/* Subroutines for manipulating rtx's in semantically interesting ways.
Copyright (C) 1987, 1991, 1994, 1995, 1996, 1997, 1998,
- 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
+ 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
Free Software Foundation, Inc.
This file is part of GCC.
#include "system.h"
#include "coretypes.h"
#include "tm.h"
-#include "toplev.h"
+#include "diagnostic-core.h"
#include "rtl.h"
#include "tree.h"
#include "tm_p.h"
#include "function.h"
#include "expr.h"
#include "optabs.h"
+#include "libfuncs.h"
#include "hard-reg-set.h"
#include "insn-config.h"
#include "ggc.h"
#include "output.h"
static rtx break_out_memory_refs (rtx);
-static void emit_stack_probe (rtx);
/* Truncate and perhaps sign-extend C as appropriate for MODE. */
if (SYMBOL_REF_DECL (s) && DECL_P (SYMBOL_REF_DECL (s)))
sa = DECL_ALIGN (SYMBOL_REF_DECL (s));
- ca = exact_log2 (INTVAL (c) & -INTVAL (c)) * BITS_PER_UNIT;
-
- align = MIN (sa, ca);
+ if (INTVAL (c) == 0)
+ align = sa;
+ else
+ {
+ ca = ctz_hwi (INTVAL (c)) * BITS_PER_UNIT;
+ align = MIN (sa, ca);
+ }
}
if (align || (MEM_P (x) && MEM_POINTER (x)))
promote_function_mode (const_tree type, enum machine_mode mode, int *punsignedp,
const_tree funtype, int for_return)
{
+ /* Called without a type node for a libcall. */
+ if (type == NULL_TREE)
+ {
+ if (INTEGRAL_MODE_P (mode))
+ return targetm.calls.promote_function_mode (NULL_TREE, mode,
+ punsignedp, funtype,
+ for_return);
+ else
+ return mode;
+ }
+
switch (TREE_CODE (type))
{
case INTEGER_TYPE: case ENUMERAL_TYPE: case BOOLEAN_TYPE:
promote_mode (const_tree type ATTRIBUTE_UNUSED, enum machine_mode mode,
int *punsignedp ATTRIBUTE_UNUSED)
{
+#ifdef PROMOTE_MODE
+ enum tree_code code;
+ int unsignedp;
+#endif
+
+ /* For libcalls this is invoked without TYPE from the backends
+ TARGET_PROMOTE_FUNCTION_MODE hooks. Don't do anything in that
+ case. */
+ if (type == NULL_TREE)
+ return mode;
+
/* FIXME: this is the same logic that was there until GCC 4.4, but we
probably want to test POINTERS_EXTEND_UNSIGNED even if PROMOTE_MODE
is not defined. The affected targets are M32C, S390, SPARC. */
#ifdef PROMOTE_MODE
- const enum tree_code code = TREE_CODE (type);
- int unsignedp = *punsignedp;
+ code = TREE_CODE (type);
+ unsignedp = *punsignedp;
switch (code)
{
static rtx
round_push (rtx size)
{
- int align = PREFERRED_STACK_BOUNDARY / BITS_PER_UNIT;
+ rtx align_rtx, alignm1_rtx;
- if (align == 1)
- return size;
-
- if (CONST_INT_P (size))
+ if (!SUPPORTS_STACK_ALIGNMENT
+ || crtl->preferred_stack_boundary == MAX_SUPPORTED_STACK_ALIGNMENT)
{
- HOST_WIDE_INT new_size = (INTVAL (size) + align - 1) / align * align;
+ int align = crtl->preferred_stack_boundary / BITS_PER_UNIT;
- if (INTVAL (size) != new_size)
- size = GEN_INT (new_size);
+ if (align == 1)
+ return size;
+
+ if (CONST_INT_P (size))
+ {
+ HOST_WIDE_INT new_size = (INTVAL (size) + align - 1) / align * align;
+
+ if (INTVAL (size) != new_size)
+ size = GEN_INT (new_size);
+ return size;
+ }
+
+ align_rtx = GEN_INT (align);
+ alignm1_rtx = GEN_INT (align - 1);
}
else
{
- /* CEIL_DIV_EXPR needs to worry about the addition overflowing,
- but we know it can't. So add ourselves and then do
- TRUNC_DIV_EXPR. */
- size = expand_binop (Pmode, add_optab, size, GEN_INT (align - 1),
- NULL_RTX, 1, OPTAB_LIB_WIDEN);
- size = expand_divmod (0, TRUNC_DIV_EXPR, Pmode, size, GEN_INT (align),
- NULL_RTX, 1);
- size = expand_mult (Pmode, size, GEN_INT (align), NULL_RTX, 1);
+ /* If crtl->preferred_stack_boundary might still grow, use
+ virtual_preferred_stack_boundary_rtx instead. This will be
+ substituted by the right value in vregs pass and optimized
+ during combine. */
+ align_rtx = virtual_preferred_stack_boundary_rtx;
+ alignm1_rtx = force_operand (plus_constant (align_rtx, -1), NULL_RTX);
}
+ /* CEIL_DIV_EXPR needs to worry about the addition overflowing,
+ but we know it can't. So add ourselves and then do
+ TRUNC_DIV_EXPR. */
+ size = expand_binop (Pmode, add_optab, size, alignm1_rtx,
+ NULL_RTX, 1, OPTAB_LIB_WIDEN);
+ size = expand_divmod (0, TRUNC_DIV_EXPR, Pmode, size, align_rtx,
+ NULL_RTX, 1);
+ size = expand_mult (Pmode, size, align_rtx, NULL_RTX, 1);
+
return size;
}
\f
/* Save the stack pointer for the purpose in SAVE_LEVEL. PSAVE is a pointer
to a previously-created save area. If no save area has been allocated,
this function will allocate one. If a save area is specified, it
- must be of the proper mode.
