https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79430
--- Comment #74 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
This part looks wrong:
@@ -19206,8 +19196,9 @@
movq %r11, 112(%rsp)
movq %rax, 96(%rsp)
call integral_over_z_part_isr.6797
-.LVL1464:
+.LVL1465:
.loc 1 3089 0
+ fldt 64(%rsp)
fldt 496(%rsp)
popq %rsi
.cfi_def_cfa_offset 1160
@@ -19215,8 +19206,6 @@
.cfi_def_cfa_offset 1152
movq 80(%rsp), %rax
movq 96(%rsp), %r11
- fldt 64(%rsp)
- fxch %st(1)
fucomi %st(1), %st
fstp %st(1)
fldt 112(%rsp)
Here, the fldt 64(%rsp) is moved before the popq %rsi - instruction,
which changes %rsp.
This is also the place where the error occured: line 3089, with
the call to integral_over_z_part_isr just before.
