octeontx2-af: Support fixed transmit scheduler topology

CN96xx initial silicon doesn't support all features pertaining to
NIX transmit scheduling and shaping.
- It supports a fixed topology of 1:1 mapped transmit
  limiters at all levels.
- Supports DWRR only at SMQ/MDQ and TL1.
- Doesn't support shaping and coloring.

This patch adds HW capability structure by which each variant
and skew of silicon can be differentiated by their supported
features. And adds support for A0 silicon's transmit scheduler
capabilities or rather limitations.

Signed-off-by: Sunil Goutham <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Author: sgoutham-marvell

drivers/net/ethernet/marvell/octeontx2/af/cgx.c

@@ -291,6 +291,35 @@ void cgx_lmac_promisc_config(int cgx_id, int lmac_id, bool enable)
 }
 EXPORT_SYMBOL(cgx_lmac_promisc_config);
 
+/* Enable or disable forwarding received pause frames to Tx block */
+void cgx_lmac_enadis_rx_pause_fwding(void *cgxd, int lmac_id, bool enable)
+{
+	struct cgx *cgx = cgxd;
+	u64 cfg;
+
+	if (!cgx)
+		return;
+
+	if (enable) {
+		cfg = cgx_read(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL);
+		cfg |= CGX_GMP_GMI_RXX_FRM_CTL_CTL_BCK;
+		cgx_write(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL, cfg);
+
+		cfg = cgx_read(cgx, lmac_id, CGXX_SMUX_RX_FRM_CTL);
+		cfg |= CGX_SMUX_RX_FRM_CTL_CTL_BCK;
+		cgx_write(cgx, lmac_id,	CGXX_SMUX_RX_FRM_CTL, cfg);
+	} else {
+		cfg = cgx_read(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL);
+		cfg &= ~CGX_GMP_GMI_RXX_FRM_CTL_CTL_BCK;
+		cgx_write(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL, cfg);
+
+		cfg = cgx_read(cgx, lmac_id, CGXX_SMUX_RX_FRM_CTL);
+		cfg &= ~CGX_SMUX_RX_FRM_CTL_CTL_BCK;
+		cgx_write(cgx, lmac_id,	CGXX_SMUX_RX_FRM_CTL, cfg);
+	}
+}
+EXPORT_SYMBOL(cgx_lmac_enadis_rx_pause_fwding);
+
 int cgx_get_rx_stats(void *cgxd, int lmac_id, int idx, u64 *rx_stat)
 {
 	struct cgx *cgx = cgxd;
@@ -331,6 +360,27 @@ int cgx_lmac_rx_tx_enable(void *cgxd, int lmac_id, bool enable)
 }
 EXPORT_SYMBOL(cgx_lmac_rx_tx_enable);
 
+int cgx_lmac_tx_enable(void *cgxd, int lmac_id, bool enable)
+{
+	struct cgx *cgx = cgxd;
+	u64 cfg, last;
+
+	if (!cgx || lmac_id >= cgx->lmac_count)
+		return -ENODEV;
+
+	cfg = cgx_read(cgx, lmac_id, CGXX_CMRX_CFG);
+	last = cfg;
+	if (enable)
+		cfg |= DATA_PKT_TX_EN;
+	else
+		cfg &= ~DATA_PKT_TX_EN;
+
+	if (cfg != last)
+		cgx_write(cgx, lmac_id, CGXX_CMRX_CFG, cfg);
+	return !!(last & DATA_PKT_TX_EN);
+}
+EXPORT_SYMBOL(cgx_lmac_tx_enable);
+
 /* CGX Firmware interface low level support */
 static int cgx_fwi_cmd_send(u64 req, u64 *resp, struct lmac *lmac)
 {

drivers/net/ethernet/marvell/octeontx2/af/cgx.h

@@ -56,6 +56,11 @@
 #define CGXX_GMP_PCS_MRX_CTL		0x30000
 #define CGXX_GMP_PCS_MRX_CTL_LBK	BIT_ULL(14)
 
