Skip to contentSkip to conversation

tools: Fix usage for hciattach command

PATCH In review [email protected] 2 participants 6 messages Sep 8, 2014 – Sep 19, 2014 Bluetooth drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/3 tools: Fix usage for hciattach command

From: Tzu-Jung Lee <[email protected]>

---
 tools/hciattach.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/hciattach.c +1 -1
index 1904ac5..542b5b3 100644
@@ -1276,7 +1276,7 @@ static void usage(void)
 {
 	printf("hciattach - HCI UART driver initialization utility\n");
 	printf("Usage:\n");
-	printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [bdaddr]\n");
+	printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [sleep|nosleep] [bdaddr]\n");
 	printf("\thciattach -l\n");
 }
 
Signature · 1 line
2.0.4
#1 lore raw

PATCH 2/3 tools: Use specified uart speed to load firmware for bcm43xx

From: Tzu-Jung Lee <[email protected]>

---
 tools/hciattach_bcm43xx.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
tools/hciattach_bcm43xx.c +22 -0
index cb4bfb9..630220e 100644
@@ -359,9 +359,27 @@ int bcm43xx_init(int fd, int speed, struct termios *ti, const char *bdaddr)
 	if (bcm43xx_locate_patch(FIRMWARE_DIR, chip_name, fw_path)) {
 		fprintf(stderr, "Patch not found, continue anyway\n");
 	} else {
+		/* Speed up firmware loading */
+		if (speed > 3000000 && bcm43xx_set_clock(fd, BCM43XX_CLOCK_48))
+			return -1;
+
+		if (bcm43xx_set_speed(fd, speed))
+			return -1;
+
+		if (set_speed(fd, ti, speed))
+			return -1;
+
 		if (bcm43xx_load_firmware(fd, fw_path))
 			return -1;
 
+		/*
+		 * The controller reset the uart speed to 115200 after
+		 * firmware loading, so we have to reset the terminal
+		 * correspondingly
+		 */
+		if (set_speed(fd, ti, 115200))
+			return -1;
+
 		if (bcm43xx_reset(fd))
 			return -1;
 	}
@@ -375,5 +393,9 @@ int bcm43xx_init(int fd, int speed, struct termios *ti, const char *bdaddr)
 	if (bcm43xx_set_speed(fd, speed))
 		return -1;
 
+	if (set_speed(fd, ti, speed))
+		return -1;
+
+
 	return 0;
 }
Signature · 1 line
2.0.4
#2 lore raw

PATCH 3/3 tools: Fix download mode delay for bcm43xx

In reply to [email protected]

From: Tzu-Jung Lee <[email protected]>

---
 tools/hciattach_bcm43xx.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/hciattach_bcm43xx.c +1 -1
index 630220e..53dbce5 100644
@@ -220,7 +220,7 @@ static int bcm43xx_set_clock(int fd, unsigned char clock)
 static int bcm43xx_load_firmware(int fd, const char *fw)
 {
 	unsigned char cmd[] = { HCI_COMMAND_PKT, 0x2e, 0xfc, 0x00 };
-	struct timespec tm_mode = { 0, 50000 };
+	struct timespec tm_mode = { 0, 50000000 };
 	struct timespec tm_ready = { 0, 2000000 };
 	unsigned char resp[CC_MIN_SIZE];
 	unsigned char tx_buf[1024];
Signature · 1 line
2.0.4
#3 lore raw
In reply to [email protected]

Hi,

Quoted 20 lines · On Tue, Sep 9, 2014 at 12:21 AM, <[email protected]> wrote:

From: Tzu-Jung Lee <[email protected]>

---
 tools/hciattach.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/hciattach.c +1 -2
index 1904ac5..542b5b3 100644
@@ -1276,7 +1276,7 @@ static void usage(void)
 {
        printf("hciattach - HCI UART driver initialization utility\n");
        printf("Usage:\n");
-       printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [bdaddr]\n");
+       printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [sleep|nosleep] [bdaddr]\n");
        printf("\thciattach -l\n");
 }
 
--
Quoted 1 line

2.0.4

It doesn't look like you are doing anything with sleep|nosleep here, also you should probably a add more information regarding what it is fixing.

Signature · 1 line
Luiz Augusto von Dentz
#4 lore raw

Hi Luiz,

Quoted 22 lines · Luiz Augusto von Dentz

Hi,

Quoted 19 lines · On Tue, Sep 9, 2014 at 12:21 AM, <[email protected]> wrote:

From: Tzu-Jung Lee <[email protected]>

---
 tools/hciattach.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/hciattach.c +1 -1
index 1904ac5..542b5b3 100644
@@ -1276,7 +1276,7 @@ static void usage(void)
 {
        printf("hciattach - HCI UART driver initialization utility\n");
        printf("Usage:\n");
-       printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [bdaddr]\n");
+       printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [sleep|nosleep] [bdaddr]\n");
        printf("\thciattach -l\n");
 }
 
Quoted 7 lines
Quoted 2 lines

--
2.0.4

It doesn't look like you are doing anything with sleep|nosleep here, also you should probably a add more information regarding what it is fixing.

