blob: 6dd6f73e978040f78931f90ca153bff10b55bdbd [file] [log] [blame]
/* Convert function calls to rtl insns, for GNU C compiler.
Copyright (C) 1989-2022 Free Software Foundation, Inc.
This file is part of GCC.
GCC is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation; either version 3, or (at your option) any later
GCC is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or
for more details.
You should have received a copy of the GNU General Public License
along with GCC; see the file COPYING3. If not see
<>. */
#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "backend.h"
#include "target.h"
#include "rtl.h"
#include "tree.h"
#include "gimple.h"
#include "predict.h"
#include "memmodel.h"
#include "tm_p.h"
#include "stringpool.h"
#include "expmed.h"
#include "optabs.h"
#include "emit-rtl.h"
#include "cgraph.h"
#include "diagnostic-core.h"
#include "fold-const.h"
#include "stor-layout.h"
#include "varasm.h"
#include "internal-fn.h"
#include "dojump.h"
#include "explow.h"
#include "calls.h"
#include "expr.h"
#include "output.h"
#include "langhooks.h"
#include "except.h"
#include "dbgcnt.h"
#include "rtl-iter.h"
#include "tree-vrp.h"
#include "tree-ssanames.h"
#include "intl.h"
#include "stringpool.h"
#include "hash-map.h"
#include "hash-traits.h"
#include "attribs.h"
#include "builtins.h"
#include "gimple-iterator.h"
#include "gimple-fold.h"
#include "attr-fnspec.h"
#include "value-query.h"
#include "tree-pretty-print.h"
#include "tree-eh.h"
/* Like PREFERRED_STACK_BOUNDARY but in units of bytes, not bits. */
/* Data structure and subroutines used within expand_call. */
struct arg_data
/* Tree node for this argument. */
tree tree_value;
/* Mode for value; TYPE_MODE unless promoted. */
machine_mode mode;
/* Current RTL value for argument, or 0 if it isn't precomputed. */
rtx value;
/* Initially-compute RTL value for argument; only for const functions. */
rtx initial_value;
/* Register to pass this argument in, 0 if passed on stack, or an
PARALLEL if the arg is to be copied into multiple non-contiguous
registers. */
rtx reg;
/* Register to pass this argument in when generating tail call sequence.
This is not the same register as for normal calls on machines with
register windows. */
rtx tail_call_reg;
/* If REG is a PARALLEL, this is a copy of VALUE pulled into the correct
form for emit_group_move. */
rtx parallel_value;
/* If REG was promoted from the actual mode of the argument expression,
indicates whether the promotion is sign- or zero-extended. */
int unsignedp;
/* Number of bytes to put in registers. 0 means put the whole arg
in registers. Also 0 if not passed in registers. */
int partial;
/* Nonzero if argument must be passed on stack.
Note that some arguments may be passed on the stack
even though pass_on_stack is zero, just because FUNCTION_ARG says so.
pass_on_stack identifies arguments that *cannot* go in registers. */
int pass_on_stack;
/* Some fields packaged up for locate_and_pad_parm. */
struct locate_and_pad_arg_data locate;
/* Location on the stack at which parameter should be stored. The store
has already been done if STACK == VALUE. */
rtx stack;
/* Location on the stack of the start of this argument slot. This can
differ from STACK if this arg pads downward. This location is known
rtx stack_slot;
/* Place that this stack area has been saved, if needed. */
rtx save_area;
/* If an argument's alignment does not permit direct copying into registers,
copy in smaller-sized pieces into pseudos. These are stored in a
block pointed to by this field. The next field says how many
word-sized pseudos we made. */
rtx *aligned_regs;
int n_aligned_regs;
/* A vector of one char per byte of stack space. A byte if nonzero if
the corresponding stack location has been used.
This vector is used to prevent a function call within an argument from
clobbering any stack already set up. */
static char *stack_usage_map;
/* Size of STACK_USAGE_MAP. */
static unsigned int highest_outgoing_arg_in_use;
/* Assume that any stack location at this byte index is used,
without checking the contents of stack_usage_map. */
static unsigned HOST_WIDE_INT stack_usage_watermark = HOST_WIDE_INT_M1U;
/* A bitmap of virtual-incoming stack space. Bit is set if the corresponding
stack location's tail call argument has been already stored into the stack.
This bitmap is used to prevent sibling call optimization if function tries
to use parent's incoming argument slots when they have been already
overwritten with tail call arguments. */
static sbitmap stored_args_map;
/* Assume that any virtual-incoming location at this byte index has been
stored, without checking the contents of stored_args_map. */
static unsigned HOST_WIDE_INT stored_args_watermark;
/* stack_arg_under_construction is nonzero when an argument may be
initialized with a constructor call (including a C function that
returns a BLKmode struct) and expand_call must take special action
to make sure the object being constructed does not overlap the
argument list for the constructor call. */
static int stack_arg_under_construction;
static void precompute_register_parameters (int, struct arg_data *, int *);
static int store_one_arg (struct arg_data *, rtx, int, int, int);
static void store_unaligned_arguments_into_pseudos (struct arg_data *, int);
static int finalize_must_preallocate (int, int, struct arg_data *,
struct args_size *);
static void precompute_arguments (int, struct arg_data *);
static void compute_argument_addresses (struct arg_data *, rtx, int);
static rtx rtx_for_function_call (tree, tree);
static void load_register_parameters (struct arg_data *, int, rtx *, int,
int, int *);
static int special_function_p (const_tree, int);
static int check_sibcall_argument_overlap_1 (rtx);
static int check_sibcall_argument_overlap (rtx_insn *, struct arg_data *, int);
static tree split_complex_types (tree);
static rtx save_fixed_argument_area (int, rtx, int *, int *);
static void restore_fixed_argument_area (rtx, rtx, int, int);
/* Return true if bytes [LOWER_BOUND, UPPER_BOUND) of the outgoing
stack region might already be in use. */
static bool
stack_region_maybe_used_p (poly_uint64 lower_bound, poly_uint64 upper_bound,
unsigned int reg_parm_stack_space)
unsigned HOST_WIDE_INT const_lower, const_upper;
const_lower = constant_lower_bound (lower_bound);
if (!upper_bound.is_constant (&const_upper))
const_upper = HOST_WIDE_INT_M1U;
if (const_upper > stack_usage_watermark)
return true;
/* Don't worry about things in the fixed argument area;
it has already been saved. */
const_lower = MAX (const_lower, reg_parm_stack_space);
const_upper = MIN (const_upper, highest_outgoing_arg_in_use);
for (unsigned HOST_WIDE_INT i = const_lower; i < const_upper; ++i)
if (stack_usage_map[i])
return true;
return false;
/* Record that bytes [LOWER_BOUND, UPPER_BOUND) of the outgoing
stack region are now in use. */
static void
mark_stack_region_used (poly_uint64 lower_bound, poly_uint64 upper_bound)
unsigned HOST_WIDE_INT const_lower, const_upper;
const_lower = constant_lower_bound (lower_bound);
if (upper_bound.is_constant (&const_upper)
&& const_upper <= highest_outgoing_arg_in_use)
for (unsigned HOST_WIDE_INT i = const_lower; i < const_upper; ++i)
stack_usage_map[i] = 1;
stack_usage_watermark = MIN (stack_usage_watermark, const_lower);
/* Force FUNEXP into a form suitable for the address of a CALL,
and return that as an rtx. Also load the static chain register
if FNDECL is a nested function.
