Author | Tokens | Token Proportion | Commits | Commit Proportion |
---|---|---|---|---|
Ratheesh Kannoth | 8002 | 88.04% | 11 | 22.00% |
Sunil Goutham | 493 | 5.42% | 13 | 26.00% |
Subbaraya Sundeep | 132 | 1.45% | 4 | 8.00% |
Linu Cherian | 83 | 0.91% | 3 | 6.00% |
Naveen Mamindlapalli | 82 | 0.90% | 2 | 4.00% |
Suman Ghosh | 69 | 0.76% | 1 | 2.00% |
Zyta Szpak | 59 | 0.65% | 1 | 2.00% |
Geetha Sowjanya | 45 | 0.50% | 1 | 2.00% |
Christina Jacob | 42 | 0.46% | 1 | 2.00% |
Santosh Shukla | 25 | 0.28% | 1 | 2.00% |
Felix Manlunas | 20 | 0.22% | 1 | 2.00% |
Tomasz Duszynski | 9 | 0.10% | 2 | 4.00% |
Sebin Sebastian | 8 | 0.09% | 1 | 2.00% |
Christophe Jaillet | 6 | 0.07% | 4 | 8.00% |
Aleksey Makarov | 6 | 0.07% | 1 | 2.00% |
Rakesh Babu | 5 | 0.06% | 1 | 2.00% |
Li Zetao | 2 | 0.02% | 1 | 2.00% |
SrujanaChalla | 1 | 0.01% | 1 | 2.00% |
Total | 9089 | 50 |
// SPDX-License-Identifier: GPL-2.0 /* Marvell RVU Admin Function driver * * Copyright (C) 2022 Marvell. * */ #include <linux/bitfield.h> #include <linux/module.h> #include <linux/pci.h> #include <linux/firmware.h> #include <linux/stddef.h> #include <linux/debugfs.h> #include "rvu_struct.h" #include "rvu_reg.h" #include "rvu.h" #include "npc.h" #include "cgx.h" #include "rvu_npc_fs.h" #include "rvu_npc_hash.h" static u64 rvu_npc_wide_extract(const u64 input[], size_t start_bit, size_t width_bits) { const u64 mask = ~(u64)((~(__uint128_t)0) << width_bits); const size_t msb = start_bit + width_bits - 1; const size_t lword = start_bit >> 6; const size_t uword = msb >> 6; size_t lbits; u64 hi, lo; if (lword == uword) return (input[lword] >> (start_bit & 63)) & mask; lbits = 64 - (start_bit & 63); hi = input[uword]; lo = (input[lword] >> (start_bit & 63)); return ((hi << lbits) | lo) & mask; } static void rvu_npc_lshift_key(u64 *key, size_t key_bit_len) { u64 prev_orig_word = 0; u64 cur_orig_word = 0; size_t extra = key_bit_len % 64; size_t max_idx = key_bit_len / 64; size_t i; if (extra) max_idx++; for (i = 0; i < max_idx; i++) { cur_orig_word = key[i]; key[i] = key[i] << 1; key[i] |= ((prev_orig_word >> 63) & 0x1); prev_orig_word = cur_orig_word; } } static u32 rvu_npc_toeplitz_hash(const u64 *data, u64 *key, size_t data_bit_len, size_t key_bit_len) { u32 hash_out = 0; u64 temp_data = 0; int i; for (i = data_bit_len - 1; i >= 0; i--) { temp_data = (data[i / 64]); temp_data = temp_data >> (i % 64); temp_data &= 0x1; if (temp_data) hash_out ^= (u32)(rvu_npc_wide_extract(key, key_bit_len - 32, 32)); rvu_npc_lshift_key(key, key_bit_len); } return hash_out; } u32 npc_field_hash_calc(u64 *ldata, struct npc_get_field_hash_info_rsp rsp, u8 intf, u8 hash_idx) { u64 hash_key[3]; u64 data_padded[2]; u32 field_hash; hash_key[0] = rsp.secret_key[1] << 31; hash_key[0] |= rsp.secret_key[2]; hash_key[1] = rsp.secret_key[1] >> 33; hash_key[1] |= rsp.secret_key[0] << 31; hash_key[2] = rsp.secret_key[0] >> 33; data_padded[0] = rsp.hash_mask[intf][hash_idx][0] & ldata[0]; data_padded[1] = rsp.hash_mask[intf][hash_idx][1] & ldata[1]; field_hash = rvu_npc_toeplitz_hash(data_padded, hash_key, 128, 159); field_hash &= FIELD_GET(GENMASK(63, 32), rsp.hash_ctrl[intf][hash_idx]); field_hash += FIELD_GET(GENMASK(31, 0), rsp.hash_ctrl[intf][hash_idx]); return field_hash; } static u64 npc_update_use_hash(struct rvu *rvu, int blkaddr, u8 intf, int lid, int lt, int ld) { u8 hdr, key; u64 cfg; cfg = rvu_read64(rvu, blkaddr, NPC_AF_INTFX_LIDX_LTX_LDX_CFG(intf, lid, lt, ld)); hdr = FIELD_GET(NPC_HDR_OFFSET, cfg); key = FIELD_GET(NPC_KEY_OFFSET, cfg); /* Update use_hash(bit-20) to 'true' and * bytesm1(bit-16:19) to '0x3' in KEX_LD_CFG */ cfg = KEX_LD_CFG_USE_HASH(0x1, 0x03, hdr, 0x1, 0x0, key); return cfg; } static void npc_program_mkex_hash_rx(struct rvu *rvu, int blkaddr, u8 intf) { struct npc_mcam_kex_hash *mkex_hash = rvu->kpu.mkex_hash; int lid, lt, ld, hash_cnt = 0; if (is_npc_intf_tx(intf)) return; /* Program HASH_CFG */ for (lid = 0; lid < NPC_MAX_LID; lid++) { for (lt = 0; lt < NPC_MAX_LT; lt++) { for (ld = 0; ld < NPC_MAX_LD; ld++) { if (mkex_hash->lid_lt_ld_hash_en[intf][lid][lt][ld]) { u64 cfg; if (hash_cnt == NPC_MAX_HASH) return; cfg = npc_update_use_hash(rvu, blkaddr, intf, lid, lt, ld); /* Set updated KEX configuration */ SET_KEX_LD(intf, lid, lt, ld, cfg); /* Set HASH configuration */ SET_KEX_LD_HASH(intf, ld, mkex_hash->hash[intf][ld]); SET_KEX_LD_HASH_MASK(intf, ld, 0, mkex_hash->hash_mask[intf][ld][0]); SET_KEX_LD_HASH_MASK(intf, ld, 1, mkex_hash->hash_mask[intf][ld][1]); SET_KEX_LD_HASH_CTRL(intf, ld, mkex_hash->hash_ctrl[intf][ld]); hash_cnt++; } } } } } static void npc_program_mkex_hash_tx(struct rvu *rvu, int blkaddr, u8 intf) { struct npc_mcam_kex_hash *mkex_hash = rvu->kpu.mkex_hash; int lid, lt, ld, hash_cnt = 0; if (is_npc_intf_rx(intf)) return; /* Program HASH_CFG */ for (lid = 0; lid < NPC_MAX_LID; lid++) { for (lt = 0; lt < NPC_MAX_LT; lt++) { for (ld = 0; ld < NPC_MAX_LD; ld++) if (mkex_hash->lid_lt_ld_hash_en[intf][lid][lt][ld]) { u64 cfg; if (hash_cnt == NPC_MAX_HASH) return; cfg = npc_update_use_hash(rvu, blkaddr, intf, lid, lt, ld); /* Set updated KEX configuration */ SET_KEX_LD(intf, lid, lt, ld, cfg); /* Set HASH configuration */ SET_KEX_LD_HASH(intf, ld, mkex_hash->hash[intf][ld]); SET_KEX_LD_HASH_MASK(intf, ld, 0, mkex_hash->hash_mask[intf][ld][0]); SET_KEX_LD_HASH_MASK(intf, ld, 1, mkex_hash->hash_mask[intf][ld][1]); SET_KEX_LD_HASH_CTRL(intf, ld, mkex_hash->hash_ctrl[intf][ld]); hash_cnt++; } } } } void npc_config_secret_key(struct rvu *rvu, int blkaddr) { struct hw_cap *hwcap = &rvu->hw->cap; struct rvu_hwinfo *hw = rvu->hw; u8 intf; if (!hwcap->npc_hash_extract) return; for (intf = 0; intf < hw->npc_intfs; intf++) { rvu_write64(rvu, blkaddr, NPC_AF_INTFX_SECRET_KEY0(intf), RVU_NPC_HASH_SECRET_KEY0); rvu_write64(rvu, blkaddr, NPC_AF_INTFX_SECRET_KEY1(intf), RVU_NPC_HASH_SECRET_KEY1); rvu_write64(rvu, blkaddr, NPC_AF_INTFX_SECRET_KEY2(intf), RVU_NPC_HASH_SECRET_KEY2); } } void npc_program_mkex_hash(struct rvu *rvu, int blkaddr) { struct npc_mcam_kex_hash *mh = rvu->kpu.mkex_hash; struct hw_cap *hwcap = &rvu->hw->cap; u8 intf, ld, hdr_offset, byte_len; struct rvu_hwinfo *hw = rvu->hw; u64 cfg; /* Check if hardware supports hash extraction */ if (!hwcap->npc_hash_extract) return; /* Check if IPv6 source/destination address * should be hash enabled. * Hashing reduces 128bit SIP/DIP fields to 32bit * so that 224 bit X2 key can be used for IPv6 based filters as well, * which in turn results in more number of MCAM entries available for * use. * * Hashing of IPV6 SIP/DIP is enabled in below scenarios * 1. If the silicon variant supports hashing feature * 2. If the number of bytes of IP addr being extracted is 4 bytes ie * 32bit. The assumption here is that if