disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Mon, 11 May 2020 09:08:41 GMT

I’ve had Leo's firmware USB840M_200527_UARTLog_Pin6.rar, from late on May 6, running on two devices for three days without ever seeing the ‘no event’ error. I’m happy to say this seems to be solved. I’ve had it running on the Fanstel dongle and the Nordic Development kit.

I'll summarize all the changes since this support case started.

1. In my application, sd_ble_gap_connect was being called with ble_gap_scan_params_t.timeout=0 so that if the connection was not established it would not receive event BLE_GAP_EVT_TIMEOUT(BLE_GAP_TIMEOUT_SRC_CONN)

2. In my application, all the sd_ble_gattc_* operations needed to watch for event BLE_GATTC_EVT_TIMEOUT to indicate the operation would not complete successfully

3. In my application, all the sd_ble_gattc_* operations needed to watch for event BLE_GAP_EVT_DISCONNECTED to indicate the operation would not complete successfully.

4. In the connectivity firmware, fixed race condition around resuming SoftDevice events, see April 24 post from Jorgen. SDK file examples\connectivity\ble_connectivity\main.c

5. In connectivity firmware, increase scheduler queue size from 16 to 64, see Jorgen post April 29, and in more detail on May 5: change SER_CONN_SCHED_QUEUE_SIZE from 16 to 64 in SDK components\serialization\connectivity\ser_conn_handlers.h

The only remaining step is to get those changes into the released SDK.

Thanks,

Paul

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

LeoChen — Thu, 07 May 2020 02:03:33 GMT

Hi,

The following link for log on P0.06, clock internal.

https://www.dropbox.com/s/q7ddx4zqw86x0q4/USB840M_200527_UARTLog_Pin6.rar?dl=0

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Wed, 06 May 2020 19:13:38 GMT

I did a quick test run with logging enabled for all SD events and function-calls in the connectivity FW. I tried to synch the clocks as good as possible (second accuracy), as the logger and application run on different computers This is the output from your application:

2020-05-06 20:58:36.352: (WARNING) restart scanning because we've gone 10 timer ticks with no advertisements

This is the log file from the connectivity FW: devzone.nordicsemi.com/.../20200506_5F00_sdcalls_5F00_events.log

From what I can see, it looks like there are ADV_REPROT events generated about every 1-2 seconds before the reset in your application, so I'm not sure if the events are not passed fast enough to your application, or if they are not processed quickly enough in your application. There is also a huge amount of RSSI_CHANGED events, which may slow things down. Do you require to be notified on every RSSI change in your application? You can set the threshold/skip count in the softdevice API call: sd_ble_gap_rssi_start().

What connection parameters do you have for the connection? From the RSSI events, it could look like the connection interval is set to 30 ms? What about connection event length, slave latency, etc? This will all affect the available radio time for doing scanning. Since you are doing advertising simultaneously, you may reach the radio time limit if your parameters are not carefully selected. Check out the Scheduling chapter in the softdevice specifications.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Wed, 06 May 2020 15:06:01 GMT

"Receiving no advertisements" means any advertisement at all, not just from the tracked phone. The timer tick is a one-second period where we aren't doing any work other than scanning - since we're only scanning, we expect we'd received advertisements. Between timer ticks we do stuff like connecting to phones, so N timer ticks can be much longer than N seconds. This issue of not receiving advertisements is a real problem that I'll need to resolve, but I don't want it to interfere with this support case, so I've uploaded modified test program bgSDKTestAppMay6.zip that doesn't reset the nRF52840.

I was able to wire the pins for logging. I removed the wire to use Leo's firmware that did RTT logging (which so far is not reproducing the issue).

