From: Jerin Jacob <jerinjacobk@gmail.com>
To: Nithin Dabilpuram <ndabilpuram@marvell.com>
Cc: Pavan Nikhilesh <pbhagavatula@marvell.com>,
Shijith Thotton <sthotton@marvell.com>,
Kiran Kumar K <kirankumark@marvell.com>,
Sunil Kumar Kori <skori@marvell.com>,
Satha Rao <skoteshwar@marvell.com>,
jerinj@marvell.com, dev@dpdk.org
Subject: Re: [PATCH v3] net/cnxk: add support for reassembly of multi-seg pkts
Date: Thu, 15 Jun 2023 13:36:36 +0530 [thread overview]
Message-ID: <CALBAE1PDXKu+q87nw6P7rUN54yN75pFP-fuq0=piP4dhdhTCyQ@mail.gmail.com> (raw)
In-Reply-To: <20230613150904.765725-1-ndabilpuram@marvell.com>
On Tue, Jun 13, 2023 at 8:39 PM Nithin Dabilpuram
<ndabilpuram@marvell.com> wrote:
>
> Add support for HW reassembly of multi-seg pkts.
> Also optimize the code for normal reassembly path.
>
> Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Updated the release notes as folllowing
diff --git a/doc/guides/rel_notes/release_23_07.rst
b/doc/guides/rel_notes/release_23_07.rst
index d6af8c25a7..e0b706b540 100644
--- a/doc/guides/rel_notes/release_23_07.rst
+++ b/doc/guides/rel_notes/release_23_07.rst
@@ -144,6 +144,7 @@ New Features
* Extended ``RTE_FLOW_ACTION_TYPE_PORT_ID`` to redirect traffic
across PF ports.
* Added support for Inline MACsec processing using rte_security framework
for CN103 platform.
+ * Added support for reassembly of multi-seg packets.
Updated the git commit as follows and applied to
dpdk-next-net-mrvl/for-next-net. Thanks
net/cnxk: support reassembly of multi-seg packets
Add support for HW reassembly of multi-seg packets.
Also optimize the code for normal reassembly path.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
> ---
> v3:
> - No changes, resend after rebase
>
> v2:
> - Removed change id
>
> drivers/event/cnxk/cn10k_worker.h | 21 +-
> drivers/net/cnxk/cn10k_rx.h | 714 +++++++++++++++---------------
> 2 files changed, 375 insertions(+), 360 deletions(-)
>
> diff --git a/drivers/event/cnxk/cn10k_worker.h b/drivers/event/cnxk/cn10k_worker.h
> index a01894ae10..f5d7d879a7 100644
> --- a/drivers/event/cnxk/cn10k_worker.h
> +++ b/drivers/event/cnxk/cn10k_worker.h
> @@ -16,7 +16,7 @@
> static __rte_always_inline void
> cn10k_wqe_to_mbuf(uint64_t wqe, const uint64_t __mbuf, uint8_t port_id,
> const uint32_t tag, const uint32_t flags,
> - const void *const lookup_mem)
> + const void *const lookup_mem, uintptr_t cpth, uintptr_t sa_base)
> {
> const uint64_t mbuf_init = 0x100010000ULL | RTE_PKTMBUF_HEADROOM |
> (flags & NIX_RX_OFFLOAD_TSTAMP_F ? 8 : 0);
> @@ -24,7 +24,7 @@ cn10k_wqe_to_mbuf(uint64_t wqe, const uint64_t __mbuf, uint8_t port_id,
>
> cn10k_nix_cqe_to_mbuf((struct nix_cqe_hdr_s *)wqe, tag,
> (struct rte_mbuf *)mbuf, lookup_mem,
> - mbuf_init | ((uint64_t)port_id) << 48, flags);
> + mbuf_init | ((uint64_t)port_id) << 48, cpth, sa_base, flags);
> }
>
> static void
> @@ -59,6 +59,7 @@ cn10k_process_vwqe(uintptr_t vwqe, uint16_t port_id, const uint32_t flags, struc
> uint16_t lmt_id, d_off;
> struct rte_mbuf **wqe;
> struct rte_mbuf *mbuf;
> + uintptr_t cpth = 0;
> uint8_t loff = 0;
> uint64_t sa_base;
> int i;
> @@ -122,13 +123,15 @@ cn10k_process_vwqe(uintptr_t vwqe, uint16_t port_id, const uint32_t flags, struc
> const uint64_t cq_w1 = *((const uint64_t *)cqe + 1);
> const uint64_t cq_w5 = *((const uint64_t *)cqe + 5);
>
> + cpth = ((uintptr_t)mbuf + (uint16_t)d_off);
> +
> mbuf = nix_sec_meta_to_mbuf_sc(cq_w1, cq_w5, sa_base, laddr,
> &loff, mbuf, d_off,
> flags, mbuf_init);
> }
>
> cn10k_nix_cqe_to_mbuf(cqe, cqe->tag, mbuf, lookup_mem,
> - mbuf_init, flags);
> + mbuf_init, cpth, sa_base, flags);
>
> if (flags & NIX_RX_OFFLOAD_TSTAMP_F)
> cn10k_sso_process_tstamp((uint64_t)wqe[0],
> @@ -149,6 +152,8 @@ static __rte_always_inline void
> cn10k_sso_hws_post_process(struct cn10k_sso_hws *ws, uint64_t *u64,
> const uint32_t flags)
> {
> + uintptr_t sa_base = 0;
> +
> u64[0] = (u64[0] & (0x3ull << 32)) << 6 |
> (u64[0] & (0x3FFull << 36)) << 4 | (u64[0] & 0xffffffff);
> if (CNXK_EVENT_TYPE_FROM_TAG(u64[0]) == RTE_EVENT_TYPE_CRYPTODEV) {
> @@ -157,6 +162,7 @@ cn10k_sso_hws_post_process(struct cn10k_sso_hws *ws, uint64_t *u64,
> u64[1] = cn10k_cpt_crypto_adapter_vector_dequeue(u64[1]);
> } else if (CNXK_EVENT_TYPE_FROM_TAG(u64[0]) == RTE_EVENT_TYPE_ETHDEV) {
> uint8_t port = CNXK_SUB_EVENT_FROM_TAG(u64[0]);
> + uintptr_t cpth = 0;
> uint64_t mbuf;
>
> mbuf = u64[1] - sizeof(struct rte_mbuf);
> @@ -174,7 +180,6 @@ cn10k_sso_hws_post_process(struct cn10k_sso_hws *ws, uint64_t *u64,
> 0x100010000ULL | RTE_PKTMBUF_HEADROOM |
> (flags & NIX_RX_OFFLOAD_TSTAMP_F ? 8 : 0);
> struct rte_mbuf *m;
> - uintptr_t sa_base;
> uint64_t iova = 0;
> uint8_t loff = 0;
> uint16_t d_off;
> @@ -190,12 +195,14 @@ cn10k_sso_hws_post_process(struct cn10k_sso_hws *ws, uint64_t *u64,
> sa_base = cnxk_nix_sa_base_get(port, ws->lookup_mem);
> sa_base &= ~(ROC_NIX_INL_SA_BASE_ALIGN - 1);
>
> + cpth = ((uintptr_t)mbuf + (uint16_t)d_off);
> + mp = (struct rte_mempool *)cnxk_nix_inl_metapool_get(port, lookup_mem);