-
- The insns are emitted after insn AFTER, if nonzero, otherwise the insns
- are emitted at the current position. */
+ must be of the proper mode. */
void
-emit_stack_save (enum save_level save_level, rtx *psave, rtx after)
+emit_stack_save (enum save_level save_level, rtx *psave)
{
rtx sa = *psave;
/* The default is that we use a move insn and save in a Pmode object. */
}
}
- if (after)
- {
- rtx seq;
-
- start_sequence ();
- do_pending_stack_adjust ();
- /* We must validize inside the sequence, to ensure that any instructions
- created by the validize call also get moved to the right place. */
- if (sa != 0)
- sa = validize_mem (sa);
- emit_insn (fcn (sa, stack_pointer_rtx));
- seq = get_insns ();
- end_sequence ();
- emit_insn_after (seq, after);
- }
- else
- {
- do_pending_stack_adjust ();
- if (sa != 0)
- sa = validize_mem (sa);
- emit_insn (fcn (sa, stack_pointer_rtx));
- }
+ do_pending_stack_adjust ();
+ if (sa != 0)
+ sa = validize_mem (sa);
+ emit_insn (fcn (sa, stack_pointer_rtx));
}
/* Restore the stack pointer for the purpose in SAVE_LEVEL. SA is the save
- area made by emit_stack_save. If it is zero, we have nothing to do.
-
- Put any emitted insns after insn AFTER, if nonzero, otherwise at
- current position. */
+ area made by emit_stack_save. If it is zero, we have nothing to do. */
void
-emit_stack_restore (enum save_level save_level, rtx sa, rtx after)
+emit_stack_restore (enum save_level save_level, rtx sa)
{
/* The default is that we use a move insn. */
rtx (*fcn) (rtx, rtx) = gen_move_insn;
discard_pending_stack_adjust ();
- if (after)
- {
- rtx seq;
-
- start_sequence ();
- emit_insn (fcn (stack_pointer_rtx, sa));
- seq = get_insns ();
- end_sequence ();
- emit_insn_after (seq, after);
- }
- else
- emit_insn (fcn (stack_pointer_rtx, sa));
+ emit_insn (fcn (stack_pointer_rtx, sa));
}
/* Invoke emit_stack_save on the nonlocal_goto_save_area for the current
integer_one_node, NULL_TREE, NULL_TREE);
r_save = expand_expr (t_save, NULL_RTX, VOIDmode, EXPAND_WRITE);
- emit_stack_save (SAVE_NONLOCAL, &r_save, NULL_RTX);
+ emit_stack_save (SAVE_NONLOCAL, &r_save);
}
\f
/* Return an rtx representing the address of an area of memory dynamically
- pushed on the stack. This region of memory is always aligned to
- a multiple of BIGGEST_ALIGNMENT.
+ pushed on the stack.
Any required stack pointer alignment is preserved.
SIZE is an rtx representing the size of the area.
- TARGET is a place in which the address can be placed.
- KNOWN_ALIGN is the alignment (in bits) that we know SIZE has. */
+ SIZE_ALIGN is the alignment (in bits) that we know SIZE has. This
+ parameter may be zero. If so, a proper value will be extracted
+ from SIZE if it is constant, otherwise BITS_PER_UNIT will be assumed.
+
+ REQUIRED_ALIGN is the alignment (in bits) required for the region
+ of memory.
+
+ If CANNOT_ACCUMULATE is set to TRUE, the caller guarantees that the
+ stack space allocated by the generated code cannot be added with itself
+ in the course of the execution of the function. It is always safe to
+ pass FALSE here and the following criterion is sufficient in order to
+ pass TRUE: every path in the CFG that starts at the allocation point and
+ loops to it executes the associated deallocation code. */
rtx
-allocate_dynamic_stack_space (rtx size, rtx target, int known_align)
+allocate_dynamic_stack_space (rtx size, unsigned size_align,
+ unsigned required_align, bool cannot_accumulate)
{
+ HOST_WIDE_INT stack_usage_size = -1;
+ rtx final_label, final_target, target;
+ unsigned extra_align = 0;
+ bool must_align;
+
/* If we're asking for zero bytes, it doesn't matter what we point
to since we can't dereference it. But return a reasonable
address anyway. */
/* Otherwise, show we're calling alloca or equivalent. */
cfun->calls_alloca = 1;
+ /* If stack usage info is requested, look into the size we are passed.