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Wed, 06 May 2020 14:00:30 GMT

[quote user="Paul Bradford"]after a prolonged period of receiving no advertisements[/quote]

Does this include all advertisements or only advertisements from the tracked phone? How long time is "10 timer ticks"? I see this message a lot when I leave the phone for a while, it kinda seems like the phone/application enters some kind of state where it pauses/terminates the application. I have tried to disable power optimization for the app on the phone. I have verified with the 's' command that the phone is actually being tracked, but sometimes it does report that it is not tracked, until I unlock the phone. I have not purposedly set any configs on the phone side, it should use the default settings of the Samsung Galaxy S7. I do not see this event on when testing Huawei P30 Pro.

Let me know if you are not able to wire the pins for logging, I can build you a version with log on P0.06.

@leochen: I was not able to reproduce the original issue with RTT logging enabled/RTT Viewer open. If you are always running this with the firmware, it could be the reason that you did not manage to reproduce the problem.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

LeoChen — Wed, 06 May 2020 03:23:04 GMT

Hi Jorgen,

Thanks for detail.

I found where cause the new problem on my code.

I am following the latest Nordic patch file and compiler again.

I already installed three dongles to try.

My log is setup to RTT Viewer.

https://www.dropbox.com/s/z5xll23kkcpaxks/USB840M_200506.rar?dl=0

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Tue, 05 May 2020 16:39:07 GMT

The reset you are seeing is induced by the test program, and has changed since prior versions. This test program uses lots of code from our product, and one thing it does is reset the nRF52840 after a prolonged period of receiving no advertisements, since we're scanning constantly and should always be seeing advertisements - we take this 5+ minutes with advertisements to mean the chip has failed, so we reset it. Note that earlier we had noticed that advertisements stopped and had tried to restart them: "restart scanning because we've gone 10 timer ticks with no advertisements". Does the firmware logging show what's going on that causes no advertisements to be received?

It may not mean anything, but the line of output from the test program "Nordic event handler received an un-handled event with ID: 30" indicates it received event BLE_GAP_EVT_SEC_REQUEST which the product does not process. I don't recall ever seeing that event. Is your test phone somehow set up to requiring pairing or bonding? If it were, it would not successfully stay connected to the test program.

When you run the test program, do you verify that shortly after starting it has connected to the phone ("phone is being tracked") as described in the README?

On the PuTTY question from earlier, I had missed the instruction about wiring pins on the DK. I'll work on this with the limited equipment I have at home.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Tue, 05 May 2020 14:32:35 GMT

Hi Paul,

I tracked down the reset that I'm seeing to a call to sd_nvic_SystemReset() in ser_conn_generic_command_process() when a SER_GENERIC_CMD_RESET command is received. This indicates that the pc-ble-driver application is issuing a reset command. This also looks to be the case in the log output from your application:

