Rework interfaces

We now use a portion of the ucode itself for version information
and to configure the base of SRAM, this will allow the driver
to use a pool allocator with less constraints.

We put the status register as part of the command register, this
speeds things up a bit and remove the need to clear it from
the host.

We also turn some SRAM stores from byte to long stores, they seem
to be faster (still room for improvement)

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

3 files changed, 122 insertions, 77 deletions

diff --git a/cf-code/cf-fsi-fw.S b/cf-code/cf-fsi-fw.S
index aa6a388..b771b79 100644
--- a/cf-code/cf-fsi-fw.S
+++ b/cf-code/cf-fsi-fw.S
@@ -161,13 +161,37 @@ _vecs:
 	.endr
 
 	/*
-	 * Main entry point
+	 * Header info
 	 */
 	.org	0x400
+_header_info:
+	.short	SYS_SIG			/* 0x00 */
+	.short	FW_VERSION		/* 0x02 */
+	.byte	API_VERSION_MAJ		/* 0x04 */
+	.byte	API_VERSION_MIN		/* 0x05 */
+	.byte	0,0			/* 0x06 pad */
+#ifdef ENABLE_TRACE
+	.long	FW_OPTION_TRACE_EN	/* 0x08 */
+#else
+	.long	0
+#endif
+
+	/*
+	 * Config area
+	*/
+	.org	0x400 + HDR_CMD_STAT_AREA
+_cmd_stat_base:
+	.long	SRAM_BASE_BE
+
+	/*
+	 * Main entry point
+	 */
+	.org	0x500
 	.global	_start
 _start:
 	/* Get base addresses */
-	movea.l	#SRAM_BASE_BE,%a1
+	lea	%pc@(_cmd_stat_base),%a0
+	movea.l	%a0@(0),%a1
 	movea.l	#GPIO_BASE,%a4
 	movea.l	%a4,%a5
 	add.l	#CLOCK_GPIO_VREG,%a5
@@ -240,18 +264,8 @@ _start:
 	/* Configure GPIOs to output */
 	bsr	config_gpio_out
 
-	/* Populate version & signature */
-	move.w	#SYS_SIG,%a1@(SYS_SIG_REG)
-	move.b	#FW_VERSION,%a1@(FW_VERS_REG)
-
-	/* This must happen last as it's the signal to the host
-	 * that we are ready
-	 */
-#ifdef ENABLE_TRACE
-	move.b	#(API_VERSION + API_VERSION_TRACE_EN),%a1@(API_VERS_REG)
-#else
-	move.b	#API_VERSION,%a1@(API_VERS_REG)
-#endif
+	/* Mark command complete, indicates we are ready */
+	move.b	#STAT_COMPLETE,%a1@(CMD_STAT_REG)
 
 	/*
 	 * Main command loop
@@ -266,7 +280,7 @@ main_loop:
 	 */
 1:	move.b	%a1@(ARB_REG),%d2
 	bne	arbitration_request
-	move.l	%a1@(CMD_REG),%d2
+	move.l	%a1@(CMD_STAT_REG),%d2
 	tst.b	%d2
 	bne	command_request
 	stop	#0x2000
@@ -301,11 +315,9 @@ command_request:
 	/* Mask interrupts */
 	move.w	#0x2007,%sr
 
-	/* Mark ourselves as sending a command */
-	move.b	#STAT_SENDING,%a1@(STAT_REG)
-
-	/* Clear command register */
-	move.b	#CMD_NONE,%a1@(CMD_REG + 3)
+	/* Clear the command/status register */
+	moveq.l	#0,%d0
+	move.l	%d0,%a1@(CMD_STAT_REG)
 
 	/* Start command ? */
 	cmpi.b	#CMD_COMMAND,%d2
@@ -320,7 +332,7 @@ command_request:
 	beq	start_idle_clocks
 