+ We need to do so this early to avoid the obfuscation that may be
+ introduced later by the various alignment operations. */
+ if (flag_stack_usage_info)
+ {
+ if (CONST_INT_P (size))
+ stack_usage_size = INTVAL (size);
+ else if (REG_P (size))
+ {
+ /* Look into the last emitted insn and see if we can deduce
+ something for the register. */
+ rtx insn, set, note;
+ insn = get_last_insn ();
+ if ((set = single_set (insn)) && rtx_equal_p (SET_DEST (set), size))
+ {
+ if (CONST_INT_P (SET_SRC (set)))
+ stack_usage_size = INTVAL (SET_SRC (set));
+ else if ((note = find_reg_equal_equiv_note (insn))
+ && CONST_INT_P (XEXP (note, 0)))
+ stack_usage_size = INTVAL (XEXP (note, 0));
+ }
+ }
+
+ /* If the size is not constant, we can't say anything. */
+ if (stack_usage_size == -1)
+ {
+ current_function_has_unbounded_dynamic_stack_size = 1;
+ stack_usage_size = 0;
+ }
+ }
+
/* Ensure the size is in the proper mode. */
if (GET_MODE (size) != VOIDmode && GET_MODE (size) != Pmode)
size = convert_to_mode (Pmode, size, 1);
+ /* Adjust SIZE_ALIGN, if needed. */
+ if (CONST_INT_P (size))
+ {
+ unsigned HOST_WIDE_INT lsb;
+
+ lsb = INTVAL (size);
+ lsb &= -lsb;
+
+ /* Watch out for overflow truncating to "unsigned". */
+ if (lsb > UINT_MAX / BITS_PER_UNIT)
+ size_align = 1u << (HOST_BITS_PER_INT - 1);
+ else
+ size_align = (unsigned)lsb * BITS_PER_UNIT;
+ }
+ else if (size_align < BITS_PER_UNIT)
+ size_align = BITS_PER_UNIT;
+
/* We can't attempt to minimize alignment necessary, because we don't
know the final value of preferred_stack_boundary yet while executing
this code. */
- crtl->preferred_stack_boundary = PREFERRED_STACK_BOUNDARY;
+ if (crtl->preferred_stack_boundary < PREFERRED_STACK_BOUNDARY)
+ crtl->preferred_stack_boundary = PREFERRED_STACK_BOUNDARY;
/* We will need to ensure that the address we return is aligned to
- BIGGEST_ALIGNMENT. If STACK_DYNAMIC_OFFSET is defined, we don't
+ REQUIRED_ALIGN. If STACK_DYNAMIC_OFFSET is defined, we don't
always know its final value at this point in the compilation (it
might depend on the size of the outgoing parameter lists, for
example), so we must align the value to be returned in that case.
(Note that STACK_DYNAMIC_OFFSET will have a default nonzero value if
STACK_POINTER_OFFSET or ACCUMULATE_OUTGOING_ARGS are defined).
We must also do an alignment operation on the returned value if
- the stack pointer alignment is less strict that BIGGEST_ALIGNMENT.
+ the stack pointer alignment is less strict than REQUIRED_ALIGN.
If we have to align, we must leave space in SIZE for the hole
that might result from the alignment operation. */
+ must_align = (crtl->preferred_stack_boundary < required_align);
+ if (must_align)
+ {
+ if (required_align > PREFERRED_STACK_BOUNDARY)
+ extra_align = PREFERRED_STACK_BOUNDARY;
+ else if (required_align > STACK_BOUNDARY)
+ extra_align = STACK_BOUNDARY;
+ else
+ extra_align = BITS_PER_UNIT;
+ }
+
+ /* ??? STACK_POINTER_OFFSET is always defined now. */
#if defined (STACK_DYNAMIC_OFFSET) || defined (STACK_POINTER_OFFSET)
-#define MUST_ALIGN 1
-#else
-#define MUST_ALIGN (PREFERRED_STACK_BOUNDARY < BIGGEST_ALIGNMENT)
+ must_align = true;
+ extra_align = BITS_PER_UNIT;
#endif
- if (MUST_ALIGN)
- size
- = force_operand (plus_constant (size,
- BIGGEST_ALIGNMENT / BITS_PER_UNIT - 1),
- NULL_RTX);
+ if (must_align)
+ {
+ unsigned extra = (required_align - extra_align) / BITS_PER_UNIT;
+
+ size = plus_constant (size, extra);
+ size = force_operand (size, NULL_RTX);
+
+ if (flag_stack_usage_info)
+ stack_usage_size += extra;
+
+ if (extra && size_align > extra_align)
+ size_align = extra_align;
+ }
#ifdef SETJMP_VIA_SAVE_AREA
/* If setjmp restores regs from a save area in the stack frame,
would use reg notes to store the "optimized" size and fix things
up later. These days we know this information before we ever
start building RTL so the reg notes are unnecessary. */
- if (!cfun->calls_setjmp)
- {
- int align = PREFERRED_STACK_BOUNDARY / BITS_PER_UNIT;
-
- /* ??? Code below assumes that the save area needs maximal
- alignment. This constraint may be too strong. */
- gcc_assert (PREFERRED_STACK_BOUNDARY == BIGGEST_ALIGNMENT);
-
- if (CONST_INT_P (size))
- {
- HOST_WIDE_INT new_size = INTVAL (size) / align * align;
-
- if (INTVAL (size) != new_size)
- size = GEN_INT (new_size);
- }
- else
- {
- /* Since we know overflow is not possible, we avoid using
- CEIL_DIV_EXPR and use TRUNC_DIV_EXPR instead. */
- size = expand_divmod (0, TRUNC_DIV_EXPR, Pmode, size,
- GEN_INT (align), NULL_RTX, 1);
- size = expand_mult (Pmode, size,
- GEN_INT (align), NULL_RTX, 1);
- }
- }
- else
+ if (cfun->calls_setjmp)
{
rtx dynamic_offset
= expand_binop (Pmode, sub_optab, virtual_stack_dynamic_rtx,
size = expand_binop (Pmode, add_optab, size, dynamic_offset,
NULL_RTX, 1, OPTAB_LIB_WIDEN);
+
+ /* The above dynamic offset cannot be computed statically at this
+ point, but it will be possible to do so after RTL expansion is
+ done. Record how many times we will need to add it. */
+ if (flag_stack_usage_info)
+ current_function_dynamic_alloc_count++;
+
+ /* ??? Can we infer a minimum of STACK_BOUNDARY here? */
+ size_align = BITS_PER_UNIT;
}
#endif /* SETJMP_VIA_SAVE_AREA */
insns. Since this is an extremely rare event, we have no reliable
way of knowing which systems have this problem. So we avoid even
momentarily mis-aligning the stack. */
+ if (size_align % MAX_SUPPORTED_STACK_ALIGNMENT != 0)
+ {
+ size = round_push (size);
+
+ if (flag_stack_usage_info)
+ {
+ int align = crtl->preferred_stack_boundary / BITS_PER_UNIT;
+ stack_usage_size = (stack_usage_size + align - 1) / align * align;
+ }
+ }
+
+ target = gen_reg_rtx (Pmode);
+
+ /* The size is supposed to be fully adjusted at this point so record it
+ if stack usage info is requested. */
+ if (flag_stack_usage_info)
+ {
+ current_function_dynamic_stack_size += stack_usage_size;
- /* If we added a variable amount to SIZE,
- we can no longer assume it is aligned. */
-#if !defined (SETJMP_VIA_SAVE_AREA)
- if (MUST_ALIGN || known_align % PREFERRED_STACK_BOUNDARY != 0)
+ /* ??? This is gross but the only safe stance in the absence
+ of stack usage oriented flow analysis. */
+ if (!cannot_accumulate)
+ current_function_has_unbounded_dynamic_stack_size = 1;
+ }
+
+ final_label = NULL_RTX;
+ final_target = NULL_RTX;
+
+ /* If we are splitting the stack, we need to ask the backend whether
+ there is enough room on the current stack. If there isn't, or if
+ the backend doesn't know how to tell is, then we need to call a
+ function to allocate memory in some other way. This memory will
+ be released when we release the current stack segment. The
+ effect is that stack allocation becomes less efficient, but at
+ least it doesn't cause a stack overflow. */
+ if (flag_split_stack)
+ {
+ rtx available_label, ask, space, func;
+
+ available_label = NULL_RTX;
+
+#ifdef HAVE_split_stack_space_check
+ if (HAVE_split_stack_space_check)
+ {
+ available_label = gen_label_rtx ();
+
+ /* This instruction will branch to AVAILABLE_LABEL if there
+ are SIZE bytes available on the stack. */
+ emit_insn (gen_split_stack_space_check (size, available_label));
+ }
#endif
- size = round_push (size);
+
+ /* The __morestack_allocate_stack_space function will allocate
+ memory using malloc. If the alignment of the memory returned
+ by malloc does not meet REQUIRED_ALIGN, we increase SIZE to
+ make sure we allocate enough space. */
+ if (MALLOC_ABI_ALIGNMENT >= required_align)
+ ask = size;
+ else
+ {
+ ask = expand_binop (Pmode, add_optab, size,
+ GEN_INT (required_align / BITS_PER_UNIT - 1),
+ NULL_RTX, 1, OPTAB_LIB_WIDEN);
+ must_align = true;
+ }
+
+ func = init_one_libfunc ("__morestack_allocate_stack_space");
+
+ space = emit_library_call_value (func, target, LCT_NORMAL, Pmode,
+ 1, ask, Pmode);
+
+ if (available_label == NULL_RTX)
+ return space;
+
+ final_target = gen_reg_rtx (Pmode);
+
+ emit_move_insn (final_target, space);
+
+ final_label = gen_label_rtx ();
+ emit_jump (final_label);
+
+ emit_label (available_label);
+ }
do_pending_stack_adjust ();
gcc_assert (!(stack_pointer_delta
% (PREFERRED_STACK_BOUNDARY / BITS_PER_UNIT)));
- /* If needed, check that we have the required amount of stack.
