korg/bluez
0
0
Fork 0
mirror of https://git.kernel.org/pub/scm/bluetooth/bluez.git synced 2024-12-02 00:24:25 +08:00
Code Issues Packages Projects Releases Wiki Activity
14,458 Commits 1 Branch 320 Tags 39 MiB
0eb8474255
Commit Graph

14458 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Johan Hedberg
b01bc4c867 core: Only set EIR data pointer after confirming it points to valid memory 
Even though we do not access the memory it's still safer not to have any
pointers to it until we know it's valid.
 2013-01-22 15:48:36 +02:00
Johan Hedberg
252222a414 monitor: Add support for passkey notify mgmt events 2013-01-22 15:39:55 +02:00
Johan Hedberg
cdc215e1b4 core: Fix canceling pairing through mgmt 
The mgmt_reply() function should be used instead of mgmt_send() since
otherwise the request obeys the usual command queuing which wont work
due to the pending mgmt_pair_device command.
 2013-01-22 14:57:18 +02:00
Johan Hedberg
f46bba7faa tools: Add mgmt_cancel_pair_device support to btmgmt 2013-01-22 14:52:04 +02:00
Syam Sidhardhan
e755bd90d2 a2dp: Fix invalid memory access during abort_ind() 
There is an invalid memory access w.r.t to the callback
during the Abort_Ind finalize_setup_errno().

We should terminate the variable arguments with NULL.

Log:
bluetoothd[3353]: audio/avdtp.c:avdtp_parse_cmd() Received ABORT_CMD
bluetoothd[3353]: audio/a2dp.c:abort_ind() Source 0xb6f5ecc8: Abort_Ind
bluetoothd[3353]: audio/a2dp.c:setup_ref() 0xb6f63200: ref=2
bluetoothd[3353]: audio/transport.c:media_transport_remove() Transport
 /org/bluez/3353/hci0/dev_BC_47_60_F5_88_89/fd1 Owner :1.0
bluetoothd[3353]: audio/transport.c:media_transport_release() Transport
 /org/bluez/3353/hci0/dev_BC_47_60_F5_88_89/fd1: read lock released
bluetoothd[3353]: audio/transport.c:media_transport_release() Transport
 /org/bluez/3353/hci0/dev_BC_47_60_F5_88_89/fd1: write lock released
bluetoothd[3353]: audio/transport.c:media_request_reply() Request
 Acquire Reply Input/output error
bluetoothd[3353]: audio/transport.c:media_owner_free() Owner :1.0
bluetoothd[3353]: audio/transport.c:media_owner_remove() Owner :1.0 Request Acquire
bluetoothd[3353]: audio/a2dp.c:a2dp_sep_unlock() SEP 0xb6f5ecc8 unlocked
bluetoothd[3353]: audio/a2dp.c:setup_unref() 0xb6f63200: ref=1
[sys_assert]START of sighandler
[sys-assert]exepath = bluetoothd
[sys-assert]processname = bluetoothd
[sys_assert]this thread is main thread. pid=3353
[sys-assert]cs timestr 1358524835
bluetoothd[3353]: crashed [1358524835] processname=bluetoothd, pid=3353, tid=3353, signal=11
[sys-assert]start print_node_to_file
sighandler = 0xb6e8cfc9, g_sig_oldact[i] = (nil)
[sys_assert]END of sighandler
Segmentation fault (core dumped)
 2013-01-22 13:30:30 +02:00
Syam Sidhardhan
cb377fa61d a2dp: Fix invalid memory access during suspend_ind() 
There is a possible invalid memory access during suspend_ind().
We should terminate the variable arguments with NULL.
 2013-01-22 13:30:30 +02:00
Syam Sidhardhan
3021f18098 health: Fix possible use after free 
A pointer to freed memory is dereferenced if we call function
hdp_get_dcpsm_cb() with out any earlier reference.
 2013-01-22 13:30:30 +02:00
Johan Hedberg
da5596731b core: Fix mgmt_get_connections response size checking 
Each address entry is of the size mgmt_addr_info and not bdaddr_t.
 2013-01-22 13:18:32 +02:00
Marcel Holtmann
4dbe677ad2 monitor: Use explicit filter policy strings for LE create connection 2013-01-22 02:09:16 -08:00
Marcel Holtmann
180899b102 monitor: Decode LE filter policy for all cases 2013-01-22 01:51:02 -08:00
Szymon Janc
781f3cde67 monitor: Fix possible crash on unknown command 
If unknown command is received monitor would crash due to dereference
of NULL opcode_data pointer.
 2013-01-21 11:38:17 -08:00
Alexandros Antonopoulos
1f3d21d46f core: Fix crash when an invalid uuid is passed to Device1.DisconnectProfile 
If the user calls Device1.DisconnectProfile with an invalid profile
uuid disconnect_profile still tries to parse the uuid resulting in
a SIGSEGV
 2013-01-21 17:58:54 +02:00
Alexandros Antonopoulos
d42cbe8580 avctp: Fix size of read() for browsing channel callbacks 
The read() function should attempt to read browsing->imtu bytes
and not sizeof(browsing->imtu).
 2013-01-21 17:53:19 +02:00
Alexandros Antonopoulos
d23fb32b8e avctp: Fix request timeout after a channel is destroyed 
When the control channel is destroyed if there is a pending request
(chan->p) then the channel queue is deleted in avctp_channel_destroy
but the timer is still alive resulting in a SIGSEGV
 2013-01-21 17:53:18 +02:00
Alexandros Antonopoulos
879efa424a avctp: Fix avctp_unregister_browsing_pdu_handler inner loop 
Inner loop should access the sessions data and not the servers
data
 2013-01-21 17:53:18 +02:00
Luiz Augusto von Dentz
feccaf0498 player: Fix not returning current position properly 
Current position should be returned for Get/GetAll not last position
notified otherwise clients that are started after playback are
not able to synchronize its position properly as it does no know how
for how long the playback has advanced.
 2013-01-21 14:55:36 +02:00
Marcel Holtmann
f426775b84 shared: Remove also queued requests and replies for index cancel 2013-01-20 20:12:07 -08:00
Marcel Holtmann
ab0ff42d43 shared: Fix issue with management destroy handling 2013-01-20 19:10:23 -08:00
Marcel Holtmann
65214a997e core: Remove pointless GFunc casts from device handling 2013-01-20 18:54:33 -08:00
Marcel Holtmann
a3d31b58d2 shared: Mark and skip notification callbacks when become invalid 2013-01-20 18:45:03 -08:00
Marcel Holtmann
e3c028a44d shared: Remove all pending requests for the specified index 2013-01-20 18:36:10 -08:00
Marcel Holtmann
c229ba89a1 gdbus: Check for valid path before handling properties 2013-01-20 18:25:12 -08:00
Marcel Holtmann
b63458353e core: Don't take device reference for connection requests 
The device reference is not needed since the adapter handling actually
owns the device object anyway. So do not bother taking another reference.

In addition actually remove the devices from the list when the adapter
is removed. Otherwise they stay around and just causes problems.
 2013-01-20 16:14:32 -08:00
Marcel Holtmann
338efa13d4 core: Keep the DBusConnection around for device handling 
It is required to have the connection around so device objects can
be deregistered. This happens triggered from the adapter handling.
 2013-01-20 16:13:31 -08:00
Marcel Holtmann
f1e7f81efe core: Use same reference counting as adapter handling 2013-01-20 16:13:05 -08:00
Marcel Holtmann
8332abd8fe core: Remove pointless protection for device path at unregister 
If such a protection is really needed, then it might should be better
handled properly within GDBus. So remove it and if it causes an issue,
fix it then.
 2013-01-20 15:31:17 -08:00
Marcel Holtmann
a399286dc5 core: Use a global DBusConnection variable 2013-01-20 15:29:50 -08:00
Marcel Holtmann
2e89402a25 core: Add device init and cleanup functions 2013-01-20 15:21:08 -08:00
Marcel Holtmann
9390708d29 core: Add some extra debugs for device creation 2013-01-20 15:15:29 -08:00
Marcel Holtmann
f710762d3f core: Make the default UUID search list const 2013-01-20 15:15:14 -08:00
Marcel Holtmann
70c733459f monitor: Add support for decoding LE white list commands 2013-01-20 10:09:31 -08:00
Marcel Holtmann
6fe39e7f9f monitor: Add support for decoding LE features 2013-01-20 09:52:06 -08:00
Marcel Holtmann
27383772dd monitor: Add support for decoding LE event mask 2013-01-20 09:43:15 -08:00
Marcel Holtmann
fd303a4ef1 monitor: Add support for decoding LE supported states 2013-01-20 09:35:49 -08:00
Marcel Holtmann
cdb2fd58d3 core: Provide skeleton for LE connection trigger 2013-01-20 01:02:31 -08:00
Marcel Holtmann
dee5bc60cb core: Handle invalidated RSSI signals a bit better 2013-01-20 01:01:27 -08:00
Marcel Holtmann
3f3152f15e core: Avoid massive signals for changes below 8 dBm 
Instead of blasting clients with RSSI changed signals, slow this down
a little bit and only report changes of 8 dBm and more.
 2013-01-20 00:33:03 -08:00
Marcel Holtmann
f179ff75b7 core: Allow passive scanning to update existing devices 
During the passive scan, do not create new devices or update temporary
ones. However update of existing paired devices is desired.
 2013-01-20 00:09:19 -08:00
Marcel Holtmann
99d34b8ceb core: Check for LE controller before cancelling passive scanning 2013-01-19 23:48:59 -08:00
Marcel Holtmann
3f2d53c64e core: Avoid pointless casting of GCompareFunc for compare functions 2013-01-19 23:42:28 -08:00
Marcel Holtmann
bf7446cbcf core: Use delay between passive scanning iterations 
Keeping the conroller active for passive scanning with the current
kernel interface seems inefficient since it automatically disables
the discovery procedure. So instead of restarting it right away,
introduce a small delay to give the controller more chances to sleep.
 2013-01-19 23:18:45 -08:00
Marcel Holtmann
c16853b3fd core: Add support for triggering passive scanning 
This is the first step for passive scanning support. Create triggers
at various location to allow entering passive scanning.
 2013-01-19 20:58:01 -08:00
Marcel Holtmann
f0ccb6d471 core: Handle existing start discovery gracefully 
In case a discovery already exists, keep using it if is has the same
type and only if not, stop it and start a new one.
 2013-01-19 19:13:49 -08:00
Marcel Holtmann
7b897d2d38 core: Simplify discovery start and stop handling 
To make the discovery start, stop and restart logic even simple, just
trigger start of discovery when client gets registered. It is almost
impossible to handle all corner cases, so just trigger another restart
of the discovery with a larger timeout in case of an error.
 2013-01-19 18:41:08 -08:00
Marcel Holtmann
407579a962 core: Fix multiple issues with discovery handling 
This patch is the first attempt in fixing the discovery issues for
dual-mode BR/EDR + LE controllers. It creates a clean chain for
start/stop discovery client handling and also for the internal
handling of suspend, resume and restart.

However this patch removes the LE connection handling. And that needs
to be put back with a separate patch.
 2013-01-19 16:01:45 -08:00
Szymon Janc
bbf2080f64 tools: Use proper test data in mgmt_unblock_device test 
Copy and paste error. Although both data values are same use proper
one for clarity.
 2013-01-19 09:55:39 +02:00
Marcel Holtmann
d1c281d89a core: Use discovery found list instead of nested struct 2013-01-18 21:00:38 -08:00
Marcel Holtmann
a97fef5405 core: Remove pending_mode session variable 2013-01-18 20:49:58 -08:00
Marcel Holtmann
dde1b9580e core: Remove unneeded session reference counting 2013-01-18 17:36:33 -08:00
Marcel Holtmann
5e4f1b8da3 core: Remove session_req_type variable and parameter 2013-01-18 17:31:07 -08:00