search

ARMmbed / mbed-os

Arm Mbed OS is a platform operating system designed for the internet of things
https://mbed.com
Other
4.64k stars 2.97k forks source link

[BLE] When pairing with Android device (Android 10) the pairing request has to be confirmed twice #14654

Open kalle16lab opened 3 years ago

kalle16lab commented 3 years ago

Description of defect

When running the example provided Here and attempting to pair the device using nRF Connect app the pairing request confirmation dialog appears twice in a row. The vendor provided Cordio based BLE examples successfully pair on first try.

Target(s) affected by this defect ?

Sparkfun Artemis has been tested but issue does not seem to be Ambiq MCU specific.

Toolchain(s) (name and version) displaying this defect ?

GNU Arm Embedded Toolchain 9-2020-q2-update 9.3.1 20200408 (release)

What version of Mbed-os are you using (tag or sha) ?

mbed commit sha: 26c6b75d3b3af8c50238f372591bea9f58187bed

What version(s) of tools are you using. List all that apply (E.g. mbed-cli)

Mbed Studio 1.4.1

How is this defect reproduced ?

Compile the example and attempt to pair device to an Android phone using nRF Connect app.

ciarmcom commented 3 years ago

Thank you for raising this detailed GitHub issue. I am now notifying our internal issue triagers. Internal Jira reference: https://jira.arm.com/browse/IOTOSM-3961

idea--list commented 3 years ago

I find the same with the BLE_Advertising example found here.

The original example is non-connectable, however modifying line 69 like ble::advertising_type_t::CONNECTABLE_UNDIRECTED, we can convert it to a connectable example.

After doing so we have to connect twice in nRF Connect app in order to find any data under the CLIENT tab. So it seems the bug might have its roots somewhere else than just the SecurityManager API.

UPDATE: nRF Connect logs Error 133 (0x85): GATT ERROR

paul-szczepanek-arm commented 3 years ago

I'm having trouble reproducing this problem. In the BLE_SecurityAndPrivacy demo I connect and pair (and bond) once. After that the other device will disconnect automatically. When I reconnect again the bond is found and link is encrypted successfully without pairing. I'll try with the advertising example and more boards.

@kalle16lab when you run the example with two boards, do you get a normal log (pair once, and then uses bond information from then on)?

kalle16lab commented 3 years ago

I had not attempted to reconnect to device after pairing but I discovered that reconnecting fails.

The following trace contains boot of device, a pairing event (confirmation dialog on android device appears twice an I accepted twice) and afterwards I attempted to reconnect and this does not succeed.

Also from the main loop the following code has been removed and privacy has been disabled since the controller doesn't support random addresses.

/*
printf("\r\n * Device is a central *\r\n\r\n");
SecurityCentral central(ble, queue);
central.run();
*/
[INFO][BLCO]: WSF Cordio tracing enabled
[INFO][BLE ]: Allocated 9000 bytes for Cordio
[INFO][BLE ]: BLE features enabled: EXTENDED_ADVERTISING GATT_CLIENT GATT_SERVER PERIODIC_ADVERTISING PHY_MANAGEMENT PRIVACY SECURE_CONNECTIONS SECURITY SIGNING WHITELIST
[INFO][BLE ]: BLE roles enabled: PERIPHERAL CENTRAL
[INFO][BLE ]: Initialising Cordio host stack