- Take into account what has already been checked. */
- if (flag_stack_check == GENERIC_STACK_CHECK)
+ /* If needed, check that we have the required amount of stack. Take into
+ account what has already been checked. */
+ if (STACK_CHECK_MOVING_SP)
+ ;
+ else if (flag_stack_check == GENERIC_STACK_CHECK)
probe_stack_range (STACK_OLD_CHECK_PROTECT + STACK_CHECK_MAX_FRAME_SIZE,
size);
else if (flag_stack_check == STATIC_BUILTIN_STACK_CHECK)
probe_stack_range (STACK_CHECK_PROTECT, size);
- /* Don't use a TARGET that isn't a pseudo or is the wrong mode. */
- if (target == 0 || !REG_P (target)
- || REGNO (target) < FIRST_PSEUDO_REGISTER
- || GET_MODE (target) != Pmode)
- target = gen_reg_rtx (Pmode);
-
- mark_reg_pointer (target, known_align);
-
/* Perform the required allocation from the stack. Some systems do
this differently than simply incrementing/decrementing from the
stack pointer, such as acquiring the space by calling malloc(). */
#ifdef HAVE_allocate_stack
if (HAVE_allocate_stack)
{
- enum machine_mode mode = STACK_SIZE_MODE;
- insn_operand_predicate_fn pred;
-
+ struct expand_operand ops[2];
/* We don't have to check against the predicate for operand 0 since
TARGET is known to be a pseudo of the proper mode, which must
- be valid for the operand. For operand 1, convert to the
- proper mode and validate. */
- if (mode == VOIDmode)
- mode = insn_data[(int) CODE_FOR_allocate_stack].operand[1].mode;
-
- pred = insn_data[(int) CODE_FOR_allocate_stack].operand[1].predicate;
- if (pred && ! ((*pred) (size, mode)))
- size = copy_to_mode_reg (mode, convert_to_mode (mode, size, 1));
-
- emit_insn (gen_allocate_stack (target, size));
+ be valid for the operand. */
+ create_fixed_operand (&ops[0], target);
+ create_convert_operand_to (&ops[1], size, STACK_SIZE_MODE, true);
+ expand_insn (CODE_FOR_allocate_stack, 2, ops);
}
else
#endif
{
+ int saved_stack_pointer_delta;
+
#ifndef STACK_GROWS_DOWNWARD
emit_move_insn (target, virtual_stack_dynamic_rtx);
#endif
emit_label (space_available);
}
- anti_adjust_stack (size);
+ saved_stack_pointer_delta = stack_pointer_delta;
+ if (flag_stack_check && STACK_CHECK_MOVING_SP)
+ anti_adjust_stack_and_probe (size, false);
+ else
+ anti_adjust_stack (size);
+ /* Even if size is constant, don't modify stack_pointer_delta.
+ The constant size alloca should preserve
+ crtl->preferred_stack_boundary alignment. */
+ stack_pointer_delta = saved_stack_pointer_delta;
#ifdef STACK_GROWS_DOWNWARD
emit_move_insn (target, virtual_stack_dynamic_rtx);
#endif
}
- if (MUST_ALIGN)
+ /* Finish up the split stack handling. */
+ if (final_label != NULL_RTX)
+ {
+ gcc_assert (flag_split_stack);
+ emit_move_insn (final_target, target);
+ emit_label (final_label);
+ target = final_target;
+ }
+
+ if (must_align)
{
/* CEIL_DIV_EXPR needs to worry about the addition overflowing,
but we know it can't. So add ourselves and then do
TRUNC_DIV_EXPR. */
target = expand_binop (Pmode, add_optab, target,
- GEN_INT (BIGGEST_ALIGNMENT / BITS_PER_UNIT - 1),
+ GEN_INT (required_align / BITS_PER_UNIT - 1),
NULL_RTX, 1, OPTAB_LIB_WIDEN);
target = expand_divmod (0, TRUNC_DIV_EXPR, Pmode, target,
- GEN_INT (BIGGEST_ALIGNMENT / BITS_PER_UNIT),
+ GEN_INT (required_align / BITS_PER_UNIT),
NULL_RTX, 1);
target = expand_mult (Pmode, target,
- GEN_INT (BIGGEST_ALIGNMENT / BITS_PER_UNIT),
+ GEN_INT (required_align / BITS_PER_UNIT),
NULL_RTX, 1);
}
+ /* Now that we've committed to a return value, mark its alignment. */
+ mark_reg_pointer (target, required_align);
+
/* Record the new stack level for nonlocal gotos. */
if (cfun->nonlocal_goto_save_area != 0)
update_nonlocal_goto_save_area ();
static GTY(()) rtx stack_check_libfunc;
void
-set_stack_check_libfunc (rtx libfunc)
+set_stack_check_libfunc (const char *libfunc_name)
{
- stack_check_libfunc = libfunc;
+ gcc_assert (stack_check_libfunc == NULL_RTX);
+ stack_check_libfunc = gen_rtx_SYMBOL_REF (Pmode, libfunc_name);
}
\f
/* Emit one stack probe at ADDRESS, an address within the stack. */
-static void
+void
emit_stack_probe (rtx address)
{
rtx memref = gen_rtx_MEM (word_mode, address);
MEM_VOLATILE_P (memref) = 1;
- if (STACK_CHECK_PROBE_LOAD)
- emit_move_insn (gen_reg_rtx (word_mode), memref);
+ /* See if we have an insn to probe the stack. */
+#ifdef HAVE_probe_stack
+ if (HAVE_probe_stack)
+ emit_insn (gen_probe_stack (memref));
else
+#endif
emit_move_insn (memref, const0_rtx);
}
/* Probe a range of stack addresses from FIRST to FIRST+SIZE, inclusive.
