[Bug target/32431] New: ICE in df_refs_verify, at df-scan.c:4066

[Bug target/32431] New: ICE in df_refs_verify, at df-scan.c:4066

[rask at sygehus dot dk]

I have configured GCC this way:
$ /n/08/rask/src/gcc/configure --target m68hc11-unknown-none --disable-gdb
--disable-nls --with-newlib

The attached testcase causes an ICE:
$ ./xgcc -B./ ~/muldi3.c -S -dp -o /dev/null -O2
/n/08/rask/src/gcc/libgcc/../gcc/libgcc2.c: In function '__muldi3':
/n/08/rask/src/gcc/libgcc/../gcc/libgcc2.c:566: internal compiler error: in
df_refs_verify, at df-scan.c:4066


-- 
           Summary: ICE in df_refs_verify, at df-scan.c:4066
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: rask at sygehus dot dk
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: m68hc11-unknown-none


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] ICE in df_refs_verify, at df-scan.c:4066

[rask at sygehus dot dk]

------- Comment #1 from rask at sygehus dot dk  2007-06-20 16:58 -------
Created an attachment (id=13747)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=13747&action=view)
Preprocessed testcase


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] ICE in df_refs_verify, at df-scan.c:4066

[zadeck at naturalbridge dot com]

------- Comment #2 from zadeck at naturalbridge dot com  2007-06-21 12:29 -------
Subject: Re:  ICE in df_refs_verify, at df-scan.c:4066

rask at sygehus dot dk wrote:
> ------- Comment #1 from rask at sygehus dot dk  2007-06-20 16:58 -------
> Created an attachment (id=13747)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=13747&action=view)
>  --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=13747&action=view)
> Preprocessed testcase
>
>
>   
:ADDPATCH MIDDLE-END:

The underlying problem is that the later phases of compilation on the
m68hc11 create shared rtl.  The reorg phase needs df and also needs the
rtl to be unshared. This causes problems because the unsharing process
changes insns in ways that mess up df's scanning information.

There are two ways to solve this problem:

1) Go in and fix the places that create shared rtl in this back end.
2) Make the rtl unsharing df-ready.

I chose plan (2) because I know how to do this and do not know how to do
(1).
The astute middle end/gwp maintainer may say that (1) is the right thing
to do and so the sharing police (namely honza) should investigate this
pr.  If that is the proper plan of action, then this patch may go into
the trash.

I tested this on x86-64 which does call unshare_all_rtl_again (from
ifcvt) to verify that this does not cause any problems and it does fix
the immediate bug here. 

Ok for trunk?

Kenny


2007-06-21  Kenneth Zadeck <zadeck@naturalbridge.com>

        PR middle-end/32431
    * emit-rtl.c (unshare_all_rtl_in_chain): Now rescans insns and
    notes if unsharing happens.
    (copy_rtx_if_shared_1): Returns true if sharing was found.
    config/m68hc11/m68hc11.c: (m68hc11_reorg): Move reinitialization
    of cfg to before unsharing rtl.

Index: emit-rtl.c
===================================================================
--- emit-rtl.c  (revision 125916)
+++ emit-rtl.c  (working copy)
@@ -184,7 +184,7 @@ static int reg_attrs_htab_eq (const void
 static reg_attrs *get_reg_attrs (tree, int);
 static tree component_ref_for_mem_expr (tree);
 static rtx gen_const_vector (enum machine_mode, int);
-static void copy_rtx_if_shared_1 (rtx *orig);
+static bool copy_rtx_if_shared_1 (rtx *orig);

