib_frame_scalar_args (frame);
u8 qid = tf->queue_id;
avf_txq_t *txq = vec_elt_at_index (ad->txqs, qid);
u16 next;
u16 mask = txq->size - 1;
u32 *buffers = vlib_frame_vector_args (frame);
u16 n_enq, n_left, n_desc, *slot;
u16 n_retry = 2;
if (tf->shared_queue)
clib_spinlock_lock (&txq->lock);
n_left = frame->n_vectors;
retry:
next = txq->next;
/* release consumed bufs */
if (txq->n_enqueued)
{
i32 complete_slot = -1;
while (1)
{
u16 *slot = clib_ring_get_first (txq->rs_slots);
if (slot == 0)
break;
if (avf_tx_desc_get_dtyp (txq->descs + slot[0]) != 0x0F)
break;
complete_slot = slot[0];
clib_ring_deq (txq->rs_slots);
}
if (complete_slot >= 0)
{
u16 first, mask, n_free;
mask = txq->size - 1;
first = (txq->next - txq->n_enqueued) & mask;
n_free = (complete_slot + 1 - first) & mask;
txq->n_enqueued -= n_free;
vlib_buffer_free_from_ring_no_next (vm, txq->bufs, first, txq->size,
n_free);
}
}
n_desc = 0;
if (ad->flags & AVF_DEVICE_F_VA_DMA)
n_enq = avf_tx_prepare (vm, node, txq, buffers, n_left, &n_desc, 1);
else
n_enq = avf_tx_prepare (vm, node, txq, buffers, n_left, &n_desc, 0);
if (n_desc)
{
if (PREDICT_TRUE (next + n_desc <= txq->size))
{
/* no wrap */
avf_tx_copy_desc (txq->descs + next, txq->tmp_descs, n_desc);
vlib_buffer_copy_indices (txq->bufs + next, txq->tmp_bufs, n_desc);
}
else
{
/* wrap */
u32 n_not_wrap = txq->size - next;
avf_tx_copy_desc (txq->descs + next, txq->tmp_descs, n_not_wrap);
avf_tx_copy_desc (txq->descs, txq->tmp_descs + n_not_wrap,
n_desc - n_not_wrap);
vlib_buffer_copy_indices (txq->bufs + next, txq->tmp_bufs,
n_not_wrap);
vlib_buffer_copy_indices (txq->bufs, txq->tmp_bufs + n_not_wrap,
n_desc - n_not_wrap);
}
next += n_desc;
if ((slot = clib_ring_enq (txq->rs_slots)))
{
u16 rs_slot = slot[0] = (next - 1) & mask;
txq->descs[rs_slot].qword[1] |= AVF_TXD_CMD_RS;
}
txq->next = next & mask;
avf_tail_write (txq->qtx_tail, txq->next);
txq->n_enqueued += n_desc;
n_left -= n_enq;
}
if (n_left)
{
buffers += n_enq;
if (n_retry--)
goto retry;
vlib_buffer_free (vm, buffers, n_left);
vlib_error_count (vm, node->node_index,
AVF_TX_ERROR_NO_FREE_SLOTS, n_left);
}
if (tf->shared_queue)
clib_spinlock_unlock (&txq->lock);
return frame->n_vectors - n_left;
}
/*
* fd.io coding-style-patch-verification: ON
*
* Local Variables:
* eval: (c-set-style "gnu")
* End:
*/
Reference in New Issue View Git Blame Copy Permalink