[PATCH net-next v7 05/11] net: bcmasp: Add support for wake on net filters

From: Justin Chen
Date: Wed Jun 14 2023 - 18:31:00 EST


Add support for wake on network filters. The max match is 256 bytes.

Signed-off-by: Justin Chen <justin.chen@xxxxxxxxxxxx>
---
drivers/net/ethernet/broadcom/asp2/bcmasp.c | 595 +++++++++++++++++++++
drivers/net/ethernet/broadcom/asp2/bcmasp.h | 40 ++
.../net/ethernet/broadcom/asp2/bcmasp_ethtool.c | 131 ++++-
drivers/net/ethernet/broadcom/asp2/bcmasp_intf.c | 3 +
4 files changed, 768 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/broadcom/asp2/bcmasp.c b/drivers/net/ethernet/broadcom/asp2/bcmasp.c
index 9e3977856b71..ba9842c760d5 100644
--- a/drivers/net/ethernet/broadcom/asp2/bcmasp.c
+++ b/drivers/net/ethernet/broadcom/asp2/bcmasp.c
@@ -130,6 +130,597 @@ void bcmasp_flush_rx_port(struct bcmasp_intf *intf)
rx_ctrl_core_wl(priv, mask, priv->hw_info->rx_ctrl_flush);
}

