[Bug rtl-optimization/115912] [15 regression] Harfbuzz testsuite fails (mvar_partial_instance test) since r15-1901-g98914f9eba5f19

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115912

--- Comment #19 from Sam James <sjames at gcc dot gnu.org> ---
Created attachment 58688
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=58688&action=edit
pass_diff_31_32.txt.xz

(In reply to Sam James from comment #18)
> The first point of divergence seems to be:
> [...]
> i.e. arrayZ gets corrupted?

This might be wrong, it looks like it's consistent at return for the first call
to subset.

(In reply to Andrew Pinski from comment #17)
> If anything maybe something else afterwards (not ext-dce) optimizes something incorrectly.

Not sure how useful it is in this format, but here's the full dumps diff
between 31 & 32.