> + meta_aura = mp ? mp->pool_id : m->pool->pool_id;
> +
> mbuf = (uint64_t)nix_sec_meta_to_mbuf_sc(
> cq_w1, cq_w5, sa_base, (uintptr_t)&iova, &loff,
> (struct rte_mbuf *)mbuf, d_off, flags,
> mbuf_init | ((uint64_t)port) << 48);
> - mp = (struct rte_mempool *)cnxk_nix_inl_metapool_get(port, lookup_mem);
> - meta_aura = mp ? mp->pool_id : m->pool->pool_id;
>
> if (loff)
> roc_npa_aura_op_free(meta_aura, 0, iova);
> @@ -203,7 +210,7 @@ cn10k_sso_hws_post_process(struct cn10k_sso_hws *ws, uint64_t *u64,
>
> u64[0] = CNXK_CLR_SUB_EVENT(u64[0]);
> cn10k_wqe_to_mbuf(u64[1], mbuf, port, u64[0] & 0xFFFFF, flags,
> - ws->lookup_mem);
> + ws->lookup_mem, cpth, sa_base);
> if (flags & NIX_RX_OFFLOAD_TSTAMP_F)
> cn10k_sso_process_tstamp(u64[1], mbuf,
> ws->tstamp[port]);
> diff --git a/drivers/net/cnxk/cn10k_rx.h b/drivers/net/cnxk/cn10k_rx.h
> index 68c8c73790..b7074906a6 100644
> --- a/drivers/net/cnxk/cn10k_rx.h
> +++ b/drivers/net/cnxk/cn10k_rx.h
> @@ -144,18 +144,126 @@ nix_sec_flush_meta(uintptr_t laddr, uint16_t lmt_id, uint8_t loff,
> roc_lmt_submit_steorl(lmt_id, pa);
> }
>
> +#if defined(RTE_ARCH_ARM64)
> +static __rte_always_inline uint64_t
> +nix_sec_reass_frags_get(const struct cpt_parse_hdr_s *hdr, struct rte_mbuf **next_mbufs)
> +{
> + const struct cpt_frag_info_s *finfo;
> + uint32_t offset = hdr->w2.fi_offset;
> + const uint64_t *frag_ptr;
> + uint64x2_t frags23;
> + uint16x4_t fsz_w1;
> +
> + /* offset of 0 implies 256B, otherwise it implies offset*8B */
> + offset = (((offset - 1) & 0x1f) + 1) * 8;
> + finfo = RTE_PTR_ADD(hdr, offset);
> + frag_ptr = (const uint64_t *)(finfo + 1);
> + frags23 = vrev64q_u8(vld1q_u64(frag_ptr));
> +
> + next_mbufs[0] = ((struct rte_mbuf *)rte_be_to_cpu_64(hdr->frag1_wqe_ptr) - 1);
> + next_mbufs[1] = ((struct rte_mbuf *)vgetq_lane_u64(frags23, 0) - 1);
> + next_mbufs[2] = ((struct rte_mbuf *)vgetq_lane_u64(frags23, 1) - 1);
> +
> + fsz_w1 = vdup_n_u64(finfo->w1.u64);
> + fsz_w1 = vrev16_u8(fsz_w1);
> + return vget_lane_u64(fsz_w1, 0);
> +}
> +
> +static __rte_always_inline void
> +nix_sec_reass_first_frag_update(struct rte_mbuf *head, const uint8_t *m_ipptr,
> + uint64_t fsz, uint64_t cq_w1, uint16_t *ihl)
> +{
> + union nix_rx_parse_u *rx = (union nix_rx_parse_u *)((uintptr_t)(head + 1) + 8);
> + uint16_t fragx_sum = vaddv_u16(vdup_n_u64(fsz));
> + uint8_t lcptr = rx->lcptr;
> + uint16_t tot_len;
> + uint32_t cksum;
> + uint8_t *ipptr;
> +
> + ipptr = (uint8_t *)head->buf_addr + head->data_off + lcptr;
> + /* Find the L3 header length and update inner pkt based on meta lc type */
> + if (((cq_w1 >> 40) & 0xF) == NPC_LT_LC_IP) {
> + const struct rte_ipv4_hdr *m_hdr = (const struct rte_ipv4_hdr *)m_ipptr;
> + struct rte_ipv4_hdr *hdr = (struct rte_ipv4_hdr *)ipptr;
> +
> + *ihl = (m_hdr->version_ihl & 0xf) << 2;
> +
> + hdr->fragment_offset = 0;
> + tot_len = rte_cpu_to_be_16(fragx_sum + *ihl);
> + hdr->total_length = tot_len;
> + /* Perform incremental checksum based on meta pkt ip hdr */
> + cksum = m_hdr->hdr_checksum;
> + cksum += m_hdr->fragment_offset;
> + cksum += 0xFFFF;
> + cksum += m_hdr->total_length;
> + cksum += (uint16_t)(~tot_len);
> + cksum = (cksum & 0xFFFF) + ((cksum & 0xFFFF0000) >> 16);
> + hdr->hdr_checksum = cksum;
> +
> + head->pkt_len = lcptr + *ihl + fragx_sum;
> + } else {
> + struct rte_ipv6_hdr *hdr = (struct rte_ipv6_hdr *)ipptr;
> + size_t ext_len = sizeof(struct rte_ipv6_hdr);
> + uint8_t *nxt_hdr = (uint8_t *)hdr;
> + int nh = hdr->proto;
> +
> + *ihl = 0;
> + while (nh != -EINVAL) {
> + nxt_hdr += ext_len;
> + *ihl += ext_len;
> + nh = rte_ipv6_get_next_ext(nxt_hdr, nh, &ext_len);
> + }
> +
> + /* Remove the frag header by moving header 8 bytes forward */
> + hdr->payload_len = rte_cpu_to_be_16(fragx_sum + *ihl -
> + 8 - sizeof(struct rte_ipv6_hdr));
> +
> + rte_memcpy(rte_pktmbuf_mtod_offset(head, void *, 8),
> + rte_pktmbuf_mtod(head, void *),
> + lcptr + sizeof(struct rte_ipv6_hdr));
> +
> + head->data_len -= 8;
> + head->data_off += 8;
> + head->pkt_len = lcptr + *ihl - 8 + fragx_sum;
> + }
> +}
> +
> +#else
> +static __rte_always_inline uint64_t
> +nix_sec_reass_frags_get(const struct cpt_parse_hdr_s *hdr, struct rte_mbuf **next_mbufs)
> +{
> + RTE_SET_USED(hdr);
> + next_mbufs[0] = NULL;
> + next_mbufs[1] = NULL;
> + next_mbufs[2] = NULL;
> + return 0;
> +}
> +
> +static __rte_always_inline void
> +nix_sec_reass_first_frag_update(struct rte_mbuf *head, const uint8_t *m_ipptr,
> + uint64_t fsz, uint64_t cq_w1, uint16_t *ihl)
> +{
> + RTE_SET_USED(head);
> + RTE_SET_USED(m_ipptr);
> + RTE_SET_USED(fsz);
> + RTE_SET_USED(cq_w1);
> + *ihl = 0;
> +}
> +#endif
> +
> static struct rte_mbuf *
> nix_sec_attach_frags(const struct cpt_parse_hdr_s *hdr,
> + struct rte_mbuf *head,
> struct cn10k_inb_priv_data *inb_priv,
> const uint64_t mbuf_init)
> {
> - struct rte_mbuf *head, *mbuf, *mbuf_prev;
> - uint32_t offset = hdr->w2.fi_offset;
> + uint8_t num_frags = hdr->w0.num_frags;
> + struct rte_mbuf *next_mbufs[3];
> union nix_rx_parse_u *frag_rx;