 * Device is a peripheral *

[INFO][BLPR]: Initialize privacy PAL
[INFO][BLE ]: Initialising BLE instance
[INFO][BLHC]: CordioHCIDriver initializing
[INFO][BLE ]: Reset sequence complete. Controller supported features:
[INFO][BLE ]: 21:00:00:00
[INFO][BLE ]: Initialising extended features
[INFO][BLGS]: Initialize GattServer
[INFO][BLGS]: Add Generic Access Service to the Gatt Server: handle=1
[INFO][BLGS]: Add Device Name characteristic: handle=3
[INFO][BLGS]: Add Appearance characteristic: handle=5
[INFO][BLGS]: Add Peripheral Preferred Connection characteristic: handle=7
[INFO][BLGS]: Add Generic Attribute Service to the Gatt Server: handle=8
[INFO][BLGS]: Add Service Changed characteristic: handle=10
[INFO][BLGP]: Get address - type=PUBLIC, address=00:00:00:00:00:00
Device address: 66:77:88:23:bb:ef
[INFO][BLSM]: SM init: bondable=true, mitm=false, iocaps=IO_CAPS_NONE, passkey=0, signing=false, db_path=0x35984
[INFO][BLSM]: Initialize PAL SM
[INFO][BLSM]: Initialize database. Path = /fs/bt_sec_db
[WARN][BLDB]: Security DB didn't exist, creating new one
[ERR ][BLDB]: Failed to create security DB
[INFO][BLDB]: Using memory security DB (capacity 5 entries) - no persistence across reset
[INFO][BLSM]: Set stack I/O capabilities: NO_INPUT_NO_OUTPUT
[INFO][BLSM]: Set stack display passkey: 30:30:30:30:30:30
[INFO][BLSM]: Initialize resolving list
[INFO][BLSM]: Identity list retrieved: 0 entries to add to the resolving list
[INFO][BLSM]: Clear resolving list
[INFO][BLSM]: Init identity
[INFO][BLSM]: Stack generated random data: af:8e:7a:05:d6:51:e5:17
[INFO][BLSM]: Stack generated random data: 3a:24:17:2d:05:58:a3:1f
[INFO][BLSM]: Random data generated: af:8e:7a:05:d6:51:e5:17:3a:24:17:2d:05:58:a3:1f
[INFO][BLSM]: Set stack local IRK: af:8e:7a:05:d6:51:e5:17:3a:24:17:2d:05:58:a3:1f
[INFO][BLSM]: Set local irk: af:8e:7a:05:d6:51:e5:17:3a:24:17:2d:05:58:a3:1f
[INFO][BLPR]: PAL start generation of RPA from local irk: af:8e:7a:05:d6:51:e5:17:3a:24:17:2d:05:58:a3:1f
[INFO][BLSM]: Set stack local identity address: address = 00:00:00:00:00:00, public = false
[INFO][BLSM]: Set manual handling of pairing request: true
[INFO][BLSM]: Set role reversal hint: true
[INFO][BLGP]: Set peripheral privacy configuration - use_non_resolvable_random_address=false, resolution_strategy=PERFORM_PAIRING_PROCEDURE
[INFO][BLGP]: Advertising set 0: set advertising data - payload=02:01:06:0d:09:53:65:63:75:72:69:74:79:44:65:6d:6f, minimiseFragmentation=true, scan_response=false
[INFO][BLGP]: Advertising set 0: set extended advertising parameters - _advType=CONNECTABLE_UNDIRECTED, _minInterval=640ms, _maxInterval=1280ms, _peerAddressType=PUBLIC, _ownAddressType=RANDOM, _policy=NO_FILTER, _primaryPhy=LE_1M, _secondaryPhy=LE_1M, _peerAddress=00:00:00:00:00:00, _txPower=127, _maxSkip=0, _channel37=true, _channel38=true, _channel39=true, _anonymous=false, _notifyOnScan=false, _legacyPDU=true, _includeHeaderTxPower=false
[INFO][BLDM]: Legacy PDU w/undirected advertising event
[INFO][BLDM]: Advertising set 0: update direct advertising parameters - advertising_type=0, peer_address=00:00:00:00:00:00, peer_address_type=PUBLIC
[INFO][BLGP]: Advertising set 0: set advertising parameters - _advType=CONNECTABLE_UNDIRECTED, _minInterval=640ms, _maxInterval=1280ms, _peerAddressType=PUBLIC, _ownAddressType=PUBLIC, _policy=NO_FILTER, _primaryPhy=LE_1M, _secondaryPhy=LE_1M, _peerAddress=00:00:00:00:00:00, _txPower=127, _maxSkip=0, _channel37=true, _channel38=true, _channel39=true, _anonymous=false, _notifyOnScan=false, _legacyPDU=true, _includeHeaderTxPower=false
[INFO][BLDM]: Advertising set 0: update direct advertising parameters - advertising_type=0, peer_address=00:00:00:00:00:00, peer_address_type=PUBLIC
[INFO][BLGP]: Advertising set 0: start advertising - maxDuration=10000ms, maxEvents=0
[INFO][BLDM]: Set random address - address=00:00:00:00:00:00
[INFO][BLDM]: Advertising enable - enable=true
Advertising...
[INFO][BLPR]: Privacy handling: DM_PRIV_GENERATE_ADDR_IND
[INFO][BLSM]: Resolvable private address generated: 0c:87:c5:3d:18:7c
[INFO][BLGP]: Private address generated - connectable=true