>bgsdkTestApp.exe -u27527585-e5bb-4697-b0af-0e92785043b6 -f5 -v0 -cCOM4
use challenge/response frequency of 5 seconds
logging level 0
'd' to disconnect any connected devices
'q' to quit program
'r' to do a hard reset of the dongle - after this the program exits
's' to show status of all monitored phones
'+' to increase logging level
'-' to decrease logging level
2020-05-05 15:49:48.999: (WARNING) Nordic event handler received an un-handled event with ID:  30
2020-05-05 15:50:31.955: (WARNING) restart scanning because we've gone 10 timer ticks with no advertisements
2020-05-05 15:50:33.799: (ERROR) Error:  Failed to decode event, error code is 14/0xe.
2020-05-05 15:50:33.924: (WARNING) bgsdk::details::AdvDataParser::Parse MAC =  42:f5:68:c6:cd:d1 failed to parse 1e ff 06 00 01 09 20 02 9b 89 59 dc cd f2 0f a9 46 2f b7 61 7a  invalid advertisement: section length = 30 results in end index 31 > size 21
2020-05-05 15:50:47.656: (WARNING) restart scanning because we've gone 10 timer ticks with no advertisements
2020-05-05 15:50:48.968: (ERROR) Error:  Failed to decode event, error code is 14/0xe.
2020-05-05 15:50:49.281: (ERROR) Error:  Failed to decode event, error code is 14/0xe.
2020-05-05 15:55:40.263: (WARNING) trigger reset hardware_radio_error because we've gone 300 timer ticks with no advertisements, indicating the Bluegiga hardware failed
IMPBGSDK reported HardwareFailure (for Nordic this happens when we don't detect advertisements for a long time)
IMPBGSDK reported HardwareSoftResetInitiated WITH DeviceRemoval (for Nordic this happens when we don't detect advertisements for a long time)
2020-05-05 15:55:40.910: (ERROR) Error:  serial port write operation on port COM4 failed. Error: The device does not recognize the command.[22]
2020-05-05 15:55:42.218: (WARNING) resetting Nordic failed:  0x802a NRF_ERROR_SD_RPC_H5_TRANSPORT_NO_RESPONSE
2020-05-05 15:55:43.773: (ERROR) Error:  Error purging UART 22
2020-05-05 15:55:43.779: (WARNING) std::exception during keepalive: sd_ble_gattc_read failed (Error: 32773) : 0x8005 NRF_ERROR_SD_RPC_NO_RESPONSE
2020-05-05 15:55:43.784: (ERROR) bgsdk::BeaconController::Impl::StopAdvertising Failed to stop advertising:  Nordic adapter closed (Error: 36865) : Unknown error
2020-05-05 15:55:43.786: (ERROR) bgsdk::BeaconController::Impl::StartAdvertising Failed to start advertising:  Nordic adapter closed (Error: 36865) : Unknown error
2020-05-05 15:55:43.787: (WARNING) start scanning caused exception:  Nordic adapter closed (Error: 36865) : Unknown error
2020-05-05 15:55:43.787: (WARNING) Event wait terminates because of global reset

This was seen when running your latest application (bgSDKTestAppMay4.zip). I did not see similar output in the old application. Did you make any changes to the application to make the chip reset? I will try to test with the Imprivata_bgTestApp.zip application to see if I'm still able to reproduce.

Best regards,
Jørgen

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Tue, 05 May 2020 12:09:25 GMT

Did you connect a wire between P0.31 and P0.06 on the DK, as I described in the last change-point below?

[quote user="joh2"]When running on DK, a wire must be connected between P0.31 and P0.06 to output logs. [/quote]

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Tue, 05 May 2020 11:41:45 GMT

I loaded 200505_ble_connectivity_s140_usb_hci_pca10056_lfrc_mergedsoftdevice.hex into the Nordic DK, and I'm running the test program. But when I start PuTTY to log the JLINK serial port, I see no output at all. I'm using the same procedure I've been using the last few days, including BAUD rate 115200.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Tue, 05 May 2020 10:52:39 GMT

I reproduced the issue with the hex-files from Leo as well. How exactly did you change the scheduler queue size? As far as I can see, this is still 16 in the latest project you uploaded (nRF5_SDK_15.3.0_59ac345_PC_Connect_200505.rar):

// In ser_conn_handlers.h:
/** Maximum number of events in the application scheduler queue. */
#ifdef S112
#define SER_CONN_SCHED_QUEUE_SIZE             8u
#else
#define SER_CONN_SCHED_QUEUE_SIZE             16u
#endif

// In main.c:
APP_SCHED_INIT(SER_CONN_SCHED_MAX_EVENT_DATA_SIZE, SER_CONN_SCHED_QUEUE_SIZE);

And when debugging, it is set to 16 when app_scheduler is initialized:

I created a clean patched SDK, with the appropriate fixes and a few changes. Can you please try building using this? These are the changes (can also be seen in attached patch-file:devzone.nordicsemi.com/.../changes.patch)