user wants 8bytes of LSB of * IP addr or full 16 bytes then his intention is not to use 32bit * hash. */ for (intf = 0; intf < hw->npc_intfs; intf++) { for (ld = 0; ld < NPC_MAX_LD; ld++) { cfg = rvu_read64(rvu, blkaddr, NPC_AF_INTFX_LIDX_LTX_LDX_CFG(intf, NPC_LID_LC, NPC_LT_LC_IP6, ld)); hdr_offset = FIELD_GET(NPC_HDR_OFFSET, cfg); byte_len = FIELD_GET(NPC_BYTESM, cfg); /* Hashing of IPv6 source/destination address should be * enabled if, * hdr_offset == 8 (offset of source IPv6 address) or * hdr_offset == 24 (offset of destination IPv6) * address) and the number of byte to be * extracted is 4. As per hardware configuration * byte_len should be == actual byte_len - 1. * Hence byte_len is checked against 3 but nor 4. */ if ((hdr_offset == 8 || hdr_offset == 24) && byte_len == 3) mh->lid_lt_ld_hash_en[intf][NPC_LID_LC][NPC_LT_LC_IP6][ld] = true; } } /* Update hash configuration if the field is hash enabled */ for (intf = 0; intf < hw->npc_intfs; intf++) { npc_program_mkex_hash_rx(rvu, blkaddr, intf); npc_program_mkex_hash_tx(rvu, blkaddr, intf); } } void npc_update_field_hash(struct rvu *rvu, u8 intf, struct mcam_entry *entry, int blkaddr, u64 features, struct flow_msg *pkt, struct flow_msg *mask, struct flow_msg *opkt, struct flow_msg *omask) { struct npc_mcam_kex_hash *mkex_hash = rvu->kpu.mkex_hash; struct npc_get_field_hash_info_req req; struct npc_get_field_hash_info_rsp rsp; u64 ldata[2], cfg; u32 field_hash; u8 hash_idx; if (!rvu->hw->cap.npc_hash_extract) { dev_dbg(rvu->dev, "%s: Field hash extract feature is not supported\n", __func__); return; } req.intf = intf; rvu_mbox_handler_npc_get_field_hash_info(rvu, &req, &rsp); for (hash_idx = 0; hash_idx < NPC_MAX_HASH; hash_idx++) { cfg = rvu_read64(rvu, blkaddr, NPC_AF_INTFX_HASHX_CFG(intf, hash_idx)); if ((cfg & BIT_ULL(11)) && (cfg & BIT_ULL(12))) { u8 lid = (cfg & GENMASK_ULL(10, 8)) >> 8; u8 ltype = (cfg & GENMASK_ULL(7, 4)) >> 4; u8 ltype_mask = cfg & GENMASK_ULL(3, 0); if (mkex_hash->lid_lt_ld_hash_en[intf][lid][ltype][hash_idx]) { switch (ltype & ltype_mask) { /* If hash extract enabled is supported for IPv6 then * 128 bit IPv6 source and destination addressed * is hashed to 32 bit value. */ case NPC_LT_LC_IP6: /* ld[0] == hash_idx[0] == Source IPv6 * ld[1] == hash_idx[1] == Destination IPv6 */ if ((features & BIT_ULL(NPC_SIP_IPV6)) && !hash_idx) { u32 src_ip[IPV6_WORDS]; be32_to_cpu_array(src_ip, pkt->ip6src, IPV6_WORDS); ldata[1] = (u64)src_ip[0] << 32 | src_ip[1]; ldata[0] = (u64)src_ip[2] << 32 | src_ip[3]; field_hash = npc_field_hash_calc(ldata, rsp, intf, hash_idx); npc_update_entry(rvu, NPC_SIP_IPV6, entry, field_hash, 0, GENMASK(31, 0), 0, intf); memcpy(&opkt->ip6src, &pkt->ip6src, sizeof(pkt->ip6src)); memcpy(&omask->ip6src, &mask->ip6src, sizeof(mask->ip6src)); } else if ((features & BIT_ULL(NPC_DIP_IPV6)) && hash_idx) { u32 dst_ip[IPV6_WORDS]; be32_to_cpu_array(dst_ip, pkt->ip6dst, IPV6_WORDS); ldata[1] = (u64)dst_ip[0] << 32 | dst_ip[1]; ldata[0] = (u64)dst_ip[2] << 32 | dst_ip[3]; field_hash = npc_field_hash_calc(ldata, rsp, intf, hash_idx); npc_update_entry(rvu, NPC_DIP_IPV6, entry, field_hash, 0, GENMASK(31, 0), 0, intf); memcpy(&opkt->ip6dst, &pkt->ip6dst, sizeof(pkt->ip6dst)); memcpy(&omask->ip6dst, &mask->ip6dst, sizeof(mask->ip6dst)); } break; } } } } } int rvu_mbox_handler_npc_get_field_hash_info(struct rvu *rvu, struct npc_get_field_hash_info_req *req, struct npc_get_field_hash_info_rsp *rsp) { u64 *secret_key = rsp->secret_key; u8 intf = req->intf; int i, j, blkaddr; blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); if (blkaddr < 0) { dev_err(rvu->dev, "%s: NPC block not implemented\n", __func__); return -EINVAL; } secret_key[0] = rvu_read64(rvu, blkaddr, NPC_AF_INTFX_SECRET_KEY0(intf)); secret_key[1] = rvu_read64(rvu, blkaddr, NPC_AF_INTFX_SECRET_KEY1(intf)); secret_key[2] = rvu_read64(rvu, blkaddr, NPC_AF_INTFX_SECRET_KEY2(intf)); for (i = 0; i < NPC_MAX_HASH; i++) { for (j = 0; j < NPC_MAX_HASH_MASK; j++) { rsp->hash_mask[NIX_INTF_RX][i][j] = GET_KEX_LD_HASH_MASK(NIX_INTF_RX, i, j); rsp->hash_mask[NIX_INTF_TX][i][j] = GET_KEX_LD_HASH_MASK(NIX_INTF_TX, i, j); } } for (i = 0; i < NPC_MAX_INTF; i++) for (j = 0; j < NPC_MAX_HASH; j++) rsp->hash_ctrl[i][j] = GET_KEX_LD_HASH_CTRL(i, j); return 0; } /** * rvu_exact_prepare_mdata - Make mdata for mcam entry * @mac: MAC address * @chan: Channel number. * @ctype: Channel Type. * @mask: LDATA mask. * Return: Meta data */ static u64 rvu_exact_prepare_mdata(u8 *mac, u16 chan, u16 ctype, u64 mask) { u64 ldata = ether_addr_to_u64(mac); /* Please note that mask is 48bit which excludes chan and ctype. * Increase mask bits if we need to include them as well. */ ldata |= ((u64)chan << 48); ldata |= ((u64)ctype << 60); ldata &= mask; ldata = ldata << 2; return ldata; } /** * rvu_exact_calculate_hash - calculate hash index to mem table. * @rvu: resource virtualization unit. * @chan: Channel number * @ctype: Channel type. * @mac: MAC address * @mask: HASH mask. * @table_depth: Depth of table. * Return: Hash value */ static u32 rvu_exact_calculate_hash(struct rvu *rvu, u16 chan, u16 ctype, u8 *mac, u64 mask, u32 table_depth) { struct npc_exact_table *table = rvu->hw->table; u64 hash_key[2]; u64 key_in[2]; u64 ldata; u32 hash; key_in[0] = RVU_NPC_HASH_SECRET_KEY0; key_in[1] = RVU_NPC_HASH_SECRET_KEY2; hash_key[0] = key_in[0] << 31; hash_key[0] |= key_in[1]; hash_key[1] = key_in[0] >> 33; ldata = rvu_exact_prepare_mdata(mac, chan, ctype, mask); dev_dbg(rvu->dev, "%s: ldata=0x%llx hash_key0=0x%llx hash_key2=0x%llx\n", __func__, ldata, hash_key[1], hash_key[0]); hash = rvu_npc_toeplitz_hash(&ldata, (u64 *)hash_key, 64, 95); hash &= table->mem_table.hash_mask; hash += table->mem_table.hash_offset; dev_dbg(rvu->dev, "%s: hash=%x\n", __func__, hash); return hash; } /** * rvu_npc_exact_alloc_mem_table_entry - find free entry in 4 way table. * @rvu: resource virtualization unit. * @way: Indicate way to table. * @index: Hash index to 4 way table. * @hash: Hash value. * * Searches 4 way table using hash index. Returns 0 on success. * Return: 0 upon success. */ static int rvu_npc_exact_alloc_mem_table_entry(struct rvu *rvu, u8 *way, u32 *index, unsigned int hash) { struct npc_exact_table *table; int depth, i; table = rvu->hw->table; depth = table->mem_table.depth; /* Check all the 4 ways for a free slot. */ mutex_lock(&table->lock); for (i = 0; i < table->mem_table.ways; i++) { if (test_bit(hash + i * depth, table->mem_table.bmap)) continue; set_bit(hash + i * depth, table->mem_table.bmap); mutex_unlock(&table->lock); dev_dbg(rvu->dev, "%s: mem table entry alloc success (way=%d index=%d)\n", __func__, i, hash); *way = i; *index = hash; return 0; } mutex_unlock(&table->lock); dev_dbg(rvu->dev, "%s: No space in 4 way exact way, weight=%u\n", __func__, bitmap_weight(table->mem_table.bmap, table->mem_table.depth)); return -ENOSPC; } /** * rvu_npc_exact_free_id - Free seq id from bitmat. * @rvu: Resource virtualization unit. * @seq_id: Sequence identifier to be freed. */ static void rvu_npc_exact_free_id(struct rvu *rvu, u32 seq_id) { struct npc_exact_table *table; table = rvu->hw->table; mutex_lock(&table->lock); clear_bit(seq_id, table->id_bmap); mutex_unlock(&table->lock); dev_dbg(rvu->dev, "%s: freed id %d\n", __func__, seq_id); } /** * rvu_npc_exact_alloc_id - Alloc seq id from bitmap. * @rvu: Resource virtualization unit. * @seq_id: Sequence identifier. * Return: True or false. */ static bool rvu_npc_exact_alloc_id(struct rvu *rvu, u32 *seq_id) { struct npc_exact_table *table; u32 idx; table = rvu->hw->table; mutex_lock(&table->lock); idx = find_first_zero_bit(table->id_bmap, table->tot_ids); if (idx == table->tot_ids) { mutex_unlock(&table->lock); dev_err(rvu->dev, "%s: No space in id bitmap (%d)\n", __func__, table->tot_ids); return false; } /* Mark bit map to indicate that slot is used.*/ set_bit(idx, table->id_bmap); mutex_unlock(&table->lock); *seq_id = idx; dev_dbg(rvu->dev, "%s: Allocated id (%d)\n", __func__, *seq_id); return true; } /** * rvu_npc_exact_alloc_cam_table_entry - find free slot in fully associative table. * @rvu: resource virtualization unit. * @index: Index to exact CAM table. * Return: 0 upon success; else error number. */ static int rvu_npc_exact_alloc_cam_table_entry(struct rvu *rvu, int *index) { struct npc_exact_table *table; u32 idx; table = rvu->hw->table; mutex_lock(&table->lock); idx = find_first_zero_bit(table->cam_table.bmap, table->cam_table.depth); if (idx == table->cam_table.depth) { mutex_unlock(&table->lock); dev_info(rvu->dev, "%s: No space in exact cam table, weight=%u\n", __func__, bitmap_weight(table->cam_table.bmap, table->cam_table.depth)); return -ENOSPC; } /* Mark bit map to indicate that slot is used.*/ set_bit(idx, table->cam_table.bmap); mutex_unlock(&table->lock); *index = idx; dev_dbg(rvu->dev, "%s: cam table entry alloc success (index=%d)\n", __func__, idx); return 0; } /** * rvu_exact_prepare_table_entry - Data for exact match table entry. * @rvu: Resource virtualization unit. * @enable: Enable/Disable entry * @ctype: Software defined channel type. Currently set as 0. * @chan: Channel number. * @mac_addr: Destination mac address. * Return: mdata for exact match table. */ static u64 rvu_exact_prepare_table_entry(struct rvu *rvu, bool enable, u8 ctype, u16 chan, u8 *mac_addr) { u64 ldata = ether_addr_to_u64(mac_addr); /* Enable or disable */ u64 mdata = FIELD_PREP(GENMASK_ULL(63, 63), enable ? 1 : 0); /* Set Ctype */ mdata |= FIELD_PREP(GENMASK_ULL(61, 60), ctype); /* Set chan */ mdata |= FIELD_PREP(GENMASK_ULL(59, 48), chan); /* MAC address */ mdata |= FIELD_PREP(GENMASK_ULL(47, 0), ldata); return mdata; } /** * rvu_exact_config_secret_key - Configure secret key. * @rvu: Resource virtualization unit. */ static void rvu_exact_config_secret_key(struct rvu *rvu) { int blkaddr; blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); rvu_write64(rvu, blkaddr, NPC_AF_INTFX_EXACT_SECRET0(NIX_INTF_RX), RVU_NPC_HASH_SECRET_KEY0); rvu_write64(rvu, blkaddr, NPC_AF_INTFX_EXACT_SECRET1(NIX_INTF_RX), RVU_NPC_HASH_SECRET_KEY1); rvu_write64(rvu, blkaddr, NPC_AF_INTFX_EXACT_SECRET2(NIX_INTF_RX), RVU_NPC_HASH_SECRET_KEY2); } /** * rvu_exact_config_search_key - Configure search key * @rvu: Resource virtualization unit. */ static void rvu_exact_config_search_key(struct rvu *rvu) { int blkaddr; u64 reg_val; blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); /* HDR offset */ reg_val = FIELD_PREP(GENMASK_ULL(39, 32), 0); /* BYTESM1, number of bytes - 1 */ reg_val |= FIELD_PREP(GENMASK_ULL(18, 16), ETH_ALEN - 1); /* Enable LID and set LID to NPC_LID_LA */ reg_val |= FIELD_PREP(GENMASK_ULL(11, 11), 1); reg_val |= FIELD_PREP(GENMASK_ULL(10, 8), NPC_LID_LA); /* Clear layer type based extraction */ /* Disable LT_EN */ reg_val |= FIELD_PREP(GENMASK_ULL(12, 12), 0); /* Set LTYPE_MATCH to 0 */ reg_val |= FIELD_PREP(GENMASK_ULL(7, 4), 0); /* Set LTYPE_MASK to 0 */ reg_val |= FIELD_PREP(GENMASK_ULL(3, 0), 0); rvu_write64(rvu, blkaddr, NPC_AF_INTFX_EXACT_CFG(NIX_INTF_RX), reg_val); } /** * rvu_exact_config_result_ctrl - Set exact table hash control * @rvu: Resource virtualization unit. * @depth: Depth of Exact match table. * * Sets mask and offset for hash for mem table. */ static void rvu_exact_config_result_ctrl(struct rvu *rvu, uint32_t depth) { int blkaddr; u64 reg = 0; blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); /* Set mask. Note that depth is a power of 2 */ rvu->hw->table->mem_table.hash_mask = (depth - 1); reg |= FIELD_PREP(GENMASK_ULL(42, 32), (depth - 1)); /* Set offset as 0 */ rvu->hw->table->mem_table.hash_offset = 0; reg |= FIELD_PREP(GENMASK_ULL(10, 0), 0); /* Set reg for RX */ rvu_write64(rvu, blkaddr, NPC_AF_INTFX_EXACT_RESULT_CTL(NIX_INTF_RX), reg); /* Store hash mask and offset for s/w algorithm */ } /** * rvu_exact_config_table_mask - Set exact table mask. * @rvu: Resource virtualization unit. */ static void rvu_exact_config_table_mask(struct rvu *rvu) { int blkaddr; u64 mask = 0; blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); /* Don't use Ctype */ mask |= FIELD_PREP(GENMASK_ULL(61, 60), 0); /* Set chan */ mask |= GENMASK_ULL(59, 48); /* Full ldata */ mask |= GENMASK_ULL(47, 0); /* Store mask for s/w hash calcualtion */ rvu->hw->table->mem_table.mask = mask; /* Set mask for RX.*/ rvu_write64(rvu, blkaddr, NPC_AF_INTFX_EXACT_MASK(NIX_INTF_RX), mask); } /** * rvu_npc_exact_get_max_entries - Get total number of entries in table. * @rvu: resource virtualization unit. * Return: Maximum table entries possible. */ u32 rvu_npc_exact_get_max_entries(struct rvu *rvu) { struct npc_exact_table *table; table = rvu->hw->table; return table->tot_ids; } /** * rvu_npc_exact_has_match_table - Checks support for exact match. * @rvu: resource virtualization unit. * Return: True if exact match table is supported/enabled. */ bool rvu_npc_exact_has_match_table(struct rvu *rvu) { return rvu->hw->cap.npc_exact_match_enabled; } /** * __rvu_npc_exact_find_entry_by_seq_id - find entry by id * @rvu: resource virtualization unit. * @seq_id: Sequence identifier. * * Caller should acquire the lock. * Return: Pointer to table entry. */ static struct npc_exact_table_entry * __rvu_npc_exact_find_entry_by_seq_id(struct rvu *rvu, u32 seq_id) { struct npc_exact_table *table = rvu->hw->table; struct npc_exact_table_entry *entry = NULL; struct list_head *lhead; lhead = &table->lhead_gbl; /* traverse to find the matching entry */ list_for_each_entry(entry, lhead, glist) { if (entry->seq_id != seq_id) continue; return entry; } return NULL; } /** * rvu_npc_exact_add_to_list - Add entry to list * @rvu: resource virtualization unit. * @opc_type: OPCODE to select MEM/CAM table. * @ways: MEM table ways. * @index: Index in MEM/CAM table. * @cgx_id: CGX identifier. * @lmac_id: LMAC identifier. * @mac_addr: MAC address. * @chan: Channel number. * @ctype: Channel Type. * @seq_id: Sequence identifier * @cmd: True if function is called by ethtool cmd * @mcam_idx: NPC mcam index of DMAC entry in NPC mcam. * @pcifunc: pci function * Return: 0 upon success. */ static int rvu_npc_exact_add_to_list(struct rvu *rvu, enum npc_exact_opc_type opc_type, u8 ways, u32 index, u8 cgx_id, u8 lmac_id, u8 *mac_addr, u16 chan, u8 ctype, u32 *seq_id, bool cmd, u32 mcam_idx, u16 pcifunc) { struct npc_exact_table_entry *entry, *tmp, *iter; struct npc_exact_table *table = rvu->hw->table; struct list_head *lhead, *pprev; WARN_ON(ways >= NPC_EXACT_TBL_MAX_WAYS); if (!rvu_npc_exact_alloc_id(rvu, seq_id)) { dev_err(rvu->dev, "%s: Generate seq id failed\n", __func__); return -EFAULT; } entry = kmalloc(sizeof(*entry), GFP_KERNEL); if (!entry) { rvu_npc_exact_free_id(rvu, *seq_id); dev_err(rvu->dev, "%s: Memory allocation failed\n", __func__); return -ENOMEM; } mutex_lock(&table->lock); switch (opc_type) { case NPC_EXACT_OPC_CAM: lhead = &table->lhead_cam_tbl_entry; table->cam_tbl_entry_cnt++; break; case NPC_EXACT_OPC_MEM: lhead = &table->lhead_mem_tbl_entry[ways]; table->mem_tbl_entry_cnt++; break; default: mutex_unlock(&table->lock); kfree(entry); rvu_npc_exact_free_id(rvu, *seq_id); dev_err(rvu->dev, "%s: Unknown opc type%d\n", __func__, opc_type); return -EINVAL; } /* Add to global list */ INIT_LIST_HEAD(&entry->glist); list_add_tail(&entry->glist, &table->lhead_gbl); INIT_LIST_HEAD(&entry->list); entry->index = index; entry->ways = ways; entry->opc_type = opc_type; entry->pcifunc = pcifunc; ether_addr_copy(entry->mac, mac_addr); entry->chan = chan; entry->ctype = ctype; entry->cgx_id = cgx_id; entry->lmac_id = lmac_id; entry->seq_id = *seq_id; entry->mcam_idx = mcam_idx; entry->cmd = cmd; pprev = lhead; /* Insert entry in ascending order of index */ list_for_each_entry_safe(iter, tmp, lhead, list) { if (index < iter->index) break; pprev = &iter->list; } /* Add to each table list */ list_add(&entry->list, pprev); mutex_unlock(&table->lock); return 0; } /** * rvu_npc_exact_mem_table_write - Wrapper for register write * @rvu: resource virtualization unit. * @blkaddr: Block address * @ways: ways for MEM table. * @index: Index in MEM * @mdata: Meta data to be written to register. */ static void rvu_npc_exact_mem_table_write(struct rvu *rvu, int blkaddr, u8 ways, u32 index, u64 mdata) { rvu_write64(rvu, blkaddr, NPC_AF_EXACT_MEM_ENTRY(ways, index), mdata); } /** * rvu_npc_exact_cam_table_write - Wrapper for register write * @rvu: resource virtualization unit. * @blkaddr: Block address * @index: Index in MEM * @mdata: Meta data to be written to register. */ static void rvu_npc_exact_cam_table_write(struct rvu *rvu, int blkaddr, u32 index, u64 mdata) { rvu_write64(rvu, blkaddr, NPC_AF_EXACT_CAM_ENTRY(index), mdata); } /** * rvu_npc_exact_dealloc_table_entry - dealloc table entry * @rvu: resource virtualization unit. * @opc_type: OPCODE for selection of table(MEM or CAM) * @ways: ways if opc_type is MEM table. * @index: Index of MEM or CAM table. * Return: 0 upon success. */ static int rvu_npc_exact_dealloc_table_entry(struct rvu *rvu, enum npc_exact_opc_type opc_type, u8 ways, u32 index) { int blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); struct npc_exact_table *table; u8 null_dmac[6] = { 0 }; int depth; /* Prepare entry with all fields set to zero */ u64 null_mdata = rvu_exact_prepare_table_entry(rvu, false, 0, 0, null_dmac); table = rvu->hw->table; depth = table->mem_table.depth; mutex_lock(&table->lock); switch (opc_type) { case NPC_EXACT_OPC_CAM: /* Check whether entry is used already */ if (!test_bit(index, table->cam_table.bmap)) { mutex_unlock(&table->lock); dev_err(rvu->dev, "%s: Trying to free an unused entry ways=%d index=%d\n", __func__, ways, index); return -EINVAL; } rvu_npc_exact_cam_table_write(rvu, blkaddr, index, null_mdata); clear_bit(index, table->cam_table.bmap); break; case NPC_EXACT_OPC_MEM: /* Check whether entry is used already */ if (!test_bit(index + ways * depth, table->mem_table.bmap)) { mutex_unlock(&table->lock); dev_err(rvu->dev, "%s: Trying to free an unused entry index=%d\n", __func__, index); return -EINVAL; } rvu_npc_exact_mem_table_write(rvu, blkaddr, ways, index, null_mdata); clear_bit(index + ways * depth, table->mem_table.bmap); break; default: mutex_unlock(&table->lock); dev_err(rvu->dev, "%s: invalid opc type %d", __func__, opc_type); return -ENOSPC; } mutex_unlock(&table->lock); dev_dbg(rvu->dev, "%s: Successfully deleted entry (index=%d, ways=%d opc_type=%d\n", __func__, index, ways, opc_type); return 0; } /** * rvu_npc_exact_alloc_table_entry - Allociate an entry * @rvu: resource virtualization unit. * @mac: MAC address. * @chan: Channel number. * @ctype: Channel Type. * @index: Index of MEM table or CAM table. * @ways: Ways. Only valid for MEM table. * @opc_type: OPCODE to select table (MEM or CAM) * * Try allocating a slot from MEM table. If all 4 ways * slot are full for a hash index, check availability in * 32-entry CAM table for allocation. * Return: 0 upon success. */ static int rvu_npc_exact_alloc_table_entry(struct rvu *rvu, char *mac, u16 chan, u8 ctype, u32 *index, u8 *ways, enum npc_exact_opc_type *opc_type) { struct npc_exact_table *table; unsigned int hash; int err; table = rvu->hw->table; /* Check in 4-ways mem entry for free slote */ hash = rvu_exact_calculate_hash(rvu, chan, ctype, mac, table->mem_table.mask, table->mem_table.depth); err = rvu_npc_exact_alloc_mem_table_entry(rvu, ways, index, hash); if (!err) { *opc_type = NPC_EXACT_OPC_MEM; dev_dbg(rvu->dev, "%s: inserted in 4 ways hash table ways=%d, index=%d\n", __func__, *ways, *index); return 0; } dev_dbg(rvu->dev, "%s: failed to insert in 4 ways hash table\n", __func__); /* wayss is 0 for cam table */ *ways = 0; err = rvu_npc_exact_alloc_cam_table_entry(rvu, index); if (!err) { *opc_type = NPC_EXACT_OPC_CAM; dev_dbg(rvu->dev, "%s: inserted in fully associative hash table index=%u\n", __func__, *index); return 0; } dev_err(rvu->dev, "%s: failed to insert in fully associative hash table\n", __func__); return -ENOSPC; } /** * rvu_npc_exact_save_drop_rule_chan_and_mask - Save drop rules info in data base. * @rvu: resource virtualization unit. * @drop_mcam_idx: Drop rule index in NPC mcam. * @chan_val: Channel value. * @chan_mask: Channel Mask. * @pcifunc: pcifunc of interface. * Return: True upon success. */ static bool rvu_npc_exact_save_drop_rule_chan_and_mask(struct rvu *rvu, int drop_mcam_idx, u64 chan_val, u64 chan_mask, u16 pcifunc) { struct npc_exact_table *table; int i; table = rvu->hw->table; for (i = 0; i < NPC_MCAM_DROP_RULE_MAX; i++) { if (!table->drop_rule_map[i].valid) break; if (table->drop_rule_map[i].chan_val != (u16)chan_val) continue; if (table->drop_rule_map[i].chan_mask != (u16)chan_mask) continue; return false; } if (i == NPC_MCAM_DROP_RULE_MAX) return false; table->drop_rule_map[i].drop_rule_idx = drop_mcam_idx; table->drop_rule_map[i].chan_val = (u16)chan_val; table->drop_rule_map[i].chan_mask = (u16)chan_mask; table->drop_rule_map[i].pcifunc = pcifunc; table->drop_rule_map[i].valid = true; return true; } /** * rvu_npc_exact_calc_drop_rule_chan_and_mask - Calculate Channel number and mask. * @rvu: resource virtualization unit. * @intf_type: Interface type (SDK, LBK or CGX) * @cgx_id: CGX identifier. * @lmac_id: LAMC identifier. * @val: Channel number. * @mask: Channel mask. * Return: True upon success. */ static bool rvu_npc_exact_calc_drop_rule_chan_and_mask(struct rvu *rvu, u8 intf_type, u8 cgx_id, u8 lmac_id, u64 *val, u64 *mask) { u16 chan_val, chan_mask; /* No support for SDP and LBK */ if (intf_type != NIX_INTF_TYPE_CGX) return false; chan_val = rvu_nix_chan_cgx(rvu, cgx_id, lmac_id, 0); chan_mask = 0xfff; if (val) *val = chan_val; if (mask) *mask = chan_mask; return true; } /** * rvu_npc_exact_drop_rule_to_pcifunc - Retrieve pcifunc * @rvu: resource virtualization unit. * @drop_rule_idx: Drop rule index in NPC mcam. * * Debugfs (exact_drop_cnt) entry displays pcifunc for interface * by retrieving the pcifunc value from data base. * Return: Drop rule index. */ u16 rvu_npc_exact_drop_rule_to_pcifunc(struct rvu *rvu, u32 drop_rule_idx) { struct npc_exact_table *table; int i; table = rvu->hw->table; for (i = 0; i < NPC_MCAM_DROP_RULE_MAX; i++) { if (!table->drop_rule_map[i].valid) break; if (table->drop_rule_map[i].drop_rule_idx != drop_rule_idx) continue; return table->drop_rule_map[i].pcifunc; } dev_err(rvu->dev, "%s: drop mcam rule index (%d) >= NPC_MCAM_DROP_RULE_MAX\n", __func__, drop_rule_idx); return -1; } /** * rvu_npc_exact_get_drop_rule_info - Get drop rule information. * @rvu: resource virtualization unit. * @intf_type: Interface type (CGX, SDP or LBK) * @cgx_id: CGX identifier. * @lmac_id: LMAC identifier. * @drop_mcam_idx: NPC mcam drop rule index. * @val: Channel value. * @mask: Channel mask. * @pcifunc: pcifunc of interface corresponding to the drop rule. * Return: True upon success. */ static bool rvu_npc_exact_get_drop_rule_info(struct rvu *rvu, u8 intf_type, u8 cgx_id, u8 lmac_id, u32 *drop_mcam_idx, u64 *val, u64 *mask, u16 *pcifunc) { struct npc_exact_table *table; u64 chan_val, chan_mask; bool rc; int i; table = rvu->hw->table; if (intf_type != NIX_INTF_TYPE_CGX) { dev_err(rvu->dev, "%s: No drop rule for LBK/SDP mode\n", __func__); return false; } rc = rvu_npc_exact_calc_drop_rule_chan_and_mask(rvu, intf_type, cgx_id, lmac_id, &chan_val, &chan_mask); if (!rc) return false; for (i = 0; i < NPC_MCAM_DROP_RULE_MAX; i++) { if (!table->drop_rule_map[i].valid) break; if (table->drop_rule_map[i].chan_val != (u16)chan_val) continue; if (val) *val = table->drop_rule_map[i].chan_val; if (mask) *mask = table->drop_rule_map[i].chan_mask; if (pcifunc) *pcifunc = table->drop_rule_map[i].pcifunc; *drop_mcam_idx = i; return true; } if (i == NPC_MCAM_DROP_RULE_MAX) { dev_err(rvu->dev, "%s: drop mcam rule index (%d) >= NPC_MCAM_DROP_RULE_MAX\n", __func__, *drop_mcam_idx); return false; } dev_err(rvu->dev, "%s: Could not retrieve for cgx=%d, lmac=%d\n", __func__, cgx_id, lmac_id); return false; } /** * __rvu_npc_exact_cmd_rules_cnt_update - Update number dmac rules against a drop rule. * @rvu: resource virtualization unit. * @drop_mcam_idx: NPC mcam drop rule index. * @val: +1 or -1. * @enable_or_disable_cam: If no exact match rules against a drop rule, disable it. * * when first exact match entry against a drop rule is added, enable_or_disable_cam * is set to true. When last exact match entry against a drop rule is deleted, * enable_or_disable_cam is set to true. * Return: Number of rules */ static u16 __rvu_npc_exact_cmd_rules_cnt_update(struct rvu *rvu, int drop_mcam_idx, int val, bool *enable_or_disable_cam) { struct npc_exact_table *table; u16 *cnt, old_cnt; bool promisc; table = rvu->hw->table; promisc = table->promisc_mode[drop_mcam_idx]; cnt = &table->cnt_cmd_rules[drop_mcam_idx]; old_cnt = *cnt; *cnt += val; if (!enable_or_disable_cam) goto done; *enable_or_disable_cam = false; if (promisc) goto done; /* If all rules are deleted and not already in promisc mode; * disable cam */ if (!*cnt && val < 0) { *enable_or_disable_cam = true; goto done; } /* If rule got added and not already in promisc mode; enable cam */ if (!old_cnt && val > 0) { *enable_or_disable_cam = true; goto done; } done: return *cnt; } /** * rvu_npc_exact_del_table_entry_by_id - Delete and free table entry. * @rvu: resource virtualization unit. * @seq_id: Sequence identifier of the entry. * * Deletes entry from linked lists and free up slot in HW MEM or CAM * table. * Return: 0 upon success. */ static int rvu_npc_exact_del_table_entry_by_id(struct rvu *rvu, u32 seq_id) { struct npc_exact_table_entry *entry = NULL; struct npc_exact_table *table; bool disable_cam = false; u32 drop_mcam_idx = -1; int *cnt; bool rc; table = rvu->hw->table; mutex_lock(&table->lock); /* Lookup for entry which needs to be updated */ entry = __rvu_npc_exact_find_entry_by_seq_id(rvu, seq_id); if (!entry) { dev_dbg(rvu->dev, "%s: failed to find entry for id=%d\n", __func__, seq_id); mutex_unlock(&table->lock); return -ENODATA; } cnt = (entry->opc_type == NPC_EXACT_OPC_CAM) ? &table->cam_tbl_entry_cnt : &table->mem_tbl_entry_cnt; /* delete from lists */ list_del_init(&entry->list); list_del_init(&entry->glist); (*cnt)--; rc = rvu_npc_exact_get_drop_rule_info(rvu, NIX_INTF_TYPE_CGX, entry->cgx_id, entry->lmac_id, &drop_mcam_idx, NULL, NULL, NULL); if (!rc) { dev_dbg(rvu->dev, "%s: failed to retrieve drop info for id=0x%x\n", __func__, seq_id); mutex_unlock(&table->lock); return -ENODATA; } if (entry->cmd) __rvu_npc_exact_cmd_rules_cnt_update(rvu, drop_mcam_idx, -1, &disable_cam); /* No dmac filter rules; disable drop on hit rule */ if (disable_cam) { rvu_npc_enable_mcam_by_entry_index(rvu, drop_mcam_idx, NIX_INTF_RX, false); dev_dbg(rvu->dev, "%s: Disabling mcam idx %d\n", __func__, drop_mcam_idx); } mutex_unlock(&table->lock); rvu_npc_exact_dealloc_table_entry(rvu, entry->opc_type, entry->ways, entry->index); rvu_npc_exact_free_id(rvu, seq_id); dev_dbg(rvu->dev, "%s: delete entry success for