OSDN Git Service

android-x86/external-bluetooth-bluez.git
14 years agoadd sbc_primitives_mmx.c for x86 eclair-x86
Chih-Wei Huang [Fri, 9 Oct 2009 11:21:45 +0000 (19:21 +0800)]
add sbc_primitives_mmx.c for x86

14 years agoreconcile main tree with open-source eclair
The Android Open Source Project [Fri, 5 Feb 2010 16:09:33 +0000 (08:09 -0800)]
reconcile main tree with open-source eclair

14 years agoreconcile android-2.1_r1 snapshot
The Android Open Source Project [Fri, 29 Jan 2010 22:07:48 +0000 (14:07 -0800)]
reconcile android-2.1_r1 snapshot

14 years agoandroid-2.1_r1 snapshot
The Android Open Source Project [Tue, 12 Jan 2010 23:18:17 +0000 (15:18 -0800)]
android-2.1_r1 snapshot

14 years agoUse monotonically increasing clock for A2DP timing.
Nick Pelly [Thu, 17 Dec 2009 11:12:26 +0000 (03:12 -0800)]
Use monotonically increasing clock for A2DP timing.

This is to avoid skips when the system time changes.

14 years agoStop trying to send A2DP packets quickly to catch-up on missed write()'s after
Nick Pelly [Tue, 15 Dec 2009 00:55:17 +0000 (16:55 -0800)]
Stop trying to send A2DP packets quickly to catch-up on missed write()'s after
falling behind by 200ms, or on poll() timeout.

This allows us to recover more quickly from a situation where an A2DP headset
rejects A2DP packets for a short time. We now reset to regular packet metering
faster.

14 years agoChanges flush timeout from 60 ms to 120 ms.
Nick Pelly [Thu, 10 Dec 2009 00:20:28 +0000 (16:20 -0800)]
Changes flush timeout from 60 ms to 120 ms.

Less A2DP drop-outs due to unnecessary flushing.

BCM recommend 120-150ms for lazy flushing.

14 years agoFix liba2dp.c compiler warnings.
Nick Pelly [Wed, 9 Dec 2009 09:25:41 +0000 (01:25 -0800)]
Fix liba2dp.c compiler warnings.

14 years agoSet A2DP L2CAP socket to flushable while streaming A2DP, and set the automatic
Nick Pelly [Wed, 9 Dec 2009 09:23:43 +0000 (01:23 -0800)]
Set A2DP L2CAP socket to flushable while streaming A2DP, and set the automatic
flush timeout.

Remove the (ahead < 0) logic that was designed to skip forwards when streaming
was lagging. This logic was not working properly, and ACL flushing does its
job.

14 years agoAdd L2CAP_LM_FLUSHABLE to Bluez header.
Nick Pelly [Wed, 9 Dec 2009 09:18:22 +0000 (01:18 -0800)]
Add L2CAP_LM_FLUSHABLE to Bluez header.

14 years agoAdd ACL flow control state (MTU, packets available, packets total) for debug.
Nick Pelly [Wed, 9 Dec 2009 09:09:36 +0000 (01:09 -0800)]
Add ACL flow control state (MTU, packets available, packets total) for debug.

Accessed with the debug tool 'hcitool con'

14 years agoeclair snapshot
Jean-Baptiste Queru [Fri, 13 Nov 2009 02:45:21 +0000 (18:45 -0800)]
eclair snapshot

14 years agoSend SBC frames if output buffer is nearly full.
Yu Kang Ku [Thu, 5 Nov 2009 18:49:48 +0000 (10:49 -0800)]
Send SBC frames if output buffer is nearly full.

This is to address an A2DP issue seen with the Mazda carkit, where A2DP audio
cannot be played. The MTU size (8200) requested by this carkit is much greater
than the output buffer size (2048) allocated by the Bluez A2DP profile. The
output buffer being full is giving an SBC encode error. This patch checks for
the buffer full condition.

