This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug target/28126] gcc moves an expensive instruction outside of a conditional



------- Comment #2 from kkojima at gcc dot gnu dot org  2006-06-22 04:16 -------
Although I don't know much about the instruction scheduling, I had
a similar problem on SH and it was workarounded with emitting blockage
insns.  The patch below might work for you, though I'm not sure if
it's the right thing to do.  Ian's comment in the gcc list
  http://gcc.gnu.org/ml/gcc/2006-06/msg00669.html
suggests us that there are more than one problem.

--- ORIG/gcc-4_1-branch/gcc/config/mips/mips.c  2006-06-20 10:53:21.000000000
+0900
+++ TMP/gcc-4_1-branch/gcc/config/mips/mips.c   2006-06-22 12:49:26.000000000
+0900
@@ -1995,6 +1995,7 @@ mips_legitimize_tls_address (rtx loc)
     case TLS_MODEL_INITIAL_EXEC:
       tmp1 = gen_reg_rtx (Pmode);
       tmp2 = mips_unspec_address (loc, SYMBOL_GOTTPREL);
+      emit_insn (gen_blockage ());
       if (Pmode == DImode)
        {
          emit_insn (gen_tls_get_tp_di (v1));
@@ -2011,6 +2012,7 @@ mips_legitimize_tls_address (rtx loc)

     case TLS_MODEL_LOCAL_EXEC:

+      emit_insn (gen_blockage ());
       if (Pmode == DImode)
        emit_insn (gen_tls_get_tp_di (v1));
       else


-- 

kkojima at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |kkojima at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28126


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]