[valgrind] [Bug 523790] s390: inefficient IR generated for branch insn

https://bugs.kde.org/show_bug.cgi?id=3D523790

--- Comment #1 from Florian Krohm <[email protected]> ---
Not a bug and not specific to s390. Caused by maybe_loop_unroll_BB. There i=
s a
post-pass  do_deadcode_BB, cprop_BB but it does not help.

Here is a more realistic code snippet.

lghi %r9,15                          r9 =3D 15
lghi %r7,0                            r7 =3D 0
0: aghi %r7,5                 L1:  r7 +=3D 5
cgrjne %r7,%r9,0b               if (r7 !=3D r9) goto L1

I would have expected the loop to be replaced with r7 =3D 15. But r7 =3D 0 =
and r7
+=3D 5 are in different super blocks and optimisation
is considering only one SB at a time.

--=20
You are receiving this mail because:
You are watching all bug changes.=