 	/* Error */
-	move.b	#STAT_ERR_INVAL_CMD,%a1@(STAT_REG)
+	move.b	#STAT_ERR_INVAL_CMD,%a1@(CMD_STAT_REG)
 	bra	main_loop
 
 	/*
@@ -389,7 +401,7 @@ start_command:
 	beq	1f
 	subq.l	#1,%d3
 	bne	0b
-	move.b	#STAT_ERR_MTOE,%a1@(STAT_REG)
+	move.b	#STAT_ERR_MTOE,%a1@(CMD_STAT_REG)
 	bra	send_delay
 
 1:	/* Got start bit, clock in slave ID and response tag */
@@ -454,7 +466,9 @@ start_command:
 	move.b	%d4,%a1@(STAT_RCRC)
 
 	/* Mark command complete */
-	move.b	#STAT_COMPLETE,%a1@(STAT_REG)
+	moveq.l	#STAT_COMPLETE,%d0
+	byterev	%d0
+	move.l	%d0,%a1@(CMD_STAT_REG)
 
 send_delay:
 	/* Send delay after every command */
@@ -485,8 +499,11 @@ start_break:
 	move.l	#FSI_POST_BREAK_CLOCKS,%d3
 	clock_out_zeros %d3
 
-	/* That's it */
-	move.b	#STAT_COMPLETE,%a1@(STAT_REG)
+	/* Mark command complete */
+	moveq.l	#STAT_COMPLETE,%d0
+	byterev	%d0
+	move.l	%d0,%a1@(CMD_STAT_REG)
+
 	bra	main_loop
 
 start_idle_clocks:
@@ -498,8 +515,11 @@ start_idle_clocks:
 	move.b	%d2,%d3
 	clock_out_zeros %d3
 
-	/* That's it */
-	move.b	#STAT_COMPLETE,%a1@(STAT_REG)
+	/* Mark command complete */
+	moveq.l	#STAT_COMPLETE,%d0
+	byterev	%d0
+	move.l	%d0,%a1@(CMD_STAT_REG)
+
 	bra	main_loop
 
 config_gpio_out:
@@ -549,7 +569,7 @@ bad_exception:
 	sub.l	%d1,%d0
 	lsr.l	#2,%d0
 	move.b	%d0,%a1@(BAD_INT_VEC)
-	move.b	#STAT_ERR_INVAL_IRQ,%a1@(STAT_REG)
+	move.b	#STAT_ERR_INVAL_IRQ,%a1@(CMD_STAT_REG)
 	halt
 
 	/* Bad exception stubs */
diff --git a/cf-fsi-fw.h b/cf-fsi-fw.h
index 4942901..0c4453f 100644
--- a/cf-fsi-fw.h
+++ b/cf-fsi-fw.h
@@ -2,16 +2,52 @@
 #define __CF_FSI_FW_H
 
 /*
- *  SRAM layout: Main part
+ * uCode file layout
+ *
+ * 0000...03ff : m68k exception vectors
+ * 0400...04ff : Header info & boot config block
+ * 0500....... : Code & stack
  */
 