+static void bcmasp_netfilt_hw_en_wake(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt)
+{
+ rx_filter_core_wl(priv, ASP_RX_FILTER_NET_OFFSET_L3_1(64),
+ ASP_RX_FILTER_NET_OFFSET(nfilt->hw_index));
+
+ rx_filter_core_wl(priv, ASP_RX_FILTER_NET_OFFSET_L2(32) |
+ ASP_RX_FILTER_NET_OFFSET_L3_0(32) |
+ ASP_RX_FILTER_NET_OFFSET_L3_1(96) |
+ ASP_RX_FILTER_NET_OFFSET_L4(32),
+ ASP_RX_FILTER_NET_OFFSET(nfilt->hw_index + 1));
+
+ rx_filter_core_wl(priv, ASP_RX_FILTER_NET_CFG_CH(nfilt->port + 8) |
+ ASP_RX_FILTER_NET_CFG_EN |
+ ASP_RX_FILTER_NET_CFG_L2_EN |
+ ASP_RX_FILTER_NET_CFG_L3_EN |
+ ASP_RX_FILTER_NET_CFG_L4_EN |
+ ASP_RX_FILTER_NET_CFG_L3_FRM(2) |
+ ASP_RX_FILTER_NET_CFG_L4_FRM(2) |
+ ASP_RX_FILTER_NET_CFG_UMC(nfilt->port),
+ ASP_RX_FILTER_NET_CFG(nfilt->hw_index));
+
+ rx_filter_core_wl(priv, ASP_RX_FILTER_NET_CFG_CH(nfilt->port + 8) |
+ ASP_RX_FILTER_NET_CFG_EN |
+ ASP_RX_FILTER_NET_CFG_L2_EN |
+ ASP_RX_FILTER_NET_CFG_L3_EN |
+ ASP_RX_FILTER_NET_CFG_L4_EN |
+ ASP_RX_FILTER_NET_CFG_L3_FRM(2) |
+ ASP_RX_FILTER_NET_CFG_L4_FRM(2) |
+ ASP_RX_FILTER_NET_CFG_UMC(nfilt->port),
+ ASP_RX_FILTER_NET_CFG(nfilt->hw_index + 1));
+}
+
+#define MAX_WAKE_FILTER_SIZE 256
+enum asp_netfilt_reg_type {
+ ASP_NETFILT_MATCH = 0,
+ ASP_NETFILT_MASK,
+ ASP_NETFILT_MAX
+};
+
+static int bcmasp_netfilt_get_reg_offset(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt,
+ enum asp_netfilt_reg_type reg_type,
+ u32 offset)
+{
+ u32 block_index, filter_sel;
+
+ if (offset < 32) {
+ block_index = ASP_RX_FILTER_NET_L2;
+ filter_sel = nfilt->hw_index;
+ } else if (offset < 64) {
+ block_index = ASP_RX_FILTER_NET_L2;
+ filter_sel = nfilt->hw_index + 1;
+ } else if (offset < 96) {
+ block_index = ASP_RX_FILTER_NET_L3_0;
+ filter_sel = nfilt->hw_index;
+ } else if (offset < 128) {
+ block_index = ASP_RX_FILTER_NET_L3_0;
+ filter_sel = nfilt->hw_index + 1;
+ } else if (offset < 160) {
+ block_index = ASP_RX_FILTER_NET_L3_1;
+ filter_sel = nfilt->hw_index;
+ } else if (offset < 192) {
+ block_index = ASP_RX_FILTER_NET_L3_1;
+ filter_sel = nfilt->hw_index + 1;
+ } else if (offset < 224) {
+ block_index = ASP_RX_FILTER_NET_L4;
+ filter_sel = nfilt->hw_index;
+ } else if (offset < 256) {
+ block_index = ASP_RX_FILTER_NET_L4;
+ filter_sel = nfilt->hw_index + 1;
+ } else {
+ return -EINVAL;
+ }
+
+ switch (reg_type) {
+ case ASP_NETFILT_MATCH:
+ return ASP_RX_FILTER_NET_PAT(filter_sel, block_index,
+ (offset % 32));
+ case ASP_NETFILT_MASK:
+ return ASP_RX_FILTER_NET_MASK(filter_sel, block_index,
+ (offset % 32));
+ default:
+ return -EINVAL;
+ }
+}
+
+static void bcmasp_netfilt_wr(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt,
+ enum asp_netfilt_reg_type reg_type,
+ u32 val, u32 offset)
+{
+ int reg_offset;
+
+ /* HW only accepts 4 byte aligned writes */
+ if (!IS_ALIGNED(offset, 4) || offset > MAX_WAKE_FILTER_SIZE)
+ return;
+
+ reg_offset = bcmasp_netfilt_get_reg_offset(priv, nfilt, reg_type,
+ offset);
+
+ rx_filter_core_wl(priv, val, reg_offset);
+}
+
+static u32 bcmasp_netfilt_rd(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt,
+ enum asp_netfilt_reg_type reg_type,
+ u32 offset)
+{
+ int reg_offset;
+
+ /* HW only accepts 4 byte aligned writes */
+ if (!IS_ALIGNED(offset, 4) || offset > MAX_WAKE_FILTER_SIZE)
+ return 0;
+
+ reg_offset = bcmasp_netfilt_get_reg_offset(priv, nfilt, reg_type,
+ offset);
+
+ return rx_filter_core_rl(priv, reg_offset);
+}
+
+static int bcmasp_netfilt_wr_m_wake(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt,
+ u32 offset, void *match, void *mask,
+ size_t size)
+{
+ u32 shift, mask_val = 0, match_val = 0;
+ bool first_byte = true;
+
+ if ((offset + size) > MAX_WAKE_FILTER_SIZE)
+ return -EINVAL;
+
+ while (size--) {
+ /* The HW only accepts 4 byte aligned writes, so if we
+ * begin unaligned or if remaining bytes less than 4,
+ * we need to read then write to avoid losing current
+ * register state
+ */
+ if (first_byte && (!IS_ALIGNED(offset, 4) || size < 3)) {
+ match_val = bcmasp_netfilt_rd(priv, nfilt,
+ ASP_NETFILT_MATCH,
+ ALIGN_DOWN(offset, 4));
+ mask_val = bcmasp_netfilt_rd(priv, nfilt,
+ ASP_NETFILT_MASK,
+ ALIGN_DOWN(offset, 4));
+ }
+
+ shift = (3 - (offset % 4)) * 8;
+ match_val &= ~GENMASK(shift + 7, shift);