I probably should have separated these 3 patches to 2 different sets though they are all trivial fixes.

The first one attempts to reveal the missing [speed | nospeed] to the user:

Currently, the argument parsing code in the hciattach.c expect very specific order of the arguments:

    hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed]

<tty> <type | id> [speed] [flow|noflow] [bdaddr]

For example, if an user would like to specify the BDADDR, none of the following works:

hciattach TTY any BDADDR
hciattach TTY any SPEED BDADDR
hciattach TTY any SPEED NOFLOW BDADDR

The user has to put the BDADDR exactly in the 4th argument after <type | id>, ex:

hciattach TTY any SPEED NOFLOW NOSLEEP BDADDR

And this is impossible to figure out without looking at the source code. The patch only helps a little (hope so) by revealing the missing argument to the user.

    hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed]

<tty> <type | id> [speed] [flow|noflow] [sleep|nosleep] [bdaddr]

Though I do agree the information is still too little, and we probably should rewrite the argument parsing code to remove the enforcement of order.
Otherwise, the precise usage syntax for the current logic should be:

hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>
<type | id> [speed]
hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>
<type | id> [speed flow|noflow]
hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>
<type | id> [speed flow|noflow] sleep|nosleep]
hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>
<type | id> [speed flow|noflow sleep|nosleep bdaddr]

Even though, user still need to figure out what suppose to be put in the flow/sleep if he only wants to change the bdaddr without changing other default settings.

The second patch speed up the firmware loading speed by raising the UART baudrate before (and after) loading firmware.
This helps our project reduce the hciattach from 7+ seconds, to 750 ms. The tricky part for the controller is that it drops the UART back to 115200 after firmware loading.
So we still need to update the baudrate again after firmware is loaded.

Thanks.
Roy

Quoted 2 lines

--
Luiz Augusto von Dentz

#5 lore raw

Hi Luiz,

Any comment for the patch 2/3 ?

Thanks.
Roy

Quoted 96 lines · Tzu-Jung Lee

Hi Luiz,

Quoted 22 lines · Luiz Augusto von Dentz

Hi,

Quoted 19 lines · On Tue, Sep 9, 2014 at 12:21 AM, <[email protected]> wrote:

From: Tzu-Jung Lee <[email protected]>

---
 tools/hciattach.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/hciattach.c +1 -1
index 1904ac5..542b5b3 100644
@@ -1276,7 +1276,7 @@ static void usage(void)
 {
        printf("hciattach - HCI UART driver initialization utility\n");
        printf("Usage:\n");
-       printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [bdaddr]\n");
+       printf("\thciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty> <type | id> [speed] [flow|noflow] [sleep|nosleep] [bdaddr]\n");
        printf("\thciattach -l\n");
 }
 
Quoted 7 lines
Quoted 2 lines

--
2.0.4

It doesn't look like you are doing anything with sleep|nosleep here, also you should probably a add more information regarding what it is fixing.

I probably should have separated these 3 patches to 2 different sets though they are all trivial fixes.

The first one attempts to reveal the missing [speed | nospeed] to the user:

Currently, the argument parsing code in the hciattach.c expect very specific order of the arguments:

    hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed]

<tty> <type | id> [speed] [flow|noflow] [bdaddr]

For example, if an user would like to specify the BDADDR, none of the following works:

hciattach TTY any BDADDR
hciattach TTY any SPEED BDADDR
hciattach TTY any SPEED NOFLOW BDADDR

The user has to put the BDADDR exactly in the 4th argument after <type | id>, ex:

hciattach TTY any SPEED NOFLOW NOSLEEP BDADDR

And this is impossible to figure out without looking at the source code. The patch only helps a little (hope so) by revealing the missing argument to the user.

    hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed]

<tty> <type | id> [speed] [flow|noflow] [sleep|nosleep] [bdaddr]

Though I do agree the information is still too little, and we probably should rewrite the argument parsing code to remove the enforcement of order.
Otherwise, the precise usage syntax for the current logic should be:

   hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>

<type | id> [speed]
hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>
<type | id> [speed flow|noflow]
hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>
<type | id> [speed flow|noflow] sleep|nosleep]
hciattach [-n] [-p] [-b] [-r] [-t timeout] [-s initial_speed] <tty>
<type | id> [speed flow|noflow sleep|nosleep bdaddr]

Even though, user still need to figure out what suppose to be put in the flow/sleep if he only wants to change the bdaddr without changing other default settings.

The second patch speed up the firmware loading speed by raising the UART baudrate before (and after) loading firmware.
This helps our project reduce the hciattach from 7+ seconds, to 750 ms. The tricky part for the controller is that it drops the UART back to 115200 after firmware loading.
So we still need to update the baudrate again after firmware is loaded.

Thanks.
Roy

Quoted 2 lines

--
Luiz Augusto von Dentz