+cpt_fc_ciph_set_key(void *ctx, cipher_type_t type, const uint8_t *key,
+ uint16_t key_len, uint8_t *salt)
+{
+ struct cpt_ctx *cpt_ctx = ctx;
+ mc_fc_context_t *fctx = &cpt_ctx->fctx;
+ int ret;
+
+ ret = cpt_fc_ciph_set_type(type, cpt_ctx, key_len);
+ if (unlikely(ret))
+ return -1;
+
+ if (cpt_ctx->fc_type == FC_GEN) {
+ /*
+ * We need to always say IV is from DPTR as user can
+ * sometimes iverride IV per operation.
+ */
+ fctx->enc.iv_source = CPT_FROM_DPTR;
+
+ if (cpt_ctx->auth_key_len > 64)
+ return -1;
+ }
+
+ switch (type) {
+ case PASSTHROUGH:
+ cpt_fc_ciph_set_key_passthrough(cpt_ctx, fctx);
+ goto success;
+ case DES3_CBC:
+ /* CPT performs DES using 3DES with the 8B DES-key
+ * replicated 2 more times to match the 24B 3DES-key.
+ * Eg. If org. key is "0x0a 0x0b", then new key is
+ * "0x0a 0x0b 0x0a 0x0b 0x0a 0x0b"
+ */
+ if (key_len == 8) {
+ /* Skipping the first 8B as it will be copied
+ * in the regular code flow
+ */
+ memcpy(fctx->enc.encr_key+key_len, key, key_len);
+ memcpy(fctx->enc.encr_key+2*key_len, key, key_len);
+ }
+ break;
+ case DES3_ECB:
+ /* For DES3_ECB IV need to be from CTX. */
+ fctx->enc.iv_source = CPT_FROM_CTX;
+ break;
+ case AES_CBC:
+ case AES_ECB:
+ case AES_CFB:
+ case AES_CTR:
+ cpt_fc_ciph_set_key_set_aes_key_type(fctx, key_len);
+ break;
+ case AES_GCM:
+ /* Even though iv source is from dptr,
+ * aes_gcm salt is taken from ctx
+ */
+ if (salt) {
+ memcpy(fctx->enc.encr_iv, salt, 4);
+ /* Assuming it was just salt update
+ * and nothing else
+ */
+ if (!key)
+ goto success;
+ }
+ cpt_fc_ciph_set_key_set_aes_key_type(fctx, key_len);
+ break;
+ case AES_XTS:
+ key_len = key_len / 2;
+ cpt_fc_ciph_set_key_set_aes_key_type(fctx, key_len);
+
+ /* Copy key2 for XTS into ipad */
+ memset(fctx->hmac.ipad, 0, sizeof(fctx->hmac.ipad));
+ memcpy(fctx->hmac.ipad, &key[key_len], key_len);
+ break;
+ case SNOW3G_UEA2:
+ cpt_fc_ciph_set_key_snow3g_uea2(cpt_ctx, key, key_len);
+ goto success;
+ case ZUC_EEA3:
+ cpt_fc_ciph_set_key_zuc_eea3(cpt_ctx, key, key_len);
+ goto success;
+ case KASUMI_F8_ECB:
+ cpt_fc_ciph_set_key_kasumi_f8_ecb(cpt_ctx, key, key_len);
+ goto success;
+ case KASUMI_F8_CBC:
+ cpt_fc_ciph_set_key_kasumi_f8_cbc(cpt_ctx, key, key_len);
+ goto success;
+ default:
+ return -1;
+ }
+
+ /* Only for FC_GEN case */
+
+ /* For GMAC auth, cipher must be NULL */
+ if (cpt_ctx->hash_type != GMAC_TYPE)
+ fctx->enc.enc_cipher = type;
+
+ memcpy(fctx->enc.encr_key, key, key_len);
+
+success:
+ cpt_ctx->enc_cipher = type;
+
+ return 0;
+}
+
+static __rte_always_inline uint32_t
+fill_sg_comp(sg_comp_t *list,
+ uint32_t i,
+ phys_addr_t dma_addr,
+ uint32_t size)
+{
+ sg_comp_t *to = &list[i>>2];
+
+ to->u.s.len[i%4] = rte_cpu_to_be_16(size);
+ to->ptr[i%4] = rte_cpu_to_be_64(dma_addr);
+ i++;
+ return i;
+}
+
+static __rte_always_inline uint32_t
+fill_sg_comp_from_buf(sg_comp_t *list,
+ uint32_t i,
+ buf_ptr_t *from)
+{
+ sg_comp_t *to = &list[i>>2];
+
+ to->u.s.len[i%4] = rte_cpu_to_be_16(from->size);
+ to->ptr[i%4] = rte_cpu_to_be_64(from->dma_addr);
+ i++;
+ return i;
+}
+
+static __rte_always_inline uint32_t
+fill_sg_comp_from_buf_min(sg_comp_t *list,
+ uint32_t i,
+ buf_ptr_t *from,
+ uint32_t *psize)
+{
+ sg_comp_t *to = &list[i >> 2];
+ uint32_t size = *psize;
+ uint32_t e_len;
+
+ e_len = (size > from->size) ? from->size : size;
+ to->u.s.len[i % 4] = rte_cpu_to_be_16(e_len);
+ to->ptr[i % 4] = rte_cpu_to_be_64(from->dma_addr);
+ *psize -= e_len;
+ i++;
+ return i;
+}
+
+/*
+ * This fills the MC expected SGIO list
+ * from IOV given by user.
+ */
+static __rte_always_inline uint32_t
+fill_sg_comp_from_iov(sg_comp_t *list,
+ uint32_t i,
+ iov_ptr_t *from, uint32_t from_offset,
+ uint32_t *psize, buf_ptr_t *extra_buf,
+ uint32_t extra_offset)
+{
+ int32_t j;
+ uint32_t extra_len = extra_buf ? extra_buf->size : 0;
+ uint32_t size = *psize;
+ buf_ptr_t *bufs;
+
+ bufs = from->bufs;
+ for (j = 0; (j < from->buf_cnt) && size; j++) {
+ phys_addr_t e_dma_addr;
+ uint32_t e_len;
+ sg_comp_t *to = &list[i >> 2];
+
+ if (unlikely(from_offset)) {
+ if (from_offset >= bufs[j].size) {
+ from_offset -= bufs[j].size;
+ continue;
+ }
+ e_dma_addr = bufs[j].dma_addr + from_offset;
+ e_len = (size > (bufs[j].size - from_offset)) ?
+ (bufs[j].size - from_offset) : size;
+ from_offset = 0;
+ } else {
+ e_dma_addr = bufs[j].dma_addr;
+ e_len = (size > bufs[j].size) ?