id=0x%x, mca=%pM\n", __func__, seq_id, entry->mac); kfree(entry); return 0; } /** * rvu_npc_exact_add_table_entry - Adds a table entry * @rvu: resource virtualization unit. * @cgx_id: cgx identifier. * @lmac_id: lmac identifier. * @mac: MAC address. * @chan: Channel number. * @ctype: Channel Type. * @seq_id: Sequence number. * @cmd: Whether it is invoked by ethtool cmd. * @mcam_idx: NPC mcam index corresponding to MAC * @pcifunc: PCI func. * * Creates a new exact match table entry in either CAM or * MEM table. * Return: 0 upon success. */ static int rvu_npc_exact_add_table_entry(struct rvu *rvu, u8 cgx_id, u8 lmac_id, u8 *mac, u16 chan, u8 ctype, u32 *seq_id, bool cmd, u32 mcam_idx, u16 pcifunc) { int blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); enum npc_exact_opc_type opc_type; bool enable_cam = false; u32 drop_mcam_idx; u32 index; u64 mdata; bool rc; int err; u8 ways; ctype = 0; err = rvu_npc_exact_alloc_table_entry(rvu, mac, chan, ctype, &index, &ways, &opc_type); if (err) { dev_err(rvu->dev, "%s: Could not alloc in exact match table\n", __func__); return err; } /* Write mdata to table */ mdata = rvu_exact_prepare_table_entry(rvu, true, ctype, chan, mac); if (opc_type == NPC_EXACT_OPC_CAM) rvu_npc_exact_cam_table_write(rvu, blkaddr, index, mdata); else rvu_npc_exact_mem_table_write(rvu, blkaddr, ways, index, mdata); /* Insert entry to linked list */ err = rvu_npc_exact_add_to_list(rvu, opc_type, ways, index, cgx_id, lmac_id, mac, chan, ctype, seq_id, cmd, mcam_idx, pcifunc); if (err) { rvu_npc_exact_dealloc_table_entry(rvu, opc_type, ways, index); dev_err(rvu->dev, "%s: could not add to exact match table\n", __func__); return err; } rc = rvu_npc_exact_get_drop_rule_info(rvu, NIX_INTF_TYPE_CGX, cgx_id, lmac_id, &drop_mcam_idx, NULL, NULL, NULL); if (!rc) { rvu_npc_exact_dealloc_table_entry(rvu, opc_type, ways, index); dev_dbg(rvu->dev, "%s: failed to get drop rule info cgx=%d lmac=%d\n", __func__, cgx_id, lmac_id); return -EINVAL; } if (cmd) __rvu_npc_exact_cmd_rules_cnt_update(rvu, drop_mcam_idx, 1, &enable_cam); /* First command rule; enable drop on hit rule */ if (enable_cam) { rvu_npc_enable_mcam_by_entry_index(rvu, drop_mcam_idx, NIX_INTF_RX, true); dev_dbg(rvu->dev, "%s: Enabling mcam idx %d\n", __func__, drop_mcam_idx); } dev_dbg(rvu->dev, "%s: Successfully added entry (index=%d, dmac=%pM, ways=%d opc_type=%d\n", __func__, index, mac, ways, opc_type); return 0; } /** * rvu_npc_exact_update_table_entry - Update exact match table. * @rvu: resource virtualization unit. * @cgx_id: CGX identifier. * @lmac_id: LMAC identifier. * @old_mac: Existing MAC address entry. * @new_mac: New MAC address entry. * @seq_id: Sequence identifier of the entry. * * Updates MAC address of an entry. If entry is in MEM table, new * hash value may not match with old one. * Return: 0 upon success. */ static int rvu_npc_exact_update_table_entry(struct rvu *rvu, u8 cgx_id, u8 lmac_id, u8 *old_mac, u8 *new_mac, u32 *seq_id) { int blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); struct npc_exact_table_entry *entry; struct npc_exact_table *table; u32 hash_index; u64 mdata; table = rvu->hw->table; mutex_lock(&table->lock); /* Lookup for entry which needs to be updated */ entry = __rvu_npc_exact_find_entry_by_seq_id(rvu, *seq_id); if (!entry) { mutex_unlock(&table->lock); dev_dbg(rvu->dev, "%s: failed to find entry for cgx_id=%d lmac_id=%d old_mac=%pM\n", __func__, cgx_id, lmac_id, old_mac); return -ENODATA; } /* If entry is in mem table and new hash index is different than old * hash index, we cannot update the entry. Fail in these scenarios. */ if (entry->opc_type == NPC_EXACT_OPC_MEM) { hash_index = rvu_exact_calculate_hash(rvu, entry->chan, entry->ctype, new_mac, table->mem_table.mask, table->mem_table.depth); if (hash_index != entry->index) { dev_dbg(rvu->dev, "%s: Update failed due to index mismatch(new=0x%x, old=%x)\n", __func__, hash_index, entry->index); mutex_unlock(&table->lock); return -EINVAL; } } mdata = rvu_exact_prepare_table_entry(rvu, true, entry->ctype, entry->chan, new_mac); if (entry->opc_type == NPC_EXACT_OPC_MEM) rvu_npc_exact_mem_table_write(rvu, blkaddr, entry->ways, entry->index, mdata); else rvu_npc_exact_cam_table_write(rvu, blkaddr, entry->index, mdata); /* Update entry fields */ ether_addr_copy(entry->mac, new_mac); *seq_id = entry->seq_id; dev_dbg(rvu->dev, "%s: Successfully updated entry (index=%d, dmac=%pM, ways=%d opc_type=%d\n", __func__, entry->index, entry->mac, entry->ways, entry->opc_type); dev_dbg(rvu->dev, "%s: Successfully updated entry (old mac=%pM new_mac=%pM\n", __func__, old_mac, new_mac); mutex_unlock(&table->lock); return 0; } /** * rvu_npc_exact_promisc_disable - Disable promiscuous mode. * @rvu: resource virtualization unit. * @pcifunc: pcifunc * * Drop rule is against each PF. We dont support DMAC filter for * VF. * Return: 0 upon success */ int rvu_npc_exact_promisc_disable(struct rvu *rvu, u16 pcifunc) { struct npc_exact_table *table; int pf = rvu_get_pf(pcifunc); u8 cgx_id, lmac_id; u32 drop_mcam_idx; bool *promisc; bool rc; table = rvu->hw->table; rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id); rc = rvu_npc_exact_get_drop_rule_info(rvu, NIX_INTF_TYPE_CGX, cgx_id, lmac_id, &drop_mcam_idx, NULL, NULL, NULL); if (!rc) { dev_dbg(rvu->dev, "%s: failed to get drop rule info cgx=%d lmac=%d\n", __func__, cgx_id, lmac_id); return -EINVAL; } mutex_lock(&table->lock); promisc = &table->promisc_mode[drop_mcam_idx]; if (!*promisc) { mutex_unlock(&table->lock); dev_dbg(rvu->dev, "%s: Err Already promisc mode disabled (cgx=%d lmac=%d)\n", __func__, cgx_id, lmac_id); return LMAC_AF_ERR_INVALID_PARAM; } *promisc = false; mutex_unlock(&table->lock); /* Enable drop rule */ rvu_npc_enable_mcam_by_entry_index(rvu, drop_mcam_idx, NIX_INTF_RX, true); dev_dbg(rvu->dev, "%s: disabled promisc mode (cgx=%d lmac=%d)\n", __func__, cgx_id, lmac_id); return 0; } /** * rvu_npc_exact_promisc_enable - Enable promiscuous mode. * @rvu: resource virtualization unit. * @pcifunc: pcifunc. * Return: 0 upon success */ int rvu_npc_exact_promisc_enable(struct rvu *rvu, u16 pcifunc) { struct npc_exact_table *table; int pf = rvu_get_pf(pcifunc); u8 cgx_id, lmac_id; u32 drop_mcam_idx; bool *promisc; bool rc; table = rvu->hw->table; rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id); rc = rvu_npc_exact_get_drop_rule_info(rvu, NIX_INTF_TYPE_CGX, cgx_id, lmac_id, &drop_mcam_idx, NULL, NULL, NULL); if (!rc) { dev_dbg(rvu->dev, "%s: failed to get drop rule info cgx=%d lmac=%d\n", __func__, cgx_id, lmac_id); return -EINVAL; } mutex_lock(&table->lock); promisc = &table->promisc_mode[drop_mcam_idx]; if (*promisc) { mutex_unlock(&table->lock); dev_dbg(rvu->dev, "%s: Already in promisc mode (cgx=%d