do not expose stack limit as GPR

[jnk0le]

rationale:

• Stack limit is not about to be frequently accessed by thread code nor it is available from raw C/C++.
• Reserving another general purpose register, increases register pressure especially in RV32E which currently have less available registers than armv7[M].
• Stack limit can be corrupted by code.
• Mapping another shadow register into GPRs and protecting it from corruption by thread code, will increase hardware complexity.
• Saving 2 cycles on push/pop in context switch is a sign of premature optimization of whole ABI for specific use case. Assuming 50MHz clockrate and 1000Hz scheduler tickrate, those 2 cycles saved per context switch accounts for 4E-5% of total cycles saved. Of course, only if rest of the code is actually not starving from missing register.

[ilg-ul]

@jnk0le, you're probably right, for RV32E dedicating a register for the stack limit might be too expensive, but generally I would not bet on RV32E.

actually, for now, I would not bet on RV32 in the microcontroller space at all, there is no interest for it, and the only developments are in the privileged space.

[ilg-ul]

I added your comments to the page.

[jnk0le]

A little correction; 2000/50M should be 0,004%, not 4E-5

[ilg-ul]

ok, page updated.