> - struct cpt_frag_info_s *finfo;
> - uint64_t *frag_ptr = NULL;
> + struct rte_mbuf *mbuf;
> uint64_t ol_flags;
> uint16_t frag_size;
> + uint8_t frag_i = 0;
> uint16_t rlen;
> uint64_t *wqe;
> int off;
> @@ -164,48 +272,37 @@ nix_sec_attach_frags(const struct cpt_parse_hdr_s *hdr,
> ol_flags = BIT_ULL(inb_priv->reass_dynflag_bit);
> ol_flags |= RTE_MBUF_F_RX_SEC_OFFLOAD;
>
> - /* offset of 0 implies 256B, otherwise it implies offset*8B */
> - offset = (((offset - 1) & 0x1f) + 1) * 8;
> - finfo = RTE_PTR_ADD(hdr, offset);
> + /* Get frags list */
> + nix_sec_reass_frags_get(hdr, next_mbufs);
>
> /* Frag-0: */
> - wqe = (uint64_t *)(rte_be_to_cpu_64(hdr->wqe_ptr));
> + wqe = (uint64_t *)(head + 1);
> rlen = ((*(wqe + 10)) >> 16) & 0xFFFF;
>
> frag_rx = (union nix_rx_parse_u *)(wqe + 1);
> - frag_size = rlen + frag_rx->lcptr - frag_rx->laptr;
> - frag_rx->pkt_lenm1 = frag_size - 1;
>
> - mbuf = (struct rte_mbuf *)((uintptr_t)wqe - sizeof(struct rte_mbuf));
> - *(uint64_t *)(&mbuf->rearm_data) = mbuf_init;
> - mbuf->data_len = frag_size;
> - mbuf->pkt_len = frag_size;
> - mbuf->ol_flags = ol_flags;
> - mbuf->next = NULL;
> - head = mbuf;
> - mbuf_prev = mbuf;
> + head->ol_flags = ol_flags;
> /* Update dynamic field with userdata */
> - *rte_security_dynfield(mbuf) = (uint64_t)inb_priv->userdata;
> + *rte_security_dynfield(head) = (uint64_t)inb_priv->userdata;
>
> - cnxk_ip_reassembly_dynfield(head, off)->nb_frags = hdr->w0.num_frags - 1;
> - cnxk_ip_reassembly_dynfield(head, off)->next_frag = NULL;
> + num_frags--;
> + mbuf = head;
>
> - /* Frag-1: */
> - if (hdr->w0.num_frags > 1) {
> - wqe = (uint64_t *)(rte_be_to_cpu_64(hdr->frag1_wqe_ptr));
> + /* Frag-1+: */
> + while (num_frags) {
> + cnxk_ip_reassembly_dynfield(mbuf, off)->next_frag = next_mbufs[frag_i];
> + cnxk_ip_reassembly_dynfield(mbuf, off)->nb_frags = num_frags;
> + mbuf = next_mbufs[frag_i];
> + wqe = (uint64_t *)(mbuf + 1);
> rlen = ((*(wqe + 10)) >> 16) & 0xFFFF;
>
> frag_rx = (union nix_rx_parse_u *)(wqe + 1);
> frag_size = rlen + frag_rx->lcptr - frag_rx->laptr;
> - frag_rx->pkt_lenm1 = frag_size - 1;
>
> - mbuf = (struct rte_mbuf *)((uintptr_t)wqe -
> - sizeof(struct rte_mbuf));
> *(uint64_t *)(&mbuf->rearm_data) = mbuf_init;
> mbuf->data_len = frag_size;
> mbuf->pkt_len = frag_size;
> mbuf->ol_flags = ol_flags;
> - mbuf->next = NULL;
>
> /* Update dynamic field with userdata */
> *rte_security_dynfield(mbuf) = (uint64_t)inb_priv->userdata;
> @@ -213,218 +310,95 @@ nix_sec_attach_frags(const struct cpt_parse_hdr_s *hdr,
> /* Mark frag as get */
> RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
>
> - cnxk_ip_reassembly_dynfield(mbuf, off)->nb_frags =
> - hdr->w0.num_frags - 2;
> - cnxk_ip_reassembly_dynfield(mbuf, off)->next_frag = NULL;
> - cnxk_ip_reassembly_dynfield(mbuf_prev, off)->next_frag = mbuf;
> - mbuf_prev = mbuf;
> + num_frags--;
> + frag_i++;
> }
> + cnxk_ip_reassembly_dynfield(mbuf, off)->nb_frags = 0;
> + cnxk_ip_reassembly_dynfield(mbuf, off)->next_frag = NULL;
>
> - /* Frag-2: */
> - if (hdr->w0.num_frags > 2) {
> - frag_ptr = (uint64_t *)(finfo + 1);
> - wqe = (uint64_t *)(rte_be_to_cpu_64(*frag_ptr));
> - rlen = ((*(wqe + 10)) >> 16) & 0xFFFF;
> -
> - frag_rx = (union nix_rx_parse_u *)(wqe + 1);
> - frag_size = rlen + frag_rx->lcptr - frag_rx->laptr;
> - frag_rx->pkt_lenm1 = frag_size - 1;
> -
> - mbuf = (struct rte_mbuf *)((uintptr_t)wqe -
> - sizeof(struct rte_mbuf));
> - *(uint64_t *)(&mbuf->rearm_data) = mbuf_init;
> - mbuf->data_len = frag_size;
> - mbuf->pkt_len = frag_size;
> - mbuf->ol_flags = ol_flags;
> - mbuf->next = NULL;
> -
> - /* Update dynamic field with userdata */
> - *rte_security_dynfield(mbuf) = (uint64_t)inb_priv->userdata;
> -
> - /* Mark frag as get */
> - RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
> -
> - cnxk_ip_reassembly_dynfield(mbuf, off)->nb_frags =
> - hdr->w0.num_frags - 3;
> - cnxk_ip_reassembly_dynfield(mbuf, off)->next_frag = NULL;
> - cnxk_ip_reassembly_dynfield(mbuf_prev, off)->next_frag = mbuf;
> - mbuf_prev = mbuf;
> - }
> -
> - /* Frag-3: */
> - if (hdr->w0.num_frags > 3) {
> - wqe = (uint64_t *)(rte_be_to_cpu_64(*(frag_ptr + 1)));
> - rlen = ((*(wqe + 10)) >> 16) & 0xFFFF;
> -
> - frag_rx = (union nix_rx_parse_u *)(wqe + 1);
> - frag_size = rlen + frag_rx->lcptr - frag_rx->laptr;
> - frag_rx->pkt_lenm1 = frag_size - 1;
> -
> - mbuf = (struct rte_mbuf *)((uintptr_t)wqe -
> - sizeof(struct rte_mbuf));
> - *(uint64_t *)(&mbuf->rearm_data) = mbuf_init;
> - mbuf->data_len = frag_size;
> - mbuf->pkt_len = frag_size;
> - mbuf->ol_flags = ol_flags;
> - mbuf->next = NULL;
> -
> - /* Mark frag as get */
> - RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
> -
> - /* Update dynamic field with userdata */
> - *rte_security_dynfield(mbuf) = (uint64_t)inb_priv->userdata;
> -
> - cnxk_ip_reassembly_dynfield(mbuf, off)->nb_frags =
> - hdr->w0.num_frags - 4;
> - cnxk_ip_reassembly_dynfield(mbuf, off)->next_frag = NULL;
> - cnxk_ip_reassembly_dynfield(mbuf_prev, off)->next_frag = mbuf;