+ bufs[j].size : size;
+ }
+
+ to->u.s.len[i % 4] = rte_cpu_to_be_16(e_len);
+ to->ptr[i % 4] = rte_cpu_to_be_64(e_dma_addr);
+
+ if (extra_len && (e_len >= extra_offset)) {
+ /* Break the data at given offset */
+ uint32_t next_len = e_len - extra_offset;
+ phys_addr_t next_dma = e_dma_addr + extra_offset;
+
+ if (!extra_offset) {
+ i--;
+ } else {
+ e_len = extra_offset;
+ size -= e_len;
+ to->u.s.len[i % 4] = rte_cpu_to_be_16(e_len);
+ }
+
+ extra_len = RTE_MIN(extra_len, size);
+ /* Insert extra data ptr */
+ if (extra_len) {
+ i++;
+ to = &list[i >> 2];
+ to->u.s.len[i % 4] =
+ rte_cpu_to_be_16(extra_len);
+ to->ptr[i % 4] =
+ rte_cpu_to_be_64(extra_buf->dma_addr);
+ size -= extra_len;
+ }
+
+ next_len = RTE_MIN(next_len, size);
+ /* insert the rest of the data */
+ if (next_len) {
+ i++;
+ to = &list[i >> 2];
+ to->u.s.len[i % 4] = rte_cpu_to_be_16(next_len);
+ to->ptr[i % 4] = rte_cpu_to_be_64(next_dma);
+ size -= next_len;
+ }
+ extra_len = 0;
+
+ } else {
+ size -= e_len;
+ }
+ if (extra_offset)
+ extra_offset -= size;
+ i++;
+ }
+
+ *psize = size;
+ return (uint32_t)i;
+}
+
+static __rte_always_inline void
+cpt_digest_gen_prep(uint32_t flags,
+ uint64_t d_lens,
+ digest_params_t *params,
+ void *op,
+ void **prep_req)
+{
+ struct cpt_request_info *req;
+ uint32_t size, i;
+ uint16_t data_len, mac_len, key_len;
+ auth_type_t hash_type;
+ buf_ptr_t *meta_p;
+ struct cpt_ctx *ctx;
+ sg_comp_t *gather_comp;
+ sg_comp_t *scatter_comp;
+ uint8_t *in_buffer;
+ uint32_t g_size_bytes, s_size_bytes;
+ uint64_t dptr_dma, rptr_dma;
+ vq_cmd_word0_t vq_cmd_w0;
+ vq_cmd_word3_t vq_cmd_w3;
+ void *c_vaddr, *m_vaddr;
+ uint64_t c_dma, m_dma;
+ opcode_info_t opcode;
+
+ ctx = params->ctx_buf.vaddr;
+ meta_p = ¶ms->meta_buf;
+
+ m_vaddr = meta_p->vaddr;
+ m_dma = meta_p->dma_addr;
+
+ /*
+ * Save initial space that followed app data for completion code &
+ * alternate completion code to fall in same cache line as app data
+ */
+ m_vaddr = (uint8_t *)m_vaddr + COMPLETION_CODE_SIZE;
+ m_dma += COMPLETION_CODE_SIZE;
+ size = (uint8_t *)RTE_PTR_ALIGN((uint8_t *)m_vaddr, 16) -
+ (uint8_t *)m_vaddr;
+ c_vaddr = (uint8_t *)m_vaddr + size;
+ c_dma = m_dma + size;
+ size += sizeof(cpt_res_s_t);
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ req = m_vaddr;
+
+ size = sizeof(struct cpt_request_info);
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ hash_type = ctx->hash_type;
+ mac_len = ctx->mac_len;
+ key_len = ctx->auth_key_len;
+ data_len = AUTH_DLEN(d_lens);
+
+ /*GP op header */
+ vq_cmd_w0.u64 = 0;
+ vq_cmd_w0.s.param2 = ((uint16_t)hash_type << 8);
+ if (ctx->hmac) {
+ opcode.s.major = CPT_MAJOR_OP_HMAC | CPT_DMA_MODE;
+ vq_cmd_w0.s.param1 = key_len;
+ vq_cmd_w0.s.dlen = data_len + ROUNDUP8(key_len);
+ } else {
+ opcode.s.major = CPT_MAJOR_OP_HASH | CPT_DMA_MODE;
+ vq_cmd_w0.s.param1 = 0;
+ vq_cmd_w0.s.dlen = data_len;
+ }
+
+ opcode.s.minor = 0;
+
+ /* Null auth only case enters the if */
+ if (unlikely(!hash_type && !ctx->enc_cipher)) {
+ opcode.s.major = CPT_MAJOR_OP_MISC;
+ /* Minor op is passthrough */
+ opcode.s.minor = 0x03;
+ /* Send out completion code only */
+ vq_cmd_w0.s.param2 = 0x1;
+ }
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ /* DPTR has SG list */
+ in_buffer = m_vaddr;
+ dptr_dma = m_dma;
+
+ ((uint16_t *)in_buffer)[0] = 0;
+ ((uint16_t *)in_buffer)[1] = 0;
+
+ /* TODO Add error check if space will be sufficient */
+ gather_comp = (sg_comp_t *)((uint8_t *)m_vaddr + 8);
+
+ /*
+ * Input gather list
+ */
+
+ i = 0;
+
+ if (ctx->hmac) {
+ uint64_t k_dma = params->ctx_buf.dma_addr +
+ offsetof(struct cpt_ctx, auth_key);
+ /* Key */
+ i = fill_sg_comp(gather_comp, i, k_dma, ROUNDUP8(key_len));
+ }
+
+ /* input data */
+ size = data_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(gather_comp, i, params->src_iov,
+ 0, &size, NULL, 0);
+ if (unlikely(size)) {
+ CPT_LOG_DP_DEBUG("Insufficient dst IOV size, short"
+ " by %dB", size);
+ return;
+ }
+ } else {
+ /*
+ * Looks like we need to support zero data
+ * gather ptr in case of hash & hmac
+ */
+ i++;
+ }
+ ((uint16_t *)in_buffer)[2] = rte_cpu_to_be_16(i);
+ g_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ /*
+ * Output Gather list
+ */
+
+ i = 0;
+ scatter_comp = (sg_comp_t *)((uint8_t *)gather_comp + g_size_bytes);
+
+ if (flags & VALID_MAC_BUF) {
+ if (unlikely(params->mac_buf.size < mac_len)) {
+ CPT_LOG_DP_ERR("Insufficient MAC size");
+ return;
+ }
+
+ size = mac_len;
+ i = fill_sg_comp_from_buf_min(scatter_comp, i,
+ ¶ms->mac_buf, &size);
+ } else {
+ size = mac_len;
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ params->src_iov, data_len,
+ &size, NULL, 0);
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient dst IOV size, short by"
+ " %dB", size);
+ return;
+ }
+ }
+
+ ((uint16_t *)in_buffer)[3] = rte_cpu_to_be_16(i);
+ s_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ size = g_size_bytes + s_size_bytes + SG_LIST_HDR_SIZE;
+
+ /* This is DPTR len incase of SG mode */
+ vq_cmd_w0.s.dlen = size;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* cpt alternate completion address saved earlier */
+ req->alternate_caddr = (uint64_t *)((uint8_t *)c_vaddr - 8);
+ *req->alternate_caddr = ~((uint64_t)COMPLETION_CODE_INIT);
+ rptr_dma = c_dma - 8;
+
+ req->ist.ei1 = dptr_dma;
+ req->ist.ei2 = rptr_dma;
+
+ /* vq command w3 */
+ vq_cmd_w3.u64 = 0;
+
+ /* 16 byte aligned cpt res address */
+ req->completion_addr = (uint64_t *)((uint8_t *)c_vaddr);
+ *req->completion_addr = COMPLETION_CODE_INIT;
+ req->comp_baddr = c_dma;
+
+ /* Fill microcode part of instruction */
+ req->ist.ei0 = vq_cmd_w0.u64;
+ req->ist.ei3 = vq_cmd_w3.u64;
+
+ req->op = op;
+
+ *prep_req = req;
+ return;
+}
+
+static __rte_always_inline void
+cpt_enc_hmac_prep(uint32_t flags,
+ uint64_t d_offs,
+ uint64_t d_lens,
+ fc_params_t *fc_params,
+ void *op,
+ void **prep_req)
+{
+ uint32_t iv_offset = 0;
+ int32_t inputlen, outputlen, enc_dlen, auth_dlen;
+ struct cpt_ctx *cpt_ctx;
+ uint32_t cipher_type, hash_type;
+ uint32_t mac_len, size;
+ uint8_t iv_len = 16;
+ struct cpt_request_info *req;
+ buf_ptr_t *meta_p, *aad_buf = NULL;
+ uint32_t encr_offset, auth_offset;
+ uint32_t encr_data_len, auth_data_len, aad_len = 0;
+ uint32_t passthrough_len = 0;
+ void *m_vaddr, *offset_vaddr;
+ uint64_t m_dma, offset_dma, ctx_dma;
+ vq_cmd_word0_t vq_cmd_w0;
+ vq_cmd_word3_t vq_cmd_w3;
+ void *c_vaddr;
+ uint64_t c_dma;
+ opcode_info_t opcode;
+
+ meta_p = &fc_params->meta_buf;
+ m_vaddr = meta_p->vaddr;
+ m_dma = meta_p->dma_addr;
+
+ encr_offset = ENCR_OFFSET(d_offs);
+ auth_offset = AUTH_OFFSET(d_offs);
+ encr_data_len = ENCR_DLEN(d_lens);
+ auth_data_len = AUTH_DLEN(d_lens);
+ if (unlikely(flags & VALID_AAD_BUF)) {
+ /*
+ * We dont support both aad
+ * and auth data separately
+ */
+ auth_data_len = 0;
+ auth_offset = 0;
+ aad_len = fc_params->aad_buf.size;
+ aad_buf = &fc_params->aad_buf;
+ }
+ cpt_ctx = fc_params->ctx_buf.vaddr;
+ cipher_type = cpt_ctx->enc_cipher;
+ hash_type = cpt_ctx->hash_type;
+ mac_len = cpt_ctx->mac_len;
+
+ /*
+ * Save initial space that followed app data for completion code &
+ * alternate completion code to fall in same cache line as app data
+ */
+ m_vaddr = (uint8_t *)m_vaddr + COMPLETION_CODE_SIZE;
+ m_dma += COMPLETION_CODE_SIZE;
+ size = (uint8_t *)RTE_PTR_ALIGN((uint8_t *)m_vaddr, 16) -
+ (uint8_t *)m_vaddr;
+
+ c_vaddr = (uint8_t *)m_vaddr + size;
+ c_dma = m_dma + size;
+ size += sizeof(cpt_res_s_t);
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* start cpt request info struct at 8 byte boundary */
+ size = (uint8_t *)RTE_PTR_ALIGN(m_vaddr, 8) -
+ (uint8_t *)m_vaddr;
+
+ req = (struct cpt_request_info *)((uint8_t *)m_vaddr + size);
+
+ size += sizeof(struct cpt_request_info);
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ if (hash_type == GMAC_TYPE)
+ encr_data_len = 0;
+
+ if (unlikely(!(flags & VALID_IV_BUF))) {
+ iv_len = 0;
+ iv_offset = ENCR_IV_OFFSET(d_offs);
+ }
+
+ if (unlikely(flags & VALID_AAD_BUF)) {
+ /*
+ * When AAD is given, data above encr_offset is pass through
+ * Since AAD is given as separate pointer and not as offset,
+ * this is a special case as we need to fragment input data
+ * into passthrough + encr_data and then insert AAD in between.