- FIRST is a constant and size is a Pmode RTX. These are offsets from the
- current stack pointer. STACK_GROWS_DOWNWARD says whether to add or
- subtract from the stack. If SIZE is constant, this is done
- with a fixed number of probes. Otherwise, we must make a loop. */
+ FIRST is a constant and size is a Pmode RTX. These are offsets from
+ the current stack pointer. STACK_GROWS_DOWNWARD says whether to add
+ or subtract them from the stack pointer. */
+
+#define PROBE_INTERVAL (1 << STACK_CHECK_PROBE_INTERVAL_EXP)
#ifdef STACK_GROWS_DOWNWARD
#define STACK_GROW_OP MINUS
+#define STACK_GROW_OPTAB sub_optab
+#define STACK_GROW_OFF(off) -(off)
#else
#define STACK_GROW_OP PLUS
+#define STACK_GROW_OPTAB add_optab
+#define STACK_GROW_OFF(off) (off)
#endif
void
if (GET_MODE (size) != VOIDmode && GET_MODE (size) != Pmode)
size = convert_to_mode (Pmode, size, 1);
- /* Next see if the front end has set up a function for us to call to
- check the stack. */
- if (stack_check_libfunc != 0)
+ /* Next see if we have a function to check the stack. */
+ if (stack_check_libfunc)
{
- rtx addr = memory_address (QImode,
+ rtx addr = memory_address (Pmode,
gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
stack_pointer_rtx,
plus_constant (size, first)));
-
- addr = convert_memory_address (ptr_mode, addr);
emit_library_call (stack_check_libfunc, LCT_NORMAL, VOIDmode, 1, addr,
- ptr_mode);
+ Pmode);
+ return;
}
- /* Next see if we have an insn to check the stack. Use it if so. */
+ /* Next see if we have an insn to check the stack. */
#ifdef HAVE_check_stack
- else if (HAVE_check_stack)
+ if (HAVE_check_stack)
{
- insn_operand_predicate_fn pred;
- rtx last_addr
- = force_operand (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
- stack_pointer_rtx,
- plus_constant (size, first)),
- NULL_RTX);
-
- pred = insn_data[(int) CODE_FOR_check_stack].operand[0].predicate;
- if (pred && ! ((*pred) (last_addr, Pmode)))
- last_addr = copy_to_mode_reg (Pmode, last_addr);
-
- emit_insn (gen_check_stack (last_addr));
+ struct expand_operand ops[1];
+ rtx addr = memory_address (Pmode,
+ gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
+ stack_pointer_rtx,
+ plus_constant (size, first)));
+
+ create_input_operand (&ops[0], addr, Pmode);
+ if (maybe_expand_insn (CODE_FOR_check_stack, 1, ops))
+ return;
}
#endif
- /* If we have to generate explicit probes, see if we have a constant
- small number of them to generate. If so, that's the easy case. */
- else if (CONST_INT_P (size)
- && INTVAL (size) < 10 * STACK_CHECK_PROBE_INTERVAL)
+ /* Otherwise we have to generate explicit probes. If we have a constant
+ small number of them to generate, that's the easy case. */
+ else if (CONST_INT_P (size) && INTVAL (size) < 7 * PROBE_INTERVAL)
{
- HOST_WIDE_INT offset;
-
- /* Start probing at FIRST + N * STACK_CHECK_PROBE_INTERVAL
- for values of N from 1 until it exceeds LAST. If only one
- probe is needed, this will not generate any code. Then probe
- at LAST. */
- for (offset = first + STACK_CHECK_PROBE_INTERVAL;
- offset < INTVAL (size);
- offset = offset + STACK_CHECK_PROBE_INTERVAL)
- emit_stack_probe (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
- stack_pointer_rtx,
- GEN_INT (offset)));
-
- emit_stack_probe (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
- stack_pointer_rtx,
- plus_constant (size, first)));
+ HOST_WIDE_INT isize = INTVAL (size), i;
+ rtx addr;
+
+ /* Probe at FIRST + N * PROBE_INTERVAL for values of N from 1 until
+ it exceeds SIZE. If only one probe is needed, this will not
+ generate any code. Then probe at FIRST + SIZE. */
+ for (i = PROBE_INTERVAL; i < isize; i += PROBE_INTERVAL)
+ {
+ addr = memory_address (Pmode,
+ plus_constant (stack_pointer_rtx,
+ STACK_GROW_OFF (first + i)));
+ emit_stack_probe (addr);
+ }
+
+ addr = memory_address (Pmode,
+ plus_constant (stack_pointer_rtx,
+ STACK_GROW_OFF (first + isize)));
+ emit_stack_probe (addr);
}