[ERR ][BLGP]: privacy enabled, not passing to event handler
[INFO][BLGP]: Legacy advertising started
[INFO][BLGP]: Connection 1 successfully completed - role=PERIPHERAL, peer_address_type=RANDOM peer_address=ec:19:ad:2f:3a:5a, connection_interval=36, connection_latency=0, supervision_timeout=500, local_resolvable_private_address=00:00:00:00:00:00, peer_resolvable_private_address=00:00:00:00:00:00
[INFO][BLDM]: Advertising enable - enable=false
[INFO][BLSM]: Connection 1 - Open: ltk=false, csrk=false, irk=false
Connected to peer: 5a:3a:2f:ad:19:ec
[INFO][BLSM]: Connection 1 - Set link security: SECURITY_MODE_ENCRYPTION_NO_MITM
[INFO][BLSM]: Connection 1 - Set encryption: ENCRYPTED
[INFO][BLSM]: Connection 1 - Enable encryption
[INFO][BLSM]: Connection 1 - Ltk not available, send slave security request
[INFO][BLSM]: Connection 1 - Initiate security request
[INFO][BLSM]: Connection 1 - Pass slave security to the stack: authentication = 11
[INFO][BLSM]: Connection 1 - Request authentication
[INFO][BLSM]: Connection 1 - ltk not mitm protected, issue slave security request
[INFO][BLSM]: Connection 1 - Initiate security request
[INFO][BLSM]: Connection 1 - Pass slave security to the stack: authentication = 15
[INFO][BLGP]: Connection 1 update successfully completed - connection_interval=6, connection_latency=0,supervision_timeout=500,
[INFO][BLGP]: Connection 1 update successfully completed - connection_interval=36, connection_latency=0,supervision_timeout=500,
[INFO][BLSM]: Handling event DM_SEC_PAIR_IND
[INFO][BLSM]: Connection 1 - pairing request received: use_oob = false, authentication = 0D, initiator_dist = 0F, responder_dist = 0F
[INFO][BLSM]: Connection 1 - Forwarding pairing request to user handler
Pairing requested - authorising
[INFO][BLSM]: Connection 1 - Accept pairing request
[INFO][BLSM]: Connection 1 - Update OOB presence: false
[INFO][BLSM]: Connection 1 - Send pairing response: oob = false, auth = 15, initiator dist = 03, responder dist = 03
[INFO][BLSM]: Handling event DM_SEC_ENCRYPT_IND
[INFO][BLSM]: Connection 1 - Link encryption: updated = ENCRYPTED
Link ENCRYPTED
[INFO][BLSM]: Handling event DM_SEC_KEY_IND
[INFO][BLSM]: Connection 1 - Local LTK distributed: 96:08:1a:73:31:9b:9a:4b:f9:19:03:45:58:2c:b5:44
[INFO][BLDB]: Write DB entry 0: local ltk 96:08:1a:73:31:9b:9a:4b:f9:19:03:45:58:2c:b5:44
[INFO][BLSM]: Connection 1 - Local EDIV/RAND distributed: EDIV = 78:fb, RAND = 13:21:6b:e1:d3:43:ab:ac
[INFO][BLDB]: Write DB entry 0: local ediv 78:fb rand 13:21:6b:e1:d3:43:ab:ac
[INFO][BLSM]: Handling event DM_SEC_KEY_IND
[INFO][BLSM]: Connection 1 - LTK distributed: 3a:bf:c0:18:38:59:7e:39:02:22:fa:06:63:d6:2a:d4
[INFO][BLDB]: Write DB entry 0: peer ltk 3a:bf:c0:18:38:59:7e:39:02:22:fa:06:63:d6:2a:d4
[INFO][BLSM]: Connection 1 - EDIV and RAND distributed: ediv = be:54, rand = df:4e:75:e7:1a:db:37:80
[INFO][BLDB]: Write DB entry 0: peer ediv be:54 rand df:4e:75:e7:1a:db:37:80
[INFO][BLSM]: Handling event DM_SEC_KEY_IND
[INFO][BLSM]: Connection 1 - Peer identity distributed: address = cd:3f:60:2e:b7:dc, public = true
[INFO][BLDB]: Write DB entry 0: public peer address cd:3f:60:2e:b7:dc
[INFO][BLSM]: Connection 1 - IRK distributed: 90:7e:eb:76:4a:f8:9f:d5:c4:73:1b:f9:2b:49:c6:e1
[INFO][BLDB]: Write DB entry 0: peer irk 90:7e:eb:76:4a:f8:9f:d5:c4:73:1b:f9:2b:49:c6:e1
[INFO][BLSM]: Handling event DM_SEC_PAIR_CMPL_IND
[INFO][BLSM]: Connection 1 - Pairing successfully completed
[INFO][BLSM]: Connection 1 - Retrieve identity address to register it into resolving list
[INFO][BLSM]: Security identity retrieved for entry 0x100086A0: address: cd:3f:60:2e:b7:dc, public: true, irk: 90:7e:eb:76:4a:f8:9f:d5:c4:73:1b:f9:2b:49:c6:e1
[INFO][BLSM]: Add device to resolving list: peer address=cd:3f:60:2e:b7:dc, type=PUBLIC, peer irk=90:7e:eb:76:4a:f8:9f:d5:c4:73:1b:f9:2b:49:c6:e1
Pairing successful
[INFO][BLGP]: Connection 1: disconnect - reason=USER_TERMINATION
[INFO][BLGP]: Disconnection 1 successfully completed - reason=22
[INFO][BLSM]: Connection 1 - Process disconnection
[INFO][BLSM]: Connection 1 - Remove stack peer CSRK
Disconnected.