Added Critical region in main.c
Increased scheduler size to 64
Removed serialization error handler, use SDK default which will output error info
Added DEBUG and DEBUG_NRF preprocessor defines to Release build configuration, to catch errors and asserts.
Added separate build configurations for different LF clocks: Release_LFXO (external crystal) and Release_LFRC (internal RC oscillator).
Enabled UART logging, and set the log level to only output error logs (debug logs adds heavy CPU load).
Changed UART log TX pin to P0.31, to allow reading logs from nRF52840 Dongle. When running on DK, a wire must be connected between P0.31 and P0.06 to output logs. P0.31 on the dongle can be connected to P0.06 on an empty DK to read logs.

devzone.nordicsemi.com/.../nRF5_5F00_SDK_5F00_15.3.0_5F00_59ac345_5F00_patched_5F00_addedfixes.zip

Precompiled hex-files:

External LF crystal - devzone.nordicsemi.com/.../200505_5F00_ble_5F00_connectivity_5F00_s140_5F00_usb_5F00_hci_5F00_pca10056_5F00_lfxo_5F00_mergedsoftdevice.hex
Internal RC oscillator - devzone.nordicsemi.com/.../200505_5F00_ble_5F00_connectivity_5F00_s140_5F00_usb_5F00_hci_5F00_pca10056_5F00_lfrc_5F00_mergedsoftdevice.hex

I have not yet received the Fanstel Dongles, so I'm not sure if these will work with that. You may need to do some changes to support the bootloader etc.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Tue, 05 May 2020 08:18:58 GMT

I tested the firmware Leo referenced. With the dev-dongle, I tried both USB840M_200505_ClockInternal.zip and USB840M_200505_ClockExternal.zip and both failed with the "serial port write operation" error after a few minutes.

With the Nordic DK I tried USB840M_200505_ClockInternal_2in1.hex and got "serial port write operation" error after 30 minutes. I attached the firmware logging file 2020-05-05-035347-NordicDK_USB840M_200505_ClockInternal_2in1.hex.txt

The firmware logging ends with

[00:00:02.638,154] <debug> app: event:BLE_GATTC_EVT_HVX

[00:00:02.643,318] <debug> sphy_hci: TX request (34 bytes)

[00:00:02.648,884] <debug> sphy_hci: Started TX packet (payload 34).

[00:00:02.655,796] <error> app: Fatal error

[00:00:02.659,891] <error> app: ERROR 4 [NRF_ERROR_NO_MEM] at :0

PC at: 0x00000000

[00:00:02.667,605] <error> app: End of error report

[00:00:02.674,529] <info> app: BLE Clock Internal,UART log,Size 64,CRITICAL_REGION,

[00:00:02.682,869] <info> app: USB power detected

[00:00:02.688,683] <info> app: USB ready

I am now testing the Nordic DK with USB840M_200505_ClockExternal_2in1.hex

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

LeoChen — Tue, 05 May 2020 04:13:47 GMT

Hi Jorgen,

The following link is source project that I am using.

https://www.dropbox.com/s/t3pcunczygxxr7k/nRF5_SDK_15.3.0_59ac345_PC_Connect_200505.rar?dl=0

I did compare with the source patched project and can't see any thing wrong.

Would you please also give me the entire SDK that you are using to compare.

The Global setting

Macro:SDK=E:\Keil_v5\ARM\Device\Nordic\nRF5_SDK_15.3.0_59ac345_PC_Connect_200505

The Project path

~\nRF5_SDK_15.3.0_59ac345_PC_Connect_200505\examples\connectivity\ble_connectivity\pca10056\ser_s140_usb_hci\ses

The HEX files that replace error handling (External Clock does not work with Fanstel USB840M).

https://www.dropbox.com/s/z6u7f227h2sogl7/USB840M0505.rar?dl=0

How long you will see the error happen?

I programmed 3 dongles and running over 4 hours.

I can't see that error on my Win 7 desktop.