- /* In the variable case, do the same as above, but in a loop. We emit loop
- notes so that loop optimization can be done. */
+ /* In the variable case, do the same as above, but in a loop. Note that we
+ must be extra careful with variables wrapping around because we might be
+ at the very top (or the very bottom) of the address space and we have to
+ be able to handle this case properly; in particular, we use an equality
+ test for the loop condition. */
else
{
- rtx test_addr
- = force_operand (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
- stack_pointer_rtx,
- GEN_INT (first + STACK_CHECK_PROBE_INTERVAL)),
- NULL_RTX);
- rtx last_addr
- = force_operand (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
- stack_pointer_rtx,
- plus_constant (size, first)),
- NULL_RTX);
- rtx incr = GEN_INT (STACK_CHECK_PROBE_INTERVAL);
+ rtx rounded_size, rounded_size_op, test_addr, last_addr, temp;
rtx loop_lab = gen_label_rtx ();
- rtx test_lab = gen_label_rtx ();
rtx end_lab = gen_label_rtx ();
- rtx temp;
- if (!REG_P (test_addr)
- || REGNO (test_addr) < FIRST_PSEUDO_REGISTER)
- test_addr = force_reg (Pmode, test_addr);
- emit_jump (test_lab);
+ /* Step 1: round SIZE to the previous multiple of the interval. */
+
+ /* ROUNDED_SIZE = SIZE & -PROBE_INTERVAL */
+ rounded_size
+ = simplify_gen_binary (AND, Pmode, size, GEN_INT (-PROBE_INTERVAL));
+ rounded_size_op = force_operand (rounded_size, NULL_RTX);
+
+
+ /* Step 2: compute initial and final value of the loop counter. */
+
+ /* TEST_ADDR = SP + FIRST. */
+ test_addr = force_operand (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
+ stack_pointer_rtx,
+ GEN_INT (first)), NULL_RTX);
+
+ /* LAST_ADDR = SP + FIRST + ROUNDED_SIZE. */
+ last_addr = force_operand (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
+ test_addr,
+ rounded_size_op), NULL_RTX);
+
+
+ /* Step 3: the loop
+
+ while (TEST_ADDR != LAST_ADDR)
+ {
+ TEST_ADDR = TEST_ADDR + PROBE_INTERVAL
+ probe at TEST_ADDR
+ }
+
+ probes at FIRST + N * PROBE_INTERVAL for values of N from 1
+ until it is equal to ROUNDED_SIZE. */
emit_label (loop_lab);
- emit_stack_probe (test_addr);
-#ifdef STACK_GROWS_DOWNWARD
-#define CMP_OPCODE GTU
- temp = expand_binop (Pmode, sub_optab, test_addr, incr, test_addr,
- 1, OPTAB_WIDEN);
-#else
-#define CMP_OPCODE LTU
- temp = expand_binop (Pmode, add_optab, test_addr, incr, test_addr,
+ /* Jump to END_LAB if TEST_ADDR == LAST_ADDR. */
+ emit_cmp_and_jump_insns (test_addr, last_addr, EQ, NULL_RTX, Pmode, 1,
+ end_lab);
+
+ /* TEST_ADDR = TEST_ADDR + PROBE_INTERVAL. */
+ temp = expand_binop (Pmode, STACK_GROW_OPTAB, test_addr,
+ GEN_INT (PROBE_INTERVAL), test_addr,
1, OPTAB_WIDEN);
-#endif
gcc_assert (temp == test_addr);
- emit_label (test_lab);
- emit_cmp_and_jump_insns (test_addr, last_addr, CMP_OPCODE,
- NULL_RTX, Pmode, 1, loop_lab);
- emit_jump (end_lab);
+ /* Probe at TEST_ADDR. */
+ emit_stack_probe (test_addr);
+
+ emit_jump (loop_lab);
+
+ emit_label (end_lab);
+
+
+ /* Step 4: probe at FIRST + SIZE if we cannot assert at compile-time
+ that SIZE is equal to ROUNDED_SIZE. */
+
+ /* TEMP = SIZE - ROUNDED_SIZE. */
+ temp = simplify_gen_binary (MINUS, Pmode, size, rounded_size);
+ if (temp != const0_rtx)
+ {
+ rtx addr;
+
+ if (CONST_INT_P (temp))
+ {
+ /* Use [base + disp} addressing mode if supported. */
+ HOST_WIDE_INT offset = INTVAL (temp);
+ addr = memory_address (Pmode,
+ plus_constant (last_addr,
+ STACK_GROW_OFF (offset)));
+ }
+ else
+ {
+ /* Manual CSE if the difference is not known at compile-time. */
+ temp = gen_rtx_MINUS (Pmode, size, rounded_size_op);
+ addr = memory_address (Pmode,
+ gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
+ last_addr, temp));
+ }
+
+ emit_stack_probe (addr);
+ }
+ }
+}
+
+/* Adjust the stack pointer by minus SIZE (an rtx for a number of bytes)
+ while probing it. This pushes when SIZE is positive. SIZE need not
+ be constant. If ADJUST_BACK is true, adjust back the stack pointer
+ by plus SIZE at the end. */
+
+void