+ */
+ if (hash_type != GMAC_TYPE) {
+ passthrough_len = encr_offset;
+ auth_offset = passthrough_len + iv_len;
+ encr_offset = passthrough_len + aad_len + iv_len;
+ auth_data_len = aad_len + encr_data_len;
+ } else {
+ passthrough_len = 16 + aad_len;
+ auth_offset = passthrough_len + iv_len;
+ auth_data_len = aad_len;
+ }
+ } else {
+ encr_offset += iv_len;
+ auth_offset += iv_len;
+ }
+
+ /* Encryption */
+ opcode.s.major = CPT_MAJOR_OP_FC;
+ opcode.s.minor = 0;
+
+ auth_dlen = auth_offset + auth_data_len;
+ enc_dlen = encr_data_len + encr_offset;
+ if (unlikely(encr_data_len & 0xf)) {
+ if ((cipher_type == DES3_CBC) || (cipher_type == DES3_ECB))
+ enc_dlen = ROUNDUP8(encr_data_len) + encr_offset;
+ else if (likely((cipher_type == AES_CBC) ||
+ (cipher_type == AES_ECB)))
+ enc_dlen = ROUNDUP16(encr_data_len) + encr_offset;
+ }
+
+ if (unlikely(hash_type == GMAC_TYPE)) {
+ encr_offset = auth_dlen;
+ enc_dlen = 0;
+ }
+
+ if (unlikely(auth_dlen > enc_dlen)) {
+ inputlen = auth_dlen;
+ outputlen = auth_dlen + mac_len;
+ } else {
+ inputlen = enc_dlen;
+ outputlen = enc_dlen + mac_len;
+ }
+
+ /* GP op header */
+ vq_cmd_w0.u64 = 0;
+ vq_cmd_w0.s.param1 = encr_data_len;
+ vq_cmd_w0.s.param2 = auth_data_len;
+ /*
+ * In 83XX since we have a limitation of
+ * IV & Offset control word not part of instruction
+ * and need to be part of Data Buffer, we check if
+ * head room is there and then only do the Direct mode processing
+ */
+ if (likely((flags & SINGLE_BUF_INPLACE) &&
+ (flags & SINGLE_BUF_HEADTAILROOM))) {
+ void *dm_vaddr = fc_params->bufs[0].vaddr;
+ uint64_t dm_dma_addr = fc_params->bufs[0].dma_addr;
+ /*
+ * This flag indicates that there is 24 bytes head room and
+ * 8 bytes tail room available, so that we get to do
+ * DIRECT MODE with limitation
+ */
+
+ offset_vaddr = (uint8_t *)dm_vaddr - OFF_CTRL_LEN - iv_len;
+ offset_dma = dm_dma_addr - OFF_CTRL_LEN - iv_len;
+
+ /* DPTR */
+ req->ist.ei1 = offset_dma;
+ /* RPTR should just exclude offset control word */
+ req->ist.ei2 = dm_dma_addr - iv_len;
+ req->alternate_caddr = (uint64_t *)((uint8_t *)dm_vaddr
+ + outputlen - iv_len);
+
+ vq_cmd_w0.s.dlen = inputlen + OFF_CTRL_LEN;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ if (likely(iv_len)) {
+ uint64_t *dest = (uint64_t *)((uint8_t *)offset_vaddr
+ + OFF_CTRL_LEN);
+ uint64_t *src = fc_params->iv_buf;
+ dest[0] = src[0];
+ dest[1] = src[1];
+ }
+
+ *(uint64_t *)offset_vaddr =
+ rte_cpu_to_be_64(((uint64_t)encr_offset << 16) |
+ ((uint64_t)iv_offset << 8) |
+ ((uint64_t)auth_offset));
+
+ } else {
+ uint32_t i, g_size_bytes, s_size_bytes;
+ uint64_t dptr_dma, rptr_dma;
+ sg_comp_t *gather_comp;
+ sg_comp_t *scatter_comp;
+ uint8_t *in_buffer;
+
+ /* This falls under strict SG mode */
+ offset_vaddr = m_vaddr;
+ offset_dma = m_dma;
+ size = OFF_CTRL_LEN + iv_len;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ opcode.s.major |= CPT_DMA_MODE;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ if (likely(iv_len)) {
+ uint64_t *dest = (uint64_t *)((uint8_t *)offset_vaddr
+ + OFF_CTRL_LEN);
+ uint64_t *src = fc_params->iv_buf;
+ dest[0] = src[0];
+ dest[1] = src[1];
+ }
+
+ *(uint64_t *)offset_vaddr =
+ rte_cpu_to_be_64(((uint64_t)encr_offset << 16) |
+ ((uint64_t)iv_offset << 8) |
+ ((uint64_t)auth_offset));
+
+ /* DPTR has SG list */
+ in_buffer = m_vaddr;
+ dptr_dma = m_dma;
+
+ ((uint16_t *)in_buffer)[0] = 0;
+ ((uint16_t *)in_buffer)[1] = 0;
+
+ /* TODO Add error check if space will be sufficient */
+ gather_comp = (sg_comp_t *)((uint8_t *)m_vaddr + 8);
+
+ /*
+ * Input Gather List
+ */
+
+ i = 0;
+
+ /* Offset control word that includes iv */
+ i = fill_sg_comp(gather_comp, i, offset_dma,
+ OFF_CTRL_LEN + iv_len);
+
+ /* Add input data */
+ size = inputlen - iv_len;
+ if (likely(size)) {
+ uint32_t aad_offset = aad_len ? passthrough_len : 0;
+
+ if (unlikely(flags & SINGLE_BUF_INPLACE)) {
+ i = fill_sg_comp_from_buf_min(gather_comp, i,
+ fc_params->bufs,
+ &size);
+ } else {
+ i = fill_sg_comp_from_iov(gather_comp, i,
+ fc_params->src_iov,
+ 0, &size,
+ aad_buf, aad_offset);
+ }
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ ((uint16_t *)in_buffer)[2] = rte_cpu_to_be_16(i);
+ g_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ /*
+ * Output Scatter list
+ */
+ i = 0;
+ scatter_comp =
+ (sg_comp_t *)((uint8_t *)gather_comp + g_size_bytes);
+
+ /* Add IV */
+ if (likely(iv_len)) {
+ i = fill_sg_comp(scatter_comp, i,
+ offset_dma + OFF_CTRL_LEN,
+ iv_len);
+ }
+
+ /* output data or output data + digest*/
+ if (unlikely(flags & VALID_MAC_BUF)) {
+ size = outputlen - iv_len - mac_len;
+ if (size) {
+ uint32_t aad_offset =
+ aad_len ? passthrough_len : 0;
+
+ if (unlikely(flags & SINGLE_BUF_INPLACE)) {
+ i = fill_sg_comp_from_buf_min(
+ scatter_comp,
+ i,
+ fc_params->bufs,
+ &size);
+ } else {
+ i = fill_sg_comp_from_iov(scatter_comp,
+ i,
+ fc_params->dst_iov,
+ 0,
+ &size,
+ aad_buf,
+ aad_offset);
+ }
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer"
+ " space, size %d needed",
+ size);
+ return;
+ }
+ }
+ /* mac_data */
+ if (mac_len) {
+ i = fill_sg_comp_from_buf(scatter_comp, i,
+ &fc_params->mac_buf);
+ }
+ } else {
+ /* Output including mac */
+ size = outputlen - iv_len;
+ if (likely(size)) {
+ uint32_t aad_offset =
+ aad_len ? passthrough_len : 0;
+
+ if (unlikely(flags & SINGLE_BUF_INPLACE)) {
+ i = fill_sg_comp_from_buf_min(
+ scatter_comp,
+ i,
+ fc_params->bufs,
+ &size);
+ } else {
+ i = fill_sg_comp_from_iov(scatter_comp,
+ i,
+ fc_params->dst_iov,
+ 0,
+ &size,
+ aad_buf,
+ aad_offset);
+ }
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer"
+ " space, size %d needed",
+ size);
+ return;
+ }
+ }
+ }
+ ((uint16_t *)in_buffer)[3] = rte_cpu_to_be_16(i);
+ s_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ size = g_size_bytes + s_size_bytes + SG_LIST_HDR_SIZE;
+
+ /* This is DPTR len incase of SG mode */
+ vq_cmd_w0.s.dlen = size;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* cpt alternate completion address saved earlier */
+ req->alternate_caddr = (uint64_t *)((uint8_t *)c_vaddr - 8);
+ *req->alternate_caddr = ~((uint64_t)COMPLETION_CODE_INIT);
+ rptr_dma = c_dma - 8;
+
+ req->ist.ei1 = dptr_dma;
+ req->ist.ei2 = rptr_dma;
+ }
+
+ ctx_dma = fc_params->ctx_buf.dma_addr +
+ offsetof(struct cpt_ctx, fctx);
+ /* vq command w3 */
+ vq_cmd_w3.u64 = 0;
+ vq_cmd_w3.s.grp = 0;
+ vq_cmd_w3.s.cptr = ctx_dma;
+
+ /* 16 byte aligned cpt res address */
+ req->completion_addr = (uint64_t *)((uint8_t *)c_vaddr);
+ *req->completion_addr = COMPLETION_CODE_INIT;
+ req->comp_baddr = c_dma;
+
+ /* Fill microcode part of instruction */
+ req->ist.ei0 = vq_cmd_w0.u64;
+ req->ist.ei3 = vq_cmd_w3.u64;
+
+ req->op = op;
+
+ *prep_req = req;
+ return;
+}
+
+static __rte_always_inline void
+cpt_dec_hmac_prep(uint32_t flags,
+ uint64_t d_offs,
+ uint64_t d_lens,
+ fc_params_t *fc_params,
+ void *op,
+ void **prep_req)
+{
+ uint32_t iv_offset = 0, size;
+ int32_t inputlen, outputlen, enc_dlen, auth_dlen;
+ struct cpt_ctx *cpt_ctx;
+ int32_t hash_type, mac_len;
+ uint8_t iv_len = 16;
+ struct cpt_request_info *req;
+ buf_ptr_t *meta_p, *aad_buf = NULL;
+ uint32_t encr_offset, auth_offset;
+ uint32_t encr_data_len, auth_data_len, aad_len = 0;
+ uint32_t passthrough_len = 0;
+ void *m_vaddr, *offset_vaddr;
+ uint64_t m_dma, offset_dma, ctx_dma;
+ opcode_info_t opcode;
+ vq_cmd_word0_t vq_cmd_w0;
+ vq_cmd_word3_t vq_cmd_w3;
+ void *c_vaddr;
+ uint64_t c_dma;
+
+ meta_p = &fc_params->meta_buf;
+ m_vaddr = meta_p->vaddr;
+ m_dma = meta_p->dma_addr;
+
+ encr_offset = ENCR_OFFSET(d_offs);
+ auth_offset = AUTH_OFFSET(d_offs);
+ encr_data_len = ENCR_DLEN(d_lens);
+ auth_data_len = AUTH_DLEN(d_lens);
+
+ if (unlikely(flags & VALID_AAD_BUF)) {
+ /*
+ * We dont support both aad
+ * and auth data separately
+ */
+ auth_data_len = 0;
+ auth_offset = 0;
+ aad_len = fc_params->aad_buf.size;
+ aad_buf = &fc_params->aad_buf;
+ }
+
+ cpt_ctx = fc_params->ctx_buf.vaddr;
+ hash_type = cpt_ctx->hash_type;
+ mac_len = cpt_ctx->mac_len;
+
+ if (hash_type == GMAC_TYPE)
+ encr_data_len = 0;
+
+ if (unlikely(!(flags & VALID_IV_BUF))) {
+ iv_len = 0;
+ iv_offset = ENCR_IV_OFFSET(d_offs);
+ }
+
+ if (unlikely(flags & VALID_AAD_BUF)) {
+ /*
+ * When AAD is given, data above encr_offset is pass through
+ * Since AAD is given as separate pointer and not as offset,
+ * this is a special case as we need to fragment input data
+ * into passthrough + encr_data and then insert AAD in between.
