[Bug bootstrap/53681] New: s390 bootstrap failure since 187965

[Bug bootstrap/53681] New: s390 bootstrap failure since 187965

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

             Bug #: 53681
           Summary: s390 bootstrap failure since 187965
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: major
          Priority: P3
         Component: bootstrap
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: krebbel@gcc.gnu.org


int
__gcov_execle (const char *path, char *arg, ...)
{
  __builtin_va_list ap, aq;
  while (__builtin_va_arg (ap, char *))
      length++;
}

cc1 -fpreprocessed t.c -quiet

t.c: In function ‘__gcov_execle’:
t.c:6:7: error: ‘length’ undeclared (first use in this function)
       length++;
       ^
t.c:6:7: note: each undeclared identifier is reported only once for each
function it appears in
t.c:5:32: internal compiler error: Segmentation fault
   while (__builtin_va_arg (ap, char *))
                                ^
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

(gdb) r
Starting program: /home/andreas/patched/gcc-head-build/gcc/cc1 -fpreprocessed
t.c -quiet
t.c: In function ‘__gcov_execle’:
t.c:6:7: error: ‘length’ undeclared (first use in this function)
       length++;
       ^
t.c:6:7: note: each undeclared identifier is reported only once for each
function it appears in

Program received signal SIGSEGV, Segmentation fault.
0x0000000082618f00 in mark_sym_for_renaming (sym=0x3fff7f3a320)
    at /home/andreas/patched/gcc-head/gcc/tree-into-ssa.c:2997
2997      bitmap_set_bit (SYMS_TO_RENAME (cfun), DECL_UID (sym));
(gdb) bt
#0  0x0000000082618f00 in mark_sym_for_renaming (sym=0x3fff7f3a320)
    at /home/andreas/patched/gcc-head/gcc/tree-into-ssa.c:2997
#1  0x0000000083c03f98 in s390_gimplify_va_arg (valist=0x3fff8026300,
type=0x3fff7f549d8, pre_p=0x3ffffffd7d0, 
    post_p=0x3ffffffa698) at
/home/andreas/patched/gcc-head/gcc/config/s390/s390.c:9047
#2  0x000000008057b192 in gimplify_va_arg_expr (expr_p=0x3fff8026290,
pre_p=0x3ffffffd7d0, post_p=0x3ffffffa698)
    at /home/andreas/patched/gcc-head/gcc/builtins.c:4509
#3  0x0000000081256160 in gimplify_expr (expr_p=0x3fff8026290,
pre_p=0x3ffffffd7d0, post_p=0x3ffffffa698, 
    gimple_test_f=0x8109c684 <is_gimple_val>, fallback=1) at
/home/andreas/patched/gcc-head/gcc/gimplify.c:7178
#4  0x000000008125b52a in gimplify_expr (expr_p=0x3fff802a020,
pre_p=0x3ffffffd7d0, post_p=0x3ffffffa698, 
    gimple_test_f=0x8109aca8 <is_gimple_condexpr>, fallback=1)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:7743
#5  0x0000000081233282 in gimplify_cond_expr (expr_p=0x3fff7ffff28,
pre_p=0x3ffffffd7d0, fallback=0)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:3240
#6  0x0000000081255808 in gimplify_expr (expr_p=0x3fff7ffff28,
pre_p=0x3ffffffd7d0, post_p=0x3ffffffbcf8, 
    gimple_test_f=0x8123f6a8 <is_gimple_stmt>, fallback=0) at
/home/andreas/patched/gcc-head/gcc/gimplify.c:7085
#7  0x00000000812474e8 in gimplify_stmt (stmt_p=0x3fff7ffff28,
seq_p=0x3ffffffd7d0)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:5662
#8  0x00000000812217e0 in gimplify_statement_list (expr_p=0x3fff802a060,
pre_p=0x3ffffffd7d0)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:1529
#9  0x00000000812597d2 in gimplify_expr (expr_p=0x3fff802a060,
pre_p=0x3ffffffd7d0, post_p=0x3ffffffcb10, 
    gimple_test_f=0x8123f6a8 <is_gimple_stmt>, fallback=0) at
/home/andreas/patched/gcc-head/gcc/gimplify.c:7514
#10 0x00000000812474e8 in gimplify_stmt (stmt_p=0x3fff802a060,
seq_p=0x3ffffffd7d0)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:5662
#11 0x000000008121f25e in gimplify_bind_expr (expr_p=0x3fff8019298,
pre_p=0x3ffffffe7d8)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:1223
#12 0x0000000081257358 in gimplify_expr (expr_p=0x3fff8019298,
pre_p=0x3ffffffe7d8, post_p=0x3ffffffdaf0, 
    gimple_test_f=0x8123f6a8 <is_gimple_stmt>, fallback=0) at
/home/andreas/patched/gcc-head/gcc/gimplify.c:7299
#13 0x00000000812474e8 in gimplify_stmt (stmt_p=0x3fff8019298,
seq_p=0x3ffffffe7d8)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:5662
#14 0x000000008125e548 in gimplify_body (fndecl=0x3fff8019200, do_parms=1
'\001')
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:8160
#15 0x000000008126070a in gimplify_function_tree (fndecl=0x3fff8019200)
    at /home/andreas/patched/gcc-head/gcc/gimplify.c:8294
#16 0x00000000809aabea in cgraph_analyze_function (node=0x3fff802b000)
    at /home/andreas/patched/gcc-head/gcc/cgraphunit.c:652
#17 0x00000000809ac59e in cgraph_analyze_functions () at
/home/andreas/patched/gcc-head/gcc/cgraphunit.c:938
#18 0x00000000809b166e in finalize_compilation_unit () at
/home/andreas/patched/gcc-head/gcc/cgraphunit.c:2086
#19 0x000000008011b6c2 in c_write_global_declarations () at
/home/andreas/patched/gcc-head/gcc/c-decl.c:10112
#20 0x00000000821d8430 in compile_file () at
/home/andreas/patched/gcc-head/gcc/toplev.c:568
#21 0x00000000821db150 in do_compile () at
/home/andreas/patched/gcc-head/gcc/toplev.c:1872
#22 0x00000000821db35c in toplev_main (argc=4, argv=0x3fffffff2b8)
    at /home/andreas/patched/gcc-head/gcc/toplev.c:1948
#23 0x0000000084920768 in main (argc=4, argv=0x3fffffff2b8) at
/home/andreas/patched/gcc-head/gcc/main.c:36