+#define CGXX_SMUX_RX_FRM_CTL		0x20020
+#define CGX_SMUX_RX_FRM_CTL_CTL_BCK	BIT_ULL(3)
+#define CGXX_GMP_GMI_RXX_FRM_CTL	0x38028
+#define CGX_GMP_GMI_RXX_FRM_CTL_CTL_BCK	BIT_ULL(3)
+
 #define CGX_COMMAND_REG			CGXX_SCRATCH1_REG
 #define CGX_EVENT_REG			CGXX_SCRATCH0_REG
 #define CGX_CMD_TIMEOUT			2200 /* msecs */
@@ -110,9 +115,11 @@ int cgx_lmac_evh_unregister(void *cgxd, int lmac_id);
 int cgx_get_tx_stats(void *cgxd, int lmac_id, int idx, u64 *tx_stat);
 int cgx_get_rx_stats(void *cgxd, int lmac_id, int idx, u64 *rx_stat);
 int cgx_lmac_rx_tx_enable(void *cgxd, int lmac_id, bool enable);
+int cgx_lmac_tx_enable(void *cgxd, int lmac_id, bool enable);
 int cgx_lmac_addr_set(u8 cgx_id, u8 lmac_id, u8 *mac_addr);
 u64 cgx_lmac_addr_get(u8 cgx_id, u8 lmac_id);
 void cgx_lmac_promisc_config(int cgx_id, int lmac_id, bool enable);
+void cgx_lmac_enadis_rx_pause_fwding(void *cgxd, int lmac_id, bool enable);
 int cgx_lmac_internal_loopback(void *cgxd, int lmac_id, bool enable);
 int cgx_get_link_info(void *cgxd, int lmac_id,
 		      struct cgx_link_user_info *linfo);

drivers/net/ethernet/marvell/octeontx2/af/mbox.h

@@ -127,6 +127,7 @@ M(ATTACH_RESOURCES,	0x002, attach_resources, rsrc_attach, msg_rsp)	\
 M(DETACH_RESOURCES,	0x003, detach_resources, rsrc_detach, msg_rsp)	\
 M(MSIX_OFFSET,		0x004, msix_offset, msg_req, msix_offset_rsp)	\
 M(VF_FLR,		0x006, vf_flr, msg_req, msg_rsp)		\
+M(GET_HW_CAP,		0x008, get_hw_cap, msg_req, get_hw_cap_rsp)	\
 /* CGX mbox IDs (range 0x200 - 0x3FF) */				\
 M(CGX_START_RXTX,	0x200, cgx_start_rxtx, msg_req, msg_rsp)	\
 M(CGX_STOP_RXTX,	0x201, cgx_stop_rxtx, msg_req, msg_rsp)		\
@@ -302,6 +303,12 @@ struct msix_offset_rsp {
 	u16  cptlf_msixoff[MAX_RVU_BLKLF_CNT];
 };
 
+struct get_hw_cap_rsp {
+	struct mbox_msghdr hdr;
+	u8 nix_fixed_txschq_mapping; /* Schq mapping fixed or flexible */
+	u8 nix_shaping;		     /* Is shaping and coloring supported */
+};
+
 /* CGX mbox message formats */
 
 struct cgx_stats_rsp {
@@ -514,6 +521,9 @@ struct nix_txsch_alloc_rsp {
 	/* Scheduler queue list allocated at each level */
 	u16 schq_contig_list[NIX_TXSCH_LVL_CNT][MAX_TXSCHQ_PER_FUNC];
 	u16 schq_list[NIX_TXSCH_LVL_CNT][MAX_TXSCHQ_PER_FUNC];
+	u8  aggr_level; /* Traffic aggregation scheduler level */
+	u8  aggr_lvl_rr_prio; /* Aggregation lvl's RR_PRIO config */
+	u8  link_cfg_lvl; /* LINKX_CFG CSRs mapped to TL3 or TL2's index ? */
 };
 
 struct nix_txsch_free_req {

drivers/net/ethernet/marvell/octeontx2/af/rvu.c

@@ -56,12 +56,31 @@ static char *mkex_profile; /* MKEX profile name */
 module_param(mkex_profile, charp, 0000);
 MODULE_PARM_DESC(mkex_profile, "MKEX profile name string");
 