CALL_FUSAGE points to a variable holding the prospective
prepare_call_address (tree fndecl_or_type, rtx funexp, rtx static_chain_value,
rtx *call_fusage, int reg_parm_seen, int flags)
/* Make a valid memory address and copy constants through pseudo-regs,
but not for a constant address if -fno-function-cse. */
if (GET_CODE (funexp) != SYMBOL_REF)
/* If it's an indirect call by descriptor, generate code to perform
runtime identification of the pointer and load the descriptor. */
if ((flags & ECF_BY_DESCRIPTOR) && !flag_trampolines)
const int bit_val = targetm.calls.custom_function_descriptors;
rtx call_lab = gen_label_rtx ();
gcc_assert (fndecl_or_type && TYPE_P (fndecl_or_type));
DECL_STATIC_CHAIN (fndecl_or_type) = 1;
rtx chain = targetm.calls.static_chain (fndecl_or_type, false);
if (GET_MODE (funexp) != Pmode)
funexp = convert_memory_address (Pmode, funexp);
/* Avoid long live ranges around function calls. */
funexp = copy_to_mode_reg (Pmode, funexp);
if (REG_P (chain))
emit_insn (gen_rtx_CLOBBER (VOIDmode, chain));
/* Emit the runtime identification pattern. */
rtx mask = gen_rtx_AND (Pmode, funexp, GEN_INT (bit_val));
emit_cmp_and_jump_insns (mask, const0_rtx, EQ, NULL_RTX, Pmode, 1,
/* Statically predict the branch to very likely taken. */
rtx_insn *insn = get_last_insn ();
if (JUMP_P (insn))
predict_insn_def (insn, PRED_BUILTIN_EXPECT, TAKEN);
/* Load the descriptor. */
rtx mem = gen_rtx_MEM (ptr_mode,
plus_constant (Pmode, funexp, - bit_val));
MEM_NOTRAP_P (mem) = 1;
mem = convert_memory_address (Pmode, mem);
emit_move_insn (chain, mem);
mem = gen_rtx_MEM (ptr_mode,
plus_constant (Pmode, funexp,
- bit_val));
MEM_NOTRAP_P (mem) = 1;
mem = convert_memory_address (Pmode, mem);
emit_move_insn (funexp, mem);
emit_label (call_lab);
if (REG_P (chain))
use_reg (call_fusage, chain);
STATIC_CHAIN_REG_P (chain) = 1;
/* Make sure we're not going to be overwritten below. */
gcc_assert (!static_chain_value);
/* If we are using registers for parameters, force the
function address into a register now. */
funexp = ((reg_parm_seen
&& targetm.small_register_classes_for_mode_p (FUNCTION_MODE))
? force_not_mem (memory_address (FUNCTION_MODE, funexp))
: memory_address (FUNCTION_MODE, funexp));
/* funexp could be a SYMBOL_REF represents a function pointer which is
of ptr_mode. In this case, it should be converted into address mode
to be a valid address for memory rtx pattern. See PR 64971. */
if (GET_MODE (funexp) != Pmode)
funexp = convert_memory_address (Pmode, funexp);
if (!(flags & ECF_SIBCALL))
if (!NO_FUNCTION_CSE && optimize && ! flag_no_function_cse)
funexp = force_reg (Pmode, funexp);
if (static_chain_value != 0
&& (TREE_CODE (fndecl_or_type) != FUNCTION_DECL
|| DECL_STATIC_CHAIN (fndecl_or_type)))
rtx chain;
chain = targetm.calls.static_chain (fndecl_or_type, false);
static_chain_value = convert_memory_address (Pmode, static_chain_value);
emit_move_insn (chain, static_chain_value);
if (REG_P (chain))
use_reg (call_fusage, chain);
STATIC_CHAIN_REG_P (chain) = 1;
return funexp;
/* Generate instructions to call function FUNEXP,
and optionally pop the results.
The CALL_INSN is the first insn generated.
FNDECL is the declaration node of the function. This is given to the
hook TARGET_RETURN_POPS_ARGS to determine whether this function pops
its own args.
FUNTYPE is the data type of the function. This is given to the hook
TARGET_RETURN_POPS_ARGS to determine whether this function pops its
own args. We used to allow an identifier for library functions, but
that doesn't work when the return type is an aggregate type and the
calling convention says that the pointer to this aggregate is to be
popped by the callee.
STACK_SIZE is the number of bytes of arguments on the stack,
ROUNDED_STACK_SIZE is that number rounded up to
PREFERRED_STACK_BOUNDARY; zero if the size is variable. This is
both to put into the call insn and to generate explicit popping
code if necessary.
STRUCT_VALUE_SIZE is the number of bytes wanted in a structure value.
It is zero if this call doesn't want a structure value.
NEXT_ARG_REG is the rtx that results from executing
targetm.calls.function_arg (&args_so_far,
function_arg_info::end_marker ());
just after all the args have had their registers assigned.
This could be whatever you like, but normally it is the first
arg-register beyond those used for args in this call,
or 0 if all the arg-registers are used in this call.
It is passed on to `gen_call' so you can put this info in the call insn.
VALREG is a hard register in which a value is returned,
or 0 if the call does not return a value.
OLD_INHIBIT_DEFER_POP is the value that `inhibit_defer_pop' had before
the args to this call were processed.
We restore `inhibit_defer_pop' to that value.
CALL_FUSAGE is either empty or an EXPR_LIST of USE expressions that
denote registers used by the called function. */
static void
emit_call_1 (rtx funexp, tree fntree ATTRIBUTE_UNUSED, tree fndecl ATTRIBUTE_UNUSED,
tree funtype ATTRIBUTE_UNUSED,
poly_int64 stack_size ATTRIBUTE_UNUSED,
poly_int64 rounded_stack_size,
poly_int64 struct_value_size ATTRIBUTE_UNUSED,
rtx next_arg_reg ATTRIBUTE_UNUSED, rtx valreg,
int old_inhibit_defer_pop, rtx call_fusage, int ecf_flags,
cumulative_args_t args_so_far ATTRIBUTE_UNUSED)
rtx rounded_stack_size_rtx = gen_int_mode (rounded_stack_size, Pmode);
rtx call, funmem, pat;
int already_popped = 0;
poly_int64 n_popped = 0;
/* Sibling call patterns never pop arguments (no sibcall(_value)_pop
patterns exist). Any popping that the callee does on return will
be from our caller's frame rather than ours. */
if (!(ecf_flags & ECF_SIBCALL))
n_popped += targetm.calls.return_pops_args (fndecl, funtype, stack_size);
n_popped += CALL_POPS_ARGS (*get_cumulative_args (args_so_far));
/* Ensure address is valid. SYMBOL_REF is already valid, so no need,
and we don't want to load it into a register as an optimization,
because prepare_call_address already did it if it should be done. */
if (GET_CODE (funexp) != SYMBOL_REF)
funexp = memory_address (FUNCTION_MODE, funexp);
funmem = gen_rtx_MEM (FUNCTION_MODE, funexp);
if (fndecl && TREE_CODE (fndecl) == FUNCTION_DECL)
tree t = fndecl;
/* Although a built-in FUNCTION_DECL and its non-__builtin
counterpart compare equal and get a shared mem_attrs, they
produce different dump output in compare-debug compilations,
if an entry gets garbage collected in one compilation, then
adds a different (but equivalent) entry, while the other
doesn't run the garbage collector at the same spot and then
