[PATCH] Add udev rule for Ralink WiFi+BT module w/ bccmd

[PATCH] Add udev rule for Ralink WiFi+BT module w/ bccmd

[Eric Miao]

Some Ralink WiFi + CSR BT module needs bluetooth radio to be
turned on explicitly with bccmd by the following command:

  bccmd psset -r 0x028c 0x0001

The HCI will be probed successfully, but due to the default
OFF state of the radio, no scan or connection can be made.
This patch adds the udev rule so that once the HCI is up,
udev will invoke bccmd with warm reset to turn on the radio.

[1] https://bugs.launchpad.net/bugs/781556
[2] https://bugs.meego.com/show_bug.cgi?id=3498
[3] http://ubuntuforums.org/showthread.php?t=1776108
[4] http://ubuntuforums.org/showthread.php?t=1703941

Signed-off-by: Eric Miao <eric.miao@canonical.com>
---
 .gitignore                    |    1 +
 Makefile.am                   |    7 ++++++-
 scripts/bluetooth-bccmd.rules |   12 ++++++++++++
 3 files changed, 19 insertions(+), 1 deletions(-)
 create mode 100644 scripts/bluetooth-bccmd.rules

diff --git a/.gitignore b/.gitignore
index 137d2e5..accff4b 100644
--- a/.gitignore
+++ b/.gitignore
@@ -39,6 +39,7 @@ sap/sap.c
 scripts/bluetooth.rules
 scripts/97-bluetooth.rules
 scripts/97-bluetooth-hid2hci.rules
+scripts/97-bluetooth-bccmd.rules
 
 sbc/sbcdec
 sbc/sbcenc
diff --git a/Makefile.am b/Makefile.am
index bd587eb..b0145dc 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -406,6 +406,10 @@ if HID2HCI
 udev_files += scripts/bluetooth-hid2hci.rules
 endif
 
+if BCCMD
+udev_files += scripts/bluetooth-bccmd.rules
+endif
+
 if PCMCIA
 udevdir = @UDEV_DIR@
 
@@ -420,7 +424,8 @@ endif
 CLEANFILES += $(rules_DATA)
 
 EXTRA_DIST += scripts/bluetooth.rules \
-		scripts/bluetooth-hid2hci.rules scripts/bluetooth-serial.rules
+		scripts/bluetooth-hid2hci.rules scripts/bluetooth-serial.rules \
+		scripts/bluetooth-bccmd.rules
 
 EXTRA_DIST += doc/manager-api.txt \
 		doc/adapter-api.txt doc/device-api.txt \
diff --git a/scripts/bluetooth-bccmd.rules b/scripts/bluetooth-bccmd.rules
new file mode 100644
index 0000000..e654abf
--- /dev/null
+++ b/scripts/bluetooth-bccmd.rules
@@ -0,0 +1,12 @@
+# do not edit this file, it will be overwritten on update
+
+ACTION!="add", GOTO="bccmd_end"
+SUBSYSTEM!="bluetooth", GOTO="bccmd_end"
+
+# Some bluetooth modules require special bccmd parameters to be set
+
+# Ralink RT5390 + CSR BS8510 WiFi/BT Combo cards on some HP laptops,
+# they need to explicitly turn on the radio followed by a warm reset
+SUBSYSTEM=="bluetooth", SUBSYSTEMS=="usb", ATTRS{idVendor}=="148f", ATTRS{idProduct}=="2000|1000", RUN+="/usr/sbin/bccmd -d $kernel psset -r 0x028c 0x0001"
+
+LABEL="bccmd_end"
-- 
1.7.9.1

Re: [PATCH] Add udev rule for Ralink WiFi+BT module w/ bccmd

[Marcel Holtmann]

Hi Eric,

> Some Ralink WiFi + CSR BT module needs bluetooth radio to be
> turned on explicitly with bccmd by the following command:
> 
>   bccmd psset -r 0x028c 0x0001
> 
> The HCI will be probed successfully, but due to the default
> OFF state of the radio, no scan or connection can be made.
> This patch adds the udev rule so that once the HCI is up,
> udev will invoke bccmd with warm reset to turn on the radio.
> 
> [1] https://bugs.launchpad.net/bugs/781556
> [2] https://bugs.meego.com/show_bug.cgi?id=3498
> [3] http://ubuntuforums.org/showthread.php?t=1776108
> [4] http://ubuntuforums.org/showthread.php?t=1703941

why are we doing this in userspace? The btusb.c driver has the option
for pre-execution HCI commands queue. Why not use that one.