> - }
> return head;
> }
>
> -static struct rte_mbuf *
> -nix_sec_reassemble_frags(const struct cpt_parse_hdr_s *hdr, uint64_t cq_w1,
> - uint64_t cq_w5, uint64_t mbuf_init)
> +static __rte_always_inline struct rte_mbuf *
> +nix_sec_reassemble_frags(const struct cpt_parse_hdr_s *hdr, struct rte_mbuf *head,
> + uint64_t cq_w1, uint64_t cq_w5, uint64_t mbuf_init)
> {
> - uint32_t fragx_sum, pkt_hdr_len, l3_hdr_size;
> - uint32_t offset = hdr->w2.fi_offset;
> - union nix_rx_parse_u *inner_rx;
> - uint16_t rlen, data_off, b_off;
> + uint8_t num_frags = hdr->w0.num_frags;
> union nix_rx_parse_u *frag_rx;
> - struct cpt_frag_info_s *finfo;
> - struct rte_mbuf *head, *mbuf;
> - uint64_t *frag_ptr = NULL;
> - rte_iova_t *inner_iova;
> + struct rte_mbuf *next_mbufs[3];
> + uint16_t data_off, b_off;
> + const uint8_t *m_ipptr;
> + uint16_t l3_hdr_size;
> + struct rte_mbuf *mbuf;
> uint16_t frag_size;
> + uint64_t fsz_w1;
> uint64_t *wqe;
>
> /* Base data offset */
> b_off = mbuf_init & 0xFFFFUL;
> mbuf_init &= ~0xFFFFUL;
>
> - /* offset of 0 implies 256B, otherwise it implies offset*8B */
> - offset = (((offset - 1) & 0x1f) + 1) * 8;
> - finfo = RTE_PTR_ADD(hdr, offset);
> + /* Get list of all fragments and frag sizes */
> + fsz_w1 = nix_sec_reass_frags_get(hdr, next_mbufs);
>
> /* Frag-0: */
> - wqe = (uint64_t *)rte_be_to_cpu_64(hdr->wqe_ptr);
> - inner_rx = (union nix_rx_parse_u *)(wqe + 1);
> - inner_iova = (rte_iova_t *)*(wqe + 9);
> + wqe = (uint64_t *)(head + 1);
>
> - /* Update only the upper 28-bits from meta pkt parse info */
> - *((uint64_t *)inner_rx) = ((*((uint64_t *)inner_rx) & ((1ULL << 36) - 1)) |
> - (cq_w1 & ~((1ULL << 36) - 1)));
> -
> - rlen = ((*(wqe + 10)) >> 16) & 0xFFFF;
> - frag_size = rlen + ((cq_w5 >> 16) & 0xFF) - (cq_w5 & 0xFF);
> - fragx_sum = rte_be_to_cpu_16(finfo->w1.frag_size0);
> - pkt_hdr_len = frag_size - fragx_sum;
> -
> - mbuf = (struct rte_mbuf *)((uintptr_t)wqe - sizeof(struct rte_mbuf));
> - *(uint64_t *)(&mbuf->rearm_data) = mbuf_init | b_off;
> - mbuf->data_len = frag_size;
> - head = mbuf;
> -
> - if (inner_rx->lctype == NPC_LT_LC_IP) {
> - struct rte_ipv4_hdr *hdr = (struct rte_ipv4_hdr *)
> - RTE_PTR_ADD(inner_iova, inner_rx->lcptr);
> -
> - l3_hdr_size = (hdr->version_ihl & 0xf) << 2;
> - } else {
> - struct rte_ipv6_hdr *hdr = (struct rte_ipv6_hdr *)
> - RTE_PTR_ADD(inner_iova, inner_rx->lcptr);
> - size_t ext_len = sizeof(struct rte_ipv6_hdr);
> - uint8_t *nxt_hdr = (uint8_t *)hdr;
> - int nh = hdr->proto;
> -
> - l3_hdr_size = 0;
> - while (nh != -EINVAL) {
> - nxt_hdr += ext_len;
> - l3_hdr_size += ext_len;
> - nh = rte_ipv6_get_next_ext(nxt_hdr, nh, &ext_len);
> - }
> - }
> + /* First fragment data len is already update by caller */
> + m_ipptr = ((const uint8_t *)hdr + ((cq_w5 >> 16) & 0xFF));
> + nix_sec_reass_first_frag_update(head, m_ipptr, fsz_w1, cq_w1, &l3_hdr_size);
> + fsz_w1 >>= 16;
>
> /* Frag-1: */
> - wqe = (uint64_t *)(rte_be_to_cpu_64(hdr->frag1_wqe_ptr));
> - frag_size = rte_be_to_cpu_16(finfo->w1.frag_size1);
> + head->next = next_mbufs[0];
> + mbuf = next_mbufs[0];
> + wqe = (uint64_t *)(mbuf + 1);
> frag_rx = (union nix_rx_parse_u *)(wqe + 1);
> + frag_size = fsz_w1 & 0xFFFF;
> + fsz_w1 >>= 16;
>
> - mbuf->next = (struct rte_mbuf *)((uintptr_t)wqe - sizeof(struct rte_mbuf));
> - mbuf = mbuf->next;
> data_off = b_off + frag_rx->lcptr + l3_hdr_size;
> *(uint64_t *)(&mbuf->rearm_data) = mbuf_init | data_off;
> mbuf->data_len = frag_size;
> - fragx_sum += frag_size;
>
> /* Mark frag as get */
> RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
>
> /* Frag-2: */
> - if (hdr->w0.num_frags > 2) {
> - frag_ptr = (uint64_t *)(finfo + 1);
> - wqe = (uint64_t *)(rte_be_to_cpu_64(*frag_ptr));
> - frag_size = rte_be_to_cpu_16(finfo->w1.frag_size2);
> + if (num_frags > 2) {
> + mbuf->next = next_mbufs[1];
> + mbuf = next_mbufs[1];
> + wqe = (uint64_t *)(mbuf + 1);
> frag_rx = (union nix_rx_parse_u *)(wqe + 1);
> + frag_size = fsz_w1 & 0xFFFF;
> + fsz_w1 >>= 16;
>
> - mbuf->next = (struct rte_mbuf *)((uintptr_t)wqe - sizeof(struct rte_mbuf));
> - mbuf = mbuf->next;
> data_off = b_off + frag_rx->lcptr + l3_hdr_size;
> *(uint64_t *)(&mbuf->rearm_data) = mbuf_init | data_off;
> mbuf->data_len = frag_size;
> - fragx_sum += frag_size;
>
> /* Mark frag as get */
> RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
> }
>
> /* Frag-3: */
> - if (hdr->w0.num_frags > 3) {
> - wqe = (uint64_t *)(rte_be_to_cpu_64(*(frag_ptr + 1)));
> - frag_size = rte_be_to_cpu_16(finfo->w1.frag_size3);
> + if (num_frags > 3) {
> + mbuf->next = next_mbufs[2];
> + mbuf = next_mbufs[2];
> + wqe = (uint64_t *)(mbuf + 1);
> frag_rx = (union nix_rx_parse_u *)(wqe + 1);
> + frag_size = fsz_w1 & 0xFFFF;
> + fsz_w1 >>= 16;
>
> - mbuf->next = (struct rte_mbuf *)((uintptr_t)wqe - sizeof(struct rte_mbuf));
> - mbuf = mbuf->next;
> data_off = b_off + frag_rx->lcptr + l3_hdr_size;
> *(uint64_t *)(&mbuf->rearm_data) = mbuf_init | data_off;
> mbuf->data_len = frag_size;
> - fragx_sum += frag_size;