+ mask_val &= ~GENMASK(shift + 7, shift);
+ match_val |= (u32)(*((u8 *)match) << shift);
+ mask_val |= (u32)(*((u8 *)mask) << shift);
+
+ /* If last byte or last byte of word, write to reg */
+ if (!size || ((offset % 4) == 3)) {
+ bcmasp_netfilt_wr(priv, nfilt, ASP_NETFILT_MATCH,
+ match_val, ALIGN_DOWN(offset, 4));
+ bcmasp_netfilt_wr(priv, nfilt, ASP_NETFILT_MASK,
+ mask_val, ALIGN_DOWN(offset, 4));
+ first_byte = true;
+ } else {
+ first_byte = false;
+ }
+
+ offset++;
+ match++;
+ mask++;
+ }
+
+ return 0;
+}
+
+static void bcmasp_netfilt_reset_hw(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt)
+{
+ int i;
+
+ for (i = 0; i < MAX_WAKE_FILTER_SIZE; i += 4) {
+ bcmasp_netfilt_wr(priv, nfilt, ASP_NETFILT_MATCH, 0, i);
+ bcmasp_netfilt_wr(priv, nfilt, ASP_NETFILT_MASK, 0, i);
+ }
+}
+
+static void bcmasp_netfilt_tcpip4_wr(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt,
+ struct ethtool_tcpip4_spec *match,
+ struct ethtool_tcpip4_spec *mask,
+ u32 offset)
+{
+ __be16 val_16, mask_16;
+
+ val_16 = htons(ETH_P_IP);
+ mask_16 = htons(0xFFFF);
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, (ETH_ALEN * 2) + offset,
+ &val_16, &mask_16, sizeof(val_16));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 1,
+ &match->tos, &mask->tos,
+ sizeof(match->tos));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 12,
+ &match->ip4src, &mask->ip4src,
+ sizeof(match->ip4src));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 16,
+ &match->ip4dst, &mask->ip4dst,
+ sizeof(match->ip4dst));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 20,
+ &match->psrc, &mask->psrc,
+ sizeof(match->psrc));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 22,
+ &match->pdst, &mask->pdst,
+ sizeof(match->pdst));
+}
+
+static void bcmasp_netfilt_tcpip6_wr(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt,
+ struct ethtool_tcpip6_spec *match,
+ struct ethtool_tcpip6_spec *mask,
+ u32 offset)
+{
+ __be16 val_16, mask_16;
+
+ val_16 = htons(ETH_P_IPV6);
+ mask_16 = htons(0xFFFF);
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, (ETH_ALEN * 2) + offset,
+ &val_16, &mask_16, sizeof(val_16));
+ val_16 = htons(match->tclass << 4);
+ mask_16 = htons(mask->tclass << 4);
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset,
+ &val_16, &mask_16, sizeof(val_16));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 8,
+ &match->ip6src, &mask->ip6src,
+ sizeof(match->ip6src));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 24,
+ &match->ip6dst, &mask->ip6dst,
+ sizeof(match->ip6dst));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 40,
+ &match->psrc, &mask->psrc,
+ sizeof(match->psrc));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 42,
+ &match->pdst, &mask->pdst,
+ sizeof(match->pdst));
+}
+
+static int bcmasp_netfilt_wr_to_hw(struct bcmasp_priv *priv,
+ struct bcmasp_net_filter *nfilt)
+{
+ struct ethtool_rx_flow_spec *fs = &nfilt->fs;
+ unsigned int offset = 0;
+ __be16 val_16, mask_16;
+ u8 val_8, mask_8;
+
+ /* Currently only supports wake filters */
+ if (!nfilt->wake_filter)
+ return -EINVAL;
+
+ bcmasp_netfilt_reset_hw(priv, nfilt);
+
+ if (fs->flow_type & FLOW_MAC_EXT) {
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, 0, &fs->h_ext.h_dest,
+ &fs->m_ext.h_dest,
+ sizeof(fs->h_ext.h_dest));
+ }
+
+ if ((fs->flow_type & FLOW_EXT) &&
+ (fs->m_ext.vlan_etype || fs->m_ext.vlan_tci)) {
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, (ETH_ALEN * 2),
+ &fs->h_ext.vlan_etype,
+ &fs->m_ext.vlan_etype,
+ sizeof(fs->h_ext.vlan_etype));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ((ETH_ALEN * 2) + 2),
+ &fs->h_ext.vlan_tci,
+ &fs->m_ext.vlan_tci,
+ sizeof(fs->h_ext.vlan_tci));
+ offset += VLAN_HLEN;
+ }
+
+ switch (fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+ case ETHER_FLOW:
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, 0,
+ &fs->h_u.ether_spec.h_dest,
+ &fs->m_u.ether_spec.h_dest,
+ sizeof(fs->h_u.ether_spec.h_dest));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_ALEN,
+ &fs->h_u.ether_spec.h_source,
+ &fs->m_u.ether_spec.h_source,
+ sizeof(fs->h_u.ether_spec.h_source));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, (ETH_ALEN * 2) + offset,
+ &fs->h_u.ether_spec.h_proto,
