Tree checking failure in jc1

Tree checking failure in jc1

[Dave Korn]

    Hi lists,

  I found a couple of new FAILs in my latest libjava testrun:

> FAIL: newarray_overflow -O3 compilation from source
> FAIL: newarray_overflow -O3 -findirect-dispatch compilation from source

  These turn out to be tree checking failures:

> In file included from <built-in>:3:0:
> newarray_overflow.java:20:0: internal compiler error: tree check: expected class
>  'type', have 'declaration' (function_decl) in put_decl_node, at java/lang.c:405

... happening ...

> /* Append to decl_buf a printable name for NODE.
>    Depending on VERBOSITY, more information about NODE
>    is printed. Read the comments of decl_printable_name in
>    langhooks.h for more.  */
> 
> static void
> put_decl_node (tree node, int verbosity)
> {
>   int was_pointer = 0;
>   if (TREE_CODE (node) == POINTER_TYPE)
>     {
>       node = TREE_TYPE (node);
>       was_pointer = 1;
>     }
>   if (DECL_P (node) && DECL_NAME (node) != NULL_TREE)
>     {
>       if (TREE_CODE (node) == FUNCTION_DECL)
> 	{
> 	  if (verbosity == 0 && DECL_NAME (node))
> 	  /* We have been instructed to just print the bare name
> 	     of the function.  */
> 	    {
> 	      put_decl_node (DECL_NAME (node), 0);
> 	      return;
> 	    }
> 
> 	  /* We want to print the type the DECL belongs to. We don't do
> 	     that when we handle constructors. */
> 	  if (! DECL_CONSTRUCTOR_P (node)
> 	      && ! DECL_ARTIFICIAL (node) && DECL_CONTEXT (node)
>               /* We want to print qualified DECL names only
>                  if verbosity is higher than 1.  */
>               && verbosity >= 1)
> 	    {
> 	      put_decl_node (TYPE_NAME (DECL_CONTEXT (node)),
>                                verbosity);
 ... here:    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  The decl pointed to by 'node' is a function_decl for a builtin:

 <function_decl 0x7ff8f580 __builtin_prefetch
    type <function_type 0x7fe52ee0
        type <void_type 0x7ff608a0 void asm_written VOID
            align 1 symtab 0 alias set -1 canonical type 0x7ff608a0
            pointer_to_this <pointer_type 0x7ff60900> chain <type_decl
0x7fdd0380 void>>
        QI
        size <integer_cst 0x7fef0108 constant 8>
        unit size <integer_cst 0x7fef0120 constant 1>
        align 8 symtab 0 alias set -1 canonical type 0x7fe52ee0
        arg-types <tree_list 0x7feff138 value <pointer_type 0x7ff609c0>
            chain <tree_list 0x7fef0b10 value <void_type 0x7ff608a0 void>>>
        pointer_to_this <pointer_type 0x7fe53000>>
    addressable public external built-in QI file <built-in> line 0 col 0
    align 8 built-in BUILT_IN_NORMAL:BUILT_IN_PREFETCH context <function_decl
0x7ff82b80 int_check> chain <var_decl 0x7fe2f900 _Utf15>>

and the DECL_CONTEXT turns out to be another function, one present in the
source of the testcase:

 <function_decl 0x7ff82b80 int_check
    type <function_type 0x7ff648c0
        type <void_type 0x7ff608a0 void asm_written VOID
            align 1 symtab 0 alias set -1 canonical type 0x7ff608a0
            pointer_to_this <pointer_type 0x7ff60900> chain <type_decl
0x7fdd0380 void>>
        QI
        size <integer_cst 0x7fef0108 constant 8>
        unit size <integer_cst 0x7fef0120 constant 1>
        align 8 symtab 0 alias set -1 canonical type 0x7ff648c0
        arg-types <tree_list 0x7fef0b10 value <void_type 0x7ff608a0 void>>
        pointer_to_this <pointer_type 0x7fe27140>>
    addressable public decl_2 QI file newarray_overflow.java line 20 col 0
align 8 context <record_type 0x7ff65520 newarray_overflow> initial <block
0x7fde6af8>
    result <result_decl 0x7fe100a0 D.246 type <void_type 0x7ff608a0 void>
        ignored VOID file newarray_overflow.java line 0 col 0
        align 1 context <function_decl 0x7ff82b80 int_check>>
    struct-function 0x7ff98df8 chain <function_decl 0x7ff82c00 object_check>>

... which is why the TYPE_NAME macro complains.

  Is it expected for a builtin to appear as if it were a nested function like
this?  If so, would it make sense to do something like replace this:

	      put_decl_node (TYPE_NAME (DECL_CONTEXT (node)),
                               verbosity);

with:

	      put_decl_node (TREE_CODE (DECL_CONTEXT (node)) == FUNCTION_DECL
                               ? DECL_CONTEXT (node)
                               : TYPE_NAME (DECL_CONTEXT (node)),
                               verbosity);

so we just treat the builtin as another layer of scope?

    cheers,
      DaveK

Re: Tree checking failure in jc1

[Dave Korn]

On 10/12/2010 20:49, Dave Korn wrote:

>   I found a couple of new FAILs in my latest libjava testrun:
> 
>> FAIL: newarray_overflow -O3 compilation from source
>> FAIL: newarray_overflow -O3 -findirect-dispatch compilation from source
> 
>   These turn out to be tree checking failures:
> 
>> In file included from <built-in>:3:0:
>> newarray_overflow.java:20:0: internal compiler error: tree check: expected class
>>  'type', have 'declaration' (function_decl) in put_decl_node, at java/lang.c:405

  This turned out to be a second bug, triggered by -fdump-tree-all attempting
to output some line of mangled gimple.  Without that flag, or ...

> 	      put_decl_node (TREE_CODE (DECL_CONTEXT (node)) == FUNCTION_DECL
>                                ? DECL_CONTEXT (node)
>                                : TYPE_NAME (DECL_CONTEXT (node)),
>                                verbosity);

... with the suggested fix, compilation proceeds to the site of the actual
crash that occurred in the testsuite run: a segfault in gimple.c#walk_gimple_op():

> Program received signal SIGSEGV, Segmentation fault.
> 0x007cfdbe in walk_gimple_op (stmt=0x7fe106e0, callback_op=0x4a0ef0 <verify_expr
> .265801>, wi=0x326c554) at /gnu/gcc/gcc/gcc/gimple.c:2841
> 2841      return !AGGREGATE_TYPE_P (type);
> (gdb) call debug_gimple_stmt (stmt)
> newarray_overflow.int_check.__builtin_prefetch (D.1284_85, 0, );
> 
> (gdb)

  I think this is the same as PR29710(*) "gnat ICEs on -fprefetch-loop-arrays
-O1", so I'll put the rest of my findings there.  Note that weird missing
third argument: it turns out that integer_three_node hasn't been initialised
at the time that issue_prefetch_ref() attempts to pass it to gimple_build_call().

    cheers,
      DaveK
-- 
(*) - http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29710