+anti_adjust_stack_and_probe (rtx size, bool adjust_back)
+{
+ /* We skip the probe for the first interval + a small dope of 4 words and
+ probe that many bytes past the specified size to maintain a protection
+ area at the botton of the stack. */
+ const int dope = 4 * UNITS_PER_WORD;
+
+ /* First ensure SIZE is Pmode. */
+ if (GET_MODE (size) != VOIDmode && GET_MODE (size) != Pmode)
+ size = convert_to_mode (Pmode, size, 1);
+
+ /* If we have a constant small number of probes to generate, that's the
+ easy case. */
+ if (CONST_INT_P (size) && INTVAL (size) < 7 * PROBE_INTERVAL)
+ {
+ HOST_WIDE_INT isize = INTVAL (size), i;
+ bool first_probe = true;
+
+ /* Adjust SP and probe at PROBE_INTERVAL + N * PROBE_INTERVAL for
+ values of N from 1 until it exceeds SIZE. If only one probe is
+ needed, this will not generate any code. Then adjust and probe
+ to PROBE_INTERVAL + SIZE. */
+ for (i = PROBE_INTERVAL; i < isize; i += PROBE_INTERVAL)
+ {
+ if (first_probe)
+ {
+ anti_adjust_stack (GEN_INT (2 * PROBE_INTERVAL + dope));
+ first_probe = false;
+ }
+ else
+ anti_adjust_stack (GEN_INT (PROBE_INTERVAL));
+ emit_stack_probe (stack_pointer_rtx);
+ }
+
+ if (first_probe)
+ anti_adjust_stack (plus_constant (size, PROBE_INTERVAL + dope));
+ else
+ anti_adjust_stack (plus_constant (size, PROBE_INTERVAL - i));
+ emit_stack_probe (stack_pointer_rtx);
+ }
+
+ /* In the variable case, do the same as above, but in a loop. Note that we
+ must be extra careful with variables wrapping around because we might be
+ at the very top (or the very bottom) of the address space and we have to
+ be able to handle this case properly; in particular, we use an equality
+ test for the loop condition. */
+ else
+ {
+ rtx rounded_size, rounded_size_op, last_addr, temp;
+ rtx loop_lab = gen_label_rtx ();
+ rtx end_lab = gen_label_rtx ();
+
+
+ /* Step 1: round SIZE to the previous multiple of the interval. */
+
+ /* ROUNDED_SIZE = SIZE & -PROBE_INTERVAL */
+ rounded_size
+ = simplify_gen_binary (AND, Pmode, size, GEN_INT (-PROBE_INTERVAL));
+ rounded_size_op = force_operand (rounded_size, NULL_RTX);
+
+
+ /* Step 2: compute initial and final value of the loop counter. */
+
+ /* SP = SP_0 + PROBE_INTERVAL. */
+ anti_adjust_stack (GEN_INT (PROBE_INTERVAL + dope));
+
+ /* LAST_ADDR = SP_0 + PROBE_INTERVAL + ROUNDED_SIZE. */
+ last_addr = force_operand (gen_rtx_fmt_ee (STACK_GROW_OP, Pmode,
+ stack_pointer_rtx,
+ rounded_size_op), NULL_RTX);
+
+
+ /* Step 3: the loop
+
+ while (SP != LAST_ADDR)
+ {
+ SP = SP + PROBE_INTERVAL
+ probe at SP
+ }
+
+ adjusts SP and probes at PROBE_INTERVAL + N * PROBE_INTERVAL for
+ values of N from 1 until it is equal to ROUNDED_SIZE. */
+
+ emit_label (loop_lab);
+
+ /* Jump to END_LAB if SP == LAST_ADDR. */
+ emit_cmp_and_jump_insns (stack_pointer_rtx, last_addr, EQ, NULL_RTX,
+ Pmode, 1, end_lab);
+
+ /* SP = SP + PROBE_INTERVAL and probe at SP. */
+ anti_adjust_stack (GEN_INT (PROBE_INTERVAL));
+ emit_stack_probe (stack_pointer_rtx);
+
+ emit_jump (loop_lab);
+
emit_label (end_lab);
- emit_stack_probe (last_addr);
+
+ /* Step 4: adjust SP and probe at PROBE_INTERVAL + SIZE if we cannot
+ assert at compile-time that SIZE is equal to ROUNDED_SIZE. */
+
+ /* TEMP = SIZE - ROUNDED_SIZE. */
+ temp = simplify_gen_binary (MINUS, Pmode, size, rounded_size);
+ if (temp != const0_rtx)
+ {
+ /* Manual CSE if the difference is not known at compile-time. */
+ if (GET_CODE (temp) != CONST_INT)
+ temp = gen_rtx_MINUS (Pmode, size, rounded_size_op);
+ anti_adjust_stack (temp);
+ emit_stack_probe (stack_pointer_rtx);
+ }
}
+
+ /* Adjust back and account for the additional first interval. */
+ if (adjust_back)
+ adjust_stack (plus_constant (size, PROBE_INTERVAL + dope));
+ else
+ adjust_stack (GEN_INT (PROBE_INTERVAL + dope));
}
-\f
+
/* Return an rtx representing the register or memory location
in which a scalar value of data type VALTYPE
was returned by a function call to function FUNC.
OSDN Git Service
