What can cause a Hardfault and how to catch it

[wim]

Hi

I am investigating some weird issue where a device sometimes locks up: it is BLE connected to my iphone app, and then I bombard it with writes to a specific characteristic, to which I get a response via a notification on another characteristic. At some point, my app sees that the BLE device does not respond anymore and disconnects, My BLE device does not see this disconnect, the 'connection LED' stays on, and the debugger stopped in the hardfault_handler - I put a breakpoint there.

Question is: what can cause a hardfault? How can I determine that? And how can I get out of a hard fault? Is the only way the watchdog timer or a powercycle?

[Susheel Nuguru]

There has been a lot of post regarding it. I am just pointing to few of them.

post 1
post 2
tutorial 1

[awneil]

Hard Faults and their handling are part of the ARM Architecture - not specific to Nordic.

I collected some references here: community.arm.com/.../5414

When working with ARM Cortex-M, you should certainly have a copy of Joseph Yiu's "Definitive Guide" book - they have sections about fault handling ...

[Ulrich Myhre]

This is a good resource, but you should also note that the SoftDevice will trigger a HardFault purposefully when a stack fault happens. Before it happens, the fault handler given to sd_softdevice_enable should be called though. If this is not the case, it's one of the default ARM HardFault reasons (unaligned access or overflow most likely).