Change-Id: Idded504d085c760baef15cd3cef84fcf57165e4f
Signed-off-by: Yu Kang Ku <kang.ku@motorola.com>
Signed-off-by: Nick Pelly <npelly@google.com>
14 years agoAdd support for sending Broadcom coexistance setup commands for A2DP links.
Nick Pelly [Sat, 31 Oct 2009 01:18:33 +0000 (18:18 -0700)]
Add support for sending Broadcom coexistance setup commands for A2DP links.

Enabled on platforms with BOARD_HAVE_BLUETOOTH_BCM.

Change-Id: Ic223955ac3580e5a9430f96e069f686a7afdbef8

14 years agoFix handling of SDP records.
Jaikumar Ganesh [Fri, 30 Oct 2009 16:29:02 +0000 (09:29 -0700)]
Fix handling of SDP records.

Delete SDP records when device is deleted.
Update cache whenever SDP is performed.
Remove SDP records even when there is no driver associated with the UUID.

Bug: 2173752
Dr No: Eastham

14 years agodo not merge - Fix for issue 2184627 cherry picked from eclair-mr2
Eric Laurent [Fri, 16 Oct 2009 16:22:50 +0000 (09:22 -0700)]
do not merge - Fix for issue 2184627 cherry picked from eclair-mr2

Bluetooth A2DP suspend-resume improvements.

This change will reduce the occurence rate of A2DP sink suspend resume failures observed in issues 21846272181005 and possibly 2189628.

Avoid lockups in case of BT device disconnection during the A2DP start process by using a timeout when reading from bluetooth command socket.
Correct a typo causing potential deadlock in wait_for_start().

14 years agoAdd DBUS interfaces AddRfcommServiceRecord() and RemoveServiceRecord().
Nick Pelly [Thu, 1 Oct 2009 02:09:58 +0000 (19:09 -0700)]
Add DBUS interfaces AddRfcommServiceRecord() and RemoveServiceRecord().

org.bluez.Adapter.AddRfcommServiceRecord takes a name, uuid and channel
and creates a record like:

Service Name: MyApplication
Service RecHandle: 0x10009
Service Class ID List:
  UUID 128: 42999bc0-ac5e-11de-8a39-0800200c9a66
Protocol Descriptor List:
  "L2CAP" (0x0100)
  "RFCOMM" (0x0003)
    Channel: 26

It returns the integer Record Handle, that can be used with
org.bluez.Adapter.RemoveServiceRecord to remove that record.

Bug: 2158900
DrNo: eastham

Change-Id: Ie375d5ccd24b5faf067a26b8a04a1449b3f56ce2

14 years agoRemove redundant call to bluetooth_init().
Nick Pelly [Wed, 30 Sep 2009 18:35:21 +0000 (11:35 -0700)]
Remove redundant call to bluetooth_init().

This is a cleanup to Ian Kent's preceeding patchset.

Change-Id: I8ff00530fdd337873ff898b7fd9009b0fc13f0b4

14 years agoAdd Bluetooth A2DP suspend/resume to DBUS API.
Zhu Lan [Tue, 23 Jun 2009 08:04:09 +0000 (16:04 +0800)]
Add Bluetooth A2DP suspend/resume to DBUS API.

Change-Id: Icef1729957ed55e491e63d74d14a2c0c93984c4d

14 years agobluez a2dp - use a sensible write timeout
Ian Kent [Mon, 28 Sep 2009 15:47:44 +0000 (23:47 +0800)]
bluez a2dp - use a sensible write timeout

The 500 msec write timeout is extremely optimistic, a 1000 msec
timeout is closer to reality.

14 years agobluez a2dp - fix wait_for_start() spurious wakeup
Ian Kent [Mon, 28 Sep 2009 15:47:44 +0000 (23:47 +0800)]
bluez a2dp - fix wait_for_start() spurious wakeup

When waiting on a condition is is possble to receive spurious
wake ups. Deal with this in wait_for_start().