+ &fs->m_u.ether_spec.h_proto,
+ sizeof(fs->h_u.ether_spec.h_proto));
+
+ break;
+ case IP_USER_FLOW:
+ val_16 = htons(ETH_P_IP);
+ mask_16 = htons(0xFFFF);
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, (ETH_ALEN * 2) + offset,
+ &val_16, &mask_16, sizeof(val_16));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 1,
+ &fs->h_u.usr_ip4_spec.tos,
+ &fs->m_u.usr_ip4_spec.tos,
+ sizeof(fs->h_u.usr_ip4_spec.tos));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 9,
+ &fs->h_u.usr_ip4_spec.proto,
+ &fs->m_u.usr_ip4_spec.proto,
+ sizeof(fs->h_u.usr_ip4_spec.proto));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 12,
+ &fs->h_u.usr_ip4_spec.ip4src,
+ &fs->m_u.usr_ip4_spec.ip4src,
+ sizeof(fs->h_u.usr_ip4_spec.ip4src));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 16,
+ &fs->h_u.usr_ip4_spec.ip4dst,
+ &fs->m_u.usr_ip4_spec.ip4dst,
+ sizeof(fs->h_u.usr_ip4_spec.ip4dst));
+ if (!fs->m_u.usr_ip4_spec.l4_4_bytes)
+ break;
+
+ /* Only supports 20 byte IPv4 header */
+ val_8 = 0x45;
+ mask_8 = 0xFF;
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset,
+ &val_8, &mask_8, sizeof(val_8));
+ bcmasp_netfilt_wr_m_wake(priv, nfilt,
+ ETH_HLEN + 20 + offset,
+ &fs->h_u.usr_ip4_spec.l4_4_bytes,
+ &fs->m_u.usr_ip4_spec.l4_4_bytes,
+ sizeof(fs->h_u.usr_ip4_spec.l4_4_bytes)
+ );
+ break;
+ case TCP_V4_FLOW:
+ val_8 = IPPROTO_TCP;
+ mask_8 = 0xFF;
+ bcmasp_netfilt_tcpip4_wr(priv, nfilt, &fs->h_u.tcp_ip4_spec,
+ &fs->m_u.tcp_ip4_spec, offset);
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 9,
+ &val_8, &mask_8, sizeof(val_8));
+ break;
+ case UDP_V4_FLOW:
+ val_8 = IPPROTO_UDP;
+ mask_8 = 0xFF;
+ bcmasp_netfilt_tcpip4_wr(priv, nfilt, &fs->h_u.udp_ip4_spec,
+ &fs->m_u.udp_ip4_spec, offset);
+
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 9,
+ &val_8, &mask_8, sizeof(val_8));
+ break;
+ case TCP_V6_FLOW:
+ val_8 = IPPROTO_TCP;
+ mask_8 = 0xFF;
+ bcmasp_netfilt_tcpip6_wr(priv, nfilt, &fs->h_u.tcp_ip6_spec,
+ &fs->m_u.tcp_ip6_spec, offset);
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 6,
+ &val_8, &mask_8, sizeof(val_8));
+ break;
+ case UDP_V6_FLOW:
+ val_8 = IPPROTO_UDP;
+ mask_8 = 0xFF;
+ bcmasp_netfilt_tcpip6_wr(priv, nfilt, &fs->h_u.udp_ip6_spec,
+ &fs->m_u.udp_ip6_spec, offset);
+ bcmasp_netfilt_wr_m_wake(priv, nfilt, ETH_HLEN + offset + 6,
+ &val_8, &mask_8, sizeof(val_8));
+ break;
+ }
+
+ bcmasp_netfilt_hw_en_wake(priv, nfilt);
+
+ return 0;
+}
+
+void bcmasp_netfilt_suspend(struct bcmasp_intf *intf)
+{
+ struct bcmasp_priv *priv = intf->parent;
+ bool write = false;
+ int ret, i;
+
+ /* Write all filters to HW */
+ for (i = 0; i < NUM_NET_FILTERS; i++) {
+ /* If the filter does not match the port, skip programming. */
+ if (!priv->net_filters[i].claimed ||
+ priv->net_filters[i].port != intf->port)
+ continue;
+
+ if (i > 0 && (i % 2) &&
+ priv->net_filters[i].wake_filter &&
+ priv->net_filters[i - 1].wake_filter)
+ continue;
+
+ ret = bcmasp_netfilt_wr_to_hw(priv, &priv->net_filters[i]);
+ if (!ret)
+ write = true;
+ }
+
+ /* Successfully programmed at least one wake filter
+ * so enable top level wake config
+ */
+ if (write)
+ rx_filter_core_wl(priv, (ASP_RX_FILTER_OPUT_EN |
+ ASP_RX_FILTER_LNR_MD |
+ ASP_RX_FILTER_GEN_WK_EN |
+ ASP_RX_FILTER_NT_FLT_EN),
+ ASP_RX_FILTER_BLK_CTRL);
+}
+
+void bcmasp_netfilt_get_all_active(struct bcmasp_intf *intf, u32 *rule_locs,
+ u32 *rule_cnt)
+{
+ struct bcmasp_priv *priv = intf->parent;
+ int j = 0, i;
+
+ for (i = 0; i < NUM_NET_FILTERS; i++) {
+ if (!priv->net_filters[i].claimed ||
+ priv->net_filters[i].port != intf->port)
+ continue;
+
+ if (i > 0 && (i % 2) &&
+ priv->net_filters[i].wake_filter &&
+ priv->net_filters[i - 1].wake_filter)
+ continue;
+
+ rule_locs[j++] = priv->net_filters[i].fs.location;
+ }
+
+ *rule_cnt = j;
+}
+
+int bcmasp_netfilt_get_active(struct bcmasp_intf *intf)
+{
+ struct bcmasp_priv *priv = intf->parent;
+ int cnt = 0, i;
+
+ for (i = 0; i < NUM_NET_FILTERS; i++) {
+ if (!priv->net_filters[i].claimed ||
+ priv->net_filters[i].port != intf->port)
+ continue;
+
+ /* Skip over a wake filter pair */
+ if (i > 0 && (i % 2) &&
+ priv->net_filters[i].wake_filter &&
+ priv->net_filters[i - 1].wake_filter)
+ continue;
+
+ cnt++;
+ }
+
