https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94439

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
This diverges during sched2, on the reduced testcase with additional
--param=min-nondebug-insn-uid=10000 I see:
-(insn 10076 10075 10055 2 (set (reg:SI 43 r15)
-        (mem/c:SI (const:DI (plus:DI (symbol_ref:DI ("cs.0") [flags 0x2] 
<var_decl 0x7f4364f38900 cs>)
+(insn 10076 10075 221 2 (set (reg:SI 40 r12)
+        (mem/c:SI (const:DI (plus:DI (symbol_ref:DI ("cs.0") [flags 0x2] 
<var_decl 0x7f4bee787900 cs>)
                     (const_int 24 [0x18]))) [1 cs+24 S4 A64]))
"builtin-complex-1.c":50:3 67 {*movsi_internal}
      (nil))
difference between -g0 and -g.

Reply via email to