We are bonded, we will only accept known devices
[INFO][BLGP]: Set peripheral privacy configuration - use_non_resolvable_random_address=false, resolution_strategy=REJECT_NON_RESOLVED_ADDRESS
[INFO][BLGP]: Advertising set 0: set advertising data - payload=02:01:06:0d:09:53:65:63:75:72:69:74:79:44:65:6d:6f, minimiseFragmentation=true, scan_response=false
[INFO][BLGP]: Advertising set 0: set advertising parameters - _advType=CONNECTABLE_UNDIRECTED, _minInterval=640ms, _maxInterval=1280ms, _peerAddressType=PUBLIC, _ownAddressType=PUBLIC, _policy=NO_FILTER, _primaryPhy=LE_1M, _secondaryPhy=LE_1M, _peerAddress=00:00:00:00:00:00, _txPower=127, _maxSkip=0, _channel37=true, _channel38=true, _channel39=true, _anonymous=false, _notifyOnScan=false, _legacyPDU=true, _includeHeaderTxPower=false
[INFO][BLDM]: Advertising set 0: update direct advertising parameters - advertising_type=0, peer_address=00:00:00:00:00:00, peer_address_type=PUBLIC
[INFO][BLGP]: Advertising set 0: start advertising - maxDuration=0ms, maxEvents=0
[INFO][BLDM]: Set random address - address=00:00:00:00:00:00
[INFO][BLDM]: Advertising enable - enable=true
Advertising...
[INFO][BLGP]: Legacy advertising started
[INFO][BLGP]: Connection 1 successfully completed - role=PERIPHERAL, peer_address_type=RANDOM peer_address=ec:19:ad:2f:3a:5a, connection_interval=36, connection_latency=0, supervision_timeout=500, local_resolvable_private_address=00:00:00:00:00:00, peer_resolvable_private_address=00:00:00:00:00:00
[INFO][BLDM]: Advertising enable - enable=false
[INFO][BLGP]: Connection 1: disconnecting - reason=AUTHENTICATION_FAILURE
[INFO][BLGP]: Disconnection 1 successfully completed - reason=5
[INFO][BLSM]: Connection 1 - Process disconnection
[ERR ][BLSM]: Connection 1 - Error, control block not available
Disconnected.