+ return cnt;
+}
+
+bool bcmasp_netfilt_check_dup(struct bcmasp_intf *intf,
+ struct ethtool_rx_flow_spec *fs)
+{
+ struct bcmasp_priv *priv = intf->parent;
+ struct ethtool_rx_flow_spec *cur;
+ size_t fs_size = 0;
+ int i;
+
+ for (i = 0; i < NUM_NET_FILTERS; i++) {
+ if (!priv->net_filters[i].claimed ||
+ priv->net_filters[i].port != intf->port)
+ continue;
+
+ cur = &priv->net_filters[i].fs;
+
+ if (cur->flow_type != fs->flow_type ||
+ cur->ring_cookie != fs->ring_cookie)
+ continue;
+
+ switch (fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+ case ETHER_FLOW:
+ fs_size = sizeof(struct ethhdr);
+ break;
+ case IP_USER_FLOW:
+ fs_size = sizeof(struct ethtool_usrip4_spec);
+ break;
+ case TCP_V6_FLOW:
+ case UDP_V6_FLOW:
+ fs_size = sizeof(struct ethtool_tcpip6_spec);
+ break;
+ case TCP_V4_FLOW:
+ case UDP_V4_FLOW:
+ fs_size = sizeof(struct ethtool_tcpip4_spec);
+ break;
+ default:
+ continue;
+ }
+
+ if (memcmp(&cur->h_u, &fs->h_u, fs_size) ||
+ memcmp(&cur->m_u, &fs->m_u, fs_size))
+ continue;
+
+ if (cur->flow_type & FLOW_EXT) {
+ if (cur->h_ext.vlan_etype != fs->h_ext.vlan_etype ||
+ cur->m_ext.vlan_etype != fs->m_ext.vlan_etype ||
+ cur->h_ext.vlan_tci != fs->h_ext.vlan_tci ||
+ cur->m_ext.vlan_tci != fs->m_ext.vlan_tci ||
+ cur->h_ext.data[0] != fs->h_ext.data[0])
+ continue;
+ }
+ if (cur->flow_type & FLOW_MAC_EXT) {
+ if (memcmp(&cur->h_ext.h_dest,
+ &fs->h_ext.h_dest, ETH_ALEN) ||
+ memcmp(&cur->m_ext.h_dest,
+ &fs->m_ext.h_dest, ETH_ALEN))
+ continue;
+ }
+
+ return true;
+ }
+
+ return false;
+}
+
+/* If no network filter found, return open filter.
+ * If no more open filters return NULL
+ */
+struct bcmasp_net_filter *bcmasp_netfilt_get_init(struct bcmasp_intf *intf,
+ int loc, bool wake_filter,
+ bool init)
+{
+ struct bcmasp_net_filter *nfilter = NULL;
+ struct bcmasp_priv *priv = intf->parent;
+ int i, open_index = -1;
+
+ /* Check whether we exceed the filter table capacity */
+ if (loc != RX_CLS_LOC_ANY && loc >= NUM_NET_FILTERS)
+ return ERR_PTR(-EINVAL);
+
+ /* If the filter location is busy (already claimed) and we are initializing
+ * the filter (insertion), return a busy error code.
+ */
+ if (loc != RX_CLS_LOC_ANY && init && priv->net_filters[loc].claimed)
+ return ERR_PTR(-EBUSY);
+
+ /* We need two filters for wake-up, so we cannot use an odd filter */
+ if (wake_filter && loc != RX_CLS_LOC_ANY && (loc % 2))
+ return ERR_PTR(-EINVAL);
+
+ /* Initialize the loop index based on the desired location or from 0 */
+ i = loc == RX_CLS_LOC_ANY ? 0 : loc;
+
+ for ( ; i < NUM_NET_FILTERS; i++) {
+ /* Found matching network filter */
+ if (!init &&
+ priv->net_filters[i].claimed &&
+ priv->net_filters[i].hw_index == i &&
+ priv->net_filters[i].port == intf->port)
+ return &priv->net_filters[i];
+
+ /* If we don't need a new filter or new filter already found */
+ if (!init || open_index >= 0)
+ continue;
+
+ /* Wake filter conslidates two filters to cover more bytes
+ * Wake filter is open if...
+ * 1. It is an even filter
+ * 2. The current and next filter is not claimed
+ */
+ if (wake_filter && !(i % 2) && !priv->net_filters[i].claimed &&
+ !priv->net_filters[i + 1].claimed)
+ open_index = i;
+ else if (!priv->net_filters[i].claimed)
+ open_index = i;
+ }
+
+ if (open_index >= 0) {
+ nfilter = &priv->net_filters[open_index];
+ nfilter->claimed = true;
+ nfilter->port = intf->port;
+ nfilter->hw_index = open_index;
+ }
+
+ if (wake_filter && open_index >= 0) {
+ /* Claim next filter */
+ priv->net_filters[open_index + 1].claimed = true;
+ priv->net_filters[open_index + 1].wake_filter = true;
+ nfilter->wake_filter = true;
+ }
+
+ return nfilter ? nfilter : ERR_PTR(-EINVAL);
+}
+
+void bcmasp_netfilt_release(struct bcmasp_intf *intf,
+ struct bcmasp_net_filter *nfilt)
+{
+ struct bcmasp_priv *priv = intf->parent;
+
+ if (nfilt->wake_filter) {
+ memset(&priv->net_filters[nfilt->hw_index + 1], 0,
+ sizeof(struct bcmasp_net_filter));
+ }
+
+ memset(nfilt, 0, sizeof(struct bcmasp_net_filter));
+}
+
static void bcmasp_addr_to_uint(unsigned char *addr, u32 *high, u32 *low)
{
*high = (u32)(addr[0] << 8 | addr[1]);
@@ -330,6 +921,9 @@ static void bcmasp_core_init_filters(struct bcmasp_priv *priv)
priv->mda_filters[i].en = 0;
}