lmac=%d)\n", __func__, cgx_id, lmac_id); return LMAC_AF_ERR_INVALID_PARAM; } *promisc = true; mutex_unlock(&table->lock); /* disable drop rule */ rvu_npc_enable_mcam_by_entry_index(rvu, drop_mcam_idx, NIX_INTF_RX, false); dev_dbg(rvu->dev, "%s: Enabled promisc mode (cgx=%d lmac=%d)\n", __func__, cgx_id, lmac_id); return 0; } /** * rvu_npc_exact_mac_addr_reset - Delete PF mac address. * @rvu: resource virtualization unit. * @req: Reset request * @rsp: Reset response. * Return: 0 upon success */ int rvu_npc_exact_mac_addr_reset(struct rvu *rvu, struct cgx_mac_addr_reset_req *req, struct msg_rsp *rsp) { int pf = rvu_get_pf(req->hdr.pcifunc); u32 seq_id = req->index; struct rvu_pfvf *pfvf; u8 cgx_id, lmac_id; int rc; rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id); pfvf = rvu_get_pfvf(rvu, req->hdr.pcifunc); rc = rvu_npc_exact_del_table_entry_by_id(rvu, seq_id); if (rc) { /* TODO: how to handle this error case ? */ dev_err(rvu->dev, "%s MAC (%pM) del PF=%d failed\n", __func__, pfvf->mac_addr, pf); return 0; } dev_dbg(rvu->dev, "%s MAC (%pM) del PF=%d success (seq_id=%u)\n", __func__, pfvf->mac_addr, pf, seq_id); return 0; } /** * rvu_npc_exact_mac_addr_update - Update mac address field with new value. * @rvu: resource virtualization unit. * @req: Update request. * @rsp: Update response. * Return: 0 upon success */ int rvu_npc_exact_mac_addr_update(struct rvu *rvu, struct cgx_mac_addr_update_req *req, struct cgx_mac_addr_update_rsp *rsp) { int pf = rvu_get_pf(req->hdr.pcifunc); struct npc_exact_table_entry *entry; struct npc_exact_table *table; struct rvu_pfvf *pfvf; u32 seq_id, mcam_idx; u8 old_mac[ETH_ALEN]; u8 cgx_id, lmac_id; int rc; if (!is_cgx_config_permitted(rvu, req->hdr.pcifunc)) return LMAC_AF_ERR_PERM_DENIED; dev_dbg(rvu->dev, "%s: Update request for seq_id=%d, mac=%pM\n", __func__, req->index, req->mac_addr); rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id); pfvf = rvu_get_pfvf(rvu, req->hdr.pcifunc); table = rvu->hw->table; mutex_lock(&table->lock); /* Lookup for entry which needs to be updated */ entry = __rvu_npc_exact_find_entry_by_seq_id(rvu, req->index); if (!entry) { dev_err(rvu->dev, "%s: failed to find entry for id=0x%x\n", __func__, req->index); mutex_unlock(&table->lock); return LMAC_AF_ERR_EXACT_MATCH_TBL_LOOK_UP_FAILED; } ether_addr_copy(old_mac, entry->mac); seq_id = entry->seq_id; mcam_idx = entry->mcam_idx; mutex_unlock(&table->lock); rc = rvu_npc_exact_update_table_entry(rvu, cgx_id, lmac_id, old_mac, req->mac_addr, &seq_id); if (!rc) { rsp->index = seq_id; dev_dbg(rvu->dev, "%s mac:%pM (pfvf:%pM default:%pM) update to PF=%d success\n", __func__, req->mac_addr, pfvf->mac_addr, pfvf->default_mac, pf); ether_addr_copy(pfvf->mac_addr, req->mac_addr); return 0; } /* Try deleting and adding it again */ rc = rvu_npc_exact_del_table_entry_by_id(rvu, req->index); if (rc) { /* This could be a new entry */ dev_dbg(rvu->dev, "%s MAC (%pM) del PF=%d failed\n", __func__, pfvf->mac_addr, pf); } rc = rvu_npc_exact_add_table_entry(rvu, cgx_id, lmac_id, req->mac_addr, pfvf->rx_chan_base, 0, &seq_id, true, mcam_idx, req->hdr.pcifunc); if (rc) { dev_err(rvu->dev, "%s MAC (%pM) add PF=%d failed\n", __func__, req->mac_addr, pf); return LMAC_AF_ERR_EXACT_MATCH_TBL_ADD_FAILED; } rsp->index = seq_id; dev_dbg(rvu->dev, "%s MAC (new:%pM, old=%pM default:%pM) del and add to PF=%d success (seq_id=%u)\n", __func__, req->mac_addr, pfvf->mac_addr, pfvf->default_mac, pf, seq_id); ether_addr_copy(pfvf->mac_addr, req->mac_addr); return 0; } /** * rvu_npc_exact_mac_addr_add - Adds MAC address to exact match table. * @rvu: resource virtualization unit. * @req: Add request. * @rsp: Add response. * Return: 0 upon success */ int rvu_npc_exact_mac_addr_add(struct rvu *rvu, struct cgx_mac_addr_add_req *req, struct cgx_mac_addr_add_rsp *rsp) { int pf = rvu_get_pf(req->hdr.pcifunc); struct rvu_pfvf *pfvf; u8 cgx_id, lmac_id; int rc = 0; u32 seq_id; rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id); pfvf = rvu_get_pfvf(rvu, req->hdr.pcifunc); rc = rvu_npc_exact_add_table_entry(rvu, cgx_id, lmac_id, req->mac_addr, pfvf->rx_chan_base, 0, &seq_id, true, -1, req->hdr.pcifunc); if (!rc) { rsp->index = seq_id; dev_dbg(rvu->dev, "%s MAC (%pM) add to PF=%d success (seq_id=%u)\n", __func__, req->mac_addr, pf, seq_id); return 0; } dev_err(rvu->dev, "%s MAC (%pM) add to PF=%d failed\n", __func__, req->mac_addr, pf); return LMAC_AF_ERR_EXACT_MATCH_TBL_ADD_FAILED; } /** * rvu_npc_exact_mac_addr_del - Delete DMAC filter * @rvu: resource virtualization unit. * @req: Delete request. * @rsp: Delete response. * Return: 0 upon success */ int rvu_npc_exact_mac_addr_del(struct rvu *rvu, struct cgx_mac_addr_del_req *req, struct msg_rsp *rsp) { int pf = rvu_get_pf(req->hdr.pcifunc); int rc; rc = rvu_npc_exact_del_table_entry_by_id(rvu, req->index); if (!rc) { dev_dbg(rvu->dev, "%s del to PF=%d success (seq_id=%u)\n", __func__, pf, req->index); return 0; } dev_err(rvu->dev, "%s del to PF=%d failed (seq_id=%u)\n", __func__, pf, req->index); return LMAC_AF_ERR_EXACT_MATCH_TBL_DEL_FAILED; } /** * rvu_npc_exact_mac_addr_set - Add PF mac address to dmac filter. * @rvu: resource virtualization unit. * @req: Set request. * @rsp: Set response. * Return: 0 upon success */ int rvu_npc_exact_mac_addr_set(struct rvu *rvu, struct cgx_mac_addr_set_or_get *req, struct cgx_mac_addr_set_or_get *rsp) { int pf = rvu_get_pf(req->hdr.pcifunc); u32 seq_id = req->index; struct rvu_pfvf *pfvf; u8 cgx_id, lmac_id; u32 mcam_idx = -1; int rc, nixlf; rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id); pfvf = &rvu->pf[pf]; /* If table does not have an entry; both update entry and del table entry API * below fails. Those are not failure conditions. */ rc = rvu_npc_exact_update_table_entry(rvu, cgx_id, lmac_id, pfvf->mac_addr, req->mac_addr, &seq_id); if (!rc) { rsp->index = seq_id; ether_addr_copy(pfvf->mac_addr, req->mac_addr); ether_addr_copy(rsp->mac_addr, req->mac_addr); dev_dbg(rvu->dev, "%s MAC (%pM) update to PF=%d success\n", __func__, req->mac_addr, pf); return 0; } /* Try deleting and adding it again */ rc = rvu_npc_exact_del_table_entry_by_id(rvu, req->index); if (rc) { dev_dbg(rvu->dev, "%s MAC (%pM) del PF=%d failed\n", __func__, pfvf->mac_addr, pf); } /* find mcam entry if exist */ rc = nix_get_nixlf(rvu, req->hdr.pcifunc, &nixlf, NULL); if (!rc) { mcam_idx = npc_get_nixlf_mcam_index(&rvu->hw->mcam, req->hdr.pcifunc, nixlf, NIXLF_UCAST_ENTRY); } rc = rvu_npc_exact_add_table_entry(rvu, cgx_id, lmac_id, req->mac_addr, pfvf->rx_chan_base, 0, &seq_id, true, mcam_idx, req->hdr.pcifunc); if (rc) { dev_err(rvu->dev, "%s MAC (%pM) add PF=%d failed\n", __func__, req->mac_addr, pf); return LMAC_AF_ERR_EXACT_MATCH_TBL_ADD_FAILED; } rsp->index = seq_id; ether_addr_copy(rsp->mac_addr, req->mac_addr); ether_addr_copy(pfvf->mac_addr, req->mac_addr); dev_dbg(rvu->dev, "%s MAC (%pM) del and add to PF=%d success (seq_id=%u)\n", __func__, req->mac_addr, pf, seq_id); return 0; } /** * rvu_npc_exact_can_disable_feature - Check if feature can be disabled. * @rvu: resource virtualization unit. * Return: True if exact match feature is supported. */ bool rvu_npc_exact_can_disable_feature(struct rvu *rvu) { struct npc_exact_table *table = rvu->hw->table; bool empty; if (!rvu->hw->cap.npc_exact_match_enabled) return false; mutex_lock(&table->lock); empty = list_empty(&table->lhead_gbl); mutex_unlock(&table->lock); return empty; } /** * rvu_npc_exact_disable_feature - Disable feature. * @rvu: resource virtualization unit. */ void rvu_npc_exact_disable_feature(struct rvu *rvu) { rvu->hw->cap.npc_exact_match_enabled = false; } /** * rvu_npc_exact_reset - Delete and free all entry which match pcifunc. * @rvu: resource virtualization unit. * @pcifunc: PCI func to match. */ void rvu_npc_exact_reset(struct rvu *rvu, u16 pcifunc) { struct npc_exact_table *table = rvu->hw->table; struct npc_exact_table_entry *tmp, *iter; u32 seq_id; mutex_lock(&table->lock); list_for_each_entry_safe(iter, tmp, &table->lhead_gbl, glist) { if (pcifunc != iter->pcifunc) continue; seq_id = iter->seq_id; dev_dbg(rvu->dev, "%s: resetting pcifun=%d seq_id=%u\n", __func__, pcifunc, seq_id); mutex_unlock(&table->lock); rvu_npc_exact_del_table_entry_by_id(rvu, seq_id); mutex_lock(&table->lock); } mutex_unlock(&table->lock); } /** * rvu_npc_exact_init - initialize exact match table * @rvu: resource virtualization unit. * * Initialize HW and SW resources to manage 4way-2K table and fully * associative 32-entry mcam table. * Return: 0 upon success. */ int rvu_npc_exact_init(struct rvu *rvu) { u64 bcast_mcast_val, bcast_mcast_mask; struct npc_exact_table *table; u64 exact_val, exact_mask; u64 chan_val, chan_mask; u8 cgx_id, lmac_id; u32 *drop_mcam_idx; u16 max_lmac_cnt; u64 npc_const3; int table_size; int blkaddr; u16 pcifunc; int err, i; u64 cfg; bool rc; /* Read NPC_AF_CONST3 and check for have exact * match functionality is present */ blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0); if (blkaddr < 0) { dev_err(rvu->dev, "%s: NPC block not implemented\n", __func__); return -EINVAL; } /* Check exact match feature is supported */ npc_const3 = rvu_read64(rvu, blkaddr, NPC_AF_CONST3); if (!(npc_const3 & BIT_ULL(62))) return 0; /* Check if kex profile has enabled EXACT match nibble */ cfg = rvu_read64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_RX)); if (!(cfg & NPC_EXACT_NIBBLE_HIT)) return 0; /* Set capability to true */ rvu->hw->cap.npc_exact_match_enabled = true; table = kzalloc(sizeof(*table), GFP_KERNEL); if (!table) return -ENOMEM; dev_dbg(rvu->dev, "%s: Memory allocation for table success\n", __func__); rvu->hw->table = table; /* Read table size, ways and depth */ table->mem_table.ways = FIELD_GET(GENMASK_ULL(19, 16), npc_const3); table->mem_table.depth = FIELD_GET(GENMASK_ULL(15, 0), npc_const3); table->cam_table.depth = FIELD_GET(GENMASK_ULL(31, 24), npc_const3); dev_dbg(rvu->dev, "%s: NPC exact match 4way_2k table(ways=%d, depth=%d)\n", __func__, table->mem_table.ways, table->cam_table.depth); /* Check if depth of table is not a sequre of 2 * TODO: why _builtin_popcount() is not working ? */ if ((table->mem_table.depth & (table->mem_table.depth - 1)) != 0) { dev_err(rvu->dev, "%s: NPC exact match 4way_2k table depth(%d) is not square of 2\n", __func__, table->mem_table.depth); return -EINVAL; } table_size = table->mem_table.depth * table->mem_table.ways; /* Allocate bitmap for 4way 2K table */ table->mem_table.bmap = devm_bitmap_zalloc(rvu->dev, table_size, GFP_KERNEL); if (!table->mem_table.bmap) return -ENOMEM; dev_dbg(rvu->dev, "%s: Allocated bitmap for 4way 2K entry table\n", __func__); /* Allocate bitmap for 32 entry mcam */ table->cam_table.bmap = devm_bitmap_zalloc(rvu->dev, 32, GFP_KERNEL); if (!table->cam_table.bmap) return -ENOMEM; dev_dbg(rvu->dev, "%s: Allocated bitmap for 32 entry cam\n", __func__); table->tot_ids = table_size + table->cam_table.depth; table->id_bmap = devm_bitmap_zalloc(rvu->dev, table->tot_ids, GFP_KERNEL); if (!table->id_bmap) return -ENOMEM; dev_dbg(rvu->dev, "%s: Allocated bitmap for id map (total=%d)\n", __func__, table->tot_ids); /* Initialize list heads for npc_exact_table entries. * This entry is used by debugfs to show entries in * exact match table. */ for (i = 0; i < NPC_EXACT_TBL_MAX_WAYS; i++) INIT_LIST_HEAD(&table->lhead_mem_tbl_entry[i]); INIT_LIST_HEAD(&table->lhead_cam_tbl_entry); INIT_LIST_HEAD(&table->lhead_gbl); mutex_init(&table->lock); rvu_exact_config_secret_key(rvu); rvu_exact_config_search_key(rvu); rvu_exact_config_table_mask(rvu); rvu_exact_config_result_ctrl(rvu, table->mem_table.depth); /* - No drop rule for LBK * - Drop rules for SDP and each LMAC. */ exact_val = !NPC_EXACT_RESULT_HIT; exact_mask = NPC_EXACT_RESULT_HIT; /* nibble - 3 2 1 0 * L3B L3M L2B L2M */ bcast_mcast_val = 0b0000; bcast_mcast_mask = 0b0011; /* Install SDP drop rule */ drop_mcam_idx = &table->num_drop_rules; max_lmac_cnt = rvu->cgx_cnt_max * rvu->hw->lmac_per_cgx + PF_CGXMAP_BASE; for (i = PF_CGXMAP_BASE; i < max_lmac_cnt; i++) { if (rvu->pf2cgxlmac_map[i] == 0xFF) continue; rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[i], &cgx_id, &lmac_id); rc = rvu_npc_exact_calc_drop_rule_chan_and_mask(rvu, NIX_INTF_TYPE_CGX, cgx_id, lmac_id, &chan_val, &chan_mask); if (!rc) { dev_err(rvu->dev, "%s: failed, info chan_val=0x%llx chan_mask=0x%llx rule_id=%d\n", __func__, chan_val, chan_mask, *drop_mcam_idx); return -EINVAL; } /* Filter rules are only for PF */ pcifunc = RVU_PFFUNC(i, 0); dev_dbg(rvu->dev, "%s:Drop rule cgx=%d lmac=%d chan(val=0x%llx, mask=0x%llx\n", __func__, cgx_id, lmac_id, chan_val, chan_mask); rc = rvu_npc_exact_save_drop_rule_chan_and_mask(rvu, table->num_drop_rules, chan_val, chan_mask, pcifunc); if (!rc) { dev_err(rvu->dev, "%s: failed to set drop info for cgx=%d, lmac=%d, chan=%llx\n", __func__, cgx_id, lmac_id, chan_val); return -EINVAL; } err = npc_install_mcam_drop_rule(rvu, *drop_mcam_idx, &table->counter_idx[*drop_mcam_idx], chan_val, chan_mask, exact_val, exact_mask, bcast_mcast_val, bcast_mcast_mask); if (err) { dev_err(rvu->dev, "failed to configure drop rule (cgx=%d lmac=%d)\n", cgx_id, lmac_id); return err; } (*drop_mcam_idx)++; } dev_info(rvu->dev, "initialized exact match table successfully\n"); return 0; }
Information contained on this website is for historical information purposes only and does not indicate or represent copyright ownership.
Created with Cregit http://github.com/cregit/cregit
Version 2.0-RC1