wifi: rtw89: pci: add new RX ring design to determine full RX ring efficiently

To make hardware efficient to determine if RX ring is full, introduce new
design that checks if reading and writing indices are equal. Comparing
to old design, initial indices of both reading and writing indices are 0
that means empty, and hardware checks full by "writing index + 1 ==
reading index". The "+1" has extra cost for hardware, so new design is
to avoid this.

Take ring size is 256 as an example, the initial reading and writing
indices are 255 and 0 respectively; the initial values mean empty. If two
indices are the same, for example 5 and 5, it means ring is full.

   wp       rp       used_cnt        state
   255      0        0               initial (ring is empty)
   255      1        1               receive 1st packet
   255      2        2               receive 2nd packet
   0        2        1               driver read 1st packet
   1        2        0               driver read 2nd packet (ring is empty)
            :
   5        5        255             ring is full

Note: 'rp' is hardware writing index
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20231026120049.9187-4-pkshih@realtek.com

drivers/net/wireless/realtek/rtw89/pci.c

@@ -41,6 +41,7 @@ static u32 rtw89_pci_dma_recalc(struct rtw89_dev *rtwdev,
 				struct rtw89_pci_dma_ring *bd_ring,
 				u32 cur_idx, bool tx)
 {
+	const struct rtw89_pci_info *info = rtwdev->pci_info;
 	u32 cnt, cur_rp, wp, rp, len;
 
 	rp = bd_ring->rp;
@@ -48,10 +49,14 @@ static u32 rtw89_pci_dma_recalc(struct rtw89_dev *rtwdev,
 	len = bd_ring->len;
 
 	cur_rp = FIELD_GET(TXBD_HW_IDX_MASK, cur_idx);
-	if (tx)
+	if (tx) {
 		cnt = cur_rp >= rp ? cur_rp - rp : len - (rp - cur_rp);
-	else
+	} else {
+		if (info->rx_ring_eq_is_full)
+			wp += 1;
+
 		cnt = cur_rp >= wp ? cur_rp - wp : len - (wp - cur_rp);
+	}
 
 	bd_ring->rp = cur_rp;
 
@@ -226,6 +231,21 @@ rtw89_skb_put_rx_data(struct rtw89_dev *rtwdev, bool fs, bool ls,
 	return true;
 }
 
+static u32 rtw89_pci_get_rx_skb_idx(struct rtw89_dev *rtwdev,
+				    struct rtw89_pci_dma_ring *bd_ring)
+{
+	const struct rtw89_pci_info *info = rtwdev->pci_info;
+	u32 wp = bd_ring->wp;
+
+	if (!info->rx_ring_eq_is_full)
+		return wp;
+
+	if (++wp >= bd_ring->len)
+		wp = 0;
+
+	return wp;
+}
+
 static u32 rtw89_pci_rxbd_deliver_skbs(struct rtw89_dev *rtwdev,
 				       struct rtw89_pci_rx_ring *rx_ring)
 {
@@ -235,12 +255,14 @@ static u32 rtw89_pci_rxbd_deliver_skbs(struct rtw89_dev *rtwdev,
 	struct sk_buff *new = rx_ring->diliver_skb;
 	struct sk_buff *skb;
 	u32 rxinfo_size = sizeof(struct rtw89_pci_rxbd_info);
+	u32 skb_idx;
 	u32 offset;
 	u32 cnt = 1;
 	bool fs, ls;
 	int ret;
 
-	skb = rx_ring->buf[bd_ring->wp];
+	skb_idx = rtw89_pci_get_rx_skb_idx(rtwdev, bd_ring);
+	skb = rx_ring->buf[skb_idx];
 	rtw89_pci_sync_skb_for_cpu(rtwdev, skb);
 
 	ret = rtw89_pci_rxbd_info_update(rtwdev, skb);
@@ -525,10 +547,12 @@ static u32 rtw89_pci_release_tx_skbs(struct rtw89_dev *rtwdev,
 	u32 cnt = 0;
 	u32 rpp_size = sizeof(struct rtw89_pci_rpp_fmt);
 	u32 rxinfo_size = sizeof(struct rtw89_pci_rxbd_info);
+	u32 skb_idx;
 	u32 offset;
 	int ret;
 
-	skb = rx_ring->buf[bd_ring->wp];
+	skb_idx = rtw89_pci_get_rx_skb_idx(rtwdev, bd_ring);
+	skb = rx_ring->buf[skb_idx];
 	rtw89_pci_sync_skb_for_cpu(rtwdev, skb);
 
 	ret = rtw89_pci_rxbd_info_update(rtwdev, skb);
@@ -1454,6 +1478,7 @@ static void rtw89_pci_reset_trx_rings(struct rtw89_dev *rtwdev)
 	struct rtw89_pci_dma_ring *bd_ring;
 	const struct rtw89_pci_bd_ram *bd_ram;
 	u32 addr_num;
+	u32 addr_idx;
 	u32 addr_bdram;
 	u32 addr_desa_l;
 	u32 val32;
@@ -1487,14 +1512,21 @@ static void rtw89_pci_reset_trx_rings(struct rtw89_dev *rtwdev)
 		rx_ring = &rtwpci->rx_rings[i];
 		bd_ring = &rx_ring->bd_ring;
 		addr_num = bd_ring->addr.num;
+		addr_idx = bd_ring->addr.idx;
 		addr_desa_l = bd_ring->addr.desa_l;
-		bd_ring->wp = 0;
+		if (info->rx_ring_eq_is_full)
+			bd_ring->wp = bd_ring->len - 1;
+		else
+			bd_ring->wp = 0;
 		bd_ring->rp = 0;
 		rx_ring->diliver_skb = NULL;
 		rx_ring->diliver_desc.ready = false;
 