shares the mem_attr with the equivalent entry. */
tree t2 = builtin_decl_explicit (DECL_FUNCTION_CODE (t));
if (t2)
t = t2;
set_mem_expr (funmem, t);
else if (fntree)
set_mem_expr (funmem, build_simple_mem_ref (CALL_EXPR_FN (fntree)));
if (ecf_flags & ECF_SIBCALL)
if (valreg)
pat = targetm.gen_sibcall_value (valreg, funmem,
next_arg_reg, NULL_RTX);
pat = targetm.gen_sibcall (funmem, rounded_stack_size_rtx,
gen_int_mode (struct_value_size, Pmode));
/* If the target has "call" or "call_value" insns, then prefer them
if no arguments are actually popped. If the target does not have
"call" or "call_value" insns, then we must use the popping versions
even if the call has no arguments to pop. */
else if (maybe_ne (n_popped, 0)
|| !(valreg
? targetm.have_call_value ()
: targetm.have_call ()))
rtx n_pop = gen_int_mode (n_popped, Pmode);
/* If this subroutine pops its own args, record that in the call insn
if possible, for the sake of frame pointer elimination. */
if (valreg)
pat = targetm.gen_call_value_pop (valreg, funmem,
next_arg_reg, n_pop);
pat = targetm.gen_call_pop (funmem, rounded_stack_size_rtx,
next_arg_reg, n_pop);
already_popped = 1;
if (valreg)
pat = targetm.gen_call_value (valreg, funmem, rounded_stack_size_rtx,
next_arg_reg, NULL_RTX);
pat = targetm.gen_call (funmem, rounded_stack_size_rtx, next_arg_reg,
gen_int_mode (struct_value_size, Pmode));
emit_insn (pat);
/* Find the call we just emitted. */
rtx_call_insn *call_insn = last_call_insn ();
/* Some target create a fresh MEM instead of reusing the one provided
above. Set its MEM_EXPR. */
call = get_call_rtx_from (call_insn);
if (call
&& MEM_EXPR (XEXP (call, 0)) == NULL_TREE
&& MEM_EXPR (funmem) != NULL_TREE)
set_mem_expr (XEXP (call, 0), MEM_EXPR (funmem));
/* Put the register usage information there. */
add_function_usage_to (call_insn, call_fusage);
/* If this is a const call, then set the insn's unchanging bit. */
if (ecf_flags & ECF_CONST)
RTL_CONST_CALL_P (call_insn) = 1;
/* If this is a pure call, then set the insn's unchanging bit. */
if (ecf_flags & ECF_PURE)
RTL_PURE_CALL_P (call_insn) = 1;
/* If this is a const call, then set the insn's unchanging bit. */
if (ecf_flags & ECF_LOOPING_CONST_OR_PURE)
/* Create a nothrow REG_EH_REGION note, if needed. */
make_reg_eh_region_note (call_insn, ecf_flags, 0);
if (ecf_flags & ECF_NORETURN)
add_reg_note (call_insn, REG_NORETURN, const0_rtx);
if (ecf_flags & ECF_RETURNS_TWICE)
add_reg_note (call_insn, REG_SETJMP, const0_rtx);
cfun->calls_setjmp = 1;
SIBLING_CALL_P (call_insn) = ((ecf_flags & ECF_SIBCALL) != 0);
/* Restore this now, so that we do defer pops for this call's args
if the context of the call as a whole permits. */
inhibit_defer_pop = old_inhibit_defer_pop;
if (maybe_ne (n_popped, 0))
if (!already_popped)
= gen_rtx_EXPR_LIST (VOIDmode,
gen_rtx_CLOBBER (VOIDmode, stack_pointer_rtx),
rounded_stack_size -= n_popped;
rounded_stack_size_rtx = gen_int_mode (rounded_stack_size, Pmode);
stack_pointer_delta -= n_popped;
add_args_size_note (call_insn, stack_pointer_delta);
/* If popup is needed, stack realign must use DRAP */
crtl->need_drap = true;
/* For noreturn calls when not accumulating outgoing args force
REG_ARGS_SIZE note to prevent crossjumping of calls with different
args sizes. */
else if (!ACCUMULATE_OUTGOING_ARGS && (ecf_flags & ECF_NORETURN) != 0)
add_args_size_note (call_insn, stack_pointer_delta);
/* If returning from the subroutine does not automatically pop the args,
we need an instruction to pop them sooner or later.
Perhaps do it now; perhaps just record how much space to pop later.
If returning from the subroutine does pop the args, indicate that the
stack pointer will be changed. */
if (maybe_ne (rounded_stack_size, 0))
if (ecf_flags & ECF_NORETURN)
/* Just pretend we did the pop. */
stack_pointer_delta -= rounded_stack_size;
else if (flag_defer_pop && inhibit_defer_pop == 0
&& ! (ecf_flags & (ECF_CONST | ECF_PURE)))
pending_stack_adjust += rounded_stack_size;
adjust_stack (rounded_stack_size_rtx);
/* When we accumulate outgoing args, we must avoid any stack manipulations.
Restore the stack pointer to its original value now. Usually
ACCUMULATE_OUTGOING_ARGS targets don't get here, but there are exceptions.
On i386 ACCUMULATE_OUTGOING_ARGS can be enabled on demand, and
popping variants of functions exist as well.
??? We may optimize similar to defer_pop above, but it is
probably not worthwhile.
??? It will be worthwhile to enable combine_stack_adjustments even for
such machines. */
else if (maybe_ne (n_popped, 0))
anti_adjust_stack (gen_int_mode (n_popped, Pmode));
/* Determine if the function identified by FNDECL is one with
special properties we wish to know about. Modify FLAGS accordingly.
For example, if the function might return more than one time (setjmp), then
Set ECF_MAY_BE_ALLOCA for any memory allocation function that might allocate
space from the stack such as alloca. */
static int
special_function_p (const_tree fndecl, int flags)
tree name_decl = DECL_NAME (fndecl);
if (maybe_special_function_p (fndecl)
&& IDENTIFIER_LENGTH (name_decl) <= 11)
const char *name = IDENTIFIER_POINTER (name_decl);
const char *tname = name;
/* We assume that alloca will always be called by name. It
makes no sense to pass it as a pointer-to-function to
anything that does not understand its behavior. */
if (IDENTIFIER_LENGTH (name_decl) == 6
&& name[0] == 'a'
&& ! strcmp (name, "alloca"))
/* Disregard prefix _ or __. */
if (name[0] == '_')
if (name[1] == '_')
tname += 2;
tname += 1;
/* ECF_RETURNS_TWICE is safe even for -ffreestanding. */
if (! strcmp (tname, "setjmp")
|| ! strcmp (tname, "sigsetjmp")
|| ! strcmp (name, "savectx")
|| ! strcmp (name, "vfork")
|| ! strcmp (name, "getcontext"))
return flags;
/* Return fnspec for DECL. */
static attr_fnspec
decl_fnspec (tree fndecl)
tree attr;
tree type = TREE_TYPE (fndecl);
if (type)
attr = lookup_attribute ("fn spec", TYPE_ATTRIBUTES (type));
if (attr)
return TREE_VALUE (TREE_VALUE (attr));
if (fndecl_built_in_p (fndecl, BUILT_IN_NORMAL))
return builtin_fnspec (fndecl);
return "";
/* Similar to special_function_p; return a set of ERF_ flags for the
function FNDECL. */
static int
decl_return_flags (tree fndecl)
attr_fnspec fnspec = decl_fnspec (fndecl);
unsigned int arg;
if (fnspec.returns_arg (&arg))
return ERF_RETURNS_ARG | arg;
if (fnspec.returns_noalias_p ())
return 0;
/* Return nonzero when FNDECL represents a call to setjmp. */
setjmp_call_p (const_tree fndecl)
return special_function_p (fndecl, 0) & ECF_RETURNS_TWICE;
