[PATCH] tools/hciattach: Add support for BCM line discipline

[PATCH] tools/hciattach: Add support for BCM line discipline

[John Keeping]

This is similar to the "bcm43xx" option but using the kernel driver has
the advantage that the wakeup and shutdown pins on the device can be
used.
---
 tools/hciattach.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/tools/hciattach.c b/tools/hciattach.c
index 59a76a7..7f8f463 100644
--- a/tools/hciattach.c
+++ b/tools/hciattach.c
@@ -1163,6 +1163,10 @@ struct uart_t uart[] = {
 	{ "amp",	0x0000, 0x0000, HCI_UART_H4, 115200, 115200,
 			AMP_DEV, DISABLE_PM, NULL, NULL, NULL },
 
+	/* Broadcom BCM with kernel support */
+	{ "bcm",        0x0000, 0x0000, HCI_UART_BCM,  115200, 115200,
+				FLOW_CTL, DISABLE_PM, NULL, NULL, NULL },
+
 	{ NULL, 0 }
 };
 
-- 
2.6.3.462.gbe2c914

Re: [PATCH] tools/hciattach: Add support for BCM line discipline

[Marcel Holtmann]

Hi John,

> This is similar to the "bcm43xx" option but using the kernel driver has
> the advantage that the wakeup and shutdown pins on the device can be
> used.

what is wrong with just using btattach?

Regards

Marcel

Re: [PATCH] tools/hciattach: Add support for BCM line discipline

[John Keeping]

On Tue, 24 Nov 2015 13:21:05 +0100, Marcel Holtmann wrote:

> > This is similar to the "bcm43xx" option but using the kernel driver
> > has the advantage that the wakeup and shutdown pins on the device
> > can be used.  
> 
> what is wrong with just using btattach?

It's marked as experimental and not installed by "make install", so I
assumed that it was only supposed to be used for testing.


Regards,
John

Re: [PATCH] tools/hciattach: Add support for BCM line discipline

[Marcel Holtmann]

Hi John,

>>> This is similar to the "bcm43xx" option but using the kernel driver
>>> has the advantage that the wakeup and shutdown pins on the device
>>> can be used.  
>> 
>> what is wrong with just using btattach?
> 
> It's marked as experimental and not installed by "make install", so I
> assumed that it was only supposed to be used for testing.

hmm, not installed by default. I do not remember the reason for that, but I am sure I had one. However btattach is so dead simple that it should be used. Seems like we should look into making sure the command line options are stable and we install it by default.

The hciattach code is the most horrific code we have in our tree. I really dislike it and there is almost zero chance to clean it up. I am not sure that I want to pile on top of it.

Regards

Marcel

Re: [PATCH] tools/hciattach: Add support for BCM line discipline

[John Keeping]

On Tue, 24 Nov 2015 13:37:22 +0100, Marcel Holtmann wrote:

> Hi John,
> 
> >>> This is similar to the "bcm43xx" option but using the kernel
> >>> driver has the advantage that the wakeup and shutdown pins on the
> >>> device can be used.    
> >> 
> >> what is wrong with just using btattach?  
> > 
> > It's marked as experimental and not installed by "make install", so
> > I assumed that it was only supposed to be used for testing.  
> 
> hmm, not installed by default.

I don't think there's any way to install it via the autotools
infrastructure since it's added to noinst_PROGRAMS in Makefile.tools
(and then only when EXPERIMENTAL is set).

>                                I do not remember the reason for that,
> but I am sure I had one. However btattach is so dead simple that it
> should be used. Seems like we should look into making sure the
> command line options are stable and we install it by default.

I'm happy to submit a patch to build (and install) btattach by default,
but I guess it will also need a man page.  I'm not sure what you mean
about the command line options, they look reasonable to me but I'm not a
bluetooth expert.


Regards,
John