+ */
+ if (hash_type != GMAC_TYPE) {
+ passthrough_len = encr_offset;
+ auth_offset = passthrough_len + iv_len;
+ encr_offset = passthrough_len + aad_len + iv_len;
+ auth_data_len = aad_len + encr_data_len;
+ } else {
+ passthrough_len = 16 + aad_len;
+ auth_offset = passthrough_len + iv_len;
+ auth_data_len = aad_len;
+ }
+ } else {
+ encr_offset += iv_len;
+ auth_offset += iv_len;
+ }
+
+ /*
+ * Save initial space that followed app data for completion code &
+ * alternate completion code to fall in same cache line as app data
+ */
+ m_vaddr = (uint8_t *)m_vaddr + COMPLETION_CODE_SIZE;
+ m_dma += COMPLETION_CODE_SIZE;
+ size = (uint8_t *)RTE_PTR_ALIGN((uint8_t *)m_vaddr, 16) -
+ (uint8_t *)m_vaddr;
+ c_vaddr = (uint8_t *)m_vaddr + size;
+ c_dma = m_dma + size;
+ size += sizeof(cpt_res_s_t);
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* start cpt request info structure at 8 byte alignment */
+ size = (uint8_t *)RTE_PTR_ALIGN(m_vaddr, 8) -
+ (uint8_t *)m_vaddr;
+
+ req = (struct cpt_request_info *)((uint8_t *)m_vaddr + size);
+
+ size += sizeof(struct cpt_request_info);
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* Decryption */
+ opcode.s.major = CPT_MAJOR_OP_FC;
+ opcode.s.minor = 1;
+
+ enc_dlen = encr_offset + encr_data_len;
+ auth_dlen = auth_offset + auth_data_len;
+
+ if (auth_dlen > enc_dlen) {
+ inputlen = auth_dlen + mac_len;
+ outputlen = auth_dlen;
+ } else {
+ inputlen = enc_dlen + mac_len;
+ outputlen = enc_dlen;
+ }
+
+ if (hash_type == GMAC_TYPE)
+ encr_offset = inputlen;
+
+ vq_cmd_w0.u64 = 0;
+ vq_cmd_w0.s.param1 = encr_data_len;
+ vq_cmd_w0.s.param2 = auth_data_len;
+
+ /*
+ * In 83XX since we have a limitation of
+ * IV & Offset control word not part of instruction
+ * and need to be part of Data Buffer, we check if
+ * head room is there and then only do the Direct mode processing
+ */
+ if (likely((flags & SINGLE_BUF_INPLACE) &&
+ (flags & SINGLE_BUF_HEADTAILROOM))) {
+ void *dm_vaddr = fc_params->bufs[0].vaddr;
+ uint64_t dm_dma_addr = fc_params->bufs[0].dma_addr;
+ /*
+ * This flag indicates that there is 24 bytes head room and
+ * 8 bytes tail room available, so that we get to do
+ * DIRECT MODE with limitation
+ */
+
+ offset_vaddr = (uint8_t *)dm_vaddr - OFF_CTRL_LEN - iv_len;
+ offset_dma = dm_dma_addr - OFF_CTRL_LEN - iv_len;
+ req->ist.ei1 = offset_dma;
+
+ /* RPTR should just exclude offset control word */
+ req->ist.ei2 = dm_dma_addr - iv_len;
+
+ req->alternate_caddr = (uint64_t *)((uint8_t *)dm_vaddr +
+ outputlen - iv_len);
+ /* since this is decryption,
+ * don't touch the content of
+ * alternate ccode space as it contains
+ * hmac.
+ */
+
+ vq_cmd_w0.s.dlen = inputlen + OFF_CTRL_LEN;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ if (likely(iv_len)) {
+ uint64_t *dest = (uint64_t *)((uint8_t *)offset_vaddr +
+ OFF_CTRL_LEN);
+ uint64_t *src = fc_params->iv_buf;
+ dest[0] = src[0];
+ dest[1] = src[1];
+ }
+
+ *(uint64_t *)offset_vaddr =
+ rte_cpu_to_be_64(((uint64_t)encr_offset << 16) |
+ ((uint64_t)iv_offset << 8) |
+ ((uint64_t)auth_offset));
+
+ } else {
+ uint64_t dptr_dma, rptr_dma;
+ uint32_t g_size_bytes, s_size_bytes;
+ sg_comp_t *gather_comp;
+ sg_comp_t *scatter_comp;
+ uint8_t *in_buffer;
+ uint8_t i = 0;
+
+ /* This falls under strict SG mode */
+ offset_vaddr = m_vaddr;
+ offset_dma = m_dma;
+ size = OFF_CTRL_LEN + iv_len;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ opcode.s.major |= CPT_DMA_MODE;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ if (likely(iv_len)) {
+ uint64_t *dest = (uint64_t *)((uint8_t *)offset_vaddr +
+ OFF_CTRL_LEN);
+ uint64_t *src = fc_params->iv_buf;
+ dest[0] = src[0];
+ dest[1] = src[1];
+ }
+
+ *(uint64_t *)offset_vaddr =
+ rte_cpu_to_be_64(((uint64_t)encr_offset << 16) |
+ ((uint64_t)iv_offset << 8) |
+ ((uint64_t)auth_offset));
+
+ /* DPTR has SG list */
+ in_buffer = m_vaddr;
+ dptr_dma = m_dma;
+
+ ((uint16_t *)in_buffer)[0] = 0;
+ ((uint16_t *)in_buffer)[1] = 0;
+
+ /* TODO Add error check if space will be sufficient */
+ gather_comp = (sg_comp_t *)((uint8_t *)m_vaddr + 8);
+
+ /*
+ * Input Gather List
+ */
+ i = 0;
+
+ /* Offset control word that includes iv */
+ i = fill_sg_comp(gather_comp, i, offset_dma,
+ OFF_CTRL_LEN + iv_len);
+
+ /* Add input data */
+ if (flags & VALID_MAC_BUF) {
+ size = inputlen - iv_len - mac_len;
+ if (size) {
+ /* input data only */
+ if (unlikely(flags & SINGLE_BUF_INPLACE)) {
+ i = fill_sg_comp_from_buf_min(
+ gather_comp, i,
+ fc_params->bufs,
+ &size);
+ } else {
+ uint32_t aad_offset = aad_len ?
+ passthrough_len : 0;
+
+ i = fill_sg_comp_from_iov(gather_comp,
+ i,
+ fc_params->src_iov,
+ 0, &size,
+ aad_buf,
+ aad_offset);
+ }
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer"
+ " space, size %d needed",
+ size);
+ return;
+ }
+ }
+
+ /* mac data */
+ if (mac_len) {
+ i = fill_sg_comp_from_buf(gather_comp, i,
+ &fc_params->mac_buf);
+ }
+ } else {
+ /* input data + mac */
+ size = inputlen - iv_len;
+ if (size) {
+ if (unlikely(flags & SINGLE_BUF_INPLACE)) {
+ i = fill_sg_comp_from_buf_min(
+ gather_comp, i,
+ fc_params->bufs,
+ &size);
+ } else {
+ uint32_t aad_offset = aad_len ?