/* Return true if STMT may be an alloca call. */
gimple_maybe_alloca_call_p (const gimple *stmt)
tree fndecl;
if (!is_gimple_call (stmt))
return false;
fndecl = gimple_call_fndecl (stmt);
if (fndecl && (special_function_p (fndecl, 0) & ECF_MAY_BE_ALLOCA))
return true;
return false;
/* Return true if STMT is a builtin alloca call. */
gimple_alloca_call_p (const gimple *stmt)
tree fndecl;
if (!is_gimple_call (stmt))
return false;
fndecl = gimple_call_fndecl (stmt);
if (fndecl && fndecl_built_in_p (fndecl, BUILT_IN_NORMAL))
switch (DECL_FUNCTION_CODE (fndecl))
return gimple_call_num_args (stmt) > 0;
return false;
/* Return true when exp contains a builtin alloca call. */
alloca_call_p (const_tree exp)
tree fndecl;
if (TREE_CODE (exp) == CALL_EXPR
&& (fndecl = get_callee_fndecl (exp))
switch (DECL_FUNCTION_CODE (fndecl))
return true;
return false;
/* Return TRUE if FNDECL is either a TM builtin or a TM cloned
function. Return FALSE otherwise. */
static bool
is_tm_builtin (const_tree fndecl)
if (fndecl == NULL)
return false;
if (decl_is_tm_clone (fndecl))
return true;
switch (DECL_FUNCTION_CODE (fndecl))
case BUILT_IN_TM_LOG_M128:
case BUILT_IN_TM_LOG_M256:
return true;
return false;
/* Detect flags (function attributes) from the function decl or type node. */
flags_from_decl_or_type (const_tree exp)
int flags = 0;
if (DECL_P (exp))
/* The function exp may have the `malloc' attribute. */
if (DECL_IS_MALLOC (exp))
flags |= ECF_MALLOC;
/* The function exp may have the `returns_twice' attribute. */
/* Process the pure and const attributes. */
if (TREE_READONLY (exp))
flags |= ECF_CONST;
if (DECL_PURE_P (exp))
flags |= ECF_PURE;
if (DECL_IS_NOVOPS (exp))
flags |= ECF_NOVOPS;
if (lookup_attribute ("leaf", DECL_ATTRIBUTES (exp)))
flags |= ECF_LEAF;
if (lookup_attribute ("cold", DECL_ATTRIBUTES (exp)))
flags |= ECF_COLD;
if (TREE_NOTHROW (exp))
flags |= ECF_NOTHROW;
if (flag_tm)
if (is_tm_builtin (exp))
flags |= ECF_TM_BUILTIN;
else if ((flags & (ECF_CONST|ECF_NOVOPS)) != 0
|| lookup_attribute ("transaction_pure",
flags |= ECF_TM_PURE;
flags = special_function_p (exp, flags);
else if (TYPE_P (exp))
if (TYPE_READONLY (exp))
flags |= ECF_CONST;
if (flag_tm
&& ((flags & ECF_CONST) != 0
|| lookup_attribute ("transaction_pure", TYPE_ATTRIBUTES (exp))))
flags |= ECF_TM_PURE;
gcc_unreachable ();
flags |= ECF_NORETURN;
if (flags & (ECF_CONST|ECF_PURE))
return flags;
/* Detect flags from a CALL_EXPR. */
call_expr_flags (const_tree t)
int flags;
tree decl = get_callee_fndecl (t);
if (decl)
flags = flags_from_decl_or_type (decl);
else if (CALL_EXPR_FN (t) == NULL_TREE)
flags = internal_fn_flags (CALL_EXPR_IFN (t));
tree type = TREE_TYPE (CALL_EXPR_FN (t));
if (type && TREE_CODE (type) == POINTER_TYPE)
flags = flags_from_decl_or_type (TREE_TYPE (type));
flags = 0;
return flags;
/* Return true if ARG should be passed by invisible reference. */
pass_by_reference (CUMULATIVE_ARGS *ca, function_arg_info arg)
if (tree type = arg.type)
/* If this type contains non-trivial constructors, then it is
forbidden for the middle-end to create any new copies. */
return true;
/* GCC post 3.4 passes *all* variable sized types by reference. */
if (!TYPE_SIZE (type) || !poly_int_tree_p (TYPE_SIZE (type)))
return true;
/* If a record type should be passed the same as its first (and only)
member, use the type and mode of that member. */
arg.type = TREE_TYPE (first_field (type));
arg.mode = TYPE_MODE (arg.type);
return targetm.calls.pass_by_reference (pack_cumulative_args (ca), arg);
/* Return true if TYPE should be passed by reference when passed to
the "..." arguments of a function. */
pass_va_arg_by_reference (tree type)
return pass_by_reference (NULL, function_arg_info (type, /*named=*/false));
/* Decide whether ARG, which occurs in the state described by CA,
should be passed by reference. Return true if so and update
ARG accordingly. */
apply_pass_by_reference_rules (CUMULATIVE_ARGS *ca, function_arg_info &arg)
if (pass_by_reference (ca, arg))
arg.type = build_pointer_type (arg.type);
arg.mode = TYPE_MODE (arg.type);
arg.pass_by_reference = true;
return true;
return false;
/* Return true if ARG, which is passed by reference, should be callee
copied instead of caller copied. */
reference_callee_copied (CUMULATIVE_ARGS *ca, const function_arg_info &arg)
if (arg.type && TREE_ADDRESSABLE (arg.type))
return false;
return targetm.calls.callee_copies (pack_cumulative_args (ca), arg);
/* Precompute all register parameters as described by ARGS, storing values
into fields within the ARGS array.
NUM_ACTUALS indicates the total number elements in the ARGS array.
Set REG_PARM_SEEN if we encounter a register parameter. */
static void
precompute_register_parameters (int num_actuals, struct arg_data *args,
int *reg_parm_seen)
int i;
*reg_parm_seen = 0;
for (i = 0; i < num_actuals; i++)
if (args[i].reg != 0 && ! args[i].pass_on_stack)
*reg_parm_seen = 1;
if (args[i].value == 0)
push_temp_slots ();
args[i].value = expand_normal (args[i].tree_value);
preserve_temp_slots (args[i].value);
pop_temp_slots ();
/* If we are to promote the function arg to a wider mode,
do it now. */
machine_mode old_mode = TYPE_MODE (TREE_TYPE (args[i].tree_value));
/* Some ABIs require scalar floating point modes to be returned
in a wider scalar integer mode. We need to explicitly
reinterpret to an integer mode of the correct precision
before extending to the desired result. */
if (SCALAR_INT_MODE_P (args[i].mode)
&& SCALAR_FLOAT_MODE_P (old_mode)
&& known_gt (GET_MODE_SIZE (args[i].mode),
GET_MODE_SIZE (old_mode)))
args[i].value = convert_float_to_wider_int (args[i].mode, old_mode,
else if (args[i].mode != old_mode)
args[i].value = convert_modes (args[i].mode, old_mode,
args[i].value, args[i].unsignedp);
/* If the value is a non-legitimate constant, force it into a
pseudo now. TLS symbols sometimes need a call to resolve. */
if (CONSTANT_P (args[i].value)
&& (!targetm.legitimate_constant_p (args[i].mode, args[i].value)
|| targetm.precompute_tls_p (args[i].mode, args[i].value)))
args[i].value = force_reg (args[i].mode, args[i].value);
/* If we're going to have to load the value by parts, pull the
parts into pseudos. The part extraction process can involve
non-trivial computation. */
if (GET_CODE (args[i].reg) == PARALLEL)
tree type = TREE_TYPE (args[i].tree_value);
= emit_group_load_into_temps (args[i].reg, args[i].value,
type, int_size_in_bytes (type));
/* If the value is expensive, and we are inside an appropriately
short loop, put the value into a pseudo and then put the pseudo
into the hard reg.