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

Andreas Krebbel <krebbel at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2
                 CC|                            |krebbel at gcc dot gnu.org,
                   |                            |matz at suse dot de

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-invalid-code
                 CC|                            |rguenth at gcc dot gnu.org

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-06-15 08:26:17 UTC ---
This is an ice-on-invalid - how can that break bootstrap?

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

--- Comment #2 from Andreas Krebbel <krebbel at gcc dot gnu.org> 2012-06-15 08:30:21 UTC ---
(In reply to comment #1)
> This is an ice-on-invalid - how can that break bootstrap?

delta was a bit too eager. Same happens with:

int
__gcov_execle (const char *path, char *arg, ...)
{
  int length = 0;

  __builtin_va_list ap, aq;
  while (__builtin_va_arg (ap, char *))
      length++;
}

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-06-15 08:42:03 UTC ---
When delta reducing ICE on valid, it is always better to add to a script second
compilation (perhaps with -O0 for speed) using some compiler where it initially
compiles fine to make sure delta doesn't turn ice-on-valid into ice-on-invalid.

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[matz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

Michael Matz <matz at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |matz at gcc dot gnu.org

--- Comment #4 from Michael Matz <matz at gcc dot gnu.org> 2012-06-15 14:21:05 UTC ---
I don't see how r187965 could cause this, but I do see the problem.
mark_sym_for_renaming (called via the s390 va_arg_expr expander) is called
during, well, gimplification from GENERIC.  At that point SSA isn't
initialized yet, so cfun->gimple_df is still NULL, and so SYMS_TO_RENAME
gives a segfault.

You either have to guard the call to mark_sym_for_renaming with
gimple_in_ssa_p(), or get rid of the call alltogether.  I don't see how
new va_arg expressions would be generated during SSA optimizers, so the latter
solution would be safe.  No other backend calls mark_sym_for_renaming either,
so just remove it.

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

--- Comment #5 from Andreas Krebbel <krebbel at gcc dot gnu.org> 2012-06-18 07:09:04 UTC ---
(In reply to comment #4)
> I don't see how r187965 could cause this, but I do see the problem.
> mark_sym_for_renaming (called via the s390 va_arg_expr expander) is called
> during, well, gimplification from GENERIC.  At that point SSA isn't
> initialized yet, so cfun->gimple_df is still NULL, and so SYMS_TO_RENAME
> gives a segfault.
> 
> You either have to guard the call to mark_sym_for_renaming with
> gimple_in_ssa_p(), or get rid of the call alltogether.  I don't see how
> new va_arg expressions would be generated during SSA optimizers, so the latter
> solution would be safe.  No other backend calls mark_sym_for_renaming either,
> so just remove it.