+ for (i = 0; i < NUM_NET_FILTERS; i++)
+ rx_filter_core_wl(priv, 0x0, ASP_RX_FILTER_NET_CFG(i));
+
/* Top level filter enable bit should be enabled at all times, set
* GEN_WAKE_CLEAR to clear the network filter wake-up which would
* otherwise be sticky
@@ -655,6 +1249,7 @@ static int bcmasp_probe(struct platform_device *pdev)
spin_lock_init(&priv->mda_lock);
spin_lock_init(&priv->clk_lock);
mutex_init(&priv->wol_lock);
+ mutex_init(&priv->net_lock);

pdata = device_get_match_data(&pdev->dev);
if (!pdata)
diff --git a/drivers/net/ethernet/broadcom/asp2/bcmasp.h b/drivers/net/ethernet/broadcom/asp2/bcmasp.h
index 7034418101a4..521c9b4b7288 100644
--- a/drivers/net/ethernet/broadcom/asp2/bcmasp.h
+++ b/drivers/net/ethernet/broadcom/asp2/bcmasp.h
@@ -106,6 +106,14 @@
#define ASP_RX_FILTER_NET_OFFSET_L3_1(val) ((val) << 16)
#define ASP_RX_FILTER_NET_OFFSET_L4(val) ((val) << 24)