For small register classes, also do this if this call uses
register parameters. This is to avoid reload conflicts while
loading the parameters registers. */
else if ((! (REG_P (args[i].value)
|| (GET_CODE (args[i].value) == SUBREG
&& REG_P (SUBREG_REG (args[i].value)))))
&& args[i].mode != BLKmode
&& (set_src_cost (args[i].value, args[i].mode,
optimize_insn_for_speed_p ())
&& ((*reg_parm_seen
&& targetm.small_register_classes_for_mode_p (args[i].mode))
|| optimize))
args[i].value = copy_to_mode_reg (args[i].mode, args[i].value);
/* The argument list is the property of the called routine and it
may clobber it. If the fixed area has been used for previous
parameters, we must save and restore it. */
static rtx
save_fixed_argument_area (int reg_parm_stack_space, rtx argblock, int *low_to_save, int *high_to_save)
unsigned int low;
unsigned int high;
/* Compute the boundary of the area that needs to be saved, if any. */
high = reg_parm_stack_space;
high += 1;
if (high > highest_outgoing_arg_in_use)
high = highest_outgoing_arg_in_use;
for (low = 0; low < high; low++)
if (stack_usage_map[low] != 0 || low >= stack_usage_watermark)
int num_to_save;
machine_mode save_mode;
int delta;
rtx addr;
rtx stack_area;
rtx save_area;
while (stack_usage_map[--high] == 0)
*low_to_save = low;
*high_to_save = high;
num_to_save = high - low + 1;
/* If we don't have the required alignment, must do this
in BLKmode. */
scalar_int_mode imode;
if (int_mode_for_size (num_to_save * BITS_PER_UNIT, 1).exists (&imode)
&& (low & (MIN (GET_MODE_SIZE (imode),
save_mode = imode;
save_mode = BLKmode;
delta = -high;
delta = low;
addr = plus_constant (Pmode, argblock, delta);
stack_area = gen_rtx_MEM (save_mode, memory_address (save_mode, addr));
set_mem_align (stack_area, PARM_BOUNDARY);
if (save_mode == BLKmode)
save_area = assign_stack_temp (BLKmode, num_to_save);
emit_block_move (validize_mem (save_area), stack_area,
GEN_INT (num_to_save), BLOCK_OP_CALL_PARM);
save_area = gen_reg_rtx (save_mode);
emit_move_insn (save_area, stack_area);
return save_area;
return NULL_RTX;
static void
restore_fixed_argument_area (rtx save_area, rtx argblock, int high_to_save, int low_to_save)
machine_mode save_mode = GET_MODE (save_area);
int delta;
rtx addr, stack_area;
delta = -high_to_save;
delta = low_to_save;
addr = plus_constant (Pmode, argblock, delta);
stack_area = gen_rtx_MEM (save_mode, memory_address (save_mode, addr));
set_mem_align (stack_area, PARM_BOUNDARY);
if (save_mode != BLKmode)
emit_move_insn (stack_area, save_area);
emit_block_move (stack_area, validize_mem (save_area),
GEN_INT (high_to_save - low_to_save + 1),
/* If any elements in ARGS refer to parameters that are to be passed in
registers, but not in memory, and whose alignment does not permit a
direct copy into registers. Copy the values into a group of pseudos
which we will later copy into the appropriate hard registers.
Pseudos for each unaligned argument will be stored into the array
args[argnum].aligned_regs. The caller is responsible for deallocating
the aligned_regs array if it is nonzero. */
static void
store_unaligned_arguments_into_pseudos (struct arg_data *args, int num_actuals)
int i, j;
for (i = 0; i < num_actuals; i++)
if (args[i].reg != 0 && ! args[i].pass_on_stack
&& GET_CODE (args[i].reg) != PARALLEL
&& args[i].mode == BLKmode
&& MEM_P (args[i].value)
&& (MEM_ALIGN (args[i].value)
int bytes = int_size_in_bytes (TREE_TYPE (args[i].tree_value));
int endian_correction = 0;
if (args[i].partial)
gcc_assert (args[i].partial % UNITS_PER_WORD == 0);
args[i].n_aligned_regs = args[i].partial / UNITS_PER_WORD;
= (bytes + UNITS_PER_WORD - 1) / UNITS_PER_WORD;
args[i].aligned_regs = XNEWVEC (rtx, args[i].n_aligned_regs);
/* Structures smaller than a word are normally aligned to the
least significant byte. On a BYTES_BIG_ENDIAN machine,
this means we must skip the empty high order bytes when
calculating the bit offset. */
if (bytes < UNITS_PER_WORD
&& (BLOCK_REG_PADDING (args[i].mode,
TREE_TYPE (args[i].tree_value), 1)
endian_correction = BITS_PER_WORD - bytes * BITS_PER_UNIT;
for (j = 0; j < args[i].n_aligned_regs; j++)
rtx reg = gen_reg_rtx (word_mode);
rtx word = operand_subword_force (args[i].value, j, BLKmode);
int bitsize = MIN (bytes * BITS_PER_UNIT, BITS_PER_WORD);
args[i].aligned_regs[j] = reg;
word = extract_bit_field (word, bitsize, 0, 1, NULL_RTX,
word_mode, word_mode, false, NULL);
/* There is no need to restrict this code to loading items
in TYPE_ALIGN sized hunks. The bitfield instructions can
load up entire word sized registers efficiently.
??? This may not be needed anymore.
We use to emit a clobber here but that doesn't let later
passes optimize the instructions we emit. By storing 0 into
the register later passes know the first AND to zero out the
bitfield being set in the register is unnecessary. The store
of 0 will be deleted as will at least the first AND. */
emit_move_insn (reg, const0_rtx);
bytes -= bitsize / BITS_PER_UNIT;
store_bit_field (reg, bitsize, endian_correction, 0, 0,
word_mode, word, false, false);
/* Issue an error if CALL_EXPR was flagged as requiring
tall-call optimization. */
maybe_complain_about_tail_call (tree call_expr, const char *reason)
gcc_assert (TREE_CODE (call_expr) == CALL_EXPR);
if (!CALL_EXPR_MUST_TAIL_CALL (call_expr))
error_at (EXPR_LOCATION (call_expr), "cannot tail-call: %s", reason);
/* Fill in ARGS_SIZE and ARGS array based on the parameters found in
NUM_ACTUALS is the total number of parameters.
N_NAMED_ARGS is the total number of named arguments.
STRUCT_VALUE_ADDR_VALUE is the implicit argument for a struct return
value, or null.
FNDECL is the tree code for the target of this call (if known)
ARGS_SO_FAR holds state needed by the target to know where to place
the next argument.
REG_PARM_STACK_SPACE is the number of bytes of stack space reserved
for arguments which are passed in registers.
OLD_STACK_LEVEL is a pointer to an rtx which olds the old stack level
and may be modified by this routine.
OLD_PENDING_ADJ, MUST_PREALLOCATE and FLAGS are pointers to integer
flags which may be modified by this routine.
MAY_TAILCALL is cleared if we encounter an invisible pass-by-reference
that requires allocation of stack space.
CALL_FROM_THUNK_P is true if this call is the jump from a thunk to
the thunked-to function. */
static void
initialize_argument_information (int num_actuals ATTRIBUTE_UNUSED,
struct arg_data *args,
struct args_size *args_size,
int n_named_args ATTRIBUTE_UNUSED,
tree exp, tree struct_value_addr_value,
tree fndecl, tree fntype,
cumulative_args_t args_so_far,
int reg_parm_stack_space,
rtx *old_stack_level,
poly_int64_pod *old_pending_adj,
int *must_preallocate, int *ecf_flags,
bool *may_tailcall, bool call_from_thunk_p)
CUMULATIVE_ARGS *args_so_far_pnt = get_cumulative_args (args_so_far);
location_t loc = EXPR_LOCATION (exp);
/* Count arg position in order args appear. */
int argpos;
int i;
args_size->constant = 0;
args_size->var = 0;
/* In this loop, we consider args in the order they are written.