>
> /* Mark frag as get */
> RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
> }
>
> - if (inner_rx->lctype == NPC_LT_LC_IP) {
> - struct rte_ipv4_hdr *hdr = (struct rte_ipv4_hdr *)
> - RTE_PTR_ADD(inner_iova, inner_rx->lcptr);
> -
> - hdr->fragment_offset = 0;
> - hdr->total_length = rte_cpu_to_be_16(fragx_sum + l3_hdr_size);
> - hdr->hdr_checksum = 0;
> - hdr->hdr_checksum = rte_ipv4_cksum(hdr);
> -
> - inner_rx->pkt_lenm1 = pkt_hdr_len + fragx_sum - 1;
> - } else {
> - /* Remove the frag header by moving header 8 bytes forward */
> - struct rte_ipv6_hdr *hdr = (struct rte_ipv6_hdr *)
> - RTE_PTR_ADD(inner_iova, inner_rx->lcptr);
> -
> - hdr->payload_len = rte_cpu_to_be_16(fragx_sum + l3_hdr_size -
> - 8 - sizeof(struct rte_ipv6_hdr));
> -
> - rte_memcpy(rte_pktmbuf_mtod_offset(head, void *, 8),
> - rte_pktmbuf_mtod(head, void *),
> - inner_rx->lcptr + sizeof(struct rte_ipv6_hdr));
> -
> - inner_rx->pkt_lenm1 = pkt_hdr_len + fragx_sum - 8 - 1;
> - head->data_len -= 8;
> - head->data_off += 8;
> - }
> - mbuf->next = NULL;
> - head->pkt_len = inner_rx->pkt_lenm1 + 1;
> - head->nb_segs = hdr->w0.num_frags;
> -
> + head->nb_segs = num_frags;
> return head;
> }
>
> @@ -432,7 +406,7 @@ static __rte_always_inline struct rte_mbuf *
> nix_sec_meta_to_mbuf_sc(uint64_t cq_w1, uint64_t cq_w5, const uint64_t sa_base,
> uintptr_t laddr, uint8_t *loff, struct rte_mbuf *mbuf,
> uint16_t data_off, const uint16_t flags,
> - const uint64_t mbuf_init)
> + uint64_t mbuf_init)
> {
> const void *__p = (void *)((uintptr_t)mbuf + (uint16_t)data_off);
> const struct cpt_parse_hdr_s *hdr = (const struct cpt_parse_hdr_s *)__p;
> @@ -445,60 +419,70 @@ nix_sec_meta_to_mbuf_sc(uint64_t cq_w1, uint64_t cq_w5, const uint64_t sa_base,
> void *inb_sa;
> uint64_t w0;
>
> - if ((flags & NIX_RX_REAS_F) && (cq_w1 & BIT(11))) {
> - /* Get SPI from CPT_PARSE_S's cookie(already swapped) */
> - w0 = hdr->w0.u64;
> - sa_idx = w0 >> 32;
> -
> - inb_sa = roc_nix_inl_ot_ipsec_inb_sa(sa_base, sa_idx);
> - inb_priv = roc_nix_inl_ot_ipsec_inb_sa_sw_rsvd(inb_sa);
> -
> - if (!hdr->w0.num_frags) {
> - /* No Reassembly or inbound error */
> - inner = (struct rte_mbuf *)
> - (rte_be_to_cpu_64(hdr->wqe_ptr) -
> - sizeof(struct rte_mbuf));
> -
> - /* Update dynamic field with userdata */
> - *rte_security_dynfield(inner) =
> - (uint64_t)inb_priv->userdata;
> -
> - /* Get ucc from cpt parse header */
> - ucc = hdr->w3.hw_ccode;
> -
> - /* Calculate inner packet length as
> - * IP total len + l2 len
> - */
> - ip = (uintptr_t)hdr + ((cq_w5 >> 16) & 0xFF);
> - ip += ((cq_w1 >> 40) & 0x6);
> - len = rte_be_to_cpu_16(*(uint16_t *)ip);
> - len += ((cq_w5 >> 16) & 0xFF) - (cq_w5 & 0xFF);
> - len += (cq_w1 & BIT(42)) ? 40 : 0;
> -
> - inner->pkt_len = len;
> - inner->data_len = len;
> - *(uint64_t *)(&inner->rearm_data) = mbuf_init;
> -
> - inner->ol_flags = ((CPT_COMP_HWGOOD_MASK & (1U << ucc)) ?
> - RTE_MBUF_F_RX_SEC_OFFLOAD :
> - (RTE_MBUF_F_RX_SEC_OFFLOAD |
> - RTE_MBUF_F_RX_SEC_OFFLOAD_FAILED));
> -
> - ucc = hdr->w3.uc_ccode;
> -
> - if (ucc && ucc < 0xED) {
> - inner->ol_flags |= RTE_MBUF_F_RX_SEC_OFFLOAD_FAILED;
> - } else {
> - ucc += 3; /* To make codes in 0xFx series except 0 */
> - inner->ol_flags |= ((ucc & 0xF0) == 0xF0) ?
> - ((NIX_RX_SEC_UCC_CONST >> ((ucc & 0xF) << 3))
> - & 0xFF) << 1 : RTE_MBUF_F_RX_IP_CKSUM_GOOD;
> - }
> - } else if ((!(hdr->w0.err_sum) || roc_ie_ot_ucc_is_success(hdr->w3.uc_ccode)) &&
> - !(hdr->w0.reas_sts)) {
> + if (!(cq_w1 & BIT(11)))
> + return mbuf;
> +
> + inner = (struct rte_mbuf *)(rte_be_to_cpu_64(hdr->wqe_ptr) -
> + sizeof(struct rte_mbuf));
> +
> + /* Store meta in lmtline to free
> + * Assume all meta's from same aura.
> + */
> + *(uint64_t *)(laddr + (*loff << 3)) = (uint64_t)mbuf;
> + *loff = *loff + 1;
> +
> + /* Get SPI from CPT_PARSE_S's cookie(already swapped) */
> + w0 = hdr->w0.u64;
> + sa_idx = w0 >> 32;
> +
> + inb_sa = roc_nix_inl_ot_ipsec_inb_sa(sa_base, sa_idx);
> + inb_priv = roc_nix_inl_ot_ipsec_inb_sa_sw_rsvd(inb_sa);
> +
> + /* Update dynamic field with userdata */
> + *rte_security_dynfield(inner) = (uint64_t)inb_priv->userdata;
> +
> + /* Get ucc from cpt parse header */
> + ucc = hdr->w3.hw_ccode;
> +
> + /* Calculate inner packet length as IP total len + l2 len */
> + ip = (uintptr_t)hdr + ((cq_w5 >> 16) & 0xFF);
> + ip += ((cq_w1 >> 40) & 0x6);
> + len = rte_be_to_cpu_16(*(uint16_t *)ip);
> + len += ((cq_w5 >> 16) & 0xFF) - (cq_w5 & 0xFF);
> + len += (cq_w1 & BIT(42)) ? 40 : 0;
> +
> + inner->pkt_len = len;
> + inner->data_len = len;
> + *(uint64_t *)(&inner->rearm_data) = mbuf_init;
> +
> + inner->ol_flags = ((CPT_COMP_HWGOOD_MASK & (1U << ucc)) ?
> + RTE_MBUF_F_RX_SEC_OFFLOAD :
> + (RTE_MBUF_F_RX_SEC_OFFLOAD |
> + RTE_MBUF_F_RX_SEC_OFFLOAD_FAILED));
> +
> + ucc = hdr->w3.uc_ccode;
> +
> + if (ucc && ucc < 0xED) {
> + inner->ol_flags |= RTE_MBUF_F_RX_SEC_OFFLOAD_FAILED;
> + } else {
> + ucc += 3; /* To make codes in 0xFx series except 0 */
> + inner->ol_flags |= ((ucc & 0xF0) == 0xF0) ?
