On Fri, Apr 8, 2011 at 5:04 AM, Diego Novillo <[email protected]> wrote:
> On 04/03/2011 08:54 PM, Xinliang David Li wrote:
>
>> +
>> + fprintf (file, "# BLOCK %d", bb->index);
>> + if (bb->frequency)
>> + fprintf (file, " freq:%d", bb->frequency);
>> + if (bb->count)
>> + fprintf (file, " count:" HOST_WIDEST_INT_PRINT_DEC,
>> + bb->count);
>> + fprintf (file, " seq:%d", bb_seqn++);
>
> What is this sequence number useful for?
Keep track of the physical order of BBs laid out in the object file --
good for diagnosing bb reorder problems.
>
>
>> /* Output the insns. */
>> for (insn = first; insn;)
>> {
>> @@ -1721,8 +1788,19 @@ final (rtx first, FILE *file, int optimi
>> insn_current_address = INSN_ADDRESSES (INSN_UID (insn));
>> #endif /* HAVE_ATTR_length */
>>
>> + dump_basic_block_info (file, insn);
>
> Pass start_to_bb, end_to_bb and bb_seqn++ as arguments. No need to have
> globals. You may want to convert start_to_bb and end_to_bb to VEC(), but in
> the way you're using them, it may not make much difference in clarity.
Those are used to keep global states, not for the purpose parameter passing.
>
>
>> - if (flag_debug_asm)
>> - fprintf (asm_out_file, "\t%s basic block %d\n",
>> - ASM_COMMENT_START, NOTE_BASIC_BLOCK (insn)->index);
>> -
>
> There's a good number of test cases that use -dA. I suppose that removing
> this and adding the new output did not cause new regresions?
>
No. The change has been in google's internal compiler for a long time.
Thanks,
David
>
> OK with those changes.
>
>
> Diego.
>