+ passthrough_len : 0;
+
+ if (unlikely(!fc_params->src_iov)) {
+ CPT_LOG_DP_ERR("Bad input args");
+ return;
+ }
+
+ i = fill_sg_comp_from_iov(
+ gather_comp, i,
+ fc_params->src_iov,
+ 0, &size,
+ aad_buf,
+ aad_offset);
+ }
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer"
+ " space, size %d needed",
+ size);
+ return;
+ }
+ }
+ }
+ ((uint16_t *)in_buffer)[2] = rte_cpu_to_be_16(i);
+ g_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ /*
+ * Output Scatter List
+ */
+
+ i = 0;
+ scatter_comp =
+ (sg_comp_t *)((uint8_t *)gather_comp + g_size_bytes);
+
+ /* Add iv */
+ if (iv_len) {
+ i = fill_sg_comp(scatter_comp, i,
+ offset_dma + OFF_CTRL_LEN,
+ iv_len);
+ }
+
+ /* Add output data */
+ size = outputlen - iv_len;
+ if (size) {
+ if (unlikely(flags & SINGLE_BUF_INPLACE)) {
+ /* handle single buffer here */
+ i = fill_sg_comp_from_buf_min(scatter_comp, i,
+ fc_params->bufs,
+ &size);
+ } else {
+ uint32_t aad_offset = aad_len ?
+ passthrough_len : 0;
+
+ if (unlikely(!fc_params->dst_iov)) {
+ CPT_LOG_DP_ERR("Bad input args");
+ return;
+ }
+
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ fc_params->dst_iov, 0,
+ &size, aad_buf,
+ aad_offset);
+ }
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+
+ ((uint16_t *)in_buffer)[3] = rte_cpu_to_be_16(i);
+ s_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ size = g_size_bytes + s_size_bytes + SG_LIST_HDR_SIZE;
+
+ /* This is DPTR len incase of SG mode */
+ vq_cmd_w0.s.dlen = size;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* cpt alternate completion address saved earlier */
+ req->alternate_caddr = (uint64_t *)((uint8_t *)c_vaddr - 8);
+ *req->alternate_caddr = ~((uint64_t)COMPLETION_CODE_INIT);
+ rptr_dma = c_dma - 8;
+ size += COMPLETION_CODE_SIZE;
+
+ req->ist.ei1 = dptr_dma;
+ req->ist.ei2 = rptr_dma;
+ }
+
+ ctx_dma = fc_params->ctx_buf.dma_addr +
+ offsetof(struct cpt_ctx, fctx);
+ /* vq command w3 */
+ vq_cmd_w3.u64 = 0;
+ vq_cmd_w3.s.grp = 0;
+ vq_cmd_w3.s.cptr = ctx_dma;
+
+ /* 16 byte aligned cpt res address */
+ req->completion_addr = (uint64_t *)((uint8_t *)c_vaddr);
+ *req->completion_addr = COMPLETION_CODE_INIT;
+ req->comp_baddr = c_dma;
+
+ /* Fill microcode part of instruction */
+ req->ist.ei0 = vq_cmd_w0.u64;
+ req->ist.ei3 = vq_cmd_w3.u64;
+
+ req->op = op;
+
+ *prep_req = req;
+ return;
+}
+
+static __rte_always_inline void
+cpt_zuc_snow3g_enc_prep(uint32_t req_flags,
+ uint64_t d_offs,
+ uint64_t d_lens,
+ fc_params_t *params,
+ void *op,
+ void **prep_req)
+{
+ uint32_t size;
+ int32_t inputlen, outputlen;
+ struct cpt_ctx *cpt_ctx;
+ uint32_t mac_len = 0;
+ uint8_t snow3g, j;
+ struct cpt_request_info *req;
+ buf_ptr_t *buf_p;
+ uint32_t encr_offset = 0, auth_offset = 0;
+ uint32_t encr_data_len = 0, auth_data_len = 0;
+ int flags, iv_len = 16;
+ void *m_vaddr, *c_vaddr;
+ uint64_t m_dma, c_dma, offset_ctrl;
+ uint64_t *offset_vaddr, offset_dma;
+ uint32_t *iv_s, iv[4];
+ vq_cmd_word0_t vq_cmd_w0;
+ vq_cmd_word3_t vq_cmd_w3;
+ opcode_info_t opcode;
+
+ buf_p = ¶ms->meta_buf;
+ m_vaddr = buf_p->vaddr;
+ m_dma = buf_p->dma_addr;
+
+ cpt_ctx = params->ctx_buf.vaddr;
+ flags = cpt_ctx->zsk_flags;
+ mac_len = cpt_ctx->mac_len;
+ snow3g = cpt_ctx->snow3g;
+
+ /*
+ * Save initial space that followed app data for completion code &
+ * alternate completion code to fall in same cache line as app data
+ */
+ m_vaddr = (uint8_t *)m_vaddr + COMPLETION_CODE_SIZE;
+ m_dma += COMPLETION_CODE_SIZE;
+ size = (uint8_t *)RTE_PTR_ALIGN((uint8_t *)m_vaddr, 16) -
+ (uint8_t *)m_vaddr;
+
+ c_vaddr = (uint8_t *)m_vaddr + size;
+ c_dma = m_dma + size;
+ size += sizeof(cpt_res_s_t);
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* Reserve memory for cpt request info */
+ req = m_vaddr;
+
+ size = sizeof(struct cpt_request_info);
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ opcode.s.major = CPT_MAJOR_OP_ZUC_SNOW3G;
+
+ /* indicates CPTR ctx, operation type, KEY & IV mode from DPTR */
+
+ opcode.s.minor = ((1 << 7) | (snow3g << 5) | (0 << 4) |
+ (0 << 3) | (flags & 0x7));
+
+ if (flags == 0x1) {
+ /*
+ * Microcode expects offsets in bytes
+ * TODO: Rounding off
+ */
+ auth_data_len = AUTH_DLEN(d_lens);
+
+ /* EIA3 or UIA2 */
+ auth_offset = AUTH_OFFSET(d_offs);
+ auth_offset = auth_offset / 8;
+
+ /* consider iv len */
+ auth_offset += iv_len;
+
+ inputlen = auth_offset + (RTE_ALIGN(auth_data_len, 8) / 8);
+ outputlen = mac_len;
+
+ offset_ctrl = rte_cpu_to_be_64((uint64_t)auth_offset);
+
+ } else {
+ /* EEA3 or UEA2 */
+ /*
+ * Microcode expects offsets in bytes
+ * TODO: Rounding off
+ */
+ encr_data_len = ENCR_DLEN(d_lens);
+
+ encr_offset = ENCR_OFFSET(d_offs);
+ encr_offset = encr_offset / 8;
+ /* consider iv len */
+ encr_offset += iv_len;
+
+ inputlen = encr_offset + (RTE_ALIGN(encr_data_len, 8) / 8);
+ outputlen = inputlen;
+
+ /* iv offset is 0 */
+ offset_ctrl = rte_cpu_to_be_64((uint64_t)encr_offset << 16);
+ }
+
+ /* IV */
+ iv_s = (flags == 0x1) ? params->auth_iv_buf :
+ params->iv_buf;
+
+ if (snow3g) {
+ /*
+ * DPDK seems to provide it in form of IV3 IV2 IV1 IV0
+ * and BigEndian, MC needs it as IV0 IV1 IV2 IV3
+ */
+
+ for (j = 0; j < 4; j++)
+ iv[j] = iv_s[3 - j];
+ } else {
+ /* ZUC doesn't need a swap */
+ for (j = 0; j < 4; j++)
+ iv[j] = iv_s[j];
+ }
+
+ /*
+ * GP op header, lengths are expected in bits.