We fill up ARGS from the back. */
i = num_actuals - 1;
int j = i;
call_expr_arg_iterator iter;
tree arg;
if (struct_value_addr_value)
args[j].tree_value = struct_value_addr_value;
argpos = 0;
FOR_EACH_CALL_EXPR_ARG (arg, iter, exp)
tree argtype = TREE_TYPE (arg);
if (targetm.calls.split_complex_arg
&& argtype
&& TREE_CODE (argtype) == COMPLEX_TYPE
&& targetm.calls.split_complex_arg (argtype))
tree subtype = TREE_TYPE (argtype);
args[j].tree_value = build1 (REALPART_EXPR, subtype, arg);
args[j].tree_value = build1 (IMAGPART_EXPR, subtype, arg);
args[j].tree_value = arg;
/* I counts args in order (to be) pushed; ARGPOS counts in order written. */
for (argpos = 0; argpos < num_actuals; i--, argpos++)
tree type = TREE_TYPE (args[i].tree_value);
int unsignedp;
/* Replace erroneous argument with constant zero. */
if (type == error_mark_node || !COMPLETE_TYPE_P (type))
args[i].tree_value = integer_zero_node, type = integer_type_node;
/* If TYPE is a transparent union or record, pass things the way
we would pass the first field of the union or record. We have
already verified that the modes are the same. */
type = TREE_TYPE (first_field (type));
/* Decide where to pass this arg.
args[i].reg is nonzero if all or part is passed in registers.
args[i].partial is nonzero if part but not all is passed in registers,
and the exact value says how many bytes are passed in registers.
args[i].pass_on_stack is nonzero if the argument must at least be
computed on the stack. It may then be loaded back into registers
if args[i].reg is nonzero.
These decisions are driven by the FUNCTION_... macros and must agree
with those made by */
/* See if this argument should be passed by invisible reference. */
function_arg_info arg (type, argpos < n_named_args);
if (pass_by_reference (args_so_far_pnt, arg))
const bool callee_copies
= reference_callee_copied (args_so_far_pnt, arg);
tree base;
/* If we're compiling a thunk, pass directly the address of an object
already in memory, instead of making a copy. Likewise if we want
to make the copy in the callee instead of the caller. */
if ((call_from_thunk_p || callee_copies)
&& TREE_CODE (args[i].tree_value) != WITH_SIZE_EXPR
&& ((base = get_base_address (args[i].tree_value)), true)
&& TREE_CODE (base) != SSA_NAME
&& (!DECL_P (base) || MEM_P (DECL_RTL (base))))
/* We may have turned the parameter value into an SSA name.
Go back to the original parameter so we can take the
address. */
if (TREE_CODE (args[i].tree_value) == SSA_NAME)
gcc_assert (SSA_NAME_IS_DEFAULT_DEF (args[i].tree_value));
args[i].tree_value = SSA_NAME_VAR (args[i].tree_value);
gcc_assert (TREE_CODE (args[i].tree_value) == PARM_DECL);
/* Argument setup code may have copied the value to register. We
revert that optimization now because the tail call code must
use the original location. */
if (TREE_CODE (args[i].tree_value) == PARM_DECL
&& !MEM_P (DECL_RTL (args[i].tree_value))
&& DECL_INCOMING_RTL (args[i].tree_value)
&& MEM_P (DECL_INCOMING_RTL (args[i].tree_value)))
set_decl_rtl (args[i].tree_value,
DECL_INCOMING_RTL (args[i].tree_value));
mark_addressable (args[i].tree_value);
/* We can't use sibcalls if a callee-copied argument is
stored in the current function's frame. */
if (!call_from_thunk_p && DECL_P (base) && !TREE_STATIC (base))
*may_tailcall = false;
maybe_complain_about_tail_call (exp,
"a callee-copied argument is"
" stored in the current"
" function's frame");
args[i].tree_value = build_fold_addr_expr_loc (loc,
type = TREE_TYPE (args[i].tree_value);
if (*ecf_flags & ECF_CONST)
/* We make a copy of the object and pass the address to the
function being called. */
rtx copy;
if (!COMPLETE_TYPE_P (type)
|| (flag_stack_check == GENERIC_STACK_CHECK
&& compare_tree_int (TYPE_SIZE_UNIT (type),
/* This is a variable-sized object. Make space on the stack
for it. */
rtx size_rtx = expr_size (args[i].tree_value);
if (*old_stack_level == 0)
emit_stack_save (SAVE_BLOCK, old_stack_level);
*old_pending_adj = pending_stack_adjust;
pending_stack_adjust = 0;
/* We can pass TRUE as the 4th argument because we just
saved the stack pointer and will restore it right after
the call. */
copy = allocate_dynamic_stack_space (size_rtx,
TYPE_ALIGN (type),
TYPE_ALIGN (type),
copy = gen_rtx_MEM (BLKmode, copy);
set_mem_attributes (copy, type, 1);
copy = assign_temp (type, 1, 0);
store_expr (args[i].tree_value, copy, 0, false, false);
/* Just change the const function to pure and then let
the next test clear the pure based on
callee_copies. */
if (*ecf_flags & ECF_CONST)
*ecf_flags &= ~ECF_CONST;
*ecf_flags |= ECF_PURE;
if (!callee_copies && *ecf_flags & ECF_PURE)
= build_fold_addr_expr_loc (loc, make_tree (type, copy));
type = TREE_TYPE (args[i].tree_value);
*may_tailcall = false;
maybe_complain_about_tail_call (exp,
"argument must be passed"
" by copying");
arg.pass_by_reference = true;
unsignedp = TYPE_UNSIGNED (type);
arg.type = type;
= promote_function_mode (type, TYPE_MODE (type), &unsignedp,
fndecl ? TREE_TYPE (fndecl) : fntype, 0);
args[i].unsignedp = unsignedp;
args[i].mode = arg.mode;
targetm.calls.warn_parameter_passing_abi (args_so_far, type);
args[i].reg = targetm.calls.function_arg (args_so_far, arg);
/* If this is a sibling call and the machine has register windows, the
register window has to be unwinded before calling the routine, so
arguments have to go into the incoming registers. */
if (targetm.calls.function_incoming_arg != targetm.calls.function_arg)
= targetm.calls.function_incoming_arg (args_so_far, arg);
args[i].tail_call_reg = args[i].reg;
if (args[i].reg)
args[i].partial = targetm.calls.arg_partial_bytes (args_so_far, arg);
args[i].pass_on_stack = targetm.calls.must_pass_in_stack (arg);
/* If FUNCTION_ARG returned a (parallel [(expr_list (nil) ...) ...]),
it means that we are to pass this arg in the register(s) designated
by the PARALLEL, but also to pass it in the stack. */
if (args[i].reg && GET_CODE (args[i].reg) == PARALLEL
&& XEXP (XVECEXP (args[i].reg, 0, 0), 0) == 0)
args[i].pass_on_stack = 1;
/* If this is an addressable type, we must preallocate the stack
since we must evaluate the object into its final location.
If this is to be passed in both registers and the stack, it is simpler
to preallocate. */
|| (args[i].pass_on_stack && args[i].reg != 0))
*must_preallocate = 1;
/* Compute the stack-size of this argument. */
if (args[i].reg == 0 || args[i].partial != 0
|| reg_parm_stack_space > 0
|| args[i].pass_on_stack)
locate_and_pad_parm (arg.mode, type,
args[i].reg != 0,
args[i].pass_on_stack ? 0 : args[i].partial,
fndecl, args_size, &args[i].locate);
/* The argument is passed entirely in registers. See at which
end it should be padded. */
args[i].locate.where_pad =
BLOCK_REG_PADDING (arg.mode, type,
int_size_in_bytes (type) <= UNITS_PER_WORD);
/* Update ARGS_SIZE, the total stack space for args so far. */
args_size->constant += args[i].locate.size.constant;
if (args[i].locate.size.var)
ADD_PARM_SIZE (*args_size, args[i].locate.size.var);
/* Increment ARGS_SO_FAR, which has info about which arg-registers
have been used, etc. */
/* ??? Traditionally we've passed TYPE_MODE here, instead of the
promoted_mode used for function_arg above. However, the
corresponding handling of incoming arguments in
does pass the promoted mode. */
arg.mode = TYPE_MODE (type);
targetm.calls.function_arg_advance (args_so_far, arg);
/* Update ARGS_SIZE to contain the total size for the argument block.