> + ((NIX_RX_SEC_UCC_CONST >> ((ucc & 0xF) << 3))
> + & 0xFF) << 1 : RTE_MBUF_F_RX_IP_CKSUM_GOOD;
> + }
> +
> + /* Mark meta mbuf as put */
> + RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 0);
> +
> + /* Mark inner mbuf as get */
> + RTE_MEMPOOL_CHECK_COOKIES(inner->pool, (void **)&inner, 1, 1);
> +
> + /* Skip reassembly processing when multi-seg is enabled */
> + if (!(flags & NIX_RX_MULTI_SEG_F) && (flags & NIX_RX_REAS_F) && hdr->w0.num_frags) {
> + if ((!(hdr->w0.err_sum) || roc_ie_ot_ucc_is_success(hdr->w3.uc_ccode)) &&
> + !(hdr->w0.reas_sts)) {
> /* Reassembly success */
> - inner = nix_sec_reassemble_frags(hdr, cq_w1, cq_w5,
> - mbuf_init);
> + nix_sec_reassemble_frags(hdr, inner, cq_w1, cq_w5, mbuf_init);
>
> /* Update dynamic field with userdata */
> *rte_security_dynfield(inner) =
> @@ -508,82 +492,10 @@ nix_sec_meta_to_mbuf_sc(uint64_t cq_w1, uint64_t cq_w5, const uint64_t sa_base,
> inner->ol_flags = RTE_MBUF_F_RX_SEC_OFFLOAD;
> } else {
> /* Reassembly failure */
> - inner = nix_sec_attach_frags(hdr, inb_priv, mbuf_init);
> + nix_sec_attach_frags(hdr, inner, inb_priv, mbuf_init);
> }
> -
> - /* Store meta in lmtline to free
> - * Assume all meta's from same aura.
> - */
> - *(uint64_t *)(laddr + (*loff << 3)) = (uint64_t)mbuf;
> - *loff = *loff + 1;
> -
> - /* Mark meta mbuf as put */
> - RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 0);
> -
> - /* Mark inner mbuf as get */
> - RTE_MEMPOOL_CHECK_COOKIES(inner->pool, (void **)&inner, 1, 1);
> -
> - return inner;
> - } else if (cq_w1 & BIT(11)) {
> - inner = (struct rte_mbuf *)(rte_be_to_cpu_64(hdr->wqe_ptr) -
> - sizeof(struct rte_mbuf));
> -
> - /* Get SPI from CPT_PARSE_S's cookie(already swapped) */
> - w0 = hdr->w0.u64;
> - sa_idx = w0 >> 32;
> -
> - inb_sa = roc_nix_inl_ot_ipsec_inb_sa(sa_base, sa_idx);
> - inb_priv = roc_nix_inl_ot_ipsec_inb_sa_sw_rsvd(inb_sa);
> -
> - /* Update dynamic field with userdata */
> - *rte_security_dynfield(inner) = (uint64_t)inb_priv->userdata;
> -
> - /* Get ucc from cpt parse header */
> - ucc = hdr->w3.hw_ccode;
> -
> - /* Calculate inner packet length as IP total len + l2 len */
> - ip = (uintptr_t)hdr + ((cq_w5 >> 16) & 0xFF);
> - ip += ((cq_w1 >> 40) & 0x6);
> - len = rte_be_to_cpu_16(*(uint16_t *)ip);
> - len += ((cq_w5 >> 16) & 0xFF) - (cq_w5 & 0xFF);
> - len += (cq_w1 & BIT(42)) ? 40 : 0;
> -
> - inner->pkt_len = len;
> - inner->data_len = len;
> - *(uint64_t *)(&inner->rearm_data) = mbuf_init;
> -
> - inner->ol_flags = ((CPT_COMP_HWGOOD_MASK & (1U << ucc)) ?
> - RTE_MBUF_F_RX_SEC_OFFLOAD :
> - (RTE_MBUF_F_RX_SEC_OFFLOAD |
> - RTE_MBUF_F_RX_SEC_OFFLOAD_FAILED));
> -
> - ucc = hdr->w3.uc_ccode;
> -
> - if (ucc && ucc < 0xED) {
> - inner->ol_flags |= RTE_MBUF_F_RX_SEC_OFFLOAD_FAILED;
> - } else {
> - ucc += 3; /* To make codes in 0xFx series except 0 */
> - inner->ol_flags |= ((ucc & 0xF0) == 0xF0) ?
> - ((NIX_RX_SEC_UCC_CONST >> ((ucc & 0xF) << 3))
> - & 0xFF) << 1 : RTE_MBUF_F_RX_IP_CKSUM_GOOD;
> - }
> -
> - /* Store meta in lmtline to free
> - * Assume all meta's from same aura.
> - */
> - *(uint64_t *)(laddr + (*loff << 3)) = (uint64_t)mbuf;
> - *loff = *loff + 1;
> -
> - /* Mark meta mbuf as put */
> - RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 0);
> -
> - /* Mark inner mbuf as get */
> - RTE_MEMPOOL_CHECK_COOKIES(inner->pool, (void **)&inner, 1, 1);
> -
> - return inner;
> }
> -
> - return mbuf;
> + return inner;
> }
>
> #if defined(RTE_ARCH_ARM64)
> @@ -612,11 +524,16 @@ nix_sec_meta_to_mbuf(uint64_t cq_w1, uint64_t cq_w5, uintptr_t inb_sa,
> /* Mark inner mbuf as get */
> RTE_MEMPOOL_CHECK_COOKIES(inner->pool, (void **)&inner, 1, 1);
>
> - if (flags & NIX_RX_REAS_F && hdr->w0.num_frags) {
> + if (!(flags & NIX_RX_MULTI_SEG_F) && flags & NIX_RX_REAS_F && hdr->w0.num_frags) {
> if ((!(hdr->w0.err_sum) || roc_ie_ot_ucc_is_success(hdr->w3.uc_ccode)) &&
> !(hdr->w0.reas_sts)) {
> + /* First frag len */
> + inner->pkt_len = vgetq_lane_u16(*rx_desc_field1, 2);
> + inner->data_len = vgetq_lane_u16(*rx_desc_field1, 4);
> + *(uint64_t *)(&inner->rearm_data) = mbuf_init;
> +
> /* Reassembly success */
> - nix_sec_reassemble_frags(hdr, cq_w1, cq_w5, mbuf_init);
> + nix_sec_reassemble_frags(hdr, inner, cq_w1, cq_w5, mbuf_init);
>
> /* Assume success */
> *ol_flags |= RTE_MBUF_F_RX_SEC_OFFLOAD;
> @@ -632,14 +549,8 @@ nix_sec_meta_to_mbuf(uint64_t cq_w1, uint64_t cq_w5, uintptr_t inb_sa,
> *rearm = vsetq_lane_u64(mbuf_init, *rearm, 0);
> } else {
> /* Reassembly failure */
> - nix_sec_attach_frags(hdr, inb_priv, mbuf_init);
> + nix_sec_attach_frags(hdr, inner, inb_priv, mbuf_init);
> *ol_flags |= inner->ol_flags;
> -
> - /* Update pkt_len and data_len */
> - *rx_desc_field1 = vsetq_lane_u16(inner->pkt_len,
> - *rx_desc_field1, 2);
> - *rx_desc_field1 = vsetq_lane_u16(inner->data_len,
> - *rx_desc_field1, 4);
> }
> }
> }
> @@ -697,18 +608,30 @@ nix_update_match_id(const uint16_t match_id, uint64_t ol_flags,
>
> static __rte_always_inline void
> nix_cqe_xtract_mseg(const union nix_rx_parse_u *rx, struct rte_mbuf *mbuf,
> - uint64_t rearm, const uint16_t flags)
> + uint64_t rearm, uintptr_t cpth, uintptr_t sa_base, const uint16_t flags)
