[FFe] merge bluez 4.99-1 from Debian testing

[Mathieu Trudel-Lapierre]

Bluez 4.99 introduces version 1.0 of the management API; seems like a good idea to be running a stable version of a management API for a LTS release; and includes a few bug fixes and updates to support new devices.

ver 4.99:
 Fix issue with missing retries for BNEP connection setup.
 Fix issue with not showing name if first EIR has no details.
 Fix issue with running SDP discovery for LE devices.
 Add support for GATT using 128-bit Bluetooth UUIDs.
 Add support for retrieving key size information.
 Add support for storing Long Term Keys.
 Add support for Proximity Reporter API.
 Add support for KeyboardDisplay IO capability.
 Add support for version 1.0 of management API.
 Add support for monitoring interface.

On top of these changes; there are a few commits in git post-4.99 which fixes the same issues as in bluez 4.98-2ubuntu7; but are cleaner patches that take into account additional cases where crashes could occur. These patches will be updated to match the upstream commits.

[Scott Kitterman (kitterman)]

New API?

How does this affect packages that depend and build-depend on bluez?

$ reverse-depends -b libbluetooth-dev
Reverse-Build-Depends
=====================
* amora-server
* anyremote
* asterisk
* blueman
* bluemon
* bluez-btsco
* bluez-hcidump
* bluez-tools
* brltty
* btscanner
* cwiid
* foxtrotgps
* gammu
* gnokii
* gpe-bluetooth
* gpsd
* gvfs
* gypsy
* haskell-hcwiid
* libcwiimote
* libnet-bluetooth-perl
* libopenobex
* libpam-blue
* obex-data-server
* obexd
* obexftp
* obexpushd
* ofono
* pilot-link
* pulseaudio
* pybluez
* python2.6
* python2.7
* python3.2
* qemu-linaro
* qtmobility
* scmxx
* sitplus
* totem
* transfermii
* ussp-push

[Mathieu Trudel-Lapierre (cyphermox)]

in libbluetooth3:
/usr/lib/x86_64-linux-gnu/libbluetooth.so.3.11.5 -> /usr/lib/x86_64-linux-gnu/libbluetooth.so.3.12.0

So doesn't introduce a transition. The Management API 1.0 thing is the way by which bluez speaks to the kernel to communicate with the various bluetooth devices; mgmtops or the older hciops (raw HCI sockets). The Management API is described in more detail here: http://www.bluez.org/the-management-interface/.

I think it's safe, but will obviously need careful testing of the current bluetooth setup to make sure file transfers, A2DP and HSP/HFP still work properly.

[Mathieu Trudel-Lapierre (cyphermox)]

[13:52:07] <cyphermox> holtmann: should be good with 3.2.0?
[13:57:58] <holtmann> Yes it is. However the mgmt interface of the kernel is only default enabled with 3.4.
[13:58:08] <holtmann> So 3.2 should be just fine. However it will use HCI directly.

So we can expect that's not actually changing the current bluez behavior.

[Scott Kitterman (kitterman)]

What testing has been done to verify this is safe?

[Mathieu Trudel-Lapierre (cyphermox)]

I was testing the file transfer and A2DP and HSP/HFP audio profiles.

I've just identified some issues with sending files, and skipping in the A2DP profile; so let's NAK this and revisit after Precise, I think we can cherry-pick changes from upstream where necessary.

[Scott Kitterman (kitterman)]

OK.