Regards,

Leo

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Mon, 04 May 2020 19:56:15 GMT

LeoChen: I did manage to reproduce once with today with the USB840_Clock_ACCURACY_2in1.hex file you provided on a Nordic Dongle. Unfortunately, it does not provide much information about the fatal error. I had modified this in the connectivity hex that I build with queue size of 64 and logging enabled. Can you replace this file in your SDK? It will give the location of the error and the error code on the log.

devzone.nordicsemi.com/.../ser_5F00_conn_5F00_error_5F00_handling.c

So far, I'm not able to reproduce the issue with the connectivity FW I'm building. Can you upload the SDK directory that you used, along with the path to the exact project used to build it?

@paul-bradford: I'm not sure this issue you are now seeing is the same as the original problem. They may look similar to the application, but I never saw the fatal error with my previous testing. I'm not sure if notifications are the cause of this anymore, but we need to find the origin of the fatal error in order to determine what is happening. Hopefully, you should be able to detect that quite fast with the above change I requested from Leo.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Mon, 04 May 2020 19:07:04 GMT

I uploaded a modified test program. The change is to have a command-line option to say how often to do the challenge/response, which is what triggers the notifications from the phone, which is what we suspect is causing this problem. With no option, this is done every 5 minutes. I've been testing with new command-line option -f5 which means to do it every 5 seconds.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Mon, 04 May 2020 18:42:27 GMT

I tested the USB840_Clock_ACCURACY_3in1.hex file referenced in Leo's last post. I loaded it on the Nordic DK by dragging it to the "JLink CDC UART Port". After just a few minutes my test program got the "serial port write operation" error "The device does not recognize the command". The firmware logging over that JLink port looked the same as what I uploaded on May 2 (which was also the DK and used firmware "USB840_Connect200430_3in1.hex" ):

[00:00:13.056,233] <debug> app: event:BLE_GAP_EVT_RSSI_CHANGED

[00:00:13.061,999] <debug> sphy_hci: TX request (7 bytes)

[00:00:13.067,421] <debug> sphy_hci: Started TX packet (payload 7).

[00:00:13.074,211] <error> app: Fatal error

[00:00:13.441,628] <info> app: BLE connectivity 200427

[00:00:13.447,435] <info> app: USB power detected

[00:00:13.453,251] <info> app: USB ready

I have not been able to test this firmware with the Nordic dev-dongle. I tried loading USB840_Clock_ACCURACY_3in1.hex into the dev-dongle using nRF Connect Programmer, but that leaves the dev-dongle in a state where it does not show up as a serial port when I plug it in. If there is some other way I should try to program it, please let me know.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

LeoChen — Mon, 04 May 2020 01:49:24 GMT

The Fanstel USB840M does not have 32K crystal.

The clock setting is different.

I change the clock setting back to default.

The application should all the same as patched SDK.

The changed were add CRITICAL_REGION_ENTER/ CRITICAL_REGION_EXIT and Scheduler queue size increase to 64.

These files can't work with USB840M.

Application + softdevice

https://www.dropbox.com/s/u77p0ub617i786e/USB840_Clock_ACCURACY_2in1.hex?dl=0

Application + bootloader + softdevice

https://www.dropbox.com/s/pg2hzhorhlivdoo/USB840_Clock_ACCURACY_3in1.hex?dl=0

Would you please also try these files on Nordic dongle and DK.

Leo

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Sat, 02 May 2020 17:16:28 GMT

I've been testing with the firmware file "USB840_Connect200430_3in1.hex" from the DropBox link that Leo from Fanstel referenced in the support case on April 29. This firmware has the critical section fix described by Nordic on April 24, and the queue size 64 change.

With this firmware, I get errors quickly with the Fanstel dongle and Nordic dev-dongle, and after 17 hours with the Nordic Development Kit.

Reproduced event loss problem with Nordic Development Kit on May 2 at about 8 AM eastern daylight time, after running for over 17 hours.
Test program "2020-05-02 08:02:15.759: (ERROR) no Nordic event BLE_GATTC_EVT_WRITE_RSP was received to complete the call to sd_ble_gattc_write"

The dropbox link below has USB trace using USBPcap and PuTTY serial port logging from the Development Kit.