+ */
+ vq_cmd_w0.u64 = 0;
+ vq_cmd_w0.s.param1 = encr_data_len;
+ vq_cmd_w0.s.param2 = auth_data_len;
+
+ /*
+ * In 83XX since we have a limitation of
+ * IV & Offset control word not part of instruction
+ * and need to be part of Data Buffer, we check if
+ * head room is there and then only do the Direct mode processing
+ */
+ if (likely((req_flags & SINGLE_BUF_INPLACE) &&
+ (req_flags & SINGLE_BUF_HEADTAILROOM))) {
+ void *dm_vaddr = params->bufs[0].vaddr;
+ uint64_t dm_dma_addr = params->bufs[0].dma_addr;
+ /*
+ * This flag indicates that there is 24 bytes head room and
+ * 8 bytes tail room available, so that we get to do
+ * DIRECT MODE with limitation
+ */
+
+ offset_vaddr = (uint64_t *)((uint8_t *)dm_vaddr -
+ OFF_CTRL_LEN - iv_len);
+ offset_dma = dm_dma_addr - OFF_CTRL_LEN - iv_len;
+
+ /* DPTR */
+ req->ist.ei1 = offset_dma;
+ /* RPTR should just exclude offset control word */
+ req->ist.ei2 = dm_dma_addr - iv_len;
+ req->alternate_caddr = (uint64_t *)((uint8_t *)dm_vaddr
+ + outputlen - iv_len);
+
+ vq_cmd_w0.s.dlen = inputlen + OFF_CTRL_LEN;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ if (likely(iv_len)) {
+ uint32_t *iv_d = (uint32_t *)((uint8_t *)offset_vaddr
+ + OFF_CTRL_LEN);
+ memcpy(iv_d, iv, 16);
+ }
+
+ *offset_vaddr = offset_ctrl;
+ } else {
+ uint32_t i, g_size_bytes, s_size_bytes;
+ uint64_t dptr_dma, rptr_dma;
+ sg_comp_t *gather_comp;
+ sg_comp_t *scatter_comp;
+ uint8_t *in_buffer;
+ uint32_t *iv_d;
+
+ /* save space for iv */
+ offset_vaddr = m_vaddr;
+ offset_dma = m_dma;
+
+ m_vaddr = (uint8_t *)m_vaddr + OFF_CTRL_LEN + iv_len;
+ m_dma += OFF_CTRL_LEN + iv_len;
+
+ opcode.s.major |= CPT_DMA_MODE;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ /* DPTR has SG list */
+ in_buffer = m_vaddr;
+ dptr_dma = m_dma;
+
+ ((uint16_t *)in_buffer)[0] = 0;
+ ((uint16_t *)in_buffer)[1] = 0;
+
+ /* TODO Add error check if space will be sufficient */
+ gather_comp = (sg_comp_t *)((uint8_t *)m_vaddr + 8);
+
+ /*
+ * Input Gather List
+ */
+ i = 0;
+
+ /* Offset control word followed by iv */
+
+ i = fill_sg_comp(gather_comp, i, offset_dma,
+ OFF_CTRL_LEN + iv_len);
+
+ /* iv offset is 0 */
+ *offset_vaddr = offset_ctrl;
+
+ iv_d = (uint32_t *)((uint8_t *)offset_vaddr + OFF_CTRL_LEN);
+ memcpy(iv_d, iv, 16);
+
+ /* input data */
+ size = inputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(gather_comp, i,
+ params->src_iov,
+ 0, &size, NULL, 0);
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ ((uint16_t *)in_buffer)[2] = rte_cpu_to_be_16(i);
+ g_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ /*
+ * Output Scatter List
+ */
+
+ i = 0;
+ scatter_comp =
+ (sg_comp_t *)((uint8_t *)gather_comp + g_size_bytes);
+
+ if (flags == 0x1) {
+ /* IV in SLIST only for EEA3 & UEA2 */
+ iv_len = 0;
+ }
+
+ if (iv_len) {
+ i = fill_sg_comp(scatter_comp, i,
+ offset_dma + OFF_CTRL_LEN, iv_len);
+ }
+
+ /* Add output data */
+ if (req_flags & VALID_MAC_BUF) {
+ size = outputlen - iv_len - mac_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ params->dst_iov, 0,
+ &size, NULL, 0);
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+
+ /* mac data */
+ if (mac_len) {
+ i = fill_sg_comp_from_buf(scatter_comp, i,
+ ¶ms->mac_buf);
+ }
+ } else {
+ /* Output including mac */
+ size = outputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ params->dst_iov, 0,
+ &size, NULL, 0);
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ }
+ ((uint16_t *)in_buffer)[3] = rte_cpu_to_be_16(i);
+ s_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ size = g_size_bytes + s_size_bytes + SG_LIST_HDR_SIZE;
+
+ /* This is DPTR len incase of SG mode */
+ vq_cmd_w0.s.dlen = size;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* cpt alternate completion address saved earlier */
+ req->alternate_caddr = (uint64_t *)((uint8_t *)c_vaddr - 8);
+ *req->alternate_caddr = ~((uint64_t)COMPLETION_CODE_INIT);
+ rptr_dma = c_dma - 8;
+
+ req->ist.ei1 = dptr_dma;
+ req->ist.ei2 = rptr_dma;
+ }
+
+ /* vq command w3 */
+ vq_cmd_w3.u64 = 0;
+ vq_cmd_w3.s.grp = 0;
+ vq_cmd_w3.s.cptr = params->ctx_buf.dma_addr +
+ offsetof(struct cpt_ctx, zs_ctx);
+
+ /* 16 byte aligned cpt res address */
+ req->completion_addr = (uint64_t *)((uint8_t *)c_vaddr);
+ *req->completion_addr = COMPLETION_CODE_INIT;
+ req->comp_baddr = c_dma;
+
+ /* Fill microcode part of instruction */
+ req->ist.ei0 = vq_cmd_w0.u64;
+ req->ist.ei3 = vq_cmd_w3.u64;
+
+ req->op = op;
+
+ *prep_req = req;
+ return;
+}
+
+static __rte_always_inline void
+cpt_zuc_snow3g_dec_prep(uint32_t req_flags,
+ uint64_t d_offs,
+ uint64_t d_lens,
+ fc_params_t *params,
+ void *op,
+ void **prep_req)
+{
+ uint32_t size;
+ int32_t inputlen = 0, outputlen;
+ struct cpt_ctx *cpt_ctx;
+ uint8_t snow3g, iv_len = 16;
+ struct cpt_request_info *req;
+ buf_ptr_t *buf_p;
+ uint32_t encr_offset;
+ uint32_t encr_data_len;
+ int flags;
+ void *m_vaddr, *c_vaddr;
+ uint64_t m_dma, c_dma;
+ uint64_t *offset_vaddr, offset_dma;
+ uint32_t *iv_s, iv[4], j;
+ vq_cmd_word0_t vq_cmd_w0;
+ vq_cmd_word3_t vq_cmd_w3;
+ opcode_info_t opcode;
+
+ buf_p = ¶ms->meta_buf;
+ m_vaddr = buf_p->vaddr;
+ m_dma = buf_p->dma_addr;
+
+ /*
+ * Microcode expects offsets in bytes
+ * TODO: Rounding off
+ */
+ encr_offset = ENCR_OFFSET(d_offs) / 8;
+ encr_data_len = ENCR_DLEN(d_lens);
+
+ cpt_ctx = params->ctx_buf.vaddr;
+ flags = cpt_ctx->zsk_flags;
+ snow3g = cpt_ctx->snow3g;
+ /*
+ * Save initial space that followed app data for completion code &
+ * alternate completion code to fall in same cache line as app data
+ */
+ m_vaddr = (uint8_t *)m_vaddr + COMPLETION_CODE_SIZE;
+ m_dma += COMPLETION_CODE_SIZE;
+ size = (uint8_t *)RTE_PTR_ALIGN((uint8_t *)m_vaddr, 16) -
+ (uint8_t *)m_vaddr;
+
+ c_vaddr = (uint8_t *)m_vaddr + size;
+ c_dma = m_dma + size;
+ size += sizeof(cpt_res_s_t);
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* Reserve memory for cpt request info */
+ req = m_vaddr;
+
+ size = sizeof(struct cpt_request_info);
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ opcode.s.major = CPT_MAJOR_OP_ZUC_SNOW3G;
+
+ /* indicates CPTR ctx, operation type, KEY & IV mode from DPTR */
+
+ opcode.s.minor = ((1 << 7) | (snow3g << 5) | (0 << 4) |
+ (0 << 3) | (flags & 0x7));
+
+ /* consider iv len */
+ encr_offset += iv_len;
+
+ inputlen = encr_offset +
+ (RTE_ALIGN(encr_data_len, 8) / 8);
+ outputlen = inputlen;
+
+ /* IV */
+ iv_s = params->iv_buf;
+ if (snow3g) {
+ /*
+ * DPDK seems to provide it in form of IV3 IV2 IV1 IV0
+ * and BigEndian, MC needs it as IV0 IV1 IV2 IV3
+ */
+
+ for (j = 0; j < 4; j++)
+ iv[j] = iv_s[3 - j];
+ } else {
+ /* ZUC doesn't need a swap */
+ for (j = 0; j < 4; j++)
+ iv[j] = iv_s[j];
+ }
+
+ /*
+ * GP op header, lengths are expected in bits.
