Reduce the number of separate SPI core requests when setting the UINC bit in the TEF FIFO, and instead batch them up into a single SPI core request. Tested-by: Thomas Kopp <thomas.kopp@xxxxxxxxxxxxx> Signed-off-by: Marc Kleine-Budde <mkl@xxxxxxxxxxxxxx> --- .../net/can/spi/mcp251xfd/mcp251xfd-core.c | 64 +++++++++++++++---- drivers/net/can/spi/mcp251xfd/mcp251xfd.h | 3 + 2 files changed, 55 insertions(+), 12 deletions(-) diff --git a/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c b/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c index 137c827e97e0..e2c2331cb2fa 100644 --- a/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c +++ b/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c @@ -340,6 +340,23 @@ static void mcp251xfd_ring_init(struct mcp251xfd_priv *priv) tef_ring->head = 0; tef_ring->tail = 0; + /* FIFO increment TEF tail pointer */ + addr = MCP251XFD_REG_TEFCON; + val = MCP251XFD_REG_TEFCON_UINC; + len = mcp251xfd_cmd_prepare_write_reg(priv, &tef_ring->uinc_buf, + addr, val, val); + + for (j = 0; j < ARRAY_SIZE(tef_ring->uinc_xfer); j++) { + struct spi_transfer *xfer; + + xfer = &tef_ring->uinc_xfer[j]; + xfer->tx_buf = &tef_ring->uinc_buf; + xfer->len = len; + xfer->cs_change = 1; + xfer->cs_change_delay.value = 0; + xfer->cs_change_delay.unit = SPI_DELAY_UNIT_NSECS; + } + /* TX */ tx_ring = priv->tx; tx_ring->head = 0; @@ -1231,10 +1248,8 @@ static int mcp251xfd_handle_tefif_one(struct mcp251xfd_priv *priv, const struct mcp251xfd_hw_tef_obj *hw_tef_obj) { - struct mcp251xfd_tx_ring *tx_ring = priv->tx; struct net_device_stats *stats = &priv->ndev->stats; u32 seq, seq_masked, tef_tail_masked; - int err; seq = FIELD_GET(MCP251XFD_OBJ_FLAGS_SEQ_MCP2518FD_MASK, hw_tef_obj->flags); @@ -1255,18 +1270,9 @@ mcp251xfd_handle_tefif_one(struct mcp251xfd_priv *priv, mcp251xfd_get_tef_tail(priv), hw_tef_obj->ts); stats->tx_packets++; - - /* finally increment the TEF pointer */ - err = regmap_update_bits(priv->map_reg, MCP251XFD_REG_TEFCON, - GENMASK(15, 8), - MCP251XFD_REG_TEFCON_UINC); - if (err) - return err; - priv->tef->tail++; - tx_ring->tail++; - return mcp251xfd_check_tef_tail(priv); + return 0; } static int mcp251xfd_tef_ring_update(struct mcp251xfd_priv *priv) @@ -1353,6 +1359,40 @@ static int mcp251xfd_handle_tefif(struct mcp251xfd_priv *priv) } out_netif_wake_queue: + len = i; /* number of handled goods TEFs */ + if (len) { + struct mcp251xfd_tef_ring *ring = priv->tef; + struct mcp251xfd_tx_ring *tx_ring = priv->tx; + struct spi_transfer *last_xfer; + + tx_ring->tail += len; + + /* Increment the TEF FIFO tail pointer 'len' times in + * a single SPI message. + */ + + /* Note: + * + * "cs_change == 1" on the last transfer results in an + * active chip select after the complete SPI + * message. This causes the controller to interpret + * the next register access as data. Temporary set + * "cs_change" of the last transfer to "0" to properly + * deactivate the chip select at the end of the + * message. + */ + last_xfer = &ring->uinc_xfer[len - 1]; + last_xfer->cs_change = 0; + err = spi_sync_transfer(priv->spi, ring->uinc_xfer, len); + last_xfer->cs_change = 1; + if (err) + return err; + + err = mcp251xfd_check_tef_tail(priv); + if (err) + return err; + } + mcp251xfd_ecc_tefif_successful(priv); if (mcp251xfd_get_tx_free(priv->tx)) { diff --git a/drivers/net/can/spi/mcp251xfd/mcp251xfd.h b/drivers/net/can/spi/mcp251xfd/mcp251xfd.h index 299dbf72e24b..cb6398c2a560 100644 --- a/drivers/net/can/spi/mcp251xfd/mcp251xfd.h +++ b/drivers/net/can/spi/mcp251xfd/mcp251xfd.h @@ -504,6 +504,9 @@ struct mcp251xfd_tef_ring { /* u8 obj_num equals tx_ring->obj_num */ /* u8 obj_size equals sizeof(struct mcp251xfd_hw_tef_obj) */ + + union mcp251xfd_write_reg_buf uinc_buf; + struct spi_transfer uinc_xfer[MCP251XFD_TX_OBJ_NUM_MAX]; }; struct mcp251xfd_tx_ring { -- 2.29.2