 		rtw89_write16(rtwdev, addr_num, bd_ring->len);
 		rtw89_write32(rtwdev, addr_desa_l, bd_ring->dma);
+
+		if (info->rx_ring_eq_is_full)
+			rtw89_write16(rtwdev, addr_idx, bd_ring->wp);
 	}
 }
 
@@ -3060,6 +3092,7 @@ static int rtw89_pci_alloc_rx_ring(struct rtw89_dev *rtwdev,
 				   struct rtw89_pci_rx_ring *rx_ring,
 				   u32 desc_size, u32 len, u32 rxch)
 {
+	const struct rtw89_pci_info *info = rtwdev->pci_info;
 	const struct rtw89_pci_ch_dma_addr *rxch_addr;
 	struct sk_buff *skb;
 	u8 *head;
@@ -3086,7 +3119,10 @@ static int rtw89_pci_alloc_rx_ring(struct rtw89_dev *rtwdev,
 	rx_ring->bd_ring.len = len;
 	rx_ring->bd_ring.desc_size = desc_size;
 	rx_ring->bd_ring.addr = *rxch_addr;
-	rx_ring->bd_ring.wp = 0;
+	if (info->rx_ring_eq_is_full)
+		rx_ring->bd_ring.wp = len - 1;
+	else
+		rx_ring->bd_ring.wp = 0;
 	rx_ring->bd_ring.rp = 0;
 	rx_ring->buf_sz = buf_sz;
 	rx_ring->diliver_skb = NULL;

drivers/net/wireless/realtek/rtw89/pci.h

@@ -937,6 +937,7 @@ struct rtw89_pci_info {
 	enum mac_ax_pcie_func_ctrl autok_en;
 	enum mac_ax_pcie_func_ctrl io_rcy_en;
 	enum mac_ax_io_rcy_tmr io_rcy_tmr;
+	bool rx_ring_eq_is_full;
 
 	u32 init_cfg_reg;
 	u32 txhci_en_bit;

drivers/net/wireless/realtek/rtw89/rtw8852ae.c

@@ -24,6 +24,7 @@ static const struct rtw89_pci_info rtw8852a_pci_info = {
 	.autok_en		= MAC_AX_PCIE_DISABLE,
 	.io_rcy_en		= MAC_AX_PCIE_DISABLE,
 	.io_rcy_tmr		= MAC_AX_IO_RCY_ANA_TMR_6MS,
+	.rx_ring_eq_is_full	= false,
 
 	.init_cfg_reg		= R_AX_PCIE_INIT_CFG1,
 	.txhci_en_bit		= B_AX_TXHCI_EN,

drivers/net/wireless/realtek/rtw89/rtw8852be.c

@@ -24,6 +24,7 @@ static const struct rtw89_pci_info rtw8852b_pci_info = {
 	.autok_en		= MAC_AX_PCIE_DISABLE,
 	.io_rcy_en		= MAC_AX_PCIE_DISABLE,
 	.io_rcy_tmr		= MAC_AX_IO_RCY_ANA_TMR_6MS,
+	.rx_ring_eq_is_full	= false,
 
 	.init_cfg_reg		= R_AX_PCIE_INIT_CFG1,
 	.txhci_en_bit		= B_AX_TXHCI_EN,

drivers/net/wireless/realtek/rtw89/rtw8852ce.c

@@ -33,6 +33,7 @@ static const struct rtw89_pci_info rtw8852c_pci_info = {
 	.autok_en		= MAC_AX_PCIE_DISABLE,
 	.io_rcy_en		= MAC_AX_PCIE_ENABLE,
 	.io_rcy_tmr		= MAC_AX_IO_RCY_ANA_TMR_6MS,
+	.rx_ring_eq_is_full	= false,
 
 	.init_cfg_reg		= R_AX_HAXI_INIT_CFG1,
 	.txhci_en_bit		= B_AX_TXHCI_EN_V1,

drivers/net/wireless/realtek/rtw89/rtw8922ae.c

@@ -23,6 +23,7 @@ static const struct rtw89_pci_info rtw8922a_pci_info = {
 	.autok_en		= MAC_AX_PCIE_DISABLE,
 	.io_rcy_en		= MAC_AX_PCIE_ENABLE,
 	.io_rcy_tmr		= MAC_AX_IO_RCY_ANA_TMR_DEF,
+	.rx_ring_eq_is_full	= true,
 
 	.init_cfg_reg		= R_BE_HAXI_INIT_CFG1,
 	.txhci_en_bit		= B_BE_TXDMA_EN,