+ */
+ vq_cmd_w0.u64 = 0;
+ vq_cmd_w0.s.param1 = encr_data_len;
+
+ /*
+ * In 83XX since we have a limitation of
+ * IV & Offset control word not part of instruction
+ * and need to be part of Data Buffer, we check if
+ * head room is there and then only do the Direct mode processing
+ */
+ if (likely((req_flags & SINGLE_BUF_INPLACE) &&
+ (req_flags & SINGLE_BUF_HEADTAILROOM))) {
+ void *dm_vaddr = params->bufs[0].vaddr;
+ uint64_t dm_dma_addr = params->bufs[0].dma_addr;
+ /*
+ * This flag indicates that there is 24 bytes head room and
+ * 8 bytes tail room available, so that we get to do
+ * DIRECT MODE with limitation
+ */
+
+ offset_vaddr = (uint64_t *)((uint8_t *)dm_vaddr -
+ OFF_CTRL_LEN - iv_len);
+ offset_dma = dm_dma_addr - OFF_CTRL_LEN - iv_len;
+
+ /* DPTR */
+ req->ist.ei1 = offset_dma;
+ /* RPTR should just exclude offset control word */
+ req->ist.ei2 = dm_dma_addr - iv_len;
+ req->alternate_caddr = (uint64_t *)((uint8_t *)dm_vaddr
+ + outputlen - iv_len);
+
+ vq_cmd_w0.s.dlen = inputlen + OFF_CTRL_LEN;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ if (likely(iv_len)) {
+ uint32_t *iv_d = (uint32_t *)((uint8_t *)offset_vaddr
+ + OFF_CTRL_LEN);
+ memcpy(iv_d, iv, 16);
+ }
+
+ /* iv offset is 0 */
+ *offset_vaddr = rte_cpu_to_be_64((uint64_t)encr_offset << 16);
+ } else {
+ uint32_t i, g_size_bytes, s_size_bytes;
+ uint64_t dptr_dma, rptr_dma;
+ sg_comp_t *gather_comp;
+ sg_comp_t *scatter_comp;
+ uint8_t *in_buffer;
+ uint32_t *iv_d;
+
+ /* save space for offset and iv... */
+ offset_vaddr = m_vaddr;
+ offset_dma = m_dma;
+
+ m_vaddr = (uint8_t *)m_vaddr + OFF_CTRL_LEN + iv_len;
+ m_dma += OFF_CTRL_LEN + iv_len;
+
+ opcode.s.major |= CPT_DMA_MODE;
+
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ /* DPTR has SG list */
+ in_buffer = m_vaddr;
+ dptr_dma = m_dma;
+
+ ((uint16_t *)in_buffer)[0] = 0;
+ ((uint16_t *)in_buffer)[1] = 0;
+
+ /* TODO Add error check if space will be sufficient */
+ gather_comp = (sg_comp_t *)((uint8_t *)m_vaddr + 8);
+
+ /*
+ * Input Gather List
+ */
+ i = 0;
+
+ /* Offset control word */
+
+ /* iv offset is 0 */
+ *offset_vaddr = rte_cpu_to_be_64((uint64_t)encr_offset << 16);
+
+ i = fill_sg_comp(gather_comp, i, offset_dma,
+ OFF_CTRL_LEN + iv_len);
+
+ iv_d = (uint32_t *)((uint8_t *)offset_vaddr + OFF_CTRL_LEN);
+ memcpy(iv_d, iv, 16);
+
+ /* Add input data */
+ size = inputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(gather_comp, i,
+ params->src_iov,
+ 0, &size, NULL, 0);
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ ((uint16_t *)in_buffer)[2] = rte_cpu_to_be_16(i);
+ g_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ /*
+ * Output Scatter List
+ */
+
+ i = 0;
+ scatter_comp =
+ (sg_comp_t *)((uint8_t *)gather_comp + g_size_bytes);
+
+ /* IV */
+ i = fill_sg_comp(scatter_comp, i,
+ offset_dma + OFF_CTRL_LEN,
+ iv_len);
+
+ /* Add output data */
+ size = outputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ params->dst_iov, 0,
+ &size, NULL, 0);
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ ((uint16_t *)in_buffer)[3] = rte_cpu_to_be_16(i);
+ s_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ size = g_size_bytes + s_size_bytes + SG_LIST_HDR_SIZE;
+
+ /* This is DPTR len incase of SG mode */
+ vq_cmd_w0.s.dlen = size;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* cpt alternate completion address saved earlier */
+ req->alternate_caddr = (uint64_t *)((uint8_t *)c_vaddr - 8);
+ *req->alternate_caddr = ~((uint64_t)COMPLETION_CODE_INIT);
+ rptr_dma = c_dma - 8;
+
+ req->ist.ei1 = dptr_dma;
+ req->ist.ei2 = rptr_dma;
+ }
+
+ /* vq command w3 */
+ vq_cmd_w3.u64 = 0;
+ vq_cmd_w3.s.grp = 0;
+ vq_cmd_w3.s.cptr = params->ctx_buf.dma_addr +
+ offsetof(struct cpt_ctx, zs_ctx);
+
+ /* 16 byte aligned cpt res address */
+ req->completion_addr = (uint64_t *)((uint8_t *)c_vaddr);
+ *req->completion_addr = COMPLETION_CODE_INIT;
+ req->comp_baddr = c_dma;
+
+ /* Fill microcode part of instruction */
+ req->ist.ei0 = vq_cmd_w0.u64;
+ req->ist.ei3 = vq_cmd_w3.u64;
+
+ req->op = op;
+
+ *prep_req = req;
+ return;
+}
+
+static __rte_always_inline void
+cpt_kasumi_enc_prep(uint32_t req_flags,
+ uint64_t d_offs,
+ uint64_t d_lens,
+ fc_params_t *params,
+ void *op,
+ void **prep_req)
+{
+ uint32_t size;
+ int32_t inputlen = 0, outputlen = 0;
+ struct cpt_ctx *cpt_ctx;
+ uint32_t mac_len = 0;
+ uint8_t i = 0;
+ struct cpt_request_info *req;
+ buf_ptr_t *buf_p;
+ uint32_t encr_offset, auth_offset;
+ uint32_t encr_data_len, auth_data_len;
+ int flags;
+ uint8_t *iv_s, *iv_d, iv_len = 8;
+ uint8_t dir = 0;
+ void *m_vaddr, *c_vaddr;
+ uint64_t m_dma, c_dma;
+ uint64_t *offset_vaddr, offset_dma;
+ vq_cmd_word0_t vq_cmd_w0;
+ vq_cmd_word3_t vq_cmd_w3;
+ opcode_info_t opcode;
+ uint8_t *in_buffer;
+ uint32_t g_size_bytes, s_size_bytes;
+ uint64_t dptr_dma, rptr_dma;
+ sg_comp_t *gather_comp;
+ sg_comp_t *scatter_comp;
+
+ buf_p = ¶ms->meta_buf;
+ m_vaddr = buf_p->vaddr;
+ m_dma = buf_p->dma_addr;
+
+ encr_offset = ENCR_OFFSET(d_offs) / 8;
+ auth_offset = AUTH_OFFSET(d_offs) / 8;
+ encr_data_len = ENCR_DLEN(d_lens);
+ auth_data_len = AUTH_DLEN(d_lens);
+
+ cpt_ctx = params->ctx_buf.vaddr;
+ flags = cpt_ctx->zsk_flags;
+ mac_len = cpt_ctx->mac_len;
+
+ if (flags == 0x0)
+ iv_s = params->iv_buf;
+ else
+ iv_s = params->auth_iv_buf;
+
+ dir = iv_s[8] & 0x1;
+
+ /*
+ * Save initial space that followed app data for completion code &
+ * alternate completion code to fall in same cache line as app data
+ */
+ m_vaddr = (uint8_t *)m_vaddr + COMPLETION_CODE_SIZE;
+ m_dma += COMPLETION_CODE_SIZE;
+ size = (uint8_t *)RTE_PTR_ALIGN((uint8_t *)m_vaddr, 16) -
+ (uint8_t *)m_vaddr;
+
+ c_vaddr = (uint8_t *)m_vaddr + size;
+ c_dma = m_dma + size;
+ size += sizeof(cpt_res_s_t);
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* Reserve memory for cpt request info */
+ req = m_vaddr;
+
+ size = sizeof(struct cpt_request_info);
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ opcode.s.major = CPT_MAJOR_OP_KASUMI | CPT_DMA_MODE;
+
+ /* indicates ECB/CBC, direction, ctx from cptr, iv from dptr */
+ opcode.s.minor = ((1 << 6) | (cpt_ctx->k_ecb << 5) |
+ (dir << 4) | (0 << 3) | (flags & 0x7));
+
+ /*
+ * GP op header, lengths are expected in bits.