The PuTTY output looks normal to me until the very end:
[00:00:58.343,695] <error> app: Fatal error
[00:00:58.710,920] <info> app: BLE connectivity 200427

www.dropbox.com/.../May2ReproNordicDK.zip

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

LeoChen — Thu, 30 Apr 2020 02:38:12 GMT

I increased sched queue_size, enabled UART log and compiler agian.

I have make sure the UART log has output with this BT840_USBconnectivity200430.hex.

https://www.dropbox.com/s/yqkcn95qrwto6g7/BT840_USBconnectivity200430.rar?dl=0

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Wed, 29 Apr 2020 20:34:57 GMT

I'm trying to reproduce the problem with the DK. How do I get the UART output? That shows up as JLink CDC UART Port (COM8). So if I open COM8 with Putty I should see debug output, but I see nothing. Does the baud rate matter? If so, what is it?

But if it is set to only show errors, then maybe I haven't encountered an error yet. But I don't know if I'm properly attached to the COM port.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Wed, 29 Apr 2020 19:17:41 GMT

Right, UART logging will only be available on the DK. If you are able to reproduce on a DK, please provide the logs. You can read the UART logs using a terminal software (Putty, Tera Term, RealTerm, Termite, etc). You may also be able to read UART logs from the Dongle by connecting a wire to the DK, but this would require the logging to use a different pin for UART, as the default UART TX pin is not routed out on the Dongle.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Wed, 29 Apr 2020 18:57:37 GMT

You mentioned "UART logging". I don't know how to do that. The only type of firmware logging I've used was with RTT Viewer and the DK. I don't know how to do logging with the dev-dongle.

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Wed, 29 Apr 2020 18:40:25 GMT

I've still seen the problem using connectivity_4.1.1_usb_with_s140_6.1.1_critical_region_fix_increased_sched_queue_size.hex

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Jørgen Holmefjord — Wed, 29 Apr 2020 18:33:49 GMT

Hi Paul,

The "Erase & write" function should not be available if you program the nRF52840 Dongle (PCA10059) through the USB port of the dongle. Did you connect a J-Link debugger to program the dongle through SWD? If you did a "Erase all" operation through the SWD interface, the Bootloader that comes with the dongle will have been erased. "Erase all" operations will erase all programmable flash, including UICR registers.

The MBR is part of the softdevice-hex. It will always be present whenever the softdevice is flashed to the chip. You can read more about that in the softdevice specifications.

[quote user="Paul Bradford"]So far, using these dev-dongles and new firmware, I have still seen the original 'no event' problem on two computers.[/quote]

Do you mean with the firmware with increased scheduler queue size or the one with only critical region fix?

Best regards,
Jørgen

RE: disconnecting while operations are in progress never gives BLE_GAP_EVT_DISCONNECTED event

Paul Bradford — Wed, 29 Apr 2020 18:07:42 GMT

Hi Jørgen,

I'm trying to test the fix on two Nordic dev-dongles (PCA10059), programmed using nRF Connect Programmer. Ideally I'd like to make sure the dev-dongle is in a known good state so I'd like to use the "Erase & write" feature. When I Add File connectivity_4.1.1_usb_with_s140_6.1.1_critical_region_fix_increased_sched_queue_size.hex, nRF Connect shows that it has Connectivity, SoftDevice and Master Boot record (it shows a region name "MBR or Application" with address range 0 to 0xB00 which is the Master Boot Record range). It does not show a bootloader. I'm confused about why it would have MBR, and why if it has MBR it doesn't have a bootloader.

I went ahead with "Erase & write" and am using those dev-dongles with my test program.

So far, using these dev-dongles and new firmware, I have still seen the original 'no event' problem on two computers.

-Paul