Sidenote: tracing on BLE is quite buggy and broken. Intermittently I get file accessed in ISR crashes or other seg-fault crashes.

Advertising...
[INFO][BLPR]: Privacy handling: DM_PRIV_GENERATE_ADDR_IND
[INFO][BLSM]: Resolvable private address generated: 16:2d:46:05:d8:5e
[INFO][BLGP]: Private address generated - connectable=true
[ERR ][BLGP]: privacy enabled, not passing to event handler
[INFO][BLGP]: Legacy advertising started

++ MbedOS Error Info ++
Error Status: 0x8001012F Code: 303 Module: 1
Error Message: Error - writing to a file in an ISR or critical section

Location: 0x2EDA1
Error Value: 0x1
Current Thread: rtx_idle Id: 0x10001708 Entry: 0xFAB5 StackSize: 0x200 StackMem: 0x10001A50 SP: 0x1005FD4C
For more info, visit: https://mbed.com/s/error?error=0x8001012F&tgt=AMBIQ_EVK
noonfom commented 3 years ago

@kalle16lab Thanks for reporting the issue with BLE traces. I suspect it's related to #14574. Unfortunately, that PR got blocked by #14426 and other CI-related issues. I removed the blocking commits, so we should be able to merge the fix ASAP.

AGlass0fMilk commented 3 years ago

I will preface this with the information that I'm working on the latest Mbed-OS master (14e5d307bb6cdccb554b591ab2602d8d47e0b2d0) with the fixes from #14574 and #14673

I am seeing a similar issue with Android (Samsung Galaxy S10 Lite, Model #: SM-G770U1, Android version 10).

I do not enable privacy at runtime (I have not disabled privacy at the configuration level yet) and I have a file system set up to store the bonding credentials. When I connect with the above Android phone (using PunchThrough LightBlue), the pairing screen shows up twice. Attempting to reconnect fails the first time. The second reconnection attempt shows the two pairing popups again. I assume the credentials are being purged by the Anrdoid phone after the first failed reconnection attempt.

I also tried to pair/bond with nRF Connect Desktop using an nRF52840 dongle as the interface. In this setup, the pairing goes fine the first time (no double pairing popup). A subsequent reconnection fails with the following log output in nRF Connect:

image

Here are the corresponding BLE traces on the Mbed device side (not sure if I can enable even more traces, let me know if I can):

nrf-connect-hci-different-transaction-collision.log

Traces for the Android connection: android-pairing-trace.log (Note: when I collected these logs, the first reconnection attempt just performed pairing twice again...)

@noonfom @pan- @paul-szczepanek-arm

I have several customer projects that require BLE pairing. This is one of the last outstanding issues for most of them. I am willing to work closely with you to get this fixed as quickly as possible...

I think Mbed-OS really needs some automated testing infrastructure to ensure compatibility with major operating systems over BLE...

