From: Nicolas Chautru <nicolas.chautru@intel.com>
To: dev@dpdk.org, gakhil@marvell.com, maxime.coquelin@redhat.com
Cc: trix@redhat.com, mdr@ashroe.eu, bruce.richardson@intel.com,
hemant.agrawal@nxp.com, david.marchand@redhat.com,
stephen@networkplumber.org, hernan.vargas@intel.com,
Nic Chautru <nicolas.chautru@intel.com>
Subject: [PATCH v8 10/14] baseband/acc: add support for FFT operations
Date: Fri, 7 Oct 2022 12:03:25 -0700 [thread overview]
Message-ID: <20221007190329.25381-11-nicolas.chautru@intel.com> (raw)
In-Reply-To: <20221007190329.25381-1-nicolas.chautru@intel.com>
From: Nic Chautru <nicolas.chautru@intel.com>
Added functions and capability for FFT processing
Signed-off-by: Nic Chautru <nicolas.chautru@intel.com>
---
drivers/baseband/acc/rte_acc200_pmd.c | 240 +++++++++++++++++++++++++-
1 file changed, 238 insertions(+), 2 deletions(-)
diff --git a/drivers/baseband/acc/rte_acc200_pmd.c b/drivers/baseband/acc/rte_acc200_pmd.c
index f56f9b07ab..7496f29efb 100644
--- a/drivers/baseband/acc/rte_acc200_pmd.c
+++ b/drivers/baseband/acc/rte_acc200_pmd.c
@@ -734,6 +734,21 @@ acc200_dev_info_get(struct rte_bbdev *dev,
.num_buffers_soft_out = 0,
}
},
+ {
+ .type = RTE_BBDEV_OP_FFT,
+ .cap.fft = {
+ .capability_flags =
+ RTE_BBDEV_FFT_WINDOWING |
+ RTE_BBDEV_FFT_CS_ADJUSTMENT |
+ RTE_BBDEV_FFT_DFT_BYPASS |
+ RTE_BBDEV_FFT_IDFT_BYPASS |
+ RTE_BBDEV_FFT_WINDOWING_BYPASS,
+ .num_buffers_src =
+ RTE_BBDEV_LDPC_MAX_CODE_BLOCKS,
+ .num_buffers_dst =
+ RTE_BBDEV_LDPC_MAX_CODE_BLOCKS,
+ }
+ },
RTE_BBDEV_END_OF_CAPABILITIES_LIST()
};
@@ -756,12 +771,13 @@ acc200_dev_info_get(struct rte_bbdev *dev,
d->acc_conf.q_ul_5g.num_qgroups;
dev_info->num_queues[RTE_BBDEV_OP_LDPC_ENC] = d->acc_conf.q_dl_5g.num_aqs_per_groups *
d->acc_conf.q_dl_5g.num_qgroups;
- dev_info->num_queues[RTE_BBDEV_OP_FFT] = 0;
+ dev_info->num_queues[RTE_BBDEV_OP_FFT] = d->acc_conf.q_fft.num_aqs_per_groups *
+ d->acc_conf.q_fft.num_qgroups;
dev_info->queue_priority[RTE_BBDEV_OP_TURBO_DEC] = d->acc_conf.q_ul_4g.num_qgroups;
dev_info->queue_priority[RTE_BBDEV_OP_TURBO_ENC] = d->acc_conf.q_dl_4g.num_qgroups;
dev_info->queue_priority[RTE_BBDEV_OP_LDPC_DEC] = d->acc_conf.q_ul_5g.num_qgroups;
dev_info->queue_priority[RTE_BBDEV_OP_LDPC_ENC] = d->acc_conf.q_dl_5g.num_qgroups;
- dev_info->queue_priority[RTE_BBDEV_OP_FFT] = 0;
+ dev_info->queue_priority[RTE_BBDEV_OP_FFT] = d->acc_conf.q_fft.num_qgroups;
dev_info->max_num_queues = 0;
for (i = RTE_BBDEV_OP_NONE; i <= RTE_BBDEV_OP_FFT; i++)
dev_info->max_num_queues += dev_info->num_queues[i];
@@ -2831,6 +2847,224 @@ acc200_dequeue_ldpc_dec(struct rte_bbdev_queue_data *q_data,
return i;
}
+/* Fill in a frame control word for FFT processing. */
+static inline void
+acc200_fcw_fft_fill(struct rte_bbdev_fft_op *op, struct acc_fcw_fft *fcw)
+{
+ fcw->in_frame_size = op->fft.input_sequence_size;
+ fcw->leading_pad_size = op->fft.input_leading_padding;
+ fcw->out_frame_size = op->fft.output_sequence_size;