 /* Probability of the conditional branch currently proceeded by try_split.
    Set to -1 otherwise.  */
@@ -2350,8 +2350,20 @@ unshare_all_rtl_in_chain (rtx insn)
   for (; insn; insn = NEXT_INSN (insn))
     if (INSN_P (insn))
       {
-       PATTERN (insn) = copy_rtx_if_shared (PATTERN (insn));
-       REG_NOTES (insn) = copy_rtx_if_shared (REG_NOTES (insn));
+       rtx orig = PATTERN (insn);
+
+        if (copy_rtx_if_shared_1 (&orig))
+         {
+           PATTERN (insn) = orig;
+           df_insn_rescan (insn);
+         }
+
+       orig = REG_NOTES (insn);
+       if (copy_rtx_if_shared_1 (&orig))
+         {
+           REG_NOTES (insn) = orig;
+           df_notes_rescan (insn);
+         }
       }
 }

@@ -2383,10 +2395,11 @@ copy_rtx_if_shared (rtx orig)
   return orig;
 }

-/* Mark *ORIG1 as in use, and set it to a copy of it if it was already in
-   use.  Recursively does the same for subexpressions.  */
+/* Mark *ORIG1 as in use, and set it to a copy of it if it was already
+   in use.  Recursively does the same for subexpressions.  Return true
+   if copies were made.  */

-static void
+static bool
 copy_rtx_if_shared_1 (rtx *orig1)
 {
   rtx x;
@@ -2396,13 +2409,15 @@ copy_rtx_if_shared_1 (rtx *orig1)
   const char *format_ptr;
   int copied = 0;
   int length;
+  bool changed = false;
+

   /* Repeat is used to turn tail-recursion into iteration.  */
 repeat:
   x = *orig1;

   if (x == 0)
-    return;
+    return changed;

   code = GET_CODE (x);

@@ -2421,15 +2436,15 @@ repeat:
     case CC0:
     case SCRATCH:
       /* SCRATCH must be shared because they represent distinct values.  */
-      return;
+      return changed;
     case CLOBBER:
       if (REG_P (XEXP (x, 0)) && REGNO (XEXP (x, 0)) < FIRST_PSEUDO_REGISTER)
-       return;
+       return changed;
       break;

     case CONST:
       if (shared_const_p (x))
-       return;
+       return changed;
       break;

     case INSN:
@@ -2438,7 +2453,7 @@ repeat:
     case NOTE:
     case BARRIER:
       /* The chain of insns is not being copied.  */
-      return;
+      return changed;

     default:
       break;
@@ -2451,6 +2466,7 @@ repeat:
     {
       x = shallow_copy_rtx (x);
       copied = 1;
+      changed = true;
     }
   RTX_FLAG (x, used) = 1;

@@ -2469,7 +2485,7 @@ repeat:
        {
        case 'e':
           if (last_ptr)
-            copy_rtx_if_shared_1 (last_ptr);
+            changed |= copy_rtx_if_shared_1 (last_ptr);
          last_ptr = &XEXP (x, i);
          break;

@@ -2488,7 +2504,7 @@ repeat:
              for (j = 0; j < len; j++)
                 {
                  if (last_ptr)
-                   copy_rtx_if_shared_1 (last_ptr);
+                   changed |= copy_rtx_if_shared_1 (last_ptr);
                   last_ptr = &XVECEXP (x, i, j);
                 }
            }
@@ -2501,7 +2517,7 @@ repeat:
       orig1 = last_ptr;
       goto repeat;
     }
-  return;
+  return changed;
 }

 /* Clear all the USED bits in X to allow copy_rtx_if_shared to be used
Index: config/m68hc11/m68hc11.c
===================================================================
--- config/m68hc11/m68hc11.c    (revision 125916)
+++ config/m68hc11/m68hc11.c    (working copy)
@@ -5011,6 +5011,11 @@ m68hc11_reorg (void)
   z_reg = gen_rtx_REG (HImode, HARD_Z_REGNUM);
   first = get_insns ();

+  /* Need to compute the bb_for_insn before splitting when optimizing
+     or else the df_analyze may raise a verify error.  */
+  if (optimize)
+    compute_bb_for_insn ();
+
   /* Some RTX are shared at this point.  This breaks the Z register
      replacement, unshare everything.  */
   unshare_all_rtl_again (first);
@@ -5023,9 +5028,6 @@ m68hc11_reorg (void)
   z_replacement_completed = 1;
   m68hc11_reassign_regs (first);

