iwlwifi: mvm: support Bz TX checksum offload

Support TX checksum offload for Bz devices, where we have full
checksum offload (NETIF_F_HW_CSUM) and the hardware doesn't
need to parse the IP headers or anything.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Link: https://lore.kernel.org/r/iwlwifi.20211219132536.c0f44c98b36d.I75a688f3ac80cbe824c459ece4bb67843b9fce76@changeid
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

diff --git a/drivers/net/wireless/intel/iwlwifi/cfg/22000.c b/drivers/net/wireless/intel/iwlwifi/cfg/22000.c
index 511195a2058e2a90cb45660963930c9a37bd8a61..330ef04ca51a5900cd3a3b9ea2954278f3b37cb0 100644 (file)
--- a/drivers/net/wireless/intel/iwlwifi/cfg/22000.c
+++ b/drivers/net/wireless/intel/iwlwifi/cfg/22000.c
@@ -252,7 +252,7 @@ static const struct iwl_ht_params iwl_22000_ht_params = {
        .dccm2_len = IWL_22000_DCCM2_LEN,                               \
        .smem_offset = IWL_22000_SMEM_OFFSET,                           \
        .smem_len = IWL_22000_SMEM_LEN,                                 \
-       .features = IWL_TX_CSUM_NETIF_FLAGS | NETIF_F_RXCSUM,           \
+       .features = IWL_TX_CSUM_NETIF_FLAGS_BZ | NETIF_F_RXCSUM,        \
        .apmg_not_supported = true,                                     \
        .trans.mq_rx_supported = true,                                  \
        .vht_mu_mimo_supported = true,                                  \

diff --git a/drivers/net/wireless/intel/iwlwifi/fw/api/tx.h b/drivers/net/wireless/intel/iwlwifi/fw/api/tx.h
index 4a74c0ea0f3153a0f787d0d5c893bf9bc751875b..e73cc7380a26f543080a2702124e62c8b6e62ac9 100644 (file)
--- a/drivers/net/wireless/intel/iwlwifi/fw/api/tx.h
+++ b/drivers/net/wireless/intel/iwlwifi/fw/api/tx.h
@@ -177,6 +177,17 @@ enum iwl_tx_offload_assist_flags_pos {
 #define IWL_TX_CMD_OFFLD_MH_MASK       0x1f
 #define IWL_TX_CMD_OFFLD_IP_HDR_MASK   0x3f
 
+enum iwl_tx_offload_assist_bz {
+       IWL_TX_CMD_OFFLD_BZ_RESULT_OFFS         = 0x000003ff,
+       IWL_TX_CMD_OFFLD_BZ_START_OFFS          = 0x001ff800,
+       IWL_TX_CMD_OFFLD_BZ_MH_LEN              = 0x07c00000,
+       IWL_TX_CMD_OFFLD_BZ_MH_PAD              = 0x08000000,
+       IWL_TX_CMD_OFFLD_BZ_AMSDU               = 0x10000000,
+       IWL_TX_CMD_OFFLD_BZ_ZERO2ONES           = 0x20000000,
+       IWL_TX_CMD_OFFLD_BZ_ENABLE_CSUM         = 0x40000000,
+       IWL_TX_CMD_OFFLD_BZ_PARTIAL_CSUM        = 0x80000000,
+};
+
 /* TODO: complete documentation for try_cnt and btkill_cnt */
 /**
  * struct iwl_tx_cmd - TX command struct to FW

diff --git a/drivers/net/wireless/intel/iwlwifi/iwl-config.h b/drivers/net/wireless/intel/iwlwifi/iwl-config.h
index 95be04ba93900ffd807a68b5be2935e7ff308228..e122b8b4e1fc4ca820512b52fb6929fc4af99eeb 100644 (file)
--- a/drivers/net/wireless/intel/iwlwifi/iwl-config.h
+++ b/drivers/net/wireless/intel/iwlwifi/iwl-config.h
@@ -84,7 +84,10 @@ enum iwl_nvm_type {
 #define IWL_DEFAULT_MAX_TX_POWER 22
 #define IWL_TX_CSUM_NETIF_FLAGS (NETIF_F_IPV6_CSUM | NETIF_F_IP_CSUM |\
                                 NETIF_F_TSO | NETIF_F_TSO6)
-#define IWL_CSUM_NETIF_FLAGS_MASK (IWL_TX_CSUM_NETIF_FLAGS | NETIF_F_RXCSUM)
+#define IWL_TX_CSUM_NETIF_FLAGS_BZ (NETIF_F_HW_CSUM | NETIF_F_TSO | NETIF_F_TSO6)
+#define IWL_CSUM_NETIF_FLAGS_MASK (IWL_TX_CSUM_NETIF_FLAGS | \
+                                  IWL_TX_CSUM_NETIF_FLAGS_BZ | \
+                                  NETIF_F_RXCSUM)
 