Change-Id: I678e9d7831333d9b8baa31e9b0ec2597ca9263cd

14 years agobluez a2dp - fix wait_for_start() synchronization
Ian Kent [Mon, 28 Sep 2009 15:47:44 +0000 (23:47 +0800)]
bluez a2dp - fix wait_for_start() synchronization

On entry to wait_for_start() it's possible we are in state
A2DP_STATE_NONE if a2dp_thread() has not completed bluetooth_init().
Also, we call the pthread_mutex_*() functions more frequently than
we really need to.

Move the mutex locking outside the loop and let condition wait
take care of the locking. Also move the A2DP_STATE_NONE check
below the condition wait so we can be sure the state machine
has been poked before we test it.

Change-Id: I020bea365a623e88cb1a5f7e5fccd8f8aa948518

14 years agobluez a2dp - destroy thread attr after thread create
Ian Kent [Mon, 28 Sep 2009 15:47:44 +0000 (23:47 +0800)]
bluez a2dp - destroy thread attr after thread create

Pthread resources may be used by the thread attributes object.
We should destroy this after thread creation.

14 years agobluez a2dp - destroy thread resources at exit
Ian Kent [Mon, 28 Sep 2009 15:47:44 +0000 (23:47 +0800)]
bluez a2dp - destroy thread resources at exit

The dynamically created pthread mutex and condition variables
consume pthread resources. They should be destroyed prior to
freeing the structure containing them.

14 years agobluez a2dp - fix a2dp_thread() spurious wakeup
Ian Kent [Mon, 28 Sep 2009 15:47:44 +0000 (23:47 +0800)]
bluez a2dp - fix a2dp_thread() spurious wakeup

When waiting on a condition is is possble to receive spurious
wake ups. Use the fact that when we poke the state machine the
command must have changed to deal with this.

Change-Id: I0ed2f2a15c15dea30cb1d5cf3b67c8549d6802d4

14 years agobluez a2dp - fix state machine synchronization
Ian Kent [Mon, 28 Sep 2009 15:47:44 +0000 (23:47 +0800)]
bluez a2dp - fix state machine synchronization

Using a worker thread implementation for A2DP means that if we don't
have some way to syncroniize state transitions we can get overlap of
requested command functions or lost signals. In order for the A2DP
state machine to function each command operation must complete before
the next is initiated or we will get errors caused by these out of
sequence commands. And if we signal the thread when it is not waiting
on the condition the signal will be missed.

This could be partly resolved by implementing a queue but then there
is an overhead with also implementing a wait mechamism for state
transitions. It's much easier and simpler to just hold the mutex
during command processing which also deals with the lost signals issue.
This may well not achieve the original goal of the worker thread
implementation but neither would a queue implementation. It looks
like this is just the way things are!

14 years agobluez a2dp - syncronize worker thread start
Ian Kent [Mon, 28 Sep 2009 15:47:43 +0000 (23:47 +0800)]
bluez a2dp - syncronize worker thread start

There is an order of execution assumption made when starting the
a2dp worker thread. Add an additional condition variable to allow
us to make sure we have reached the thread_wait condition wait
before returning from a2dp_init().

14 years agoRename audio_device_is_connected more appropriately
Johan Hedberg [Thu, 3 Sep 2009 16:08:07 +0000 (19:08 +0300)]
Rename audio_device_is_connected more appropriately

14 years agoFix AVDTP Abort when transport closes before response
Johan Hedberg [Thu, 3 Sep 2009 15:22:14 +0000 (18:22 +0300)]
Fix AVDTP Abort when transport closes before response

Change-Id: I5d2bced52514eb14fd8236a9262ae5c71e68f524

14 years agoFix agent_free call due to the picking up of upmerge fixes.
Jaikumar Ganesh [Tue, 29 Sep 2009 01:05:23 +0000 (18:05 -0700)]
Fix agent_free call due to the picking up of upmerge fixes.