-/* Command register:
+/*
+ * Header info & boot config area
+ *
+ * The Header info is built into the ucode and provide version and
+ * platform information.
+ *
+ * the Boot config needs to be adjusted by the ARM prior to starting
+ * the ucode if the Command/Status area isn't at 0x320000 in CF space
+ * (ie. beginning of SRAM).
+ */
+
+#define HDR_OFFSET	        0x400
+
+/* Info: Signature & version */
+#define HDR_SYS_SIG		0x00	/* 2 bytes system signature */
+#define  SYS_SIG_ROMULUS	0x526d		/* 'Rm' */
+#define  SYS_SIG_WITHERSPOON	0x5773		/* 'Ws' */
+#define HDR_FW_VERS		0x02	/* 2 bytes Major.Minor */
+#define HDR_API_VERS		0x04	/* 2 bytes Major.Minor */
+#define  API_VERSION_MAJ	1	/* Current version */
+#define  API_VERSION_MIN	1
+#define HDR_FW_OPTIONS		0x08	/* 4 bytes option flags */
+#define   FW_OPTION_TRACE_EN	0x00000001	/* FW tracing enabled */
+
+/* Boot Config: Address of Command/Status area */
+#define HDR_CMD_STAT_AREA	0x80	/* 4 bytes CF address */
+
+/*
+ *  Command/Status area layout: Main part
+ */
+
+/* Command/Status register:
  *
  * +---------------------------+
- * | rsvd | RLEN | CLEN | CMD  |
+ * | STAT | RLEN | CLEN | CMD  |
  * |   8  |   8  |   8  |   8  |
  * +---------------------------+
- *            |      |      |
+ *    |       |      |      |
+ *    status  |      |      |
  * Response len      |      |
  * (in bits)         |      |
  *                   |      |
@@ -19,8 +55,11 @@
  *         (in bits)        |
  *                          |
  *               Command code
+ *
+ * Due to the big endian layout, that means that a byte read will
+ * return the status byte
  */
-#define	CMD_REG			0x00
+#define	CMD_STAT_REG	        0x00
 #define  CMD_REG_CMD_MASK	0x000000ff
 #define  CMD_REG_CMD_SHIFT	0
 #define	  CMD_NONE		0x00
@@ -32,40 +71,24 @@
 #define  CMD_REG_CLEN_SHIFT	8
 #define  CMD_REG_RLEN_MASK	0x00ff0000
 #define  CMD_REG_RLEN_SHIFT	16
+#define  CMD_REG_STAT_MASK	0xff000000
+#define  CMD_REG_STAT_SHIFT	24
+#define	  STAT_WORKING		0x00
+#define	  STAT_COMPLETE		0x01
+#define	  STAT_ERR_INVAL_CMD	0x80
+#define	  STAT_ERR_INVAL_IRQ	0x81
+#define	  STAT_ERR_MTOE		0x82
 
-/* Status register
- *
- */
-#define	STAT_REG		0x04 /* Status */
-#define	 STAT_STOPPED		0x00
-#define	 STAT_SENDING		0x01
-#define	 STAT_COMPLETE		0x02
-#define	 STAT_ERR_INVAL_CMD	0x80
-#define	 STAT_ERR_INVAL_IRQ	0x81
-#define	 STAT_ERR_MTOE		0x82
-
-/* Response tag */
-#define	STAT_RTAG		0x05
+/* Response tag & CRC */
+#define	STAT_RTAG		0x04
 
 /* Response CRC */
-#define	STAT_RCRC		0x06
+#define	STAT_RCRC		0x05
 
 /* Echo and Send delay */
 #define	ECHO_DLY_REG		0x08
 #define	SEND_DLY_REG		0x09
 
-/* Signature & version */
-#define SYS_SIG_REG		0x0c /* 2 bytes system signature */
-#define  SYS_SIG_ROMULUS	0x526d /* 'Rm' */
-#define  SYS_SIG_WITHERSPOON	0x5773 /* 'Ws' */
-#define FW_VERS_REG		0x0e
-#define API_VERS_REG		0x0f
-
-/* Current API version */
-#define API_VERSION_MASK	0x7f
-#define API_VERSION		1
-#define API_VERSION_TRACE_EN	0x80
-
 /* Command data area
  *
  * Last byte of message must be left aligned
diff --git a/cf-fsi-test.c b/cf-fsi-test.c
index ef9435b..bf209f3 100644
--- a/cf-fsi-test.c
+++ b/cf-fsi-test.c
@@ -222,12 +222,26 @@ static void start_cf(void)
 static void load_cf_code(void)
 {
 	extern uint8_t cf_code_start, cf_code_end;
+	uint16_t sig, fw_vers, api_vers;
+	uint32_t fw_options;
 
 	uint8_t *code = &cf_code_start;
 	uint8_t *mem = cfmem;
 