 /* Antenna presence definitions */
 #define        ANT_NONE        0x0

diff --git a/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c b/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c
index 761a9b62ac222dce4f721282015281e95d790faa..efda37645fd47f1fc355babd9490a5013468453f 100644 (file)
--- a/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c
+++ b/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c
@@ -5523,6 +5523,10 @@ static bool iwl_mvm_mac_can_aggregate(struct ieee80211_hw *hw,
 {
        struct iwl_mvm *mvm = IWL_MAC80211_GET_MVM(hw);
 
+       if (mvm->trans->trans_cfg->device_family >= IWL_DEVICE_FAMILY_BZ)
+               return iwl_mvm_tx_csum_bz(mvm, head, true) ==
+                      iwl_mvm_tx_csum_bz(mvm, skb, true);
+
        /* For now don't aggregate IPv6 in AMSDU */
        if (skb->protocol != htons(ETH_P_IP))
                return false;

diff --git a/drivers/net/wireless/intel/iwlwifi/mvm/mvm.h b/drivers/net/wireless/intel/iwlwifi/mvm/mvm.h
index 1393e5de2e9aa2ac7c3f41d0ac733e18ed3efdbc..3d010afc69bcfef11c642b35cd7de1439abc1f03 100644 (file)
--- a/drivers/net/wireless/intel/iwlwifi/mvm/mvm.h
+++ b/drivers/net/wireless/intel/iwlwifi/mvm/mvm.h
@@ -1518,6 +1518,7 @@ void iwl_mvm_mac_itxq_xmit(struct ieee80211_hw *hw, struct ieee80211_txq *txq);
 unsigned int iwl_mvm_max_amsdu_size(struct iwl_mvm *mvm,
                                    struct ieee80211_sta *sta,
                                    unsigned int tid);
+u32 iwl_mvm_tx_csum_bz(struct iwl_mvm *mvm, struct sk_buff *skb, bool amsdu);
 
 #ifdef CONFIG_IWLWIFI_DEBUG
 const char *iwl_mvm_get_tx_fail_reason(u32 status);

diff --git a/drivers/net/wireless/intel/iwlwifi/mvm/tx.c b/drivers/net/wireless/intel/iwlwifi/mvm/tx.c
index 147f8df3053d9c668e8535393a4275ec173c7a00..6fa2c12f7955a86be224073361aea4059317f700 100644 (file)
--- a/drivers/net/wireless/intel/iwlwifi/mvm/tx.c
+++ b/drivers/net/wireless/intel/iwlwifi/mvm/tx.c
@@ -39,11 +39,10 @@ iwl_mvm_bar_check_trigger(struct iwl_mvm *mvm, const u8 *addr,
 #define OPT_HDR(type, skb, off) \
        (type *)(skb_network_header(skb) + (off))
 
-static u16 iwl_mvm_tx_csum(struct iwl_mvm *mvm, struct sk_buff *skb,
-                          struct ieee80211_hdr *hdr,
-                          struct ieee80211_tx_info *info,
-                          bool amsdu)
+static u16 iwl_mvm_tx_csum_pre_bz(struct iwl_mvm *mvm, struct sk_buff *skb,
+                                 struct ieee80211_tx_info *info, bool amsdu)
 {
+       struct ieee80211_hdr *hdr = (void *)skb->data;
        u16 offload_assist = 0;
 #if IS_ENABLED(CONFIG_INET)
        u16 mh_len = ieee80211_hdrlen(hdr->frame_control);
@@ -141,6 +140,54 @@ out:
        return offload_assist;
 }
 
