mcp251x.c : patch to avoid repeated frame bug
Subsystems (2)
Hello,
The MCP2515 has a silicon bug causing repeated frame transmission, see section 5 of MCP2515 Rev. B Silicon Errata Revision G (March 2007).
Basically, setting TXBnCTRL.TXREQ in either SPI mode (00 or 11) will eventually cause the bug. The workaround proposed by Microchip is to use mode 00 and send an RTS command on the SPI bus to initiate the transmission. Accordingly, I have modified the transmission routine as follows in file mcp251x.c:
net/can/mcp251x.c renamed from net/can/mcp251x.c-original +21 -1
@@ -83,6 +83,10 @@#define INSTRUCTION_LOAD_TXB (n) (0x40 + 2 * (n))#define INSTRUCTION_READ_RXB (n) (((n) == 0) ? 0x90 : 0x94)#define INSTRUCTION_RESET 0xC0 +#define RTS_TXB0 0x01 +#define RTS_TXB1 0x02 +#define RTS_TXB2 0x04 +#define INSTRUCTION_RTS (txb) (0x80 | ((txb) & 0x07))/* MPC251x registers */ #define CANSTAT 0x0e
@@ -394,9 +398,20 @@ } } + +/* +* CAN frame is now sent using instruction RTS over SPI to work around +* repeated frame problem as per MCP2515 Rev B. Silicon Errata revision G +* (March 2007) about repeated frames (see section 5 of document). +* Note that repeated frames can still occur if using SPI mode 11. It seems +* however that spi mode is forcibly set to 0, so it looks safe. +* +*/ static void mcp251x_hw_tx (struct spi_device *spi,struct can_frame *frame,int tx_buf_idx) { +struct mcp251x_priv *priv =dev_get_drvdata (&spi->dev); +int ret;u32 sid, eid, exide, rtr;u8 buf[SPI_TRANSFER_BUF_LEN ];
@@ -418,7 +433,12 @@ buf[TXBDLC_OFF ] = (rtr <<DLC_RTR_SHIFT ) | frame->can_dlc;memcpy (buf +TXBDAT_OFF , frame->data, frame->can_dlc);mcp251x_hw_tx_frame (spi, buf, frame->can_dlc, tx_buf_idx); -mcp251x_write_reg (spi,TXBCTRL (tx_buf_idx),TXBCTRL_TXREQ ); + + priv->spi_tx_buf[0 ] =INSTRUCTION_RTS (1 << tx_buf_idx); + ret =spi_write (spi, priv->spi_tx_buf,1 ); +if (ret) { +dev_err (&spi->dev,"RTS failed: ret = %d\n" , ret); + } } static void mcp251x_hw_rx_frame(struct spi_device *spi, u8 *buf,
Would it be possible to integrate this patch into mainstream mcp251x.c?
Kind Regards,
Benoît.
Marc Kleine-Budde
maintainerQuoted 10 lines · Benoit
Hello,
The MCP2515 has a silicon bug causing repeated frame transmission, see section 5 of MCP2515 Rev. B Silicon Errata Revision G (March 2007).
Basically, setting TXBnCTRL.TXREQ in either SPI mode (00 or 11) will eventually cause the bug. The workaround proposed by Microchip is to use mode 00 and send an RTS command on the SPI bus to initiate the transmission. Accordingly, I have modified the transmission routine as follows in file
mcp251x.c:
I've some questions about the code:
net/can/mcp251x.c line 88
Quoted 10 lines
---
drivers/net/can/mcp251x.c-original2012-08-26 15:00:28.000000000
+0200
+++drivers/net/can/mcp251x.c2012-08-26 20:42:58.000000000 +0200net/can/mcp251x.c renamed from net/can/mcp251x.c-original +3
-0@@ -83,6 +83,10 @@#define INSTRUCTION_LOAD_TXB (n) (0x40 + 2 * (n))#define INSTRUCTION_READ_RXB (n) (((n) == 0) ? 0x90 : 0x94)#define INSTRUCTION_RESET 0xC0 +#define RTS_TXB0 0x01 +#define RTS_TXB1 0x02 +#define RTS_TXB2 0x04
^^^^
Please use a space between define and RTS_
Quoted 14 lines
net/can/mcp251x.c renamed from net/can/mcp251x.c-original +7
-0+#define INSTRUCTION_RTS (txb) (0x80 | ((txb) & 0x07))/* MPC251x registers */ #define CANSTAT 0x0e@@ -394,9 +398,20 @@ } } + +/* +* CAN frame is now sent using instruction RTS over SPI to work around +* repeated frame problem as per MCP2515 Rev B. Silicon Errata revision G +* (March 2007) about repeated frames (see section 5 of document). +* Note that repeated frames can still occur if using SPI mode 11. It
Quoted 4 lines
seems
+ * however that spi mode is forcibly set to 0, so it looks safe.
+ *
+ */
I think we don't need this in the code it will go into the commit message. A short note before the priv->spi_tx_buf[0] should be enough.
Quoted 9 lines
static void mcp251x_hw_tx(struct
spi_device*spi, structcan_frame*frame,
inttx_buf_idx)
{net/can/mcp251x.c renamed from net/can/mcp251x.c-original +2
-0+struct mcp251x_priv *priv =dev_get_drvdata (&spi->dev); +int ret;u32 sid, eid, exide, rtr;u8 buf[SPI_TRANSFER_BUF_LEN ];
net/can/mcp251x.c line 438
Quoted 7 lines
net/can/mcp251x.c renamed from net/can/mcp251x.c-original +3
-1@@ -418,7 +433,12 @@ buf[TXBDLC_OFF ] = (rtr <<DLC_RTR_SHIFT ) | frame->can_dlc;memcpy (buf +TXBDAT_OFF , frame->data, frame->can_dlc);mcp251x_hw_tx_frame (spi, buf, frame->can_dlc, tx_buf_idx); -mcp251x_write_reg (spi,TXBCTRL (tx_buf_idx),TXBCTRL_TXREQ ); + + priv->spi_tx_buf[0 ] =INSTRUCTION_RTS (1 << tx_buf_idx); + ret =spi_write (spi, priv->spi_tx_buf,1 );
Why don't you use mcp251x_spi_trans() here?
Quoted 9 lines
net/can/mcp251x.c renamed from net/can/mcp251x.c-original +3
-0+if (ret) { +dev_err (&spi->dev,"RTS failed: ret = %d\n" , ret); + } } static void mcp251x_hw_rx_frame(struct spi_device *spi, u8 *buf,
Quoted 5 lines
Would it be possible to integrate this patch into mainstream
mcp251x.c?Kind Regards,
Benoît
I'll send an updated version of your proposed patch.
regards, Marc
Marc
Signature · 4 lines
Pengutronix e.K. | Marc Kleine-Budde | Industrial Linux Solutions | Phone: +49-231-2826-924 | Vertretung West/Dortmund | Fax: +49-5121-206917-5555 | Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |