[Bug target/128524] [m68k] dbmi/dbpl loop exits through the early-exit branch when the result is zero (jle instead of jmi/jpl)

[mikpelinux at gmail dot com via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128524

--- Comment #2 from Mikael Pettersson <mikpelinux at gmail dot com> ---
I cannot reproduce this on m68k-linux-gnu with unmodified vanilla gcc 14.4.0,
15.3.0, 16.2.0, or 17.0.0 20261107. For all four of them it generates:

        .text
        .align  2
        .globl  check_lengths
        .type   check_lengths, @function
check_lengths:
        move.l 4(%sp),%a1
        lea (2,%a1),%a0
        lea (32,%a1),%a1
        moveq #1,%d0
.L3:
        add.l %d0,%d0
        clr.l %d1
        move.w (%a0)+,%d1
        sub.l %d1,%d0
        jmi .L4
        cmp.l %a0,%a1
        jne .L3
        rts
.L4:
        moveq #-1,%d0
        rts
        .size   check_lengths, .-check_lengths

No dbCC in sight. Could be down to your patches, how you confgured gcc, or
expected differences between m68k-linux and non-linux targets.