+ */
+ vq_cmd_w0.u64 = 0;
+ vq_cmd_w0.s.param1 = encr_data_len;
+ vq_cmd_w0.s.param2 = auth_data_len;
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ /* consider iv len */
+ if (flags == 0x0) {
+ encr_offset += iv_len;
+ auth_offset += iv_len;
+ }
+
+ /* save space for offset ctrl and iv */
+ offset_vaddr = m_vaddr;
+ offset_dma = m_dma;
+
+ m_vaddr = (uint8_t *)m_vaddr + OFF_CTRL_LEN + iv_len;
+ m_dma += OFF_CTRL_LEN + iv_len;
+
+ /* DPTR has SG list */
+ in_buffer = m_vaddr;
+ dptr_dma = m_dma;
+
+ ((uint16_t *)in_buffer)[0] = 0;
+ ((uint16_t *)in_buffer)[1] = 0;
+
+ /* TODO Add error check if space will be sufficient */
+ gather_comp = (sg_comp_t *)((uint8_t *)m_vaddr + 8);
+
+ /*
+ * Input Gather List
+ */
+ i = 0;
+
+ /* Offset control word followed by iv */
+
+ if (flags == 0x0) {
+ inputlen = encr_offset + (RTE_ALIGN(encr_data_len, 8) / 8);
+ outputlen = inputlen;
+ /* iv offset is 0 */
+ *offset_vaddr = rte_cpu_to_be_64((uint64_t)encr_offset << 16);
+ } else {
+ inputlen = auth_offset + (RTE_ALIGN(auth_data_len, 8) / 8);
+ outputlen = mac_len;
+ /* iv offset is 0 */
+ *offset_vaddr = rte_cpu_to_be_64((uint64_t)auth_offset);
+ }
+
+ i = fill_sg_comp(gather_comp, i, offset_dma, OFF_CTRL_LEN + iv_len);
+
+ /* IV */
+ iv_d = (uint8_t *)offset_vaddr + OFF_CTRL_LEN;
+ memcpy(iv_d, iv_s, iv_len);
+
+ /* input data */
+ size = inputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(gather_comp, i,
+ params->src_iov, 0,
+ &size, NULL, 0);
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ ((uint16_t *)in_buffer)[2] = rte_cpu_to_be_16(i);
+ g_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ /*
+ * Output Scatter List
+ */
+
+ i = 0;
+ scatter_comp = (sg_comp_t *)((uint8_t *)gather_comp + g_size_bytes);
+
+ if (flags == 0x1) {
+ /* IV in SLIST only for F8 */
+ iv_len = 0;
+ }
+
+ /* IV */
+ if (iv_len) {
+ i = fill_sg_comp(scatter_comp, i,
+ offset_dma + OFF_CTRL_LEN,
+ iv_len);
+ }
+
+ /* Add output data */
+ if (req_flags & VALID_MAC_BUF) {
+ size = outputlen - iv_len - mac_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ params->dst_iov, 0,
+ &size, NULL, 0);
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+
+ /* mac data */
+ if (mac_len) {
+ i = fill_sg_comp_from_buf(scatter_comp, i,
+ ¶ms->mac_buf);
+ }
+ } else {
+ /* Output including mac */
+ size = outputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ params->dst_iov, 0,
+ &size, NULL, 0);
+
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ }
+ ((uint16_t *)in_buffer)[3] = rte_cpu_to_be_16(i);
+ s_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ size = g_size_bytes + s_size_bytes + SG_LIST_HDR_SIZE;
+
+ /* This is DPTR len incase of SG mode */
+ vq_cmd_w0.s.dlen = size;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* cpt alternate completion address saved earlier */
+ req->alternate_caddr = (uint64_t *)((uint8_t *)c_vaddr - 8);
+ *req->alternate_caddr = ~((uint64_t)COMPLETION_CODE_INIT);
+ rptr_dma = c_dma - 8;
+
+ req->ist.ei1 = dptr_dma;
+ req->ist.ei2 = rptr_dma;
+
+ /* vq command w3 */
+ vq_cmd_w3.u64 = 0;
+ vq_cmd_w3.s.grp = 0;
+ vq_cmd_w3.s.cptr = params->ctx_buf.dma_addr +
+ offsetof(struct cpt_ctx, k_ctx);
+
+ /* 16 byte aligned cpt res address */
+ req->completion_addr = (uint64_t *)((uint8_t *)c_vaddr);
+ *req->completion_addr = COMPLETION_CODE_INIT;
+ req->comp_baddr = c_dma;
+
+ /* Fill microcode part of instruction */
+ req->ist.ei0 = vq_cmd_w0.u64;
+ req->ist.ei3 = vq_cmd_w3.u64;
+
+ req->op = op;
+
+ *prep_req = req;
+ return;
+}
+
+static __rte_always_inline void
+cpt_kasumi_dec_prep(uint64_t d_offs,
+ uint64_t d_lens,
+ fc_params_t *params,
+ void *op,
+ void **prep_req)
+{
+ uint32_t size;
+ int32_t inputlen = 0, outputlen;
+ struct cpt_ctx *cpt_ctx;
+ uint8_t i = 0, iv_len = 8;
+ struct cpt_request_info *req;
+ buf_ptr_t *buf_p;
+ uint32_t encr_offset;
+ uint32_t encr_data_len;
+ int flags;
+ uint8_t dir = 0;
+ void *m_vaddr, *c_vaddr;
+ uint64_t m_dma, c_dma;
+ uint64_t *offset_vaddr, offset_dma;
+ vq_cmd_word0_t vq_cmd_w0;
+ vq_cmd_word3_t vq_cmd_w3;
+ opcode_info_t opcode;
+ uint8_t *in_buffer;
+ uint32_t g_size_bytes, s_size_bytes;
+ uint64_t dptr_dma, rptr_dma;
+ sg_comp_t *gather_comp;
+ sg_comp_t *scatter_comp;
+
+ buf_p = ¶ms->meta_buf;
+ m_vaddr = buf_p->vaddr;
+ m_dma = buf_p->dma_addr;
+
+ encr_offset = ENCR_OFFSET(d_offs) / 8;
+ encr_data_len = ENCR_DLEN(d_lens);
+
+ cpt_ctx = params->ctx_buf.vaddr;
+ flags = cpt_ctx->zsk_flags;
+ /*
+ * Save initial space that followed app data for completion code &
+ * alternate completion code to fall in same cache line as app data
+ */
+ m_vaddr = (uint8_t *)m_vaddr + COMPLETION_CODE_SIZE;
+ m_dma += COMPLETION_CODE_SIZE;
+ size = (uint8_t *)RTE_PTR_ALIGN((uint8_t *)m_vaddr, 16) -
+ (uint8_t *)m_vaddr;
+
+ c_vaddr = (uint8_t *)m_vaddr + size;
+ c_dma = m_dma + size;
+ size += sizeof(cpt_res_s_t);
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* Reserve memory for cpt request info */
+ req = m_vaddr;
+
+ size = sizeof(struct cpt_request_info);
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ opcode.s.major = CPT_MAJOR_OP_KASUMI | CPT_DMA_MODE;
+
+ /* indicates ECB/CBC, direction, ctx from cptr, iv from dptr */
+ opcode.s.minor = ((1 << 6) | (cpt_ctx->k_ecb << 5) |
+ (dir << 4) | (0 << 3) | (flags & 0x7));
+
+ /*
+ * GP op header, lengths are expected in bits.
+ */
+ vq_cmd_w0.u64 = 0;
+ vq_cmd_w0.s.param1 = encr_data_len;
+ vq_cmd_w0.s.opcode = opcode.flags;
+
+ /* consider iv len */
+ encr_offset += iv_len;
+
+ inputlen = iv_len + (RTE_ALIGN(encr_data_len, 8) / 8);
+ outputlen = inputlen;
+
+ /* save space for offset ctrl & iv */
+ offset_vaddr = m_vaddr;
+ offset_dma = m_dma;
+
+ m_vaddr = (uint8_t *)m_vaddr + OFF_CTRL_LEN + iv_len;
+ m_dma += OFF_CTRL_LEN + iv_len;
+
+ /* DPTR has SG list */
+ in_buffer = m_vaddr;
+ dptr_dma = m_dma;
+
+ ((uint16_t *)in_buffer)[0] = 0;
+ ((uint16_t *)in_buffer)[1] = 0;
+
+ /* TODO Add error check if space will be sufficient */
+ gather_comp = (sg_comp_t *)((uint8_t *)m_vaddr + 8);
+
+ /*
+ * Input Gather List
+ */
+ i = 0;
+
+ /* Offset control word followed by iv */
+ *offset_vaddr = rte_cpu_to_be_64((uint64_t)encr_offset << 16);
+
+ i = fill_sg_comp(gather_comp, i, offset_dma, OFF_CTRL_LEN + iv_len);
+
+ /* IV */
+ memcpy((uint8_t *)offset_vaddr + OFF_CTRL_LEN,
+ params->iv_buf, iv_len);
+
+ /* Add input data */
+ size = inputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(gather_comp, i,
+ params->src_iov,
+ 0, &size, NULL, 0);
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ ((uint16_t *)in_buffer)[2] = rte_cpu_to_be_16(i);
+ g_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ /*
+ * Output Scatter List
+ */
+
+ i = 0;
+ scatter_comp = (sg_comp_t *)((uint8_t *)gather_comp + g_size_bytes);
+
+ /* IV */
+ i = fill_sg_comp(scatter_comp, i,
+ offset_dma + OFF_CTRL_LEN,
+ iv_len);
+
+ /* Add output data */
+ size = outputlen - iv_len;
+ if (size) {
+ i = fill_sg_comp_from_iov(scatter_comp, i,
+ params->dst_iov, 0,
+ &size, NULL, 0);
+ if (unlikely(size)) {
+ CPT_LOG_DP_ERR("Insufficient buffer space,"
+ " size %d needed", size);
+ return;
+ }
+ }
+ ((uint16_t *)in_buffer)[3] = rte_cpu_to_be_16(i);
+ s_size_bytes = ((i + 3) / 4) * sizeof(sg_comp_t);
+
+ size = g_size_bytes + s_size_bytes + SG_LIST_HDR_SIZE;
+
+ /* This is DPTR len incase of SG mode */
+ vq_cmd_w0.s.dlen = size;
+
+ m_vaddr = (uint8_t *)m_vaddr + size;
+ m_dma += size;
+
+ /* cpt alternate completion address saved earlier */
+ req->alternate_caddr = (uint64_t *)((uint8_t *)c_vaddr - 8);
+ *req->alternate_caddr = ~((uint64_t)COMPLETION_CODE_INIT);
+ rptr_dma = c_dma - 8;
+
+ req->ist.ei1 = dptr_dma;
+ req->ist.ei2 = rptr_dma;
+
+ /* vq command w3 */
+ vq_cmd_w3.u64 = 0;
+ vq_cmd_w3.s.grp = 0;
+ vq_cmd_w3.s.cptr = params->ctx_buf.dma_addr +
+ offsetof(struct cpt_ctx, k_ctx);
+
+ /* 16 byte aligned cpt res address */
+ req->completion_addr = (uint64_t *)((uint8_t *)c_vaddr);
+ *req->completion_addr = COMPLETION_CODE_INIT;
+ req->comp_baddr = c_dma;
+
+ /* Fill microcode part of instruction */
+ req->ist.ei0 = vq_cmd_w0.u64;
+ req->ist.ei3 = vq_cmd_w3.u64;
+
+ req->op = op;
+
+ *prep_req = req;
+ return;
+}
+
+static __rte_always_inline void *
+cpt_fc_dec_hmac_prep(uint32_t flags,
+ uint64_t d_offs,
+ uint64_t d_lens,
+ fc_params_t *fc_params,
+ void *op)