Change-Id: Ia0a12ec45dd498572e4a4f2354d383ba5ea436ce

14 years agoFix AVDTP suspend while closing
Johan Hedberg [Wed, 16 Sep 2009 11:15:22 +0000 (14:15 +0300)]
Fix AVDTP suspend while closing

We should reject any suspend request if we've already started the close
procedure for a stream.

14 years agoFix command timeout handling when there's no stream
Johan Hedberg [Tue, 15 Sep 2009 18:12:04 +0000 (21:12 +0300)]
Fix command timeout handling when there's no stream

The media stream could be gone by the time that the AVDTP command timeout
expires (e.g. if the media transport channel gets closed prematurely). In
such a case there's no need to send a separate ABORT command but we can
directly proceed with closing the signalling channel.

14 years agoFix AVDTP ABORTING state transition
Johan Hedberg [Tue, 15 Sep 2009 14:17:07 +0000 (17:17 +0300)]
Fix AVDTP ABORTING state transition

The transition to ABORTING state when acting as initiator should happen
when we receive the ABORT reply and not when we send the initial ABORT
command. The reply handling is already correct but since there's a state
change also in the sending part the later state change request would fail.
This patch fixes the issue by removing the state change when sending the
ABORT command.

Change-Id: I9cf9f27924b7d348523125ddccff7c8ecebacf93

14 years agoFix stream handling when the media transport disconnects prematurely
Johan Hedberg [Tue, 15 Sep 2009 08:05:49 +0000 (11:05 +0300)]
Fix stream handling when the media transport disconnects prematurely

We might not have a stream anymore when receiving the reply to a AVDTP
command that assumes that there is an existing stream. So ignore the
replies to these commands if we don't have a stream anymore.

14 years agoFixed duplicate attempt from device to connect 'signalling' channel
Peter Hurley [Wed, 9 Sep 2009 17:25:35 +0000 (13:25 -0400)]
Fixed duplicate attempt from device to connect 'signalling' channel

A device may attempt to initiate a duplicate avdtp channel while
a host-initiated channel connection is still pending. This
situation is referred to as 'XCASE connect:connect' in sink.c and
source.c. This fix rejects the device-initiated connection in this
circumstance.

14 years agoFix agent destroy callback handling for errors
Johan Hedberg [Sun, 6 Sep 2009 00:18:19 +0000 (03:18 +0300)]
Fix agent destroy callback handling for errors

The destroy callback should not be called if calling the agent fails and
an error is returned to the caller of the agent.c public function.

14 years agoFix agent_authorize error return
Johan Hedberg [Sun, 6 Sep 2009 00:15:42 +0000 (03:15 +0300)]
Fix agent_authorize error return

14 years agoFix audio authorization cancelation on the same mainloop iteration
Johan Hedberg [Sat, 5 Sep 2009 20:06:23 +0000 (23:06 +0300)]
Fix audio authorization cancelation on the same mainloop iteration

If the device is trusted and the authorization is canceled before the idle
callback takes place the idle callback should be removed. This patch adds
tracking of the idle id and removes it in the case of cancelation.

Change-Id: I7a22176bb9c881349cde0730e196378c94665ecb

14 years agoMake dev->authorized private to audio/device.c
Johan Hedberg [Sat, 5 Sep 2009 19:09:34 +0000 (22:09 +0300)]
Make dev->authorized private to audio/device.c

There's no need to expose the authorized setting outside of device.c.

14 years agoFix AVDTP stream leak for invalid media transport config
Johan Hedberg [Sat, 5 Sep 2009 14:50:13 +0000 (17:50 +0300)]
Fix AVDTP stream leak for invalid media transport config

14 years agoFix race condition in authorizing audio connections
Johan Hedberg [Fri, 4 Sep 2009 16:40:19 +0000 (19:40 +0300)]
Fix race condition in authorizing audio connections

The current check for implicit authorization is if any other audio profile
is already connected. However, a the second profile might try to connect
right after we've gotten a positive authorization reply for the first one
but before the first profile has reached "connected" state. So just by
checking the connected state of other profiles we might get a false
negative for the decision of doing implicit authorization.