 	while(code < &cf_code_end)
 		writeb(*(code++), mem++);
+
+	sig = ntohs(readw(cfmem + HDR_OFFSET + HDR_SYS_SIG));
+	fw_vers = ntohs(readw(cfmem + HDR_OFFSET + HDR_FW_VERS));
+	api_vers = ntohs(readw(cfmem + HDR_OFFSET + HDR_API_VERS));
+	fw_options = ntohl(readl(cfmem + HDR_OFFSET + HDR_FW_OPTIONS));
+
+	trace_enabled = !!(fw_options & FW_OPTION_TRACE_EN);
+
+	printf("SYS_SIG=%.4x FW_VERSION=%d API_VERSION=%d.%d (trace %s)\n",
+	       sig, fw_vers, api_vers >> 8, api_vers & 0xff,
+	       trace_enabled ? "enabled" : "disabled");
+
 }
 
 static void gpio_source_arm(void)
@@ -544,9 +558,8 @@ static void dump_stuff(void)
 {
 	int i;
 
-	printf("CMD:%08x STAT:%02x RTAG=%02x RCRC=%02x RDATA=%02x #INT=%08x\n",
-	       ntohl(readl(sysreg + SRAM_BASE + CMD_REG)),
-	       readb(sysreg + SRAM_BASE + STAT_REG),
+	printf("CMD:%08x RTAG=%02x RCRC=%02x RDATA=%02x #INT=%08x\n",
+	       ntohl(readl(sysreg + SRAM_BASE + CMD_STAT_REG)),
 	       readb(sysreg + SRAM_BASE + STAT_RTAG),
 	       readb(sysreg + SRAM_BASE + STAT_RCRC),
 	       ntohl(readl(sysreg + SRAM_BASE + RSP_DATA)),
@@ -564,11 +577,8 @@ static int do_command(uint32_t op)
 	uint32_t timeout = 100000;
 	uint8_t stat;
 
-	/* Clear status reg */
-	writeb(0, sysreg + SRAM_BASE + STAT_REG);
-
 	/* Send command */
-	writel(htonl(op), sysreg + SRAM_BASE + CMD_REG);
+	writel(htonl(op), sysreg + SRAM_BASE + CMD_STAT_REG);
 
 	/* Ring doorbell */
 	writel(0x2, sysreg + CVIC_BASE + CVIC_TRIG_REG);
@@ -581,7 +591,7 @@ static int do_command(uint32_t op)
 			dump_stuff();
 			return -ETIMEDOUT;
 		}
-		stat = readb(sysreg + SRAM_BASE + STAT_REG);
+		stat = readb(sysreg + SRAM_BASE + CMD_STAT_REG);
 	} while(stat < STAT_COMPLETE || stat == 0xff);
 
 	if (stat == STAT_COMPLETE)
@@ -738,19 +748,11 @@ int main(int argc, char *argv[])
 	/* Start ColdFire */
 	start_cf();
 
-	/* Wait for ack API version register*/
+	/* Wait for status register to say command complete */
 	do {
-		val = readb(sysreg + SRAM_BASE + API_VERS_REG);
+		val = readb(sysreg + SRAM_BASE + CMD_STAT_REG);
 	} while (val == 0x00);
 
-	trace_enabled = !!(val & API_VERSION_TRACE_EN);
-
-	printf("SYS_SIG=%.4x FW_VERSION=%d API_VERSION=%d (trace %s)\n",
-	       ntohs(readw(sysreg + SRAM_BASE + SYS_SIG_REG)),
-	       readb(sysreg + SRAM_BASE + FW_VERS_REG),
-	       val & API_VERSION_MASK,
-	       trace_enabled ? "enabled" : "disabled");
-
 	/* Configure echo & send delay */
 	writeb(16, sysreg + SRAM_BASE + ECHO_DLY_REG);
 	writeb(16, sysreg + SRAM_BASE + SEND_DLY_REG);