+ fcw->leading_depad_size = op->fft.output_leading_depadding;
+ fcw->cs_window_sel = op->fft.window_index[0] +
+ (op->fft.window_index[1] << 8) +
+ (op->fft.window_index[2] << 16) +
+ (op->fft.window_index[3] << 24);
+ fcw->cs_window_sel2 = op->fft.window_index[4] +
+ (op->fft.window_index[5] << 8);
+ fcw->cs_enable_bmap = op->fft.cs_bitmap;
+ fcw->num_antennas = op->fft.num_antennas_log2;
+ fcw->idft_size = op->fft.idft_log2;
+ fcw->dft_size = op->fft.dft_log2;
+ fcw->cs_offset = op->fft.cs_time_adjustment;
+ fcw->idft_shift = op->fft.idft_shift;
+ fcw->dft_shift = op->fft.dft_shift;
+ fcw->cs_multiplier = op->fft.ncs_reciprocal;
+ if (check_bit(op->fft.op_flags,
+ RTE_BBDEV_FFT_IDFT_BYPASS)) {
+ if (check_bit(op->fft.op_flags,
+ RTE_BBDEV_FFT_WINDOWING_BYPASS))
+ fcw->bypass = 2;
+ else
+ fcw->bypass = 1;
+ } else if (check_bit(op->fft.op_flags,
+ RTE_BBDEV_FFT_DFT_BYPASS))
+ fcw->bypass = 3;
+ else
+ fcw->bypass = 0;
+}
+
+static inline int
+acc200_dma_desc_fft_fill(struct rte_bbdev_fft_op *op,
+ struct acc_dma_req_desc *desc,
+ struct rte_mbuf *input, struct rte_mbuf *output,
+ uint32_t *in_offset, uint32_t *out_offset)
+{
+ /* FCW already done */
+ acc_header_init(desc);
+ desc->data_ptrs[1].address =
+ rte_pktmbuf_iova_offset(input, *in_offset);
+ desc->data_ptrs[1].blen = op->fft.input_sequence_size * 4;
+ desc->data_ptrs[1].blkid = ACC_DMA_BLKID_IN;
+ desc->data_ptrs[1].last = 1;
+ desc->data_ptrs[1].dma_ext = 0;
+ desc->data_ptrs[2].address =
+ rte_pktmbuf_iova_offset(output, *out_offset);
+ desc->data_ptrs[2].blen = op->fft.output_sequence_size * 4;
+ desc->data_ptrs[2].blkid = ACC_DMA_BLKID_OUT_HARD;
+ desc->data_ptrs[2].last = 1;
+ desc->data_ptrs[2].dma_ext = 0;
+ desc->m2dlen = 2;
+ desc->d2mlen = 1;
+ desc->ib_ant_offset = op->fft.input_sequence_size;
+ desc->num_ant = op->fft.num_antennas_log2 - 3;
+ int num_cs = 0, i;
+ for (i = 0; i < 12; i++)
+ if (check_bit(op->fft.cs_bitmap, 1 << i))
+ num_cs++;
+ desc->num_cs = num_cs;
+ desc->ob_cyc_offset = op->fft.output_sequence_size;
+ desc->ob_ant_offset = op->fft.output_sequence_size * num_cs;
+ desc->op_addr = op;
+ return 0;
+}
+
+
+/** Enqueue one FFT operation for ACC200 device. */
+static inline int
+enqueue_fft_one_op(struct acc_queue *q, struct rte_bbdev_fft_op *op,
+ uint16_t total_enqueued_cbs)
+{
+ union acc_dma_desc *desc;
+ uint16_t desc_idx;
+ struct rte_mbuf *input, *output;
+ uint32_t in_offset, out_offset;
+ struct acc_fcw_fft *fcw;
+
+ desc_idx = (q->sw_ring_head + total_enqueued_cbs) & q->sw_ring_wrap_mask;
+ desc = q->ring_addr + desc_idx;
+ input = op->fft.base_input.data;
+ output = op->fft.base_output.data;
+ in_offset = op->fft.base_input.offset;
+ out_offset = op->fft.base_output.offset;
+ fcw = &desc->req.fcw_fft;
+
+ acc200_fcw_fft_fill(op, fcw);
+ acc200_dma_desc_fft_fill(op, &desc->req, input, output, &in_offset, &out_offset);
+#ifdef RTE_LIBRTE_BBDEV_DEBUG
+ rte_memdump(stderr, "FCW", &desc->req.fcw_fft,
+ sizeof(desc->req.fcw_fft));
+ rte_memdump(stderr, "Req Desc.", desc, sizeof(*desc));
+#endif
+ return 1;
+}
+
+/* Enqueue decode operations for ACC200 device. */
+static uint16_t
+acc200_enqueue_fft(struct rte_bbdev_queue_data *q_data,
+ struct rte_bbdev_fft_op **ops, uint16_t num)
+{
+ struct acc_queue *q;
+ int32_t aq_avail, avail;
+ uint16_t i;
+ union acc_dma_desc *desc;
+ int ret;
+
+ aq_avail = acc_aq_avail(q_data, num);
+ if (unlikely((aq_avail <= 0) || (num == 0)))
+ return 0;
+ q = q_data->queue_private;
+ avail = acc_ring_avail_enq(q);
+
+ for (i = 0; i < num; ++i) {
+ /* Check if there are available space for further processing. */
+ if (unlikely(avail < 1))
+ break;
+ avail -= 1;
+ ret = enqueue_fft_one_op(q, ops[i], i);
+ if (ret < 0)
+ break;
+ }
+
+ if (unlikely(i == 0))
+ return 0; /* Nothing to enqueue. */
+
+ /* Set SDone in last CB in enqueued ops for CB mode. */
+ desc = q->ring_addr + ((q->sw_ring_head + i - 1) & q->sw_ring_wrap_mask);
+
+ desc->req.sdone_enable = 1;
+ desc->req.irq_enable = q->irq_enable;
+ acc_dma_enqueue(q, i, &q_data->queue_stats);
+
+ /* Update stats */
+ q_data->queue_stats.enqueued_count += i;
+ q_data->queue_stats.enqueue_err_count += num - i;
+ return i;
+}
+
+
+/* Dequeue one FFT operations from ACC200 device. */
+static inline int
+dequeue_fft_one_op(struct rte_bbdev_queue_data *q_data,
+ struct acc_queue *q, struct rte_bbdev_fft_op **ref_op,
+ uint16_t dequeued_cbs, uint32_t *aq_dequeued)
+{
+ union acc_dma_desc *desc, atom_desc;
+ union acc_dma_rsp_desc rsp;
+ struct rte_bbdev_fft_op *op;
+
+ desc = q->ring_addr + ((q->sw_ring_tail + dequeued_cbs) & q->sw_ring_wrap_mask);
+ atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, __ATOMIC_RELAXED);
+
+ /* Check fdone bit */
+ if (!(atom_desc.rsp.val & ACC_FDONE))
+ return -1;
+
+ rsp.val = atom_desc.rsp.val;
+#ifdef RTE_LIBRTE_BBDEV_DEBUG
+ rte_memdump(stderr, "Resp", &desc->rsp.val,
+ sizeof(desc->rsp.val));
+#endif
+ /* Dequeue. */
+ op = desc->req.op_addr;
+
+ /* Clearing status, it will be set based on response. */
+ op->status = 0;
+ op->status |= rsp.input_err << RTE_BBDEV_DATA_ERROR;
+ op->status |= rsp.dma_err << RTE_BBDEV_DRV_ERROR;
+ op->status |= rsp.fcw_err << RTE_BBDEV_DRV_ERROR;
+ if (op->status != 0)
+ q_data->queue_stats.dequeue_err_count++;
+
+ /* Check if this is the last desc in batch (Atomic Queue). */
+ if (desc->req.last_desc_in_batch) {
+ (*aq_dequeued)++;
+ desc->req.last_desc_in_batch = 0;
+ }
+ desc->rsp.val = ACC_DMA_DESC_TYPE;
+ desc->rsp.add_info_0 = 0;
+ *ref_op = op;
+ /* One CB (op) was successfully dequeued. */
+ return 1;
+}
+
+
+/* Dequeue FFT operations from ACC200 device. */
+static uint16_t
+acc200_dequeue_fft(struct rte_bbdev_queue_data *q_data,
+ struct rte_bbdev_fft_op **ops, uint16_t num)
+{
+ struct acc_queue *q = q_data->queue_private;
+ uint16_t dequeue_num, i, dequeued_cbs = 0;
+ uint32_t avail = acc_ring_avail_deq(q);
+ uint32_t aq_dequeued = 0;