This patch adds a variable to the audio_device struct for keeping track of
if we've gotten a positive authorization reply that can be used in
addition to the "any other profiles" connected check. This variable gets
cleared when the (global audio) device state goes back to disconnected.

14 years agoFix a few more ref-leaks with AVDTP failures and unix socket clients
Johan Hedberg [Thu, 3 Sep 2009 21:45:52 +0000 (00:45 +0300)]
Fix a few more ref-leaks with AVDTP failures and unix socket clients

If any of the IPC messages cause us to create a brand new AVDTP session
and then there's failure caused by disconnection, we have to unref the
session since there will not be a unix.c callback to do it when avdtp.c
wants to clean up in the next mainloop iteration.

14 years agoFix potential AVDTP reference leak
Johan Hedberg [Thu, 3 Sep 2009 20:25:54 +0000 (23:25 +0300)]
Fix potential AVDTP reference leak

If the AVDTP connection has just recently disconnected but not yet been
notified to the mainloop then we need to clean the unix.c avdtp session
reference within start_discovery() instead of doing it when the unix
client disconnects. This is because when the AVDTP disconnection finally
gets reported in the next mainloop iteration the unix.c reference wouldn't
get cleared as is intended (since unix.c doesn't have any callbacks at
this point).

14 years agoFix authorization when the audio profiles are slow to connect
Johan Hedberg [Thu, 3 Sep 2009 16:09:52 +0000 (19:09 +0300)]
Fix authorization when the audio profiles are slow to connect

Incoming audio connections should be automatically authorized if either
HFP/HSP or A2DP is already connected for the same device.

14 years agoReject AVDTP_DISCOVER when there are no SEPs available
Johan Hedberg [Wed, 2 Sep 2009 09:21:52 +0000 (12:21 +0300)]
Reject AVDTP_DISCOVER when there are no SEPs available

14 years agoReject AVDTP START/STOP when disconnecting
Daniel Orstadius [Tue, 1 Sep 2009 10:55:44 +0000 (13:55 +0300)]
Reject AVDTP START/STOP when disconnecting

14 years agoAdd new Agent API : RequestPairingConsent.
Jaikumar Ganesh [Fri, 4 Sep 2009 20:26:40 +0000 (13:26 -0700)]
Add new Agent API : RequestPairingConsent.

This callback will be called for incoming pairing requests for 2.1
devices only when the remote IO capabilities are NoInputNoOutput
or DisplayOnly and the local IO capability is DisplayYesNo.
Currently, we were silently auto accepting.

14 years agoDo not advertise Voice Recognition support in HF AG SDP record.
Nick Pelly [Tue, 1 Sep 2009 17:20:07 +0000 (10:20 -0700)]
Do not advertise Voice Recognition support in HF AG SDP record.

14 years agoChange UART speed before Bluetooth firmware download.
James Luan [Fri, 28 Aug 2009 18:23:33 +0000 (13:23 -0500)]
Change UART speed before Bluetooth firmware download.

14 years agoDon't bail out if Discovery is in progress when getting the rfcomm channel.
Jaikumar Ganesh [Tue, 25 Aug 2009 20:47:59 +0000 (13:47 -0700)]
Don't bail out if Discovery is in progress when getting the rfcomm channel.

We should be having the SDP record in the cache. In the very rare
case, that the remote device's SDP records have changed and if a discovery
is in progress, we would have already updated the cache or if we have not
queried the device yet, we are going to fail anyways.

14 years agoWait 5 milliseconds before doing Simple Pairing auto-accept
Marcel Holtmann [Fri, 21 Aug 2009 02:50:58 +0000 (19:50 -0700)]
Wait 5 milliseconds before doing Simple Pairing auto-accept

The BITE tester is again really stupid and it can't handle dongles
that can do the DH key exchange quickly. For example the Broadcom
dongles use a fast processors and an optimized DH algorithm. So in
the end they are way too fast.

