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 3:29 AM, Richard Biener <rguenther@suse.de> wrote:
> 
> On Mon, 16 Aug 2021, Qing Zhao wrote:
> 
>> My current code for expand_DEFERRED_INIT is like the following, could you check and see whether there is any issue for it:
>> 
>> #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 || (!use_register_for_decl (lhs)))
>>    {
>>      if (TREE_CODE (lhs) == SSA_NAME)
>>        lhs = SSA_NAME_VAR (lhs);
> 
> this should not be necessary (in fact you shouldn't see a SSA_NAME
> here, if you do then using SSA_NAME_VAR is wrong)
You mean during RTL expansion phase, all SSA_NAMEs are gone already?
> 
>>    /* If this is a VLA or the variable is not in register,
>>       expand to a memset to initialize it.  */
>>      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);
>>        }
> 
> use, independent of is_vla
> 
>         mark_addressable (lhs);
>         var_addr = build_fold_addr_expr (lhs);
Okay.
> 
>> 
>>      tree value = (init_type == AUTO_INIT_PATTERN) ?
>>                    build_int_cst (unsigned_char_type_node,
>>                                   INIT_PATTERN_VALUE) :
>>                    build_zero_cst (unsigned_char_type_node);
> 
> since memset has an integer argument for the value use
> integer_zero_node for the zero case and build_int_cst (integer_type_node, 
> ...) for the pattern case

Okay.

> 
>>      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 variable is in a register, use expand_assignment might
>>       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)
>>        {
>>          if (can_native_interpret_type_p (var_type))
>>            {
>>              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);
>>            }
>>          else
>>            {
>>              tree index_type = build_index_type (size_int (total_bytes - 1));
>>              tree array_type = build_array_type (unsigned_char_type_node,
>>                                                  index_type);
>>              tree element = build_int_cst (unsigned_char_type_node,
>>                                            INIT_PATTERN_VALUE);
>>              vec<constructor_elt, va_gc> *elts = NULL;
>>              for (unsigned int i = 0; i < total_bytes; i++)
>>                CONSTRUCTOR_APPEND_ELT (elts, NULL_TREE, element);
>>              pattern = build_constructor (array_type, elts);
>>              pattern = build1 (VIEW_CONVERT_EXPR, var_type, pattern);
>>            }
>>        }
>> 
>>      tree init = (init_type == AUTO_INIT_PATTERN) ?
>>                   pattern :
>>                   build_zero_cst (var_type);
> 
> maybe conditionally initialize init instead of pattern and init?
> Thus replace pattern by init and do
> 
>        else
>          init = build_zero_cst (var_type);

You mean the following:


tree init = pattern;
If (init_type != AUTO_INIT_PATTERN)
  Init = build_zero_cst (var_type);

Or something else?

> 
> 
> the above should work, as said the RTL expansion part can possibly
> be improved but we can do this as followup as well.

Okay.

Qing
> 
>>      expand_assignment (lhs, init, false);
>>    }
>> }
>> 
>> Thanks.
>> 
>> Qing
>> 
>> 
>> 
>> 
> 
> -- 
> Richard Biener <rguenther@suse.de>
> SUSE Software Solutions Germany GmbH, Maxfeldstrasse 5, 90409 Nuernberg,
> Germany; GF: Felix Imendörffer; HRB 36809 (AG Nuernberg)