+u32 iwl_mvm_tx_csum_bz(struct iwl_mvm *mvm, struct sk_buff *skb, bool amsdu)
+{
+       struct ieee80211_hdr *hdr = (void *)skb->data;
+       u32 offload_assist = IWL_TX_CMD_OFFLD_BZ_PARTIAL_CSUM;
+       unsigned int hdrlen = ieee80211_hdrlen(hdr->frame_control);
+       unsigned int csum_start = skb_checksum_start_offset(skb);
+
+       offload_assist |= u32_encode_bits(hdrlen / 2,
+                                         IWL_TX_CMD_OFFLD_BZ_MH_LEN);
+       if (amsdu)
+               offload_assist |= IWL_TX_CMD_OFFLD_BZ_AMSDU;
+       else if (hdrlen % 4)
+               /* padding is inserted later in transport */
+               offload_assist |= IWL_TX_CMD_OFFLD_BZ_MH_PAD;
+
+       if (skb->ip_summed != CHECKSUM_PARTIAL)
+               return offload_assist;
+
+       offload_assist |= IWL_TX_CMD_OFFLD_BZ_ENABLE_CSUM |
+                         IWL_TX_CMD_OFFLD_BZ_ZERO2ONES;
+
+       /*
+        * mac80211 will always calculate checksum in software for
+        * non-fast-xmit, and so we can only do offloaded checksum
+        * for fast-xmit frames. In this case, we always have the
+        * RFC 1042 header present. skb_checksum_start_offset()
+        * returns the offset from the beginning, but the hardware
+        * needs it from after the header & SNAP header.
+        */
+       csum_start -= hdrlen + 8;
+
+       offload_assist |= u32_encode_bits(csum_start,
+                                         IWL_TX_CMD_OFFLD_BZ_START_OFFS);
+       offload_assist |= u32_encode_bits(csum_start + skb->csum_offset,
+                                         IWL_TX_CMD_OFFLD_BZ_RESULT_OFFS);
+
+       return offload_assist;
+}
+
+static u32 iwl_mvm_tx_csum(struct iwl_mvm *mvm, struct sk_buff *skb,
+                          struct ieee80211_tx_info *info,
+                          bool amsdu)
+{
+       if (mvm->trans->trans_cfg->device_family < IWL_DEVICE_FAMILY_BZ)
+               return iwl_mvm_tx_csum_pre_bz(mvm, skb, info, amsdu);
+       return iwl_mvm_tx_csum_bz(mvm, skb, amsdu);
+}
+
 /*
  * Sets most of the Tx cmd's fields
  */
@@ -240,7 +287,7 @@ void iwl_mvm_set_tx_cmd(struct iwl_mvm *mvm, struct sk_buff *skb,
        tx_cmd->sta_id = sta_id;
 
        tx_cmd->offload_assist =
-               cpu_to_le16(iwl_mvm_tx_csum(mvm, skb, hdr, info, amsdu));
+               cpu_to_le16(iwl_mvm_tx_csum_pre_bz(mvm, skb, info, amsdu));
 }
 
 static u32 iwl_mvm_get_tx_ant(struct iwl_mvm *mvm,
@@ -462,7 +509,6 @@ iwl_mvm_set_tx_params(struct iwl_mvm *mvm, struct sk_buff *skb,
        dev_cmd->hdr.cmd = TX_CMD;
 
        if (iwl_mvm_has_new_tx_api(mvm)) {
-               u16 offload_assist;
                u32 rate_n_flags = 0;
                u16 flags = 0;
                struct iwl_mvm_sta *mvmsta = sta ?
@@ -475,8 +521,6 @@ iwl_mvm_set_tx_params(struct iwl_mvm *mvm, struct sk_buff *skb,
                        amsdu = *qc & IEEE80211_QOS_CTL_A_MSDU_PRESENT;
                }
 
-               offload_assist = iwl_mvm_tx_csum(mvm, skb, hdr, info, amsdu);
-
                if (!info->control.hw_key)
                        flags |= IWL_TX_FLAGS_ENCRYPT_DIS;
 
@@ -496,6 +540,8 @@ iwl_mvm_set_tx_params(struct iwl_mvm *mvm, struct sk_buff *skb,
                if (mvm->trans->trans_cfg->device_family >=
                    IWL_DEVICE_FAMILY_AX210) {
                        struct iwl_tx_cmd_gen3 *cmd = (void *)dev_cmd->payload;
+                       u32 offload_assist = iwl_mvm_tx_csum(mvm, skb,
+                                                            info, amsdu);
 
                        cmd->offload_assist = cpu_to_le32(offload_assist);
 
@@ -509,6 +555,9 @@ iwl_mvm_set_tx_params(struct iwl_mvm *mvm, struct sk_buff *skb,
                        cmd->rate_n_flags = cpu_to_le32(rate_n_flags);
                } else {
                        struct iwl_tx_cmd_gen2 *cmd = (void *)dev_cmd->payload;
+                       u16 offload_assist = iwl_mvm_tx_csum_pre_bz(mvm, skb,
+                                                                   info,
+                                                                   amsdu);
 
                        cmd->offload_assist = cpu_to_le16(offload_assist);