+ int ret;
+
+ dequeue_num = RTE_MIN(avail, num);
+
+ for (i = 0; i < dequeue_num; ++i) {
+ ret = dequeue_fft_one_op(q_data, q, &ops[i], dequeued_cbs, &aq_dequeued);
+ if (ret <= 0)
+ break;
+ dequeued_cbs += ret;
+ }
+
+ q->aq_dequeued += aq_dequeued;
+ q->sw_ring_tail += dequeued_cbs;
+ /* Update enqueue stats. */
+ q_data->queue_stats.dequeued_count += i;
+ return i;
+}
+
/* Initialization Function */
static void
acc200_bbdev_init(struct rte_bbdev *dev, struct rte_pci_driver *drv)
@@ -2846,6 +3080,8 @@ acc200_bbdev_init(struct rte_bbdev *dev, struct rte_pci_driver *drv)
dev->enqueue_ldpc_dec_ops = acc200_enqueue_ldpc_dec;
dev->dequeue_ldpc_enc_ops = acc200_dequeue_ldpc_enc;
dev->dequeue_ldpc_dec_ops = acc200_dequeue_ldpc_dec;
+ dev->enqueue_fft_ops = acc200_enqueue_fft;
+ dev->dequeue_fft_ops = acc200_dequeue_fft;
((struct acc_device *) dev->data->dev_private)->pf_device =
!strcmp(drv->driver.name,
--
2.37.1
next prev parent reply other threads:[~2022-10-07 19:05 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-07 19:03 From: Nicolas Chautru <nicolas.chautru@intel.com> Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 01/14] baseband/acc100: remove unused registers Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 02/14] baseband/acc100: refactor to segregate common code Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 03/14] baseband/acc: rename directory from acc100 to acc Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 04/14] baseband/acc: introduce PMD for ACC200 Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 05/14] baseband/acc: add HW register definitions " Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 06/14] baseband/acc: add info get function " Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 07/14] baseband/acc: add queue configuration " Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 08/14] baseband/acc: add LDPC processing functions Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 09/14] baseband/acc: add LTE " Nicolas Chautru
2022-10-07 19:03 ` Nicolas Chautru [this message]
2022-10-07 19:03 ` [PATCH v8 11/14] baseband/acc: support interrupt Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 12/14] baseband/acc: add device status and vf2pf comms Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 13/14] baseband/acc: add PF configure companion function Nicolas Chautru
2022-10-07 19:03 ` [PATCH v8 14/14] baseband/acc: simplify meson dependency Nicolas Chautru
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20221007190329.25381-11-nicolas.chautru@intel.com \
--to=nicolas.chautru@intel.com \
--cc=bruce.richardson@intel.com \
--cc=david.marchand@redhat.com \
--cc=dev@dpdk.org \
--cc=gakhil@marvell.com \
--cc=hemant.agrawal@nxp.com \
--cc=hernan.vargas@intel.com \
--cc=maxime.coquelin@redhat.com \
--cc=mdr@ashroe.eu \
--cc=stephen@networkplumber.org \
--cc=trix@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).