> {
> + const struct cpt_parse_hdr_s *hdr = (const struct cpt_parse_hdr_s *)cpth;
> + struct cn10k_inb_priv_data *inb_priv = NULL;
> + uint8_t num_frags = 0, frag_i = 0;
> + struct rte_mbuf *next_mbufs[3];
> const rte_iova_t *iova_list;
> + bool reas_success = false;
> uint16_t later_skip = 0;
> struct rte_mbuf *head;
> const rte_iova_t *eol;
> + uint64_t cq_w5 = 0;
> + uint16_t ihl = 0;
> + uint64_t fsz = 0;
> + int dyn_off = 0;
> uint8_t nb_segs;
> + uint16_t sg_len;
> uint64_t cq_w1;
> int64_t len;
> uint64_t sg;
>
> cq_w1 = *(const uint64_t *)rx;
> + if (flags & NIX_RX_REAS_F)
> + cq_w5 = *((const uint64_t *)rx + 4);
> /* Use inner rx parse for meta pkts sg list */
> if (cq_w1 & BIT(11) && flags & NIX_RX_OFFLOAD_SECURITY_F) {
> const uint64_t *wqe = (const uint64_t *)(mbuf + 1);
> @@ -718,40 +641,69 @@ nix_cqe_xtract_mseg(const union nix_rx_parse_u *rx, struct rte_mbuf *mbuf,
> sg = *(const uint64_t *)(rx + 1);
> nb_segs = (sg >> 48) & 0x3;
>
> - if (nb_segs == 1)
> + if (nb_segs == 1 && !(flags & NIX_RX_REAS_F))
> return;
>
> /* For security we have already updated right pkt_len */
> - if (cq_w1 & BIT(11) && flags & NIX_RX_OFFLOAD_SECURITY_F)
> + if (cq_w1 & BIT(11) && flags & NIX_RX_OFFLOAD_SECURITY_F) {
> len = mbuf->pkt_len;
> - else
> +
> + /* Handle reassembly with multi segs */
> + if (flags & NIX_RX_REAS_F && hdr->w0.num_frags) {
> + void *inb_sa;
> +
> + num_frags = hdr->w0.num_frags;
> + inb_sa = roc_nix_inl_ot_ipsec_inb_sa(sa_base, hdr->w0.u64 >> 32);
> + inb_priv = roc_nix_inl_ot_ipsec_inb_sa_sw_rsvd(inb_sa);
> + ihl = 0;
> +
> + dyn_off = inb_priv->reass_dynfield_off;
> + fsz = nix_sec_reass_frags_get(hdr, next_mbufs);
> + num_frags -= 1;
> +
> + if (!(hdr->w0.reas_sts) &&
> + (!(hdr->w0.err_sum) ||
> + roc_ie_ot_ucc_is_success(hdr->w3.uc_ccode)))
> + reas_success = true;
> + }
> + } else {
> len = rx->pkt_lenm1 + 1;
> + }
> +
> mbuf->pkt_len = len - (flags & NIX_RX_OFFLOAD_TSTAMP_F ? CNXK_NIX_TIMESYNC_RX_OFFSET : 0);
> + mbuf->nb_segs = nb_segs;
> + head = mbuf;
> mbuf->data_len =
> (sg & 0xFFFF) - (flags & NIX_RX_OFFLOAD_TSTAMP_F ? CNXK_NIX_TIMESYNC_RX_OFFSET : 0);
> + eol = ((const rte_iova_t *)(rx + 1) + ((rx->desc_sizem1 + 1) << 1));
> +again:
> len -= mbuf->data_len;
> - mbuf->nb_segs = nb_segs;
> sg = sg >> 16;
> -
> - eol = ((const rte_iova_t *)(rx + 1) + ((rx->desc_sizem1 + 1) << 1));
> /* Skip SG_S and first IOVA*/
> iova_list = ((const rte_iova_t *)(rx + 1)) + 2;
> nb_segs--;
>
> - rearm = rearm & ~0xFFFF;
> later_skip = (uintptr_t)mbuf->buf_addr - (uintptr_t)mbuf;
>
> - head = mbuf;
> while (nb_segs) {
> mbuf->next = (struct rte_mbuf *)(*iova_list - later_skip);
> mbuf = mbuf->next;
>
> RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
>
> - mbuf->data_len = sg & 0xFFFF;
> - len -= sg & 0XFFFF;
> + sg_len = sg & 0XFFFF;
> + if (flags & NIX_RX_OFFLOAD_SECURITY_F) {
> + /* Adjust last mbuf data length with negative offset for
> + * security pkts if needed.
> + */
> + len -= sg_len;
> + sg_len = (len > 0) ? sg_len : (sg_len + len);
> + len = (len > 0) ? len : 0;
> + }
> +
> + mbuf->data_len = sg_len;
> sg = sg >> 16;
> - *(uint64_t *)(&mbuf->rearm_data) = rearm;
> + *(uint64_t *)(&mbuf->rearm_data) = rearm & ~0xFFFF;
> nb_segs--;
> iova_list++;
>
> @@ -763,15 +715,66 @@ nix_cqe_xtract_mseg(const union nix_rx_parse_u *rx, struct rte_mbuf *mbuf,
> }
> }
>
> - /* Adjust last mbuf data length with negative offset for security pkts if needed */
> - if (cq_w1 & BIT(11) && flags & NIX_RX_OFFLOAD_SECURITY_F && len < 0)
> - mbuf->data_len += len;
> + if ((flags & NIX_RX_REAS_F) && (cq_w1 & BIT(11)) && num_frags) {
> + struct rte_mbuf *next_frag = next_mbufs[frag_i];
> + uint16_t lcptr, ldptr = 0;
> +
> + rx = (const union nix_rx_parse_u *)((uintptr_t)(next_frag + 1) + 8);
> + lcptr = (*((const uint64_t *)rx + 4) >> 16) & 0xFF;
> + eol = ((const rte_iova_t *)(rx + 1) + ((rx->desc_sizem1 + 1) << 1));
> + sg = *(const uint64_t *)(rx + 1);
> + nb_segs = (sg >> 48) & 0x3;
> +
> + if (reas_success) {
> + /* Update first fragment info */
> + if (!frag_i) {
> + const uint8_t *ipptr;
> +
> + ipptr = ((const uint8_t *)hdr + ((cq_w5 >> 16) & 0xFF));
> + nix_sec_reass_first_frag_update(head, ipptr, fsz, cq_w1, &ihl);
> + fsz >>= 16;
> + }
> + mbuf->next = next_frag;
> + head->nb_segs += nb_segs;
> + len = fsz & 0xFFFF;
> + fsz >>= 16;
> + ldptr = ihl + lcptr;
> + } else {
> + len = ((eol[0] >> 16) & 0xFFFF) + lcptr;
> + head->ol_flags |= BIT_ULL(inb_priv->reass_dynflag_bit) |
> + RTE_MBUF_F_RX_SEC_OFFLOAD;
> + cnxk_ip_reassembly_dynfield(head, dyn_off)->next_frag = next_frag;
> + cnxk_ip_reassembly_dynfield(head, dyn_off)->nb_frags = num_frags;
> + /* Update dynamic field with userdata from prev head */
> + *rte_security_dynfield(next_frag) = *rte_security_dynfield(head);
> + head = next_frag;
> + head->pkt_len = len - (flags & NIX_RX_OFFLOAD_TSTAMP_F ?
> + CNXK_NIX_TIMESYNC_RX_OFFSET : 0);
> + head->nb_segs = nb_segs;
> + }
> + mbuf = next_frag;
> + *(uint64_t *)(&mbuf->rearm_data) = rearm + ldptr;
> + mbuf->data_len = (sg & 0xFFFF) - ldptr -
> + (flags & NIX_RX_OFFLOAD_TSTAMP_F ?
