This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug rtl-optimization/70460] [4.9/5/6 Regression] Miscompilation of glibc on i686-linux starting with r234101
- From: "jakub at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Wed, 30 Mar 2016 16:47:20 +0000
- Subject: [Bug rtl-optimization/70460] [4.9/5/6 Regression] Miscompilation of glibc on i686-linux starting with r234101
- Auto-submitted: auto-generated
- References: <bug-70460-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70460
--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
So, what about:
--- gcc/ira.c.jj 2016-03-21 10:12:32.000000000 +0100
+++ gcc/ira.c 2016-03-30 18:42:46.095091159 +0200
@@ -3885,18 +3885,27 @@ indirect_jump_optimize (void)
{
rtx_insn *def_insn = DF_REF_INSN (def);
rtx note = find_reg_note (def_insn, REG_LABEL_OPERAND, NULL_RTX);
+ rtx lab = NULL_RTX;
if (note)
{
- /* Substitute a LABEL_REF to the label given by the
- note rather than using SET_SRC of DEF_INSN.
- DEF_INSN might be loading the label constant from
- a constant pool, which isn't what we want in a
- direct branch. */
- rtx lab = gen_rtx_LABEL_REF (Pmode, XEXP (note, 0));
- if (validate_replace_rtx (SET_SRC (x), lab, insn))
- rebuild_p = true;
+ rtx set = single_set (def_insn);
+ if (set
+ && GET_CODE (SET_SRC (set)) == LABEL_REF
+ && XEXP (SET_SRC (set), 0) == XEXP (note, 0))
+ lab = SET_SRC (set);
+ else
+ {
+ rtx eqnote = find_reg_note (def_insn, REG_EQUAL,
+ NULL_RTX);
+ if (eqnote
+ && GET_CODE (XEXP (eqnote, 0)) == LABEL_REF
+ && XEXP (XEXP (eqnote, 0), 0) == XEXP (note, 0))
+ lab = XEXP (eqnote, 0);
+ }
}
+ if (lab && validate_replace_rtx (SET_SRC (x), lab, insn))
+ rebuild_p = true;
}
}
}
then? Not trying to subst arbitrary rtxes into insn, just LABEL_REFs, both
because it is unlikely to actually help and because one would need to verify it
doesn't contain say REGs or MEMs clobbered in between def_insn and insn.