[J-core] How the IRQ/timer stuff works.

Rob Landley rob at landley.net
Tue Oct 29 05:14:53 UTC 2019


On 10/28/19 6:21 PM, Joh-Tob Sch├Ąg wrote:
>> * However only traps 64-127 an be programmed (the high 2 bits of i are ignored)
>
> This claim needs to be checked. I think it only applies to linux since
> i do not remember having trouble with TRAPA or interrupts with high
> numbers. Needs checking

This is just me forwarding the messages I sent back in August, based on an hour
or two of trying to read and understand the kernel code. The conversation's
probably moved on a bit since then.

>> There's one important unanswered question, which is when an interrupt gets
>> called, where is the address of the function to jump to?
> 
> I know this. There is register called VBR, Vector Base Register and it
> contains the pointer of an array of function pointers. Nth interrupt
> executes function which is pointed to by nth array slot.

For example yes, I remember that follow-up email. :)

> It is currently not a priority. I just wanted to know whether i missed
> anything about cyclic interrupts.

The kernel code seems to work, and this was me reading the kernel code to see
what I could understand.

> BRKADD is the address of the break point but i could never got this
> working by mere poking.

Rich had some fixes to gdbserver that make breakpoints work. The upstream gdb
hasn't got that fix yet, last I checked.

Rob


More information about the J-core mailing list