There is no silly question.

I think the easiest thing is to put the desired bit patterns in memory
use LD1 as shown here:
https://github.com/cirosantilli/linux-kernel-module-cheat/blob/657c59249e56d861bb0a437a1b1c757797281910/userland/arch/aarch64/fadd_vector.S#L21

This may also be of interest:
https://github.com/cirosantilli/linux-kernel-module-cheat/blob/657c59249e56d861bb0a437a1b1c757797281910/userland/arch/aarch64/gdb_tests/floating_registers.S

On Mon, Jan 13, 2020 at 3:29 AM Gabe Black <[email protected]> wrote:
>
> Hi folks. I do most of my aarch64 assembly level programming by example, ie
> by just doing what other code I can find is already doing. I'm trying to
> test some code I wrote to checkpoint the FP state in a fast model ARM CPU,
> and to do that I need to load known values into those registers to make
> sure they end up in the right place in the checkpoint.
>
> What assembly instruction should I use to maximally load up those registers
> with a known bit pattern? Ideally it would be something where I could tell
> that the right bits ended up in the right register and in the right part of
> the register.
>
> This particular CPU (CortexA76) does not seem to have SVE, so this would
> need to be a neon type instruction. Sorry for the dumb question, but I
> figure this might be a very easy question for somebody to answer which
> would save me a lot of trial and error trying to find the right instruction
> and syntax to get it going.
>
> Thanks!
> Gabe
> _______________________________________________
> gem5-dev mailing list
> [email protected]
> http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev

Reply via email to