Re: [patch][version 6] add -ftrivial-auto-var-init and variable attribute "uninitialized" to gcc

[Qing Zhao <qing.zhao@oracle.com>]

> On Aug 17, 2021, at 3:43 AM, Richard Biener <rguenther@suse.de> wrote:
> 
> On Mon, 16 Aug 2021, Qing Zhao wrote:
> 
>> 
>> 
>>> On Aug 16, 2021, at 2:40 AM, Richard Biener <rguenther@suse.de> wrote:
>>> 
>>> On Thu, 12 Aug 2021, Qing Zhao wrote:
>>> 
>>>> Hi, Richard,
>>>> 
>>>> For RTL expansion of call to .DEFERRED_INIT, I changed my code per your suggestions like following:
>>>> 
>>>> ======================
>>>> #define INIT_PATTERN_VALUE  0xFE
>>>> static void
>>>> expand_DEFERRED_INIT (internal_fn, gcall *stmt)
>>>> {
>>>> tree lhs = gimple_call_lhs (stmt);
>>>> tree var_size = gimple_call_arg (stmt, 0);
>>>> enum auto_init_type init_type
>>>>   = (enum auto_init_type) TREE_INT_CST_LOW (gimple_call_arg (stmt, 1));
>>>> bool is_vla = (bool) TREE_INT_CST_LOW (gimple_call_arg (stmt, 2));
>>>> 
>>>> tree var_type = TREE_TYPE (lhs);
>>>> gcc_assert (init_type > AUTO_INIT_UNINITIALIZED);
>>>> 
>>>> if (is_vla || (!can_native_interpret_type_p (var_type)))
>>>>   {
>>>>   /* If this is a VLA or the type of the variable cannot be natively
>>>>      interpreted, expand to a memset to initialize it.  */
>>>>     if (TREE_CODE (lhs) == SSA_NAME)
>>>>       lhs = SSA_NAME_VAR (lhs);
>>>>     tree var_addr = NULL_TREE;
>>>>     if (is_vla)
>>>>       var_addr = TREE_OPERAND (lhs, 0);
>>>>     else
>>>>       {
>>>>        TREE_ADDRESSABLE (lhs) = 1;
>>>>        var_addr = build_fold_addr_expr (lhs);
>>>>       }
>>>>     tree value = (init_type == AUTO_INIT_PATTERN) ?
>>>>                   build_int_cst (unsigned_char_type_node,
>>>>                                  INIT_PATTERN_VALUE) :
>>>>                   build_zero_cst (unsigned_char_type_node);
>>>>     tree m_call = build_call_expr (builtin_decl_implicit (BUILT_IN_MEMSET),
>>>>                                    3, var_addr, value, var_size);
>>>>     /* Expand this memset call.  */
>>>>     expand_builtin_memset (m_call, NULL_RTX, TYPE_MODE (var_type));
>>>>   }
>>>> else
>>>>   {
>>>>   /* If this is not a VLA and the type of the variable can be natively 
>>>>      interpreted, expand to assignment to generate better code.  */
>>>>     tree pattern = NULL_TREE;
>>>>     unsigned HOST_WIDE_INT total_bytes
>>>>       = tree_to_uhwi (TYPE_SIZE_UNIT (var_type));
>>>> 
>>>>     if (init_type == AUTO_INIT_PATTERN)
>>>>       {
>>>>         unsigned char *buf = (unsigned char *) xmalloc (total_bytes);
>>>>         memset (buf, INIT_PATTERN_VALUE, total_bytes);
>>>>         pattern = native_interpret_expr (var_type, buf, total_bytes);
>>>>         gcc_assert (pattern);
>>>>       }
>>>> 
>>>>     tree init = (init_type == AUTO_INIT_PATTERN) ?
>>>>                  pattern :
>>>>                  build_zero_cst (var_type);
>>>>     expand_assignment (lhs, init, false);
>>>>   }
>>>> }
>>>> ===========================
>>>> 
>>>> Now, I used “can_native_interpret_type_p (var_type)” instead of “use_register_for_decl (lhs)” to decide 
>>>> whether to use “memset” or use “assign” to expand this function.
>>>> 
>>>> However, this exposed an bug that is very hard to be addressed:
>>>> 
>>>> *******For the testing case: test suite/gcc.dg/uninit-I.c:
