Age | Commit message (Collapse) | Author |
|
I believe with that enabled the device is clocked down or running in a
low power mode. The debugger behave much worse with this enabled, so
let's keep it disabled for now.
Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
|
|
The code would not compile if this setting was turned off. It
is now possible to set this feature to zero to reduce code
and ram usage of the controller. Also fixed the comment that
said this feature (privacy) was not supported by the nimble
controller.
|
|
backwards-compatibility-breaking changes to develop.
* develop: (290 commits)
sim compiler - replace objsize with size
Fix warnings reported by clang.
MYNEWT-329
MYNEWT-354
STM32f407 discovery board BSP
mbedtls; use smaller version of SHA256.
boot; boot loader does not need to call os_init() anymore, as bsp_init() has been exported.
boot; app does not need the dependency to mbedtls
slinky; time-based waits must use OS_TICKS_PER_SEC.
bootutil; adjust unit tests to work with status upkeep outside sys/config.
bootutil; was returning wrong image header in response when swithing images. Add boot_set_req() routine for unit test use.
boot/bootutil; remove debug console use from bootloader.
bootutil/imgmgr; output of boot now shows the fallback image.
imgmgr; automatically confirm image as good for now.
bootutil; add 'confirm' step, telling that image was confirmed as good. Otherwise next restart we'll go back to old image.
bootutil; make status element size depend on flash alignment restrictions.
boot, imgmgr; return the slot number for test image.
bootutil; move routines reading boot-copy-status from loader.c to bootutil_misc.c.
boot; return full flash location of status bytes, instead of just offset.
boot; don't use NFFS or FCB for keeping status. Interim commit.
...
|
|
This appears to be a typo. There is no tool called objsize. size works
fine.
This only affects OS X. It hasn't been a problem in the past when gcc-5
was used because the invocation of gobjdump was crashing with a message
like the following:
2016/08/22 17:50:54 [DEBUG] o=gobjdump(68189,0x7fff71813000) malloc:
*** error for object 0x1016002a0: pointer being freed was not
allocated *** set a breakpoint in malloc_error_break to debug
When gobjdump crashed, newt ignored the error, but did not invoke the
size tool.
When clang is used instead of gcc, gobjdump doesn't crash, so the
incorrect size path is reported as a build failure.
|
|
|
|
By default, we will assume that cputime counts at a 1 MHz rate.
If a different frequency is desired the user should define
HAL_CPUTIME in their target, project or bsp. See the jira ticket
for more information.
|
|
This fixes the compilation error when BLE_LL_CFG_FEAT_LE_ENCRYPTION
was set to 0. The issue here was that BLE_LL_CFG_FEAT_LE_PING cannot
be set to 1 if encryption is not defined (set to 0). Note that
LE_PING can be either defined or undefined (1 or 0) if encryption
is defined.
|
|
Merge remote-tracking branch 'bgiori/newtmgr-id-bug' into develop
* bgiori/newtmgr-id-bug:
Fixed bug where ID was always wrong in response
|
|
https://github.com/SanjeewaUom/incubator-mynewt-core into develop
This closes #97
|
|
|
|
|
|
has been exported.
|
|
|
|
|
|
|
|
swithing images. Add boot_set_req() routine for unit test use.
|
|
|
|
|
|
|
|
Otherwise next restart we'll go back to old image.
|
|
|
|
|
|
bootutil_misc.c.
|
|
|
|
|
|
|
|
Merge branch 'fix/nmgr-config' of https://github.com/simonratner/incubator-mynewt-core into develop
|
|
Merge branch 'nffs-update' of https://github.com/peterfs/incubator-mynewt-core into develop
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The changes for this ticket caused the code to ack a packet at
the link layer even though it was not handed to the host. This
is an error; the link layer should only ack data packets that
were handed to the host.
|
|
Reported by Nges.
|
|
|
|
Before, we were logging each ATT transmission as soon as a higher layer
requested the send.
Now:
* For client requests, don't log until after transmit succeeds.
* For server responses, don't log until we are sure we will attempt a
transmission. Unfortunately, due to the way the server code is
structures, we don't have the necessary information for logging after
the transmit occurs. This is still an improvement, as it prevents
spurious logging when there are invalid arguments or insufficient
resources.
|
|
Before this commit, MTU changes were only reported when we initiated the
exchange via a GATT procedure. If the peer initiated the exchange, the
application never learned of an MTU update.
Now all MTU changes are reported through the GAP event callback. The
callback passed to ble_gattc_exchange_mtu() is not very useful anymore.
Its only real purpose is to inform the user if the peer failed to
respond to the MTU request, which causes the connection to be terminated
anyway.
Relevant additions to the API:
#define BLE_GAP_EVENT_MTU 15
struct ble_gap_event {
/* [...] */
/**
* Represents a change in an L2CAP channel's MTU.
*
* Valid for the following event types:
* o BLE_GAP_EVENT_MTU
*/
struct {
/** The handle of the relevant connection. */
uint16_t conn_handle;
/**
* Indicates the channel whose MTU has been updated; either
* BLE_L2CAP_CID_ATT or the ID of a connection-oriented
* channel.
*/
uint16_t channel_id;
/* The channel's new MTU. */
uint16_t value;
} mtu;
};
|
|
|
|
Prior to this commit, setting the preferred MTU only affected future
connections.
Now, an existing connection is updated if we haven't transmitted our
preferred MTU over it yet.
|
|
uint16_t ble_att_preferred_mtu(void);
|
|
|
|
|
|
In order to fix this issue I have added the local RPA to the resolving list and
modified the code so that the RPA timer is a true LL timer that will be used
to regenerate local RPAs at the specified timeout. This allows the local RPA
to be generated early and not during an IFS interval
|
|
with all of them
|
|
with all of them
|