AGlass0fMilk commented 3 years ago

This is related to the defect I reported in #14611 @paul-szczepanek-arm

pan- commented 3 years ago

@AGlass0fMilk @kalle16lab I made some progress and observed the following things:

The workaround at the moment is to keep privacy disabled and set the peripheral privacy strategy to DO_NOT_RESOLVE.

@paul-szczepanek-arm Can you make sure that Gap::apply_peripheral_privacy_connection_policy is only applied when privacy is enabled ? Slave security request is issued according to the privacy strategy even if the privacy is not enabled.

AGlass0fMilk commented 3 years ago

@pan-

Thanks for the update. I did notice that when I configured ble.ble-feature-privacy to false the issue did not happen.

Any idea what might be causing the BLE_HCI_DIFFERENT_TRANSACTION_COLLISION error when reconnecting with nRF connect? Not that this needs to be a supported peer, but we shouldn't have problems with any peer. I still see this issue even when I disable privacy at the configuration level.

I will do some more testing and see if there are still pairing/bonding issues with iOS/Android.

Did my sniffer logs help at all? I can collect more of them if so. I'm apprehensive to share them publicly on GH because they may contain some customer details.

AGlass0fMilk commented 3 years ago

Also, what do you think about the automated integration testing with iOS/Android/desktop operating systems...?

I would target mobile OS's first. Write a simple native app for iOS and Android that performs various common use cases (eg: unpaired connection, paired connection, reading/writing characteristics, etc). For Android the automation is simple, you can probably accomplish everything through the android debug bridge (adb).

For iOS, it's obviously a bit more complicated. You need a Mac to build. I haven't done iOS development before, but from my research Xcode has support for (on-device?) unit testing and "UI testing". UI testing seems to allow you to automate interacting with UI elements (eg: pressing buttons to pair, etc). So perhaps that's a path forward...

I know there's a lot on the roadmap already, but I really think this would help find bugs before they become issues. Otherwise, we're using our userbase as "automated testing" :grin:

AGlass0fMilk commented 3 years ago

With ble.ble-feature-privacy disabled, I observe the following behavior on Android 10:

My guess is that Android bonding (at least in Android 10) assumes privacy will be enabled and therefore expects an IRK to be able to resolve a nearby random address to a bonded device... since the Mbed device does not use privacy, the Android device cannot "see" the bonded device nearby to even attempt to connect to it.

Another possibility is that the Mbed device, even with privacy disabled, is sending an incorrect device address and/or address type. This again leads to the Anrdoid device being unable to find the bonded device nearby, even though it is right there.

@pan- Thoughts?

AGlass0fMilk commented 3 years ago

@AGlass0fMilk @kalle16lab I made some progress and observed the following things:

* The issue does not affect all android versions. It is visible on Android 10 but not on Android 8 for example.

* The issue is visible when the device issue a slave security request.

* The issue doesn't seems to be related to timing, sending the slave security request 5 seconds after the connection has been established displays the pairing pop up twice.

* The issue is visible when privacy is not enabled.

The workaround at the moment is to keep privacy disabled and set the peripheral privacy strategy to DO_NOT_RESOLVE.

@paul-szczepanek-arm Can you make sure that Gap::apply_peripheral_privacy_connection_policy is only applied when privacy is enabled ? Slave security request is issued according to the privacy strategy even if the privacy is not enabled.

This workaround works well for me with Android and iOS. Although, for some reason, after a power cycle the connection fails, the Mbed device logs say:

[WARN][BLDB]: Failed to find ltk matching given ediv&rand

Which is strange because the file system is definitely working properly...

I am shutting down BLE gracefully so the bonding database file should be flushed... I will have to investigate...

I'm wondering if there's a way we can get pairing/bonding and privacy working with mobile OS's on Mbed... in some applications (eg: wearables) that is critical.

pan- commented 3 years ago

I'm wondering if there's a way we can get pairing/bonding and privacy working with mobile OS

It used to work last year when we integrated privacy, at the time it was extensively tested on iOS and Android devices. Automated integration testing with those platforms would definitely make sense but it would be hard to cover all Android I'm afraid.