+static void rvu_setup_hw_capabilities(struct rvu *rvu)
+{
+	struct rvu_hwinfo *hw = rvu->hw;
+
+	hw->cap.nix_tx_aggr_lvl = NIX_TXSCH_LVL_TL1;
+	hw->cap.nix_fixed_txschq_mapping = false;
+	hw->cap.nix_shaping = true;
+	hw->cap.nix_tx_link_bp = true;
+
+	if (is_rvu_96xx_B0(rvu)) {
+		hw->cap.nix_fixed_txschq_mapping = true;
+		hw->cap.nix_txsch_per_cgx_lmac = 4;
+		hw->cap.nix_txsch_per_lbk_lmac = 132;
+		hw->cap.nix_txsch_per_sdp_lmac = 76;
+		hw->cap.nix_shaping = false;
+		hw->cap.nix_tx_link_bp = false;
+	}
+}
+
 /* Poll a RVU block's register 'offset', for a 'zero'
  * or 'nonzero' at bits specified by 'mask'
  */
 int rvu_poll_reg(struct rvu *rvu, u64 block, u64 offset, u64 mask, bool zero)
 {
-	unsigned long timeout = jiffies + usecs_to_jiffies(100);
+	unsigned long timeout = jiffies + usecs_to_jiffies(10000);
 	void __iomem *reg;
 	u64 reg_val;
 
@@ -73,7 +92,6 @@ int rvu_poll_reg(struct rvu *rvu, u64 block, u64 offset, u64 mask, bool zero)
 		if (!zero && (reg_val & mask))
 			return 0;
 		usleep_range(1, 5);
-		timeout--;
 	}
 	return -EBUSY;
 }
@@ -1363,6 +1381,17 @@ int rvu_mbox_handler_vf_flr(struct rvu *rvu, struct msg_req *req,
 	return 0;
 }
 
+int rvu_mbox_handler_get_hw_cap(struct rvu *rvu, struct msg_req *req,
+				struct get_hw_cap_rsp *rsp)
+{
+	struct rvu_hwinfo *hw = rvu->hw;
+
+	rsp->nix_fixed_txschq_mapping = hw->cap.nix_fixed_txschq_mapping;
+	rsp->nix_shaping = hw->cap.nix_shaping;
+
+	return 0;
+}
+
 static int rvu_process_mbox_msg(struct otx2_mbox *mbox, int devid,
 				struct mbox_msghdr *req)
 {
@@ -2448,6 +2477,8 @@ static int rvu_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 
 	rvu_reset_all_blocks(rvu);
 
+	rvu_setup_hw_capabilities(rvu);
+
 	err = rvu_setup_hw_resources(rvu);
 	if (err)
 		goto err_release_regions;

drivers/net/ethernet/marvell/octeontx2/af/rvu.h

@@ -184,10 +184,12 @@ struct rvu_pfvf {
 struct nix_txsch {
 	struct rsrc_bmap schq;
 	u8   lvl;
-#define NIX_TXSCHQ_TL1_CFG_DONE       BIT_ULL(0)
+#define NIX_TXSCHQ_FREE		      BIT_ULL(1)
+#define NIX_TXSCHQ_CFG_DONE	      BIT_ULL(0)
 #define TXSCH_MAP_FUNC(__pfvf_map)    ((__pfvf_map) & 0xFFFF)
 #define TXSCH_MAP_FLAGS(__pfvf_map)   ((__pfvf_map) >> 16)
 #define TXSCH_MAP(__func, __flags)    (((__func) & 0xFFFF) | ((__flags) << 16))
+#define TXSCH_SET_FLAG(__pfvf_map, flag)    ((__pfvf_map) | ((flag) << 16))
 	u32  *pfvf_map;
 };
 
@@ -221,6 +223,20 @@ struct nix_hw {
 	struct nix_lso lso;
 };
 