To make this work, use a 5 milliseconds delay before replying to
the auto-accept case of Simple Pairing.

14 years agoChange version number to 4.47.
Jaikumar Ganesh [Wed, 19 Aug 2009 23:21:57 +0000 (16:21 -0700)]
Change version number to 4.47.

14 years agoDon't advertise AVRCP Controller support in SDP records.
Nick Pelly [Tue, 18 Aug 2009 23:43:31 +0000 (16:43 -0700)]
Don't advertise AVRCP Controller support in SDP records.

We are a AVRCP 1.0 target, not a controller.

This is identical to the change we applied to Bluez 3.x

14 years agoDon't enable the SCO server socket when not necessary
Johan Hedberg [Tue, 11 Aug 2009 07:30:16 +0000 (10:30 +0300)]
Don't enable the SCO server socket when not necessary

The SCO server socket shouldn't be enabled when neither Gateway nor
Headset support is enabled.

14 years agoUpgrade to Bluez 4.47.
Jaikumar Ganesh [Tue, 11 Aug 2009 15:39:23 +0000 (08:39 -0700)]
Upgrade to Bluez 4.47.

14 years agoModify PBAP SDP record.
Zhu Lan [Mon, 3 Aug 2009 09:19:39 +0000 (17:19 +0800)]
Modify PBAP SDP record.

Fix to pass PTS case TC_SDAS_BV_03_I(since PTS3.2 update2).
The code tried to append the UINT type value to the supported
respositories.  However, the wrong function sdp_seq_alloc()
was used. This function adds 2 redundant bytes (SDP_SEQ8 as
0x35, seq) ahead of 0x08, so that latest PTS IOPT will fail.
According to spec, PBAP attributes should report 0x08 0x01
or 0x08 0x03 right after 0x0314. The correct function to use
is sdp_data_alloc(), with clean 0x08 0x01 appended as
SDP_ATTR_SUPPORTED_REPOSITORIES.

14 years agoRelease 4.47
Marcel Holtmann [Sat, 1 Aug 2009 22:32:22 +0000 (15:32 -0700)]
Release 4.47

14 years agoUpdate library version
Marcel Holtmann [Sat, 1 Aug 2009 22:32:00 +0000 (15:32 -0700)]
Update library version

14 years agoFix calling set_limited_discoverable twice
Johan Hedberg [Fri, 31 Jul 2009 22:13:56 +0000 (01:13 +0300)]
Fix calling set_limited_discoverable twice

set_mode doesn't need to call set_limited_discoverable since
adapter_mode_changed will do it.

14 years agoUse limited discoverable mode only when 0 < discov_timeout <= 60
Johan Hedberg [Fri, 31 Jul 2009 22:05:55 +0000 (01:05 +0300)]
Use limited discoverable mode only when 0 < discov_timeout <= 60

14 years agoCache service class updates during bootup
Alok Barsode [Wed, 29 Jul 2009 07:41:14 +0000 (13:11 +0530)]
Cache service class updates during bootup

adapter->svc_cache is used to cache service class updates during
bluetoothd bringup or when adapter is down.adapter_disable_svc_cache() is
used to switch off the cache, so any successive service class updates are
directly written to the device.

14 years agoUpdate audio-api.txt including AudioSource documentation.
Luiz Augusto von Dentz [Tue, 21 Jul 2009 13:05:17 +0000 (10:05 -0300)]
Update audio-api.txt including AudioSource documentation.

14 years agoRemove deprecated method and signals for AudioSource.
Luiz Augusto von Dentz [Tue, 21 Jul 2009 13:04:47 +0000 (10:04 -0300)]
Remove deprecated method and signals for AudioSource.

14 years agoWhen RFKILL is unblocked restore the previous power setting
Marcel Holtmann [Fri, 31 Jul 2009 12:56:23 +0000 (05:56 -0700)]
When RFKILL is unblocked restore the previous power setting

14 years agoAdd functions for RFKILL and offline mode
Johan Hedberg [Fri, 31 Jul 2009 12:47:19 +0000 (15:47 +0300)]
Add functions for RFKILL and offline mode

This patch adds a couple of helper functions that are needed to cleanly
implement RFKILL and offline mode support.

14 years agoFix ipctest and its strict-aliasing problems
Marcel Holtmann [Fri, 31 Jul 2009 09:15:34 +0000 (02:15 -0700)]
Fix ipctest and its strict-aliasing problems

14 years agoAdd skeleton for RFKILL support
Marcel Holtmann [Wed, 29 Jul 2009 19:44:31 +0000 (21:44 +0200)]
Add skeleton for RFKILL support

14 years agoFix missing includes from header files
Marcel Holtmann [Wed, 29 Jul 2009 19:35:00 +0000 (21:35 +0200)]
Fix missing includes from header files

14 years agoFix serial proxy record info attributes
Johan Hedberg [Tue, 28 Jul 2009 10:37:16 +0000 (13:37 +0300)]
Fix serial proxy record info attributes

14 years agoUse 16/32-bit UUID if possible in serial proxy records
Johan Hedberg [Tue, 28 Jul 2009 07:36:26 +0000 (10:36 +0300)]
Use 16/32-bit UUID if possible in serial proxy records

14 years agoTrack sender of CreateProxy.
Luiz Augusto von Dentz [Mon, 27 Jul 2009 22:20:41 +0000 (19:20 -0300)]
Track sender of CreateProxy.

Only proxies created by serial.conf should be persistent.

14 years agoRemove the serial storage functionality
Forrest Zhao [Mon, 27 Jul 2009 05:31:30 +0000 (13:31 +0800)]
Remove the serial storage functionality

This functionality isn't needed anymore now that we have serial.conf.
Proxies created through the D-Bus interface will always be temporary not
remembered after a bluetoothd restart.

14 years agoFix whitespace
Johan Hedberg [Fri, 24 Jul 2009 07:27:35 +0000 (10:27 +0300)]
Fix whitespace

14 years agoAdd support for multiple serial proxies configuration
Forrest Zhao [Fri, 24 Jul 2009 03:18:20 +0000 (11:18 +0800)]
Add support for multiple serial proxies configuration

14 years agoFix another possible array out of bounds.
Luiz Augusto von Dentz [Wed, 22 Jul 2009 17:33:59 +0000 (14:33 -0300)]
Fix another possible array out of bounds.

Note also that the pppd_opts should be NULL terminated thus there is only
place for 39 options rather 40.

14 years agoRemove the service_classes_callback mechanism
Alok Barsode [Wed, 22 Jul 2009 12:06:54 +0000 (17:36 +0530)]
Remove the service_classes_callback mechanism

14 years agoFix bug which could cause buffer overruns.
Luiz Augusto von Dentz [Tue, 21 Jul 2009 20:13:30 +0000 (17:13 -0300)]
Fix bug which could cause buffer overruns.

14 years agoFix possible array out of bounds.
Luiz Augusto von Dentz [Tue, 21 Jul 2009 19:00:50 +0000 (16:00 -0300)]
Fix possible array out of bounds.

14 years agoFix segfault when a record is updated.
Luiz Augusto von Dentz [Tue, 21 Jul 2009 16:40:06 +0000 (13:40 -0300)]
Fix segfault when a record is updated.

Current code is freeing the new record when it shouldn't as the memory is
reused by extract_pdu_server.

14 years agoFix bogus error comparison
Johan Hedberg [Tue, 21 Jul 2009 08:50:39 +0000 (11:50 +0300)]
Fix bogus error comparison

There's no "error" variable here so the comparison was always with the
address of the public error function, which will always be != NULL.

14 years agoRelease 4.46
Marcel Holtmann [Sat, 18 Jul 2009 21:36:46 +0000 (23:36 +0200)]
Release 4.46

14 years agoUpdate library version
Marcel Holtmann [Sat, 18 Jul 2009 21:27:46 +0000 (23:27 +0200)]
Update library version

14 years agoFix possible use of a pointer after free.
Luiz Augusto von Dentz [Fri, 17 Jul 2009 13:11:24 +0000 (10:11 -0300)]
Fix possible use of a pointer after free.

14 years agoFix possible memory leak.
Luiz Augusto von Dentz [Thu, 16 Jul 2009 13:21:50 +0000 (10:21 -0300)]
Fix possible memory leak.

14 years agoFix possible leak or crash on gen_dataseq_pdu in case of out of memory.
Luiz Augusto von Dentz [Thu, 16 Jul 2009 12:24:37 +0000 (09:24 -0300)]
Fix possible leak or crash on gen_dataseq_pdu in case of out of memory.

14 years agoFix unintended fall through.
Luiz Augusto von Dentz [Thu, 16 Jul 2009 12:23:14 +0000 (09:23 -0300)]
Fix unintended fall through.

14 years agoCleanup gateway code to make Coverty happy.
Luiz Augusto von Dentz [Wed, 15 Jul 2009 20:28:43 +0000 (17:28 -0300)]
Cleanup gateway code to make Coverty happy.

14 years agoFix unhandled return.
Luiz Augusto von Dentz [Wed, 15 Jul 2009 19:52:55 +0000 (16:52 -0300)]
Fix unhandled return.

14 years agoRemove useless instruction.
Luiz Augusto von Dentz [Wed, 15 Jul 2009 19:51:08 +0000 (16:51 -0300)]
Remove useless instruction.

14 years agoFix strict-aliasing rules handling for hciemu and IP addresses
Marcel Holtmann [Thu, 16 Jul 2009 05:25:30 +0000 (07:25 +0200)]
Fix strict-aliasing rules handling for hciemu and IP addresses

14 years agoFix strict-aliasing rules handling of l2test and info requests
Marcel Holtmann [Thu, 16 Jul 2009 05:23:38 +0000 (07:23 +0200)]
Fix strict-aliasing rules handling of l2test and info requests

14 years agoUse -fno-strict-aliasing for audio subdirectory for now
Marcel Holtmann [Thu, 16 Jul 2009 05:15:19 +0000 (07:15 +0200)]
Use -fno-strict-aliasing for audio subdirectory for now

14 years agoFix another strict-aliasing rules handling with CMSG_DATA
Marcel Holtmann [Thu, 16 Jul 2009 05:13:02 +0000 (07:13 +0200)]
Fix another strict-aliasing rules handling with CMSG_DATA

14 years agoFix strict-aliasing rules handling with CMSG_DATA
Marcel Holtmann [Thu, 16 Jul 2009 06:42:06 +0000 (08:42 +0200)]
Fix strict-aliasing rules handling with CMSG_DATA

14 years agosbc: Enable neon optimizations.
Mike Lockwood [Thu, 16 Jul 2009 00:27:50 +0000 (20:27 -0400)]
sbc: Enable neon optimizations.

Signed-off-by: Mike Lockwood <lockwood@android.com>
14 years agoFix variable typo
Marcel Holtmann [Wed, 15 Jul 2009 16:15:49 +0000 (18:15 +0200)]
Fix variable typo

14 years agoAdd option to disable optimization independent from debug
Marcel Holtmann [Wed, 15 Jul 2009 16:09:02 +0000 (18:09 +0200)]
Add option to disable optimization independent from debug

14 years agodisable neon extras until they build properly
Mike Reed [Tue, 14 Jul 2009 15:33:18 +0000 (11:33 -0400)]
disable neon extras until they build properly

14 years agoFix ipctest reading socket forever.
Luiz Augusto von Dentz [Mon, 13 Jul 2009 16:33:58 +0000 (13:33 -0300)]
Fix ipctest reading socket forever.

14 years agoAdd Sink role support to audio IPC.
João Paulo Rechi Vita [Wed, 8 Jul 2009 19:25:55 +0000 (16:25 -0300)]
Add Sink role support to audio IPC.