?! You appear to have added the call to mark_sym_for_renaming with your patch.
Since it isn't mentioned in the changelog perhaps it is a leftover from
something else in your tree? So the solution is to revert that change?

svn diff -r 187964:187965 gcc/config/s390/s390.c
Index: gcc/config/s390/s390.c
===================================================================
--- gcc/config/s390/s390.c      (revision 187964)
+++ gcc/config/s390/s390.c      (revision 187965)
@@ -9044,6 +9044,7 @@
   lab_false = create_artificial_label (UNKNOWN_LOCATION);
   lab_over = create_artificial_label (UNKNOWN_LOCATION);
   addr = create_tmp_var (ptr_type_node, "addr");
+  mark_sym_for_renaming (addr);

   t = fold_convert (TREE_TYPE (reg), size_int (max_reg));
   t = build2 (GT_EXPR, boolean_type_node, reg, t);

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[rguenther at suse dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

--- Comment #6 from rguenther at suse dot de <rguenther at suse dot de> 2012-06-18 08:24:51 UTC ---
On Mon, 18 Jun 2012, krebbel at gcc dot gnu.org wrote:

> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681
> 
> --- Comment #5 from Andreas Krebbel <krebbel at gcc dot gnu.org> 2012-06-18 07:09:04 UTC ---
> (In reply to comment #4)
> > I don't see how r187965 could cause this, but I do see the problem.
> > mark_sym_for_renaming (called via the s390 va_arg_expr expander) is called
> > during, well, gimplification from GENERIC.  At that point SSA isn't
> > initialized yet, so cfun->gimple_df is still NULL, and so SYMS_TO_RENAME
> > gives a segfault.
> > 
> > You either have to guard the call to mark_sym_for_renaming with
> > gimple_in_ssa_p(), or get rid of the call alltogether.  I don't see how
> > new va_arg expressions would be generated during SSA optimizers, so the latter
> > solution would be safe.  No other backend calls mark_sym_for_renaming either,
> > so just remove it.
> 
> ?! You appear to have added the call to mark_sym_for_renaming with your patch.
> Since it isn't mentioned in the changelog perhaps it is a leftover from
> something else in your tree? So the solution is to revert that change?
> 
> svn diff -r 187964:187965 gcc/config/s390/s390.c
> Index: gcc/config/s390/s390.c
> ===================================================================
> --- gcc/config/s390/s390.c      (revision 187964)
> +++ gcc/config/s390/s390.c      (revision 187965)
> @@ -9044,6 +9044,7 @@
>    lab_false = create_artificial_label (UNKNOWN_LOCATION);
>    lab_over = create_artificial_label (UNKNOWN_LOCATION);
>    addr = create_tmp_var (ptr_type_node, "addr");
> +  mark_sym_for_renaming (addr);

I think that change is indeed bogus as-is, if you ever call this
when we are in SSA form you'd need that call, but you need to guard
it with if (gimple_in_ssa_p (cfun)).

But I'm sure we never gimplify va-arg stuff when in SSA form.

Richard.

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

Andreas Krebbel <krebbel at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED

--- Comment #7 from Andreas Krebbel <krebbel at gcc dot gnu.org> 2012-06-18 11:06:39 UTC ---
(In reply to comment #6)
Ok. I've reverted the s390.c bits of r187965.

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[matz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

Michael Matz <matz at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|FIXED                       |

--- Comment #8 from Michael Matz <matz at gcc dot gnu.org> 2012-06-18 11:08:49 UTC ---
(In reply to comment #5)
> ?! You appear to have added the call to mark_sym_for_renaming with your patch.

Yikes!

> Since it isn't mentioned in the changelog perhaps it is a leftover from
> something else in your tree? So the solution is to revert that change?

Yes, it was an unintended change (it must have been in my dev tree quite long,
I don't remember doing it).  It seems I haven't reviewed the file list
of my commit message :-/  Sorry about that breakage.

[Bug bootstrap/53681] s390 bootstrap failure since 187965

[matz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53681

Michael Matz <matz at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|FIXED                       |
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED

--- Comment #8 from Michael Matz <matz at gcc dot gnu.org> 2012-06-18 11:08:49 UTC ---
(In reply to comment #5)
> ?! You appear to have added the call to mark_sym_for_renaming with your patch.

Yikes!

> Since it isn't mentioned in the changelog perhaps it is a leftover from
> something else in your tree? So the solution is to revert that change?

Yes, it was an unintended change (it must have been in my dev tree quite long,
I don't remember doing it).  It seems I haven't reviewed the file list
of my commit message :-/  Sorry about that breakage.

--- Comment #9 from Michael Matz <matz at gcc dot gnu.org> 2012-06-18 11:09:28 UTC ---
.