>>>> 
>>>> /* { dg-do compile } */
>>>> /* { dg-options "-O2 -Wuninitialized" } */
>>>> 
>>>> int sys_msgctl (void)
>>>> {
>>>> struct { int mode; } setbuf;
>>>> return setbuf.mode;  /* { dg-warning "'setbuf\.mode' is used" } */
>>>> ==
>>>> 
>>>> ******the above auto var “setbuf” has “struct” type, which “can_native_interpret_type_p(var_type)” is false, therefore, 
>>>> Expanding this .DEFERRED_INIT call went down the “memset” expansion route. 
>>>> 
>>>> However, this structure type can be fitted into a register, therefore cannot be taken address anymore at this stage, even though I tried:
>>>> 
>>>>        TREE_ADDRESSABLE (lhs) = 1;
>>>>        var_addr = build_fold_addr_expr (lhs);
>>>> 
>>>> To create an address variable for it, the expansion still failed at expr.c: line 8412:
>>>> during RTL pass: expand
>>>> /home/opc/Work/GCC/latest-gcc/gcc/testsuite/gcc.dg/auto-init-uninit-I.c:6:24: internal compiler error: in expand_expr_addr_expr_1, at expr.c:8412
>>>> 0xd04104 expand_expr_addr_expr_1
>>>> 	../../latest-gcc/gcc/expr.c:8412
>>>> 0xd04a95 expand_expr_addr_expr
>>>> 	../../latest-gcc/gcc/expr.c:8525
>>>> 0xd13592 expand_expr_real_1(tree_node*, rtx_def*, machine_mode, expand_modifier, rtx_def**, bool)
>>>> 	../../latest-gcc/gcc/expr.c:11741
>>>> 0xd05142 expand_expr_real(tree_node*, rtx_def*, machine_mode, expand_modifier, rtx_def**, bool)
>>>> 	../../latest-gcc/gcc/expr.c:8713
>>>> 0xaed1d3 expand_expr
>>>> 	../../latest-gcc/gcc/expr.h:301
>>>> 0xaf0d89 get_memory_rtx
>>>> 	../../latest-gcc/gcc/builtins.c:1370
>>>> 0xafb4fb expand_builtin_memset_args
>>>> 	../../latest-gcc/gcc/builtins.c:4102
>>>> 0xafacde expand_builtin_memset(tree_node*, rtx_def*, machine_mode)
>>>> 	../../latest-gcc/gcc/builtins.c:3886
>>>> 0xe97fb3 expand_DEFERRED_INIT
>>>> 
>>>> ******That’s the major reason why I chose “use_register_for_decl(lhs)” to decide “memset” expansion or “assign” expansion, “memset” expansion
>>>> needs to take address of the variable, if the variable has been decided to fit into a register, then its address cannot taken anymore at this stage.
>>>> 
>>>> ******using “can_native_interpret_type_p” did make the “pattern” generation part much  cleaner and simpler, however, looks like it didn’t work correctly.
>>>> 
>>>> Based on this, I’d like to keep my previous implementation by using “use_register_for_decl” to decide whether to take “memset” expansion or “assign” expansion.
>>>> Therefore, I might still need to keep the “UGLY”  implementation of generatting “pattern” constant for different types?
>>>> 
>>>> Let me know your opinion on this.
>>> 
>>> Hmm, I think you can use use_register_for_decl(lhs) to decide to use an
>>> alternate type to generate the pattern (and feed to 
>>> can_native_interpret_type_p) by using
>>> lang_hooks.type_for_mode (TYPE_MODE (TREE_TYPE (lhs))).
>> 
>> Do you mean that the TYPE returned by “lang_hooks.type_for_mode(TYPE_MODE (TREE_TYPE (lhs))” will always satisfy “can_native_interpret_type_p”? 
>> Even for big structure types?
> 
> I meant that for use_register_for_decl (lhs) the structures will be
> always small and the structure type will have a mode that is not BLKmode
> (but for example DImode for struct { int i; int j; }).