[WARN][BLDB]: Failed to find ltk matching given ediv&rand

@paul-szczepanek-arm Can you look into that ? It doesn't seems related to the issue.

AGlass0fMilk commented 3 years ago

I'm wondering if there's a way we can get pairing/bonding and privacy working with mobile OS

It used to work last year when we integrated privacy, at the time it was extensively tested on iOS and Android devices. Automated integration testing with those platforms would definitely make sense but it would be hard to cover all Android I'm afraid.

You would think with an open specification interoperability wouldn't be an issue... but when Android or iOS (especially iOS) make mistakes/assumptions they push it onto us to workaround :sweat:

AGlass0fMilk commented 3 years ago

[WARN][BLDB]: Failed to find ltk matching given ediv&rand

@paul-szczepanek-arm Can you look into that ? It doesn't seems related to the issue.

Hold on, I don't want to throw you on a wild goose chase... let me investigate a bit.

I am seeing this after bonding with Android, then disconnecting, then reconnecting:

[WARN][BLDB]: Found ltk matching given ediv&rand but it belonged to a different identity, update entry with new identity

Maybe the Android device had cycled its random address? Not sure what's going on there.

pan- commented 3 years ago

Maybe the Android device had cycled its random address? Not sure what's going on there.

The address should cycle but not the IRK or the identity address. That's something we've been specifically testing while implementing privacy.

AGlass0fMilk commented 3 years ago

Again, I am encountering the issue where the FileSecurityDb never actually closes the file handle and thus the bonding information never gets flushed to flash...

I have tried both with an LFS instance and a FAT instance on an external SPIFBlockDevice.

This may need its own issue but I really think we need to refactor the FileSecurityDb to operate more like "open, modify, close". In some applications, there isn't always a nice shutdown procedure (eg: hard power off with a switch) and in this case you cannot leave file handles open. That's just asking for corruption.

@paul-szczepanek-arm I will work on a refactor that works for me in the short term and I can do a PR. Do you see any issue with opening and closing the file each time it needs to be modified? Was there a reason (eg: latency) that you chose to originally keep the file open until the database was destroyed?

Currently, the only way the database gets destroyed (and thus the file is closed) is when the Security Manager itself gets destroyed... which I think never ever happens since it is a static singleton (as far as I can see).

AGlass0fMilk commented 3 years ago

I guess it does get deleted in SecurityManager::reset... I'll see if this gets called during BLE shutdown.

EDIT: I cannot find anywhere that SecurityManager::reset gets called by the stack itself. This doesn't seem like a great thing to do during the course of normal BLE operation... I still think we should go to the "open, write, close" tactic with the FileSecurityDb.

EDIT 2: I now realize why the one method is called preserveBondingStateOnReset but I'm not sure why you must reset the security manager to flush the file. Shouldn't we have some sort of flushBondingState or something that doesn't modify the state of the security manager?

Also, nowhere in the mbed-os-example-ble for security and privacy is SecurityManager::reset called, nor mentioned...

AGlass0fMilk commented 3 years ago

So I added the reset and reinitialization of the security manager upon disconnection and I can now pair with Android/iOS devices and have it persist between power cycles, great.

I still see the message when reconnecting to Android:

[WARN][BLDB]: Found ltk matching given ediv&rand but it belonged to a different identity, update entry with new identity

It may only happen the first reconnection but I'll have to look into it more. It does not seem to affect the ability of the devices to reconnect and immediately encrypt the link, though. @paul-szczepanek-arm Let me know if you want me to get you more complete logs of this event.

In the meantime, we need to figure out why the pairing screen appears twice when the device has privacy enabled at runtime.

AGlass0fMilk commented 3 years ago

This definitely needs its own issue at this point, but I wanted to point out (after testing myself) that using the KVStoreSecurityDb, you don't need to call SecurityManager::reset for the bonding information to be saved... so this is an inconsistency in the API.

Also, had to fix the KVStore implementation to get it to build, see #14684