> + CNXK_NIX_TIMESYNC_RX_OFFSET : 0);
> + RTE_MEMPOOL_CHECK_COOKIES(mbuf->pool, (void **)&mbuf, 1, 1);
> + num_frags--;
> + frag_i++;
> + goto again;
> + }
> +
> + /* Update for last failure fragment */
> + if ((flags & NIX_RX_REAS_F) && frag_i && !reas_success) {
> + cnxk_ip_reassembly_dynfield(head, dyn_off)->next_frag = NULL;
> + cnxk_ip_reassembly_dynfield(head, dyn_off)->nb_frags = 0;
> + }
> }
>
> static __rte_always_inline void
> cn10k_nix_cqe_to_mbuf(const struct nix_cqe_hdr_s *cq, const uint32_t tag,
> struct rte_mbuf *mbuf, const void *lookup_mem,
> - const uint64_t val, const uint16_t flag)
> + const uint64_t val, const uintptr_t cpth, const uintptr_t sa_base,
> + const uint16_t flag)
> {
> const union nix_rx_parse_u *rx =
> (const union nix_rx_parse_u *)((const uint64_t *)cq + 1);
> @@ -825,7 +828,7 @@ cn10k_nix_cqe_to_mbuf(const struct nix_cqe_hdr_s *cq, const uint32_t tag,
> * timestamp data process.
> * Hence, timestamp flag argument is not required.
> */
> - nix_cqe_xtract_mseg(rx, mbuf, val, flag & ~NIX_RX_OFFLOAD_TSTAMP_F);
> + nix_cqe_xtract_mseg(rx, mbuf, val, cpth, sa_base, flag & ~NIX_RX_OFFLOAD_TSTAMP_F);
> }
>
> static inline uint16_t
> @@ -906,6 +909,7 @@ cn10k_nix_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t pkts,
> struct nix_cqe_hdr_s *cq;
> struct rte_mbuf *mbuf;
> uint64_t aura_handle;
> + uintptr_t cpth = 0;
> uint64_t sa_base;
> uint16_t lmt_id;
> uint64_t laddr;
> @@ -937,13 +941,15 @@ cn10k_nix_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t pkts,
> const uint64_t cq_w1 = *((const uint64_t *)cq + 1);
> const uint64_t cq_w5 = *((const uint64_t *)cq + 5);
>
> + cpth = ((uintptr_t)mbuf + (uint16_t)data_off);
> +
> mbuf = nix_sec_meta_to_mbuf_sc(cq_w1, cq_w5, sa_base, laddr,
> &loff, mbuf, data_off,
> flags, mbuf_init);
> }
>
> cn10k_nix_cqe_to_mbuf(cq, cq->tag, mbuf, lookup_mem, mbuf_init,
> - flags);
> + cpth, sa_base, flags);
> cn10k_nix_mbuf_to_tstamp(mbuf, rxq->tstamp,
> (flags & NIX_RX_OFFLOAD_TSTAMP_F),
> (uint64_t *)((uint8_t *)mbuf
> @@ -1033,6 +1039,7 @@ cn10k_nix_recv_pkts_vector(void *args, struct rte_mbuf **mbufs, uint16_t pkts,
> const uint64_t wdata = flags & NIX_RX_VWQE_F ? 0 : rxq->wdata;
> const uintptr_t desc = flags & NIX_RX_VWQE_F ? 0 : rxq->desc;
> uint64x2_t cq0_w8, cq1_w8, cq2_w8, cq3_w8, mbuf01, mbuf23;
> + uintptr_t cpth0 = 0, cpth1 = 0, cpth2 = 0, cpth3 = 0;
> uint64_t ol_flags0, ol_flags1, ol_flags2, ol_flags3;
> uint64x2_t rearm0 = vdupq_n_u64(mbuf_initializer);
> uint64x2_t rearm1 = vdupq_n_u64(mbuf_initializer);
> @@ -1333,10 +1340,6 @@ cn10k_nix_recv_pkts_vector(void *args, struct rte_mbuf **mbufs, uint16_t pkts,
> uint64_t cq1_w5 = *CQE_PTR_OFF(cq0, 1, 40, flags);
> uint64_t cq2_w5 = *CQE_PTR_OFF(cq0, 2, 40, flags);
> uint64_t cq3_w5 = *CQE_PTR_OFF(cq0, 3, 40, flags);
> - uintptr_t cpth0 = (uintptr_t)mbuf0 + d_off;
> - uintptr_t cpth1 = (uintptr_t)mbuf1 + d_off;
> - uintptr_t cpth2 = (uintptr_t)mbuf2 + d_off;
> - uintptr_t cpth3 = (uintptr_t)mbuf3 + d_off;
> uint8_t code;
>
> uint64x2_t inner0, inner1, inner2, inner3;
> @@ -1344,6 +1347,11 @@ cn10k_nix_recv_pkts_vector(void *args, struct rte_mbuf **mbufs, uint16_t pkts,
> uint16x4_t lens, l2lens, ltypes;
> uint8x8_t ucc;
>
> + cpth0 = (uintptr_t)mbuf0 + d_off;
> + cpth1 = (uintptr_t)mbuf1 + d_off;
> + cpth2 = (uintptr_t)mbuf2 + d_off;
> + cpth3 = (uintptr_t)mbuf3 + d_off;
> +
> inner0 = vld1q_u64((const uint64_t *)cpth0);
> inner1 = vld1q_u64((const uint64_t *)cpth1);
> inner2 = vld1q_u64((const uint64_t *)cpth2);
> @@ -1729,16 +1737,16 @@ cn10k_nix_recv_pkts_vector(void *args, struct rte_mbuf **mbufs, uint16_t pkts,
> */
> nix_cqe_xtract_mseg((union nix_rx_parse_u *)
> (CQE_PTR_OFF(cq0, 0, 8, flags)),
> - mbuf0, mbuf_initializer, flags);
> + mbuf0, mbuf_initializer, cpth0, sa_base, flags);
> nix_cqe_xtract_mseg((union nix_rx_parse_u *)
> (CQE_PTR_OFF(cq0, 1, 8, flags)),
> - mbuf1, mbuf_initializer, flags);
> + mbuf1, mbuf_initializer, cpth1, sa_base, flags);
> nix_cqe_xtract_mseg((union nix_rx_parse_u *)
> (CQE_PTR_OFF(cq0, 2, 8, flags)),
> - mbuf2, mbuf_initializer, flags);
> + mbuf2, mbuf_initializer, cpth2, sa_base, flags);
> nix_cqe_xtract_mseg((union nix_rx_parse_u *)
> (CQE_PTR_OFF(cq0, 3, 8, flags)),
> - mbuf3, mbuf_initializer, flags);
> + mbuf3, mbuf_initializer, cpth3, sa_base, flags);
> }
>
> /* Store the mbufs to rx_pkts */
> --
> 2.25.1
>
prev parent reply other threads:[~2023-06-15 8:07 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-13 15:09 Nithin Dabilpuram
2023-06-15 8:06 ` Jerin Jacob [this message]
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='CALBAE1PDXKu+q87nw6P7rUN54yN75pFP-fuq0=piP4dhdhTCyQ@mail.gmail.com' \
--to=jerinjacobk@gmail.com \
--cc=dev@dpdk.org \
--cc=jerinj@marvell.com \
--cc=kirankumark@marvell.com \
--cc=ndabilpuram@marvell.com \
--cc=pbhagavatula@marvell.com \
--cc=skori@marvell.com \
--cc=skoteshwar@marvell.com \
--cc=sthotton@marvell.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).