predict.c (expensive_function_p): New.
* predict.c (expensive_function_p): New. * rtl.h (expensive_function_p): Declare. * i386.c (FAST_PROLOGUE_INSN_COUNT): New constant. (use_fast_prologue_epilogue): New static variable. (expand_prologue): Set it; emit short prologues if unset. (expand_epilogue): Likewise. From-SVN: r45176
This commit is contained in:
Jan Hubicka
Jan Hubicka
parent
033c017cb2
commit
6ab16dd9ad
|
|
@ -1,3 +1,12 @@
|
|||
Sat Aug 25 23:07:35 CEST 2001 Jan Hubicka <jh@suse.cz>
|
||||
|
||||
* predict.c (expensive_function_p): New.
|
||||
* rtl.h (expensive_function_p): Declare.
|
||||
* i386.c (FAST_PROLOGUE_INSN_COUNT): New constant.
|
||||
(use_fast_prologue_epilogue): New static variable.
|
||||
(expand_prologue): Set it; emit short prologues if unset.
|
||||
(expand_epilogue): Likewise.
|
||||
|
||||
2001-08-22 Geoffrey Keating <geoffk@redhat.com>
|
||||
|
||||
* config.gcc: Add stormy16-*-elf case.
|
||||
|
|
|
|||
|
|
@ -319,6 +319,14 @@ const int x86_accumulate_outgoing_args = m_ATHLON | m_PENT4 | m_PPRO;
|
|||
const int x86_prologue_using_move = m_ATHLON | m_PENT4 | m_PPRO;
|
||||
const int x86_epilogue_using_move = m_ATHLON | m_PENT4 | m_PPRO;
|
||||
|
||||
/* In case the avreage insn count for single function invocation is
|
||||
lower than this constant, emit fast (but longer) prologue and
|
||||
epilogue code. */
|
||||
#define FAST_PROLOGUE_INSN_COUNT 30
|
||||
/* Set by prologue expander and used by epilogue expander to determine
|
||||
the style used. */
|
||||
static int use_fast_prologue_epilogue;
|
||||
|
||||
#define AT_BP(mode) (gen_rtx_MEM ((mode), hard_frame_pointer_rtx))
|
||||
|
||||
const char * const hi_reg_name[] = HI_REGISTER_NAMES;
|
||||
|
|
@ -2653,9 +2661,15 @@ ix86_expand_prologue ()
|
|||
|| current_function_uses_const_pool)
|
||||
&& !TARGET_64BIT);
|
||||
struct ix86_frame frame;
|
||||
int use_mov = (TARGET_PROLOGUE_USING_MOVE && !optimize_size);
|
||||
int use_mov = 0;
|
||||
HOST_WIDE_INT allocate;
|
||||
|
||||
if (TARGET_PROLOGUE_USING_MOVE && !optimize_size)
|
||||
{
|
||||
use_fast_prologue_epilogue
|
||||
= !expensive_function_p (FAST_PROLOGUE_INSN_COUNT);
|
||||
use_mov = use_fast_prologue_epilogue;
|
||||
}
|
||||
ix86_compute_frame_layout (&frame);
|
||||
|
||||
/* Note: AT&T enter does NOT have reversed args. Enter is probably
|
||||
|
|
@ -2794,10 +2808,11 @@ ix86_expand_epilogue (style)
|
|||
tuning in future. */
|
||||
if ((!sp_valid && frame.nregs <= 1)
|
||||
|| (TARGET_EPILOGUE_USING_MOVE && !optimize_size
|
||||
&& use_fast_prologue_epilogue
|
||||
&& (frame.nregs > 1 || frame.to_allocate))
|
||||
|| (frame_pointer_needed && !frame.nregs && frame.to_allocate)
|
||||
|| (frame_pointer_needed && TARGET_USE_LEAVE && !optimize_size
|
||||
&& frame.nregs == 1)
|
||||
&& use_fast_prologue_epilogue && frame.nregs == 1)
|
||||
|| style == 2)
|
||||
{
|
||||
/* Restore registers. We can use ebp or esp to address the memory
|
||||
|
|
@ -2844,7 +2859,7 @@ ix86_expand_epilogue (style)
|
|||
GEN_INT (frame.to_allocate
|
||||
+ frame.nregs * UNITS_PER_WORD)));
|
||||
/* If not an i386, mov & pop is faster than "leave". */
|
||||
else if (TARGET_USE_LEAVE || optimize_size)
|
||||
else if (TARGET_USE_LEAVE || optimize_size || !use_fast_prologue_epilogue)
|
||||
emit_insn (TARGET_64BIT ? gen_leave_rex64 () : gen_leave ());
|
||||
else
|
||||
{
|
||||
|
|
|
|||
|
|
@ -782,6 +782,51 @@ counts_to_freqs ()
|
|||
}
|
||||
}
|
||||
|
||||
/* Return true if function is likely to be expensive, so there is no point
|
||||
to optimizer performance of prologue, epilogue or do inlining at the
|
||||
expense of code size growth. THRESHOLD is the limit of number
|
||||
of isntructions function can execute at average to be still considered
|
||||
not expensive. */
|
||||
bool
|
||||
expensive_function_p (threshold)
|
||||
int threshold;
|
||||
{
|
||||
unsigned int sum = 0;
|
||||
int i;
|
||||
int limit;
|
||||
|
||||
/* We can not compute accurately for large thresholds due to scaled
|
||||
frequencies. */
|
||||
if (threshold > BB_FREQ_MAX)
|
||||
abort ();
|
||||
|
||||
/* Frequencies are out of range. This eighter means that function contains
|
||||
internal loop executing more than BB_FREQ_MAX times or profile feedback
|
||||
is available and function has not been executed at all. */
|
||||
if (ENTRY_BLOCK_PTR->frequency == 0)
|
||||
return true;
|
||||
|
||||
/* Maximally BB_FREQ_MAX^2 so overflow won't happen. */
|
||||
limit = ENTRY_BLOCK_PTR->frequency * threshold;
|
||||
for (i = 0; i < n_basic_blocks; i++)
|
||||
{
|
||||
basic_block bb = BASIC_BLOCK (i);
|
||||
rtx insn;
|
||||
|
||||
for (insn = bb->head; insn != NEXT_INSN (bb->end);
|
||||
insn = NEXT_INSN (insn))
|
||||
{
|
||||
if (active_insn_p (insn))
|
||||
{
|
||||
sum += bb->frequency;
|
||||
if (sum > limit)
|
||||
return true;
|
||||
}
|
||||
}
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
/* Estimate basic blocks frequency by given branch probabilities. */
|
||||
static void
|
||||
estimate_bb_frequencies (loops)
|
||||
|
|
|
|||
Loading…
Reference in New Issue