On 19.01.2012 18:40, Jakub Jelinek wrote:
On Thu, Jan 19, 2012 at 06:13:58PM +0400, Andrey Belevantsev wrote:
2012-01-19  Andrey Belevantsev<a...@ispras.ru>

        PR target/51106
        * function.c (instantiate_virtual_regs_in_insn): Use
delete_insn_and_edges when removing a wrong asm insn.

This is ok for trunk.

After Richi's RTL generation related cleanups went it, the extra cleanup_cfg call was added so we are no longer lucky to have the proper fallthru edge in this case. The PR trail has the patch to purge_dead_edges making it consider this situation, but I still prefer the below patch that fixes only the invalid asm case. The reason is that I think it unlikely that after initial RTL expansion (of which the instantiate virtual regs pass seems to be the part) we will get the problematic situation. However, I'm happy to test the PR trail patch, too.

Tested fine on x86-64, ok for trunk?


2012-04-02  Andrey Belevantsev  <a...@ispras.ru>

        PR target/51106
        PR middle-end/52650
        * function.c (instantiate_virtual_regs_in_insn): Make sure to set
        the proper fallthru bits when removing a wrong jump asm.

diff --git a/gcc/function.c b/gcc/function.c
index 3e903ef..a2638bb 100644
--- a/gcc/function.c
+++ b/gcc/function.c
@@ -1730,6 +1730,15 @@ instantiate_virtual_regs_in_insn (rtx insn)
       if (!check_asm_operands (PATTERN (insn)))
          error_for_asm (insn, "impossible constraint in %<asm%>");
+         if (JUMP_P (insn))
+           {
+             basic_block bb = BLOCK_FOR_INSN (insn);
+             edge e;
+             if (single_succ_p (bb)
+                 && !((e = single_succ_edge (bb))->flags & EDGE_FALLTHRU))
+               e->flags |= EDGE_FALLTHRU;
+           }
          delete_insn_and_edges (insn);

Reply via email to