Hi Luiz, On 08/03/17 11:19, Luiz Augusto von Dentz wrote: > Hi Felipe, > > On Wed, Mar 8, 2017 at 12:48 PM, Felipe Ferreri Tonello > <eu@xxxxxxxxxxxxxxxxx> wrote: >> Hi Luiz, >> >> Sorry for the late reply, I was unwell this last week. >> >> On 07/03/17 09:15, Luiz Augusto von Dentz wrote: >>> Hi Felipe, >>> >>> On Tue, Feb 21, 2017 at 11:40 AM, Felipe Ferreri Tonello >>> <eu@xxxxxxxxxxxxxxxxx> wrote: >>>> Hi Luiz, >>>> >>>> On 20/02/17 15:21, Luiz Augusto von Dentz wrote: >>>>> Hi Felipe, >>>>> >>>>> On Thu, Feb 16, 2017 at 8:03 PM, Felipe F. Tonello <eu@xxxxxxxxxxxxxxxxx> wrote: >>>>>> This is one of the parts of a series of patches related to connection >>>>>> parameters update. >>>>>> >>>>>> This initial implementation adds supports to GAP Peripheral Preferred >>>>>> Connection Parameters (PPCP) characteristic and the Slave Connection Interval >>>>>> Range Advertising Data (AD). >>>>> >>>>> >>>>> Are there any devices implementing these? Id like to do a some testing >>>>> but haven't find anything implementing this, or well I can generate >>>>> the PPCP but the SCRAD is probably a waste of AD since that is pretty >>>>> limited in size. >>>> >>>> PPCP not that I know of. But the AD there are devices, including in the >>>> unit-eir there is one. >>>> >>>> The PPCP and AD is important specially for MIDI devices who require >>>> low-latency. >>>> >>>> I did test it here, it will be nice if you could as well. >>> >>> Lets try to get some points resolved here, Ive briefly chatted with >>> Johan regarding this last week and he said that we might have to >>> introduce a new command to add new parameters to the kernel, as oppose >>> to reuse load. >>> >>> We might have to choose between these 2 options: >>> >>> 1. Reload all the settings, and then move on to implement the new command. >> >> Reload all settings as in bluetoothd reloading all settings? >> >> What new command? >> >>> 2. Don't bother with old kernels and just make it work with newer one >>> with proper support. >> >> New command to update the settings for one device only? > > Yep, the reason for that is that changing the existing logic of load > may break the API for older kernels as it would no longer remove > existing settings but just update. Right. I think it is better too. Could you please, then, take a look at my kernel RFC patch[1]? I'll use that patch-set to create this new MGMT command. [1] The email subject is "[RFC][PATCH v2 BlueZ 0/2] Connection Update improvements". > >>> >>>>> >>>>>> There is a problem described in #4 that due to a limitation of the BlueZ API >>>>>> it is not possible to update the current connection with specific parameters. >>>>>> I am working on a socket option API to allow something similar. Hopefully I >>>>>> will be able to write something that covers this use-case as well. This >>>>>> problem also applies to patch #6. >>>>>> >>>>>> Changes from v1: >>>>>> * Added Slave Connection Interval Range AD support; >>>>>> * Unit tests for connection interval. >>>>>> >>>>>> Felipe F. Tonello (7): >>>>>> profiles/gap: Some code cleanup >>>>>> src/adapter: Added connection parameter load/store functions >>>>>> src/device: Added function to set connection parameters >>>>>> profiles/gap: Added support for PPCP characteristic >>>>>> src/eir: Added support for Slave Connection Interval Range AD >>>>>> src/adapter: Store Slave Connection Interval Range >>>>>> unit/test-eir: Added tests for connection interval >>>>>> >>>>>> profiles/gap/gas.c | 111 +++++++++++++++++++++++++++++++++++++++++++---------- >>>>>> src/adapter.c | 46 +++++++++++++++++++++- >>>>>> src/adapter.h | 9 +++++ >>>>>> src/device.c | 16 ++++++++ >>>>>> src/device.h | 4 ++ >>>>>> src/eir.c | 7 ++++ >>>>>> src/eir.h | 3 ++ >>>>>> unit/test-eir.c | 26 +++++++++++-- >>>>>> 8 files changed, 196 insertions(+), 26 deletions(-) >>>>>> >>>>>> -- >>>>>> 2.11.1 >>>>>> -- Felipe
Attachment:
0x92698E6A.asc
Description: application/pgp-keys