Also why -r. Meaning why do we need a warmreset? And where is the magic
0x028c PS key coming from? Does anybody actually know what it is doing?

Regards

Marcel

Re: [PATCH] Add udev rule for Ralink WiFi+BT module w/ bccmd

[Eric Miao]

On Tue, Mar 20, 2012 at 1:37 AM, Marcel Holtmann <marcel@holtmann.org> wrot=
e:
> Hi Eric,
>
>> Some Ralink WiFi + CSR BT module needs bluetooth radio to be
>> turned on explicitly with bccmd by the following command:
>>
>> =C2=A0 bccmd psset -r 0x028c 0x0001
>>
>> The HCI will be probed successfully, but due to the default
>> OFF state of the radio, no scan or connection can be made.
>> This patch adds the udev rule so that once the HCI is up,
>> udev will invoke bccmd with warm reset to turn on the radio.
>>
>> [1] https://bugs.launchpad.net/bugs/781556
>> [2] https://bugs.meego.com/show_bug.cgi?id=3D3498
>> [3] http://ubuntuforums.org/showthread.php?t=3D1776108
>> [4] http://ubuntuforums.org/showthread.php?t=3D1703941
>
> why are we doing this in userspace? The btusb.c driver has the option
> for pre-execution HCI commands queue. Why not use that one.

That sounds like a better one, I'll take a look and see if it's possible
to give a revised version. Thanks Marcel.

>
> Also why -r. Meaning why do we need a warmreset? And where is the magic
> 0x028c PS key coming from? Does anybody actually know what it is doing?

The magic came from Ralink FAE actually and it was verified to work
by various people. It was believed it's the bit to turn the radio ON/OFF.
And a warmreset ensures that radio will be ON instantly, other than a
system boot. I'm afraid there is no public documentation about this,
I didn't even find this bit in the CSR BCCMD firmware notes of CSR8510,
could be a vendor specific bit.

>
> Regards
>
> Marcel
>
>

Re: [PATCH] Add udev rule for Ralink WiFi+BT module w/ bccmd

[Eric Miao]

On Tue, Mar 20, 2012 at 9:55 AM, Eric Miao <eric.miao@canonical.com> wrote:
> On Tue, Mar 20, 2012 at 1:37 AM, Marcel Holtmann <marcel@holtmann.org> wr=
ote:
>> Hi Eric,
>>
>>> Some Ralink WiFi + CSR BT module needs bluetooth radio to be
>>> turned on explicitly with bccmd by the following command:
>>>
>>> =C2=A0 bccmd psset -r 0x028c 0x0001
>>>
>>> The HCI will be probed successfully, but due to the default
>>> OFF state of the radio, no scan or connection can be made.
>>> This patch adds the udev rule so that once the HCI is up,
>>> udev will invoke bccmd with warm reset to turn on the radio.
>>>
>>> [1] https://bugs.launchpad.net/bugs/781556
>>> [2] https://bugs.meego.com/show_bug.cgi?id=3D3498
>>> [3] http://ubuntuforums.org/showthread.php?t=3D1776108
>>> [4] http://ubuntuforums.org/showthread.php?t=3D1703941
>>
>> why are we doing this in userspace? The btusb.c driver has the option
>> for pre-execution HCI commands queue. Why not use that one.
>
> That sounds like a better one, I'll take a look and see if it's possible
> to give a revised version. Thanks Marcel.

Hmm... have thought about this twice, and I still prefer the bccmd
command in user space. Converting this very straight-forward setting
w/ bccmd into HCI commands would make it rather awkward.

And so the further question would be the bluetooth-hid2hcd.rules,
will this file gradually disappear and the necessary operations be
made into the kernel code eventually?