Hi Robin,
There is one issue with this patch and the next few patches, the mask
policy for the generated vsetvl should be mu, which is currently done
wrong. I'm going to clear the expand_cond_len* functions before sending
the next version of these patches. Then these patches can directly use
expand_cond_len_unary instead of emit_vlmax_*.
One thing maybe for the next patches: It seems to me that we lump all of
the COND_... tests into the cond subdirectory when IMHO they would also
fit into the respective directories of their operations (binop, unop
etc). Right now we will have a lot of rather unrelated tests (or just
related by their use of COND_) in one dir. What do you think?
I have no idea about it, I will add some more testcases for cond
widen-binop testcases and it looks like there are too many files in the
cond directory. Maybe it's better to subdivide unop, binop and ternop
directories in the cond directory since all these testcases are designed
to test the conditional patterns.
--
Best,
Lehua