Oh, I see.

> 
>> i.e, 
>> 
>> tree var_type = TREE_TYPE(lhs);
>> tree pattern = NULL_TREE;
>> unsigned HOST_WIDE_INT total_bytes
>>        = tree_to_uhwi (TYPE_SIZE_UNIT (var_type));
>> 
>> If (use_register_for_decl(lhs)==false)
>>  {
>>    tree alt_type = lang_hooks.type_for_mode(TYPE_MODE(var_type), TYPE_UNSIGNED(var_type);
>>    If (can_native_interpret_type_p (alt_type))
>>      {
>> 	unsigned char *buf = (unsigned char *) xmalloc (total_bytes); 
>> 	memset (buf, INIT_PATTERN_VALUE, total_bytes); 
>> 	pattern = native_interpret_expr (alt_type, buf, total_bytes); 
>> 	gcc_assert (pattern); 
>>      }
>>    else
>>      gcc_unreachable ();
>>  }
>> 
>> ?
>> Don’t quite understand here. Please clarify.
> 
> For !use_register_for_decl you use memset already, but for
> use_register_for_decl not all types satisfy can_native_interpret_type_p
> (in particular all struct and union types).  But when we use a
> register for the decl then we can of course directly initialize the
> register.

So, you mean the following:

If (is_vla || (!use_register_for_decl (lhs)))
{
  /* expand as memset that is done currently.  */
}
else 
 {
   tree alt_type = lang_hooks.type_for_mode(TYPE_MODE(var_type), TYPE_UNSIGNED(var_type);
   If (can_native_interpret_type_p (alt_type))
     {
	unsigned char *buf = (unsigned char *) xmalloc (total_bytes); 
	memset (buf, INIT_PATTERN_VALUE, total_bytes); 
	pattern = native_interpret_expr (alt_type, buf, total_bytes); 
	gcc_assert (pattern); 
     }
   else
     gcc_unreachable ();

     tree init = (init_type == AUTO_INIT_PATTERN) ?
                   pattern :
                   build_zero_cst (alt_type);

    /* here build VIEW_CONVERT <alt_type> (lhs) = init;
       And then expand it.  */
 }

?

>  As said, it would be much cleaner (and maybe also easier)
> to then simply expand the RTL directly rather than going through
> expand_assignment.

Dump questions here:  

1. When building VIEW_CONVERT  <alt_type> (lhs) = init and expand it:

Is the following correct:

 lhs = build1 (VIEW_CONVERT_EXPR, alt_type, lhs);
expand_assignment (lhs, init);


which utility routines should be used to building the assignment and then expand it?

Are the above utility routines correct?


>  It's also easier to directly see whether the
> LHS is a MEM_P or a REG_P but then for the MEM_P case it's a bit
> more "awkward" to use the easy way of the generic expand code
> (esp. if we eventually want to emit actual calls to memset - do we?)

If all these are guarded by “use_register_for_decl” already, the lhs should be fit into registers, 
Under such situation, I don’t think that we want to emit actual calls to memset. That’s too expensive.

> 
>>> You can then
>>> build the assignment from the pattern as
>>> 
>>> VIEW_CONVERT <reg-type> (lhs) = pattern_cst;
>> 
>> What’s the <reg_type> in the above? The type “alt_type” returned by “lang_hooks.type_for_mode(TREE_TYPE(lhs))?
>> Do I need to build “MODIFY_EXPR” for the above? 
> 
> Yes, the lang_hook.type_for_mode result and no, you could go through
> expand_assignment.

Okay. So, still use “expand_assigment” to expand it?
> 
>> lhs =  build1 (VIEW_CONVERT_EXPR, alt_type, lhs);
>> tree  final = build2 (MODIFY_EXPR, TREE_TYPE (alt_type), lhs, pattern);
>> 
>> Then how to expand this”final"? 
>>> 
>>> note that more RTL-expand-ish would be to simply expand 'lhs' and
>>> see whether it's a REG_P or a MEM_P and decide based on that.
>> 
>> You mean that the current RTL expansion will automatically expand LHS to memset route or assignment route based on whether
>> LHS is a REG_P or MEM_P? I don’t need to explicitly code for “expand_builtin_memset” or “expand_assign”?
> 
> No.  But you are inside the expander for the internal function call
> and this is expected to generate RTL.  You can simply generate
> RTL directly without "faking" new GENERIC calls or assignments and
> expanding those.
> 
> But lets not go there for now.

Okay. I see now.

Qing
> 
> Richard.