+enum asp_rx_net_filter_block {
+ ASP_RX_FILTER_NET_L2 = 0,
+ ASP_RX_FILTER_NET_L3_0,
+ ASP_RX_FILTER_NET_L3_1,
+ ASP_RX_FILTER_NET_L4,
+ ASP_RX_FILTER_NET_BLOCK_MAX
+};
+
#define ASP_EDPKT_OFFSET 0x9c000
#define ASP_EDPKT_ENABLE 0x4
#define ASP_EDPKT_ENABLE_EN BIT(0)
@@ -307,6 +315,17 @@ struct bcmasp_intf {
unsigned int wol_irq_enabled:1;
};

+#define NUM_NET_FILTERS 32
+struct bcmasp_net_filter {
+ struct ethtool_rx_flow_spec fs;
+
+ bool claimed;
+ bool wake_filter;
+
+ int port;
+ unsigned int hw_index;
+};
+
#define NUM_MDA_FILTERS 32
struct bcmasp_mda_filter {
/* Current owner of this filter */
@@ -362,6 +381,11 @@ struct bcmasp_priv {

/* Protects accesses to ASP_CTRL_CLOCK_CTRL */
spinlock_t clk_lock;
+
+ struct bcmasp_net_filter net_filters[NUM_NET_FILTERS];
+
+ /* Network filter lock */
+ struct mutex net_lock;
};

static inline unsigned long bcmasp_intf_rx_desc_read(struct bcmasp_intf *intf)
@@ -519,4 +543,20 @@ void bcmasp_disable_all_filters(struct bcmasp_intf *intf);

void bcmasp_core_clock_set_intf(struct bcmasp_intf *intf, bool en);

+struct bcmasp_net_filter *bcmasp_netfilt_get_init(struct bcmasp_intf *intf,
+ int loc, bool wake_filter,
+ bool init);
+
+bool bcmasp_netfilt_check_dup(struct bcmasp_intf *intf,
+ struct ethtool_rx_flow_spec *fs);
+
+void bcmasp_netfilt_release(struct bcmasp_intf *intf,
+ struct bcmasp_net_filter *nfilt);
+
+int bcmasp_netfilt_get_active(struct bcmasp_intf *intf);
+
+void bcmasp_netfilt_get_all_active(struct bcmasp_intf *intf, u32 *rule_locs,
+ u32 *rule_cnt);
+
+void bcmasp_netfilt_suspend(struct bcmasp_intf *intf);
#endif
diff --git a/drivers/net/ethernet/broadcom/asp2/bcmasp_ethtool.c b/drivers/net/ethernet/broadcom/asp2/bcmasp_ethtool.c
index ae24a1f74d49..eddd1c43f00e 100644
--- a/drivers/net/ethernet/broadcom/asp2/bcmasp_ethtool.c
+++ b/drivers/net/ethernet/broadcom/asp2/bcmasp_ethtool.c
@@ -30,7 +30,7 @@ static void bcmasp_set_msglevel(struct net_device *dev, u32 level)
intf->msg_enable = level;
}

-#define BCMASP_SUPPORTED_WAKE (WAKE_MAGIC | WAKE_MAGICSECURE)
+#define BCMASP_SUPPORTED_WAKE (WAKE_MAGIC | WAKE_MAGICSECURE | WAKE_FILTER)
static void bcmasp_get_wol(struct net_device *dev, struct ethtool_wolinfo *wol)
{
struct bcmasp_intf *intf = netdev_priv(dev);
@@ -64,6 +64,133 @@ static int bcmasp_set_wol(struct net_device *dev, struct ethtool_wolinfo *wol)
return 0;
}

