On 6/11/25 3:25 AM, Kito Cheng wrote:
LGTM, but I would like to make sure either Jeff or Patrick is OK too :)
I think it's correct. Not sure why it was written in that original
form, but conceptually the register value written by the "sc" is used at
the "bnez" and dies at that point. So that's the better choice for the
scratch register when compared to the output of the "lr" which is live
on loop exit.
Unless there's some very non-obvious reason (which would definitely be
deserving of a comment in this code) I think this should get installed.
Jeff