+/* RVU block's capabilities or functionality,
+ * which vary by silicon version/skew.
+ */
+struct hw_cap {
+	/* Transmit side supported functionality */
+	u8	nix_tx_aggr_lvl; /* Tx link's traffic aggregation level */
+	u16	nix_txsch_per_cgx_lmac; /* Max Q's transmitting to CGX LMAC */
+	u16	nix_txsch_per_lbk_lmac; /* Max Q's transmitting to LBK LMAC */
+	u16	nix_txsch_per_sdp_lmac; /* Max Q's transmitting to SDP LMAC */
+	bool	nix_fixed_txschq_mapping; /* Schq mapping fixed or flexible */
+	bool	nix_shaping;		 /* Is shaping and coloring supported */
+	bool	nix_tx_link_bp;		 /* Can link backpressure TL queues ? */
+};
+
 struct rvu_hwinfo {
 	u8	total_pfs;   /* MAX RVU PFs HW supports */
 	u16	total_vfs;   /* Max RVU VFs HW supports */
@@ -232,7 +248,7 @@ struct rvu_hwinfo {
 	u8	sdp_links;
 	u8	npc_kpus;          /* No of parser units */
 
-
+	struct hw_cap    cap;
 	struct rvu_block block[BLK_COUNT]; /* Block info */
 	struct nix_hw    *nix0;
 	struct npc_pkind pkind;
@@ -317,14 +333,23 @@ static inline u64 rvupf_read64(struct rvu *rvu, u64 offset)
 	return readq(rvu->pfreg_base + offset);
 }
 
-static inline bool is_rvu_9xxx_A0(struct rvu *rvu)
+/* Silicon revisions */
+static inline bool is_rvu_96xx_A0(struct rvu *rvu)
 {
 	struct pci_dev *pdev = rvu->pdev;
 
 	return (pdev->revision == 0x00) &&
 		(pdev->subsystem_device == PCI_SUBSYS_DEVID_96XX);
 }
 
+static inline bool is_rvu_96xx_B0(struct rvu *rvu)
+{
+	struct pci_dev *pdev = rvu->pdev;
+
+	return ((pdev->revision == 0x00) || (pdev->revision == 0x01)) &&
+		(pdev->subsystem_device == PCI_SUBSYS_DEVID_96XX);
+}
+
 /* Function Prototypes
  * RVU
  */
@@ -383,6 +408,7 @@ int rvu_cgx_init(struct rvu *rvu);
 int rvu_cgx_exit(struct rvu *rvu);
 void *rvu_cgx_pdata(u8 cgx_id, struct rvu *rvu);
 int rvu_cgx_config_rxtx(struct rvu *rvu, u16 pcifunc, bool start);
+void rvu_cgx_enadis_rx_bp(struct rvu *rvu, int pf, bool enable);
 int rvu_cgx_nix_cuml_stats(struct rvu *rvu, void *cgxd, int lmac_id, int index,
 			   int rxtxflag, u64 *stat);
 /* NPA APIs */
@@ -400,6 +426,7 @@ int rvu_nix_reserve_mark_format(struct rvu *rvu, struct nix_hw *nix_hw,
 void rvu_nix_freemem(struct rvu *rvu);
 int rvu_get_nixlf_count(struct rvu *rvu);
 void rvu_nix_lf_teardown(struct rvu *rvu, u16 pcifunc, int blkaddr, int npalf);
+int nix_get_nixlf(struct rvu *rvu, u16 pcifunc, int *nixlf);
 
 /* NPC APIs */
 int rvu_npc_init(struct rvu *rvu);

drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c

@@ -348,6 +348,24 @@ int rvu_cgx_exit(struct rvu *rvu)
 	return 0;
 }
 
+void rvu_cgx_enadis_rx_bp(struct rvu *rvu, int pf, bool enable)
+{
+	u8 cgx_id, lmac_id;
+	void *cgxd;
+
+	if (!is_pf_cgxmapped(rvu, pf))
+		return;
+
+	rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id);
+	cgxd = rvu_cgx_pdata(cgx_id, rvu);
+
+	/* Set / clear CTL_BCK to control pause frame forwarding to NIX */
+	if (enable)
+		cgx_lmac_enadis_rx_pause_fwding(cgxd, lmac_id, true);
+	else
+		cgx_lmac_enadis_rx_pause_fwding(cgxd, lmac_id, false);
+}
+
 int rvu_cgx_config_rxtx(struct rvu *rvu, u16 pcifunc, bool start)
 {
 	int pf = rvu_get_pf(pcifunc);