+static int bcmasp_flow_insert(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+ struct bcmasp_intf *intf = netdev_priv(dev);
+ struct bcmasp_net_filter *nfilter;
+ u32 loc = cmd->fs.location;
+ bool wake = false;
+
+ if (cmd->fs.ring_cookie == RX_CLS_FLOW_WAKE)
+ wake = true;
+
+ /* Currently only supports WAKE filters */
+ if (!wake)
+ return -EOPNOTSUPP;
+
+ switch (cmd->fs.flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+ case ETHER_FLOW:
+ case IP_USER_FLOW:
+ case TCP_V4_FLOW:
+ case UDP_V4_FLOW:
+ case TCP_V6_FLOW:
+ case UDP_V6_FLOW:
+ break;
+ default:
+ return -EOPNOTSUPP;
+ }
+
+ /* Check if filter already exists */
+ if (bcmasp_netfilt_check_dup(intf, &cmd->fs))
+ return -EINVAL;
+
+ nfilter = bcmasp_netfilt_get_init(intf, loc, wake, true);
+ if (IS_ERR(nfilter))
+ return PTR_ERR(nfilter);
+
+ /* Return the location where we did insert the filter */
+ cmd->fs.location = nfilter->hw_index;
+ memcpy(&nfilter->fs, &cmd->fs, sizeof(struct ethtool_rx_flow_spec));
+
+ /* Since we only support wake filters, defer register programming till
+ * suspend time.
+ */
+ return 0;
+}
+
+static int bcmasp_flow_delete(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+ struct bcmasp_intf *intf = netdev_priv(dev);
+ struct bcmasp_net_filter *nfilter;
+
+ nfilter = bcmasp_netfilt_get_init(intf, cmd->fs.location, false, false);
+ if (IS_ERR(nfilter))
+ return PTR_ERR(nfilter);
+
+ bcmasp_netfilt_release(intf, nfilter);
+
+ return 0;
+}
+
+static int bcmasp_flow_get(struct bcmasp_intf *intf, struct ethtool_rxnfc *cmd)
+{
+ struct bcmasp_net_filter *nfilter;
+
+ nfilter = bcmasp_netfilt_get_init(intf, cmd->fs.location, false, false);
+ if (IS_ERR(nfilter))
+ return PTR_ERR(nfilter);
+
+ memcpy(&cmd->fs, &nfilter->fs, sizeof(nfilter->fs));
+
+ cmd->data = NUM_NET_FILTERS;
+
+ return 0;
+}
+
+static int bcmasp_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+ struct bcmasp_intf *intf = netdev_priv(dev);
+ int ret = -EOPNOTSUPP;
+
+ mutex_lock(&intf->parent->net_lock);
+
+ switch (cmd->cmd) {
+ case ETHTOOL_SRXCLSRLINS:
+ ret = bcmasp_flow_insert(dev, cmd);
+ break;
+ case ETHTOOL_SRXCLSRLDEL:
+ ret = bcmasp_flow_delete(dev, cmd);
+ break;
+ default:
+ break;
+ }
+
+ mutex_unlock(&intf->parent->net_lock);
+
+ return ret;
+}
+
+static int bcmasp_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd,
+ u32 *rule_locs)
+{
+ struct bcmasp_intf *intf = netdev_priv(dev);
+ int err = 0;
+
+ mutex_lock(&intf->parent->net_lock);
+
+ switch (cmd->cmd) {
+ case ETHTOOL_GRXCLSRLCNT:
+ cmd->rule_cnt = bcmasp_netfilt_get_active(intf);
+ /* We support specifying rule locations */
+ cmd->data |= RX_CLS_LOC_SPECIAL;
+ break;
+ case ETHTOOL_GRXCLSRULE:
+ err = bcmasp_flow_get(intf, cmd);
+ break;
+ case ETHTOOL_GRXCLSRLALL:
+ bcmasp_netfilt_get_all_active(intf, rule_locs, &cmd->rule_cnt);
+ cmd->data = NUM_NET_FILTERS;
+ break;
+ default:
+ err = -EOPNOTSUPP;
+ break;
+ }
+
+ mutex_unlock(&intf->parent->net_lock);
+
+ return err;
+}
+
const struct ethtool_ops bcmasp_ethtool_ops = {
.get_drvinfo = bcmasp_get_drvinfo,
.get_link = ethtool_op_get_link,
@@ -73,4 +200,6 @@ const struct ethtool_ops bcmasp_ethtool_ops = {
.set_msglevel = bcmasp_set_msglevel,
.get_wol = bcmasp_get_wol,
.set_wol = bcmasp_set_wol,
+ .get_rxnfc = bcmasp_get_rxnfc,
+ .set_rxnfc = bcmasp_set_rxnfc,
};
diff --git a/drivers/net/ethernet/broadcom/asp2/bcmasp_intf.c b/drivers/net/ethernet/broadcom/asp2/bcmasp_intf.c
index 35b9d6390db3..0c59f4025965 100644
--- a/drivers/net/ethernet/broadcom/asp2/bcmasp_intf.c
+++ b/drivers/net/ethernet/broadcom/asp2/bcmasp_intf.c
@@ -1320,6 +1320,9 @@ static void bcmasp_suspend_to_wol(struct bcmasp_intf *intf)
}
umac_wl(intf, reg, UMC_MPD_CTRL);

+ if (intf->wolopts & WAKE_FILTER)
+ bcmasp_netfilt_suspend(intf);
+
/* UniMAC receive needs to be turned on */
umac_enable_set(intf, UMC_CMD_RX_EN, 1);

--
2.7.4

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature