GitLab

GATT Service is contained between start and end handle. If
characteristic definition is at end handle, and it's value definition is
after end handle, it will not be properly discovered. That's because we
use value_handle instead of attribute_handle to identify
characteristics.

As a workaround, increase service boundary if value is defined after
it's definition.

Bug: 29253825
Change-Id: Ib145aea4f5cf38a1fbb977c301136e16f8f900f7

Added a log message inside l2c_link_hci_disc_comp() to help capture
a NULL pointer p_lcb->p_fixed_ccbs[xx] that triggers SIGSEGV crash.
Also, updated the typedef of bdstr_t and moved it to
file "btcore/include/bdaddr.h"

Bug: 28765514
Change-Id: Iacd9376cfb9e18b58e18865581da8edad9f68821

Change-Id: I91a19b2cfbc96d7567e0e0914058619ea364c1d4

Bug: 27778668
Change-Id: Id7454450c1ce2164001a1f9ca0b91f18d84e3fd6

This patch changes how GATTC is structured. Up till now, it
contained tBTA_GATTC_CACHE and tBTA_GATTC_CACHE_ATTR. Those
structures were so ugly, someone hide them inside internal
header, to make sure noone would ever access them from outside
GATTC. They are now replaced with:
tBTA_GATTC_SERVICE,
tBTA_GATTC_CHARACTERISTIC,
tBTA_GATTC_DESCRIPTOR and
tBTA_GATTC_INCLUDED_SVC.

Those looks much better, and were made globally avaliable.
Thanks to way they're build, we no longer need set of access
methods, which were also very ugly:
BTA_GATTC_GetFirstChar
BTA_GATTC_GetNextChar
BTA_GATTC_GetFirstIncludedService
etc.

This patch breaks HID, DO NOT SUBMIT without HID refactor.

Bug: 27455533
Change-Id: Ic42cfff175e0cc1a0d8e1a1216e2b4b756cbf77d

Now that the instance id is equal to handle, we no longer need this
mapping function.

Bug: 27455533
Change-Id: I6760f8fc81460df8520d4ebf1899ed256ce5ad1e

Currently we generate an id for each GATT attribute. Those ids
are used to differentiate between attributes that have the same
UUID. We need to get rid of those ids because it won't work with
partial GATT database updates due to "service change"
notifications. Instead, attribute handles, which are unique
across the whole GATT database will be used.

This is first patch in series to achieve that. It gets rid
of id generation, and just assigns handle in place of ids inside
of gatt cache. Upcoming patches will change the GATT cache API
and eventually get rid of "id" completly.

Bug: 27455533
Change-Id: I74c10872ecc223329b8ac7fef4085963bf272624

Currently when storing GATT cache into file, we save it piece
after piece, max 10 elements at once. This is not necessary,
we have enough space to just save all at once.
Also the logic doesn't need to be so complicated - no need
to state machine for such simple task.

Bug: 27455533
Change-Id: I82923dbedfb317835e45e13e0ea28151eb524d37

Bug: 27455533
Change-Id: Ie5661577320810468dd635022027a1e1a2f70a3e

Bug: 27455533
Change-Id: I653b563d3b3c233ffe31b0a35d396a66db363459

 * Remove unused functions: -Wno-unused-function
 * Fix logical operations that are missing parentheses:
   -Wno-logical-not-parentheses, -Wno-parentheses
 * Add missing braces in initializers: -Wno-missing-braces
 * Remove unused variables: -Wno-unused-variable
 * Fix printf()-like formatting issues: -Wno-format
 * Avoid using unitialized variables:
   -Wno-maybe-uninitialized, -Wno-uninitialized
 * Fix a bug inside bta_gattc_get_gatt_db_impl()
   when the processed GATT DB attribute type is unknown.
 * Fix warnings about missing field initializers:
   -Wno-missing-field-initializers
 * Re-implement macro COMPILE_ASSERT(COND) to fix a compilation warning:
   -Wno-non-literal-null-conversion
 * Fix sign mismatch comparison warnings:
   -Wno-sign-compare
 * Fix warnings related to enum conversion:
   -Wno-enum-conversion
 * Fix warnings related to incompatible pointer types:
   -Wno-incompatible-pointer-types
 * Fix warnings related to enum conversion:
   -Wno-enum-con...

This typo was introduced in commit 48db2d25 3years ago.
This cause unnecesary purge of cache and rediscovery even though
discovery was successfull.

Change-Id: I828e148ad489af1e231d33d7807bbe0e46f445c5

Change-Id: Iaf0280853278a4f28dde5b952b646096cfb12d97

Removed explicit checks for NULL pointer returns after calls
to osi_calloc() and osi_malloc(), because those are not needed.
If the memory allocation fails, osi_calloc() and osi_malloc()
will trigger an assert.

Bug: 27048759
Change-Id: I2791eb2f69c08f991f8fcdef10e101a41568cd95

Fix possible memory leak when allocating reused server cache control
block.

Change-Id: I3aab727e4f60e55530f49b0b05ccf2c891c72760

Removed the alternative buffer allocation osi_getbuf() / osi_freebuf()
and use instead osi_malloc() / osi_free().

Correspondingly, replaced usage of osi_freebuf_and_reset()
with osi_free_and_reset().

Bug: 24914560
Change-Id: I7a9599ba7fa900321f087da684428133eb0ddd6b

Change-Id: I60674c47246d8fe2094fe78ce31b752fa2acb89e

Change-Id: I624aeaa0c562b5c7404ad37411ad209beac3a5e1

 * Allow to call osi_freebuf(ptr) on NULL pointers. This simplifies
   the code: a notable number of "if (foo != NULL)" checks are removed.
 * Add new function osi_free_and_reset(p_ptr) that frees the buffer,
   and explicitly resets the pointer to NULL.
   This prevents unintended usage of free memory.
 * Add corresponding function osi_freebuf_and_reset(p_ptr)
 * Minor cleanup around usages of osi_free() and osi_freebuf()

Also:
 * Removed unused function btif_gattc_cleanup()
 * Replaced usage of the following functions with osi_freebuf_and_reset()
   - mca_free_buf()
   - utl_freebuf()
   - btif_hl_free_buf()
 * Replaced usage of rc_supported_event_free() with osi_freebuf()
 * Replaced usage of btif_hl_get_buf() with osi_getbuf()
 * Eliminate some of the osi_get_buf_size() calls

Bug: 22948224
Change-Id: Ife860658b26274da6f228d7353cb0f1531587337

Change-Id: Ie02dd72a2d4a423ab532ec84a4d588ab83f3e22b

Change-Id: I41e95f880bd37647278ad9dba4e12c2b54db273b

Currently getting GATT database from HAL requires multiple calls
and iterating over each element. This is long, unnecessary
complicated process, error prone process. This patch adds new
method, get_gatt_db that can be used to grab whole GATT database
for remote device just in one call.

Change-Id: Ib5d88fbc28d09d5c191e7c0152ca61d6a34f7a81

The revert includes a compilation fix as well
(missing include of a header file).

This reverts commit 3a4082d7.

Change-Id: I5bb1f468d2c64839037afe5ed2d3ed5d1b21dca2

This reverts commit 12091a32.

(cherry picked from commit 3a4082d7)

Change-Id: I2bc9969a7bd5d434b68ca05a4899a7f52ffb4f3e

This reverts commit 12091a32.

Change-Id: I01deaa5024bfef1782887fd8f8137f83916d8be0

1. Prevent GATT operations to proceed if queue is not empty

This change returns a FALSE from GATT enqueue operation, in case
there is already a GATT command enqueued. This simple change will
a. Prevent incorrect memory freeing of the currently queued command.
b. Prevent incorrect dequeing of the ongoing command and enqueing
another command which will never be executed.
c. Double free of memory causing segmentation fault

2. Prevent performing pending operations when link is down

This change prevents performing the execution of pending operations
when the link is disconnected. The pending operation in that case
is bound to fail thus freeing the memory pointed to by the p_q_cmd
which is cleaned up again by the discovery completion routine and
thus causing double free.

Bug:24178843
Change-Id: Ief2756f289a7db73d251ef7e247774dd3f7fc413

Prevents the crash which happens when device is
disconnected during service discovery and the
services list is not yet properly allocated
memory and populated. This change frees up the
services list after discovery complete only when
it is not null.

Change-Id: I0ced14d2ee133a332fa2fd7c456267438813a600

Right now sdp_conn_id and p_sdp_db are stored in static global
bta_gattc_cb between call to bta_gattc_sdp_service_disc and
bta_gattc_sdp_callback. If multiple instances of SDP discovery
are running simultaneously, they override this field, and free
same memory multiple times. This patch fixes that by making sure
sdp_conn_id and p_sdp_db are unique for each SDP discovery.

Bug: 25801255
Change-Id: I8ec52229e906e6b8748db7504f77e1f4d7006fbe

* Moved the following GKI buffer allocation functions to OSI:
  - GKI_getbuf() -> osi_getbuf()
  - GKI_freebuf() -> osi_freebuf()
  - GKI_get_buf_size() -> osi_get_buf_size()

  For now we need the osi_getbuf() / osi_freebuf() allocation wrapper,
  because we need to be able to call osi_get_buf_size() on the allocated
  buffer.
  In the future those should be replaced with osi_malloc() / osi_free().
  Currently, the osi_malloc() buffer size internal allocation tracker
 does not always track the size, hence we need the osi_getbuf() wrapper.

* Replaced GKI_MAX_BUF_SIZE with BT_DEFAULT_BUFFER_SIZE

* Added new file include/bt_common.h that can be usee to include
  few files that should be included alost everywhere (e.g. bt_target.h"
  NOTE: This file might be removed in the future and we should include
  everywhere the right set of header files.

* Removed some of the GKI-related references

* Removed file include/gki_target.h

Change-Id: Ie87830e731...

* Added new functions to OSI:
  - fixed_queue_init()
  - fixed_queue_length()
  - fixed_queue_try_remove_from_queue()
  - fixed_queue_try_peek_last()

* Renamed fixed_queue_try_peek() to fixed_queue_try_peek_first()

* Replaced usage of GKI queue functions with OSI fixed_queue functions:
  - GKI_init_q() -> fixed_queue_new(SIZE_MAX)
    NOTE: unlike GKI_init_q(), fixed_queue_new() allocates memory /
    state that needs to be released by calling fixed_queue_free()
  - GKI_enqueue() -> fixed_queue_enqueue()
  - GKI_dequeue() -> fixed_queue_try_dequeue()
    NOTE: fixed_queue_try_dequeue() is non-blocking
  - GKI_queue_length() -> fixed_queue_length()
  - GKI_queue_is_empty() -> fixed_queue_is_empty()
  - GKI_getfirst() -> fixed_queue_try_peek_first()
  - GKI_getlast() -> fixed_queue_try_peek_last()
  - GKI_remove_from_queue() -> fixed_queue_try_remove_from_queue()
  - Queue elements iteration.
    In the fixed_queue implementation we have to use the underlying
    list_t mechanism to iterate over the elements.
    OLD:
      p = GKI_getfirst(queue);
      ...
      while ((p = GKI_getnext(p) != NULL) {
         ...
      }
    NEW:
      list_t *list = fixed_queue_get_list(queue);
      for (const list_node_t *node = list_begin(list);
           node != list_end(list); node = list_next(node)) {
          p = list_node(node);
      }

* Remove initialization of the GKI module, because it is not needed
  anymore

* Removed unused files in GKI:
  gki/common/gki_common.h
  gki/ulinux/gki_int.h
  gki/ulinux/gki_ulinux.c

Change-Id: I3ff9464db75252d6faf7476a9ca67c88e535c51c

* Replace usage of function GKI_getpoolbuf() with GKI_getbuf()
* Remove usage of function GKI_poolutilization()
* Remove usage of function GKI_poolfreecount()

Change-Id: Ide938192b878bbfb4912642c903fce548f2b5368

If there are errors when processing GATT related events (e.g.,
configuring the MTU), don't use recursive calls into
bta_gattc_sm_execute(), because it breaks the free-ing of some
of the memory.

Bug: 23756301
Change-Id: I3c685170e868ffbf4e488d2bb5a31904e3f7b39d

* Moved the following GKI buffer allocation functions to OSI:
  - GKI_getbuf() -> osi_getbuf()
  - GKI_freebuf() -> osi_freebuf()
  - GKI_get_buf_size() -> osi_get_buf_size()

  For now we need the osi_getbuf() / osi_freebuf() allocation wrapper,
  because we need to be able to call osi_get_buf_size() on the allocated
  buffer.
  In the future those should be replaced with osi_malloc() / osi_free().
  Currently, the osi_malloc() buffer size internal allocation tracker
 does not always track the size, hence we need the osi_getbuf() wrapper.

* Replaced GKI_MAX_BUF_SIZE with BT_DEFAULT_BUFFER_SIZE

* Added new file include/bt_common.h that can be usee to include
  few files that should be included alost everywhere (e.g. bt_target.h"
  NOTE: This file might be removed in the future and we should include
  everywhere the right set of header files.

* Removed some of the GKI-related references

* Removed file include/gki_target.h

Change-Id: Ie87830e731...

* Added new functions to OSI:
  - fixed_queue_init()
  - fixed_queue_length()
  - fixed_queue_try_remove_from_queue()
  - fixed_queue_try_peek_last()

* Renamed fixed_queue_try_peek() to fixed_queue_try_peek_first()

* Replaced usage of GKI queue functions with OSI fixed_queue functions:
  - GKI_init_q() -> fixed_queue_new(SIZE_MAX)
    NOTE: unlike GKI_init_q(), fixed_queue_new() allocates memory /
    state that needs to be released by calling fixed_queue_free()
  - GKI_enqueue() -> fixed_queue_enqueue()
  - GKI_dequeue() -> fixed_queue_try_dequeue()
    NOTE: fixed_queue_try_dequeue() is non-blocking
  - GKI_queue_length() -> fixed_queue_length()
  - GKI_queue_is_empty() -> fixed_queue_is_empty()
  - GKI_getfirst() -> fixed_queue_try_peek_first()
  - GKI_getlast() -> fixed_queue_try_peek_last()
  - GKI_remove_from_queue() -> fixed_queue_try_remove_from_queue()
  - Queue elements iteration.
    In the fixed_queue implementation we have to use the underlying
    list_t mechanism to iterate over the elements.
    OLD:
      p = GKI_getfirst(queue);
      ...
      while ((p = GKI_getnext(p) != NULL) {
         ...
      }
    NEW:
      list_t *list = fixed_queue_get_list(queue);
      for (const list_node_t *node = list_begin(list);
           node != list_end(list); node = list_next(node)) {
          p = list_node(node);
      }

* Remove initialization of the GKI module, because it is not needed
  anymore

* Removed unused files in GKI:
  gki/common/gki_common.h
  gki/ulinux/gki_int.h
  gki/ulinux/gki_ulinux.c

Change-Id: I3ff9464db75252d6faf7476a9ca67c88e535c51c

* Replace usage of function GKI_getpoolbuf() with GKI_getbuf()
* Remove usage of function GKI_poolutilization()
* Remove usage of function GKI_poolfreecount()

Change-Id: Ide938192b878bbfb4912642c903fce548f2b5368

If there are errors when processing GATT related events (e.g.,
configuring the MTU), don't use recursive calls into
bta_gattc_sm_execute(), because it breaks the free-ing of some
of the memory.

Bug: 23756301
Change-Id: I3c685170e868ffbf4e488d2bb5a31904e3f7b39d

Fix the order of includes across a bunch of files, and declare LOG_TAG
at the top of every file in which it is used.

Consistently use bt_ as a LOG_TAG prefix.
Fix issues with LOG macro usage.
Remove unused includes and double-includes (when in related .h)

Add includes as necessary to compile cleanly (problems surfaced by
reordering includes)

Change-Id: Ic55520c8302b06bbc1942c10c448b20844669da6

This patch fixes an issue that is reproducible in highly noisy environments
(massive deployment of BLE beacons), through repeated connect/disconnect
attempts on a remote peripheral while scanning for beacons in the background.
The state machine in bta/gatt has a special control flow for handling disconnect
requests during discovery, which in this case failed to resolve the original
request by issuing an HCI_Disconnect command. This is now fixed by always
explicitly triggering the connection close sequence once the discovery state has
been cleaned up.

This patch also includes a fix for a crash that occurred as a side-effect of the
scenario described above.

Bug: 22350508
Change-Id: Ie9cbd3c8f54239b142bfb8dde80d9581ae70ed43

Fixes build breakages related to r.android.com/156982

Change-Id: Ib1143c41fe05a17c296226998afdb41a8cb6294a

Fixes build breakages related to r.android.com/156982

Change-Id: Ib1143c41fe05a17c296226998afdb41a8cb6294a