-  if (optimize)
-    compute_bb_for_insn ();
-
   /* After some splitting, there are some opportunities for CSE pass.
      This happens quite often when 32-bit or above patterns are split.  */
   if (optimize > 0 && split_done)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3 Regression] ICE in df_refs_verify, at df-scan.c:4066

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
           Keywords|                            |build, ice-on-valid-code
            Summary|ICE in df_refs_verify, at   |[4.3 Regression] ICE in
                   |df-scan.c:4066              |df_refs_verify, at df-
                   |                            |scan.c:4066
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3 Regression] ICE in df_refs_verify, at df-scan.c:4066

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3 Regression] ICE in df_refs_verify, at df-scan.c:4066

[zadeck at naturalbridge dot com]

------- Comment #3 from zadeck at naturalbridge dot com  2007-08-02 19:19 -------
Given that the rtl passes are moving to not allow illegally shared rtl, i do
not believe that the resolution of this bug has anything to do with the
dataflow port.

If this bug is to be resolved, it will be done by cleaning up this back end.  


-- 

zadeck at naturalbridge dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|zadeck at naturalbridge dot |
                   |com                         |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[rguenth at gcc dot gnu dot org]

------- Comment #4 from rguenth at gcc dot gnu dot org  2008-01-13 14:55 -------
Does this bug still happen?  There were a lot of changes in the m68k backend
since
the bug report was opened.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[steven at gcc dot gnu dot org]

------- Comment #5 from steven at gcc dot gnu dot org  2008-01-13 16:04 -------
The m68k and m68hc11 are not the same backend.


-- 

steven at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2008-01-13 16:04:08
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[jsm28 at gcc dot gnu dot org]

------- Comment #6 from jsm28 at gcc dot gnu dot org  2008-03-15 00:41 -------
Update milestone after 4.3.0 release.


-- 

jsm28 at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.0                       |4.3.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[rguenth at gcc dot gnu dot org]

------- Comment #7 from rguenth at gcc dot gnu dot org  2008-06-06 14:57 -------
4.3.1 is being released, adjusting target milestone.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.1                       |4.3.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[jsm28 at gcc dot gnu dot org]

------- Comment #8 from jsm28 at gcc dot gnu dot org  2008-08-27 22:02 -------
4.3.2 is released, changing milestones to 4.3.3.


-- 

jsm28 at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.2                       |4.3.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[rguenth at gcc dot gnu dot org]

------- Comment #9 from rguenth at gcc dot gnu dot org  2009-01-24 10:19 -------
GCC 4.3.3 is being released, adjusting target milestone.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.3                       |4.3.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4/4.5 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[bonzini at gnu dot org]

------- Comment #10 from bonzini at gnu dot org  2009-03-31 13:22 -------
The bug is still there FWIW.


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2008-01-13 16:04:08         |2009-03-31 13:22:46
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4/4.5 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[bonzini at gnu dot org]

------- Comment #11 from bonzini at gnu dot org  2009-04-09 10:34 -------
*** Bug 35655 has been marked as a duplicate of this bug. ***


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mstein dot lists at
                   |                            |googlemail dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4/4.5 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[rguenth at gcc dot gnu dot org]

------- Comment #12 from rguenth at gcc dot gnu dot org  2009-08-04 12:28 -------
GCC 4.3.4 is being released, adjusting target milestone.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.4                       |4.3.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431

[Bug target/32431] [4.3/4.4/4.5/4.6 Regression][m68hc11] ICE in df_refs_verify, at df-scan.c:4066

[rguenth at gcc dot gnu dot org]

------- Comment #13 from rguenth at gcc dot gnu dot org  2010-05-22 18:11 -------
GCC 4.3.5 is being released, adjusting target milestone.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.5                       |4.3.6


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32431