Return the original constant component of the argument block's size.
REG_PARM_STACK_SPACE holds the number of bytes of stack space reserved
for arguments passed in registers. */
static poly_int64
compute_argument_block_size (int reg_parm_stack_space,
struct args_size *args_size,
int preferred_stack_boundary ATTRIBUTE_UNUSED)
poly_int64 unadjusted_args_size = args_size->constant;
/* For accumulate outgoing args mode we don't need to align, since the frame
will be already aligned. Align to STACK_BOUNDARY in order to prevent
backends from generating misaligned frame sizes. */
if (ACCUMULATE_OUTGOING_ARGS && preferred_stack_boundary > STACK_BOUNDARY)
preferred_stack_boundary = STACK_BOUNDARY;
/* Compute the actual size of the argument block required. The variable
and constant sizes must be combined, the size may have to be rounded,
and there may be a minimum required size. */
if (args_size->var)
args_size->var = ARGS_SIZE_TREE (*args_size);
args_size->constant = 0;
preferred_stack_boundary /= BITS_PER_UNIT;
if (preferred_stack_boundary > 1)
/* We don't handle this case yet. To handle it correctly we have
to add the delta, round and subtract the delta.
Currently no machine description requires this support. */
gcc_assert (multiple_p (stack_pointer_delta,
args_size->var = round_up (args_size->var, preferred_stack_boundary);
if (reg_parm_stack_space > 0)
= size_binop (MAX_EXPR, args_size->var,
ssize_int (reg_parm_stack_space));
/* The area corresponding to register parameters is not to count in
the size of the block we need. So make the adjustment. */
if (! OUTGOING_REG_PARM_STACK_SPACE ((!fndecl ? fntype : TREE_TYPE (fndecl))))
= size_binop (MINUS_EXPR, args_size->var,
ssize_int (reg_parm_stack_space));
preferred_stack_boundary /= BITS_PER_UNIT;
if (preferred_stack_boundary < 1)
preferred_stack_boundary = 1;
args_size->constant = (aligned_upper_bound (args_size->constant
+ stack_pointer_delta,
- stack_pointer_delta);
args_size->constant = upper_bound (args_size->constant,
if (! OUTGOING_REG_PARM_STACK_SPACE ((!fndecl ? fntype : TREE_TYPE (fndecl))))
args_size->constant -= reg_parm_stack_space;
return unadjusted_args_size;
/* Precompute parameters as needed for a function call.
FLAGS is mask of ECF_* constants.
NUM_ACTUALS is the number of arguments.
ARGS is an array containing information for each argument; this
routine fills in the INITIAL_VALUE and VALUE fields for each
precomputed argument. */
static void
precompute_arguments (int num_actuals, struct arg_data *args)
int i;
/* If this is a libcall, then precompute all arguments so that we do not
get extraneous instructions emitted as part of the libcall sequence. */
/* If we preallocated the stack space, and some arguments must be passed
on the stack, then we must precompute any parameter which contains a
function call which will store arguments on the stack.
Otherwise, evaluating the parameter may clobber previous parameters
which have already been stored into the stack. (we have code to avoid
such case by saving the outgoing stack arguments, but it results in
worse code) */
for (i = 0; i < num_actuals; i++)
tree type;
machine_mode mode;
if (TREE_CODE (args[i].tree_value) != CALL_EXPR)
/* If this is an addressable type, we cannot pre-evaluate it. */
type = TREE_TYPE (args[i].tree_value);
gcc_assert (!TREE_ADDRESSABLE (type));
args[i].initial_value = args[i].value
= expand_normal (args[i].tree_value);
mode = TYPE_MODE (type);
if (mode != args[i].mode)
int unsignedp = args[i].unsignedp;
= convert_modes (args[i].mode, mode,
args[i].value, args[i].unsignedp);
/* CSE will replace this only if it contains args[i].value
pseudo, so convert it down to the declared mode using
a SUBREG. */
if (REG_P (args[i].value)
&& GET_MODE_CLASS (args[i].mode) == MODE_INT
&& promote_mode (type, mode, &unsignedp) != args[i].mode)
= gen_lowpart_SUBREG (mode, args[i].value);
SUBREG_PROMOTED_VAR_P (args[i].initial_value) = 1;
SUBREG_PROMOTED_SET (args[i].initial_value, args[i].unsignedp);
/* Given the current state of MUST_PREALLOCATE and information about
arguments to a function call in NUM_ACTUALS, ARGS and ARGS_SIZE,
compute and return the final value for MUST_PREALLOCATE. */
static int
finalize_must_preallocate (int must_preallocate, int num_actuals,
struct arg_data *args, struct args_size *args_size)
/* See if we have or want to preallocate stack space.
If we would have to push a partially-in-regs parm
before other stack parms, preallocate stack space instead.
If the size of some parm is not a multiple of the required stack
alignment, we must preallocate.
If the total size of arguments that would otherwise create a copy in
a temporary (such as a CALL) is more than half the total argument list
size, preallocation is faster.
Another reason to preallocate is if we have a machine (like the m88k)
where stack alignment is required to be maintained between every
pair of insns, not just when the call is made. However, we assume here
that such machines either do not have push insns (and hence preallocation
would occur anyway) or the problem is taken care of with
if (! must_preallocate)
int partial_seen = 0;
poly_int64 copy_to_evaluate_size = 0;
int i;
for (i = 0; i < num_actuals && ! must_preallocate; i++)
if (args[i].partial > 0 && ! args[i].pass_on_stack)
partial_seen = 1;
else if (partial_seen && args[i].reg == 0)
must_preallocate = 1;
if (TYPE_MODE (TREE_TYPE (args[i].tree_value)) == BLKmode
&& (TREE_CODE (args[i].tree_value) == CALL_EXPR
|| TREE_CODE (args[i].tree_value) == TARGET_EXPR
|| TREE_CODE (args[i].tree_value) == COND_EXPR
|| TREE_ADDRESSABLE (TREE_TYPE (args[i].tree_value))))
+= int_size_in_bytes (TREE_TYPE (args[i].tree_value));
if (maybe_ne (args_size->constant, 0)
&& maybe_ge (copy_to_evaluate_size * 2, args_size->constant))
must_preallocate = 1;
return must_preallocate;
/* If we preallocated stack space, compute the address of each argument
and store it into the ARGS array.
We need not ensure it is a valid memory address here; it will be
validized when it is used.
ARGBLOCK is an rtx for the address of the outgoing arguments. */
static void
compute_argument_addresses (struct arg_data *args, rtx argblock, int num_actuals)
if (argblock)
rtx arg_reg = argblock;
int i;
poly_int64 arg_offset = 0;
if (GET_CODE (argblock) == PLUS)
arg_reg = XEXP (argblock, 0);
arg_offset = rtx_to_poly_int64 (XEXP (argblock, 1));
for (i = 0; i < num_actuals; i++)
rtx offset = ARGS_SIZE_RTX (args[i].locate.offset);
rtx slot_offset = ARGS_SIZE_RTX (args[i].locate.slot_offset);
rtx addr;
unsigned int align, boundary;
poly_uint64 units_on_stack = 0;
machine_mode partial_mode = VOIDmode;
/* Skip this parm if it will not be passed on the stack. */
if (! args[i].pass_on_stack
&& args[i].reg != 0
&& args[i].partial == 0)
if (TYPE_EMPTY_P (TREE_TYPE (args[i].tree_value)))
addr = simplify_gen_binary (PLUS, Pmode, arg_reg, offset);
addr = plus_constant (Pmode, addr, arg_offset);
if (args[i].partial != 0)
/* Only part of the parameter is being passed on the stack.
Generate a simple memory reference of the correct size. */
units_on_stack = args[i].locate.size.constant;
poly_uint64 bits_on_stack = units_on_stack * BITS_PER_UNIT;
partial_mode = int_mode_for_size (bits_on_stack, 1).else_blk ();
args[i].stack = gen_rtx_MEM (partial_mode, addr);
set_mem_size (args[i].stack, units_on_stack);
args[i].stack = gen_rtx_MEM (args[i].mode, addr);
set_mem_attributes (args[i].stack,
TREE_TYPE (args[i].tree_value), 1);
align = BITS_PER_UNIT;
boundary = args[i].locate.boundary;
poly_int64 offset_val;
if (args[i].locate.where_pad != PAD_DOWNWARD)
align = boundary;
else if (poly_int_rtx_p (offset, &offset_val))
align = least_bit_hwi (boundary);
unsigned int offset_align
= known_alignment (offset_val) * BITS_PER_UNIT;
if (offset_align != 0)
align = MIN (align, offset_align);
set_mem_align (args[i].stack, align);
addr = simplify_gen_binary (PLUS, Pmode, arg_reg, slot_offset);
addr = plus_constant (Pmode, addr, arg_offset);
if (args[i].partial != 0)
/* Only part of the parameter is being passed on the stack.
Generate a simple memory reference of the correct size.
args[i].stack_slot = gen_rtx_MEM (partial_mode, addr);
set_mem_size (args[i].stack_slot, units_on_stack);
args[i].stack_slot = gen_rtx_MEM (args[i].mode, addr);
set_mem_attributes (args[i].stack_slot,
TREE_TYPE (args[i].tree_value), 1);
set_mem_align (args[i].stack_slot, args[i].locate.boundary);
/* Function incoming arguments may overlap with sibling call
outgoing arguments and we cannot allow reordering of reads
from function arguments with stores to outgoing arguments
of sibling calls. */
set_mem_alias_set (args[i].stack, 0);
set_mem_alias_set (args[i].stack_slot, 0);
/* Given a FNDECL and EXP, return an rtx suitable for use as a target address
in a call instruction.
FNDECL is the tree node for the target function. For an indirect call
ADDR is the operand 0 of CALL_EXPR for this call. */
static rtx
rtx_for_function_call (tree fndecl, tree addr)
rtx funexp;
/* Get the function to call, in the form of RTL. */
if (fndecl)
if (!TREE_USED (fndecl) && fndecl != current_function_decl)
TREE_USED (fndecl) = 1;
/* Get a SYMBOL_REF rtx for the function address. */
funexp = XEXP (DECL_RTL (fndecl), 0);
/* Generate an rtx (probably a pseudo-register) for the address. */
push_temp_slots ();
funexp = expand_normal (addr);
pop_temp_slots (); /* FUNEXP can't be BLKmode. */
return funexp;
/* Return the static chain for this function, if any. */
rtx_for_static_chain (const_tree fndecl_or_type, bool incoming_p)
if (DECL_P (fndecl_or_type) && !DECL_STATIC_CHAIN (fndecl_or_type))
return NULL;
return targetm.calls.static_chain (fndecl_or_type, incoming_p);
/* Internal state for internal_arg_pointer_based_exp and its helpers. */
static struct
/* Last insn that has been scanned by internal_arg_pointer_based_exp_scan,
or NULL_RTX if none has been scanned yet. */
rtx_insn *scan_start;
/* Vector indexed by REGNO - FIRST_PSEUDO_REGISTER, recording if a pseudo is
based on crtl->args.internal_arg_pointer. The element is NULL_RTX if the
pseudo isn't based on it, a CONST_INT offset if the pseudo is based on it
with fixed offset, or PC if this is with variable or unknown offset. */
vec<rtx> cache;
} internal_arg_pointer_exp_state;
static rtx internal_arg_pointer_based_exp (const_rtx, bool);
/* Helper function for internal_arg_pointer_based_exp. Scan insns in
the tail call sequence, starting with first insn that hasn't been
scanned yet, and note for each pseudo on the LHS whether it is based
on crtl->args.internal_arg_pointer or not, and what offset from that
that pointer it has. */
static void
internal_arg_pointer_based_exp_scan (void)
rtx_insn *insn, *scan_start = internal_arg_pointer_exp_state.scan_start;
if (scan_start == NULL_RTX)
insn = get_insns ();
insn = NEXT_INSN (scan_start);
while (insn)
rtx set = single_set (insn);
if (set && REG_P (SET_DEST (set)) && !HARD_REGISTER_P (SET_DEST (set)))
rtx val = NULL_RTX;
unsigned int idx = REGNO (SET_DEST (set)) - FIRST_PSEUDO_REGISTER;
/* Punt on pseudos set multiple times. */
if (idx < internal_arg_pointer_exp_state.cache.length ()
&& (internal_arg_pointer_exp_state.cache[idx]
val = pc_rtx;
val = internal_arg_pointer_based_exp (SET_SRC (set), false);
if (val != NULL_RTX)
if (idx >= internal_arg_pointer_exp_state.cache.length ())
.safe_grow_cleared (idx + 1, true);
internal_arg_pointer_exp_state.cache[idx] = val;
if (NEXT_INSN (insn) == NULL_RTX)
scan_start = insn;
insn = NEXT_INSN (insn);
internal_arg_pointer_exp_state.scan_start = scan_start;
/* Compute whether RTL is based on crtl->args.internal_arg_pointer. Return
NULL_RTX if RTL isn't based on it, a CONST_INT offset if RTL is based on
it with fixed offset, or PC if this is with variable or unknown offset.
TOPLEVEL is true if the function is invoked at the topmost level. */
static rtx
internal_arg_pointer_based_exp (const_rtx rtl, bool toplevel)
if (CONSTANT_P (rtl))
return NULL_RTX;
if (rtl == crtl->args.internal_arg_pointer)
return const0_rtx;
if (REG_P (rtl) && HARD_REGISTER_P (rtl))
return NULL_RTX;
poly_int64 offset;
if (GET_CODE (rtl) == PLUS && poly_int_rtx_p (XEXP (rtl, 1), &offset))
rtx val = internal_arg_pointer_based_exp (XEXP (rtl, 0), toplevel);
if (val == NULL_RTX || val == pc_rtx)
return val;
return plus_constant (Pmode, val, offset);
/* When called at the topmost level, scan pseudo assignments in between the
last scanned instruction in the tail call sequence and the latest insn
in that sequence. */
if (toplevel)
internal_arg_pointer_based_exp_scan ();
if (REG_P (rtl))
unsigned int idx = REGNO (rtl) - FIRST_PSEUDO_REGISTER;
if (idx < internal_arg_pointer_exp_state.cache.length ())
return internal_arg_pointer_exp_state.cache[idx];
return NULL_RTX;
subrtx_iterator::array_type array;
FOR_EACH_SUBRTX (iter, array, rtl, NONCONST)
const_rtx x = *iter;
if (REG_P (x) && internal_arg_pointer_based_exp (x, false) != NULL_RTX)
return pc_rtx;
if (MEM_P (x))
iter.skip_subrtxes ();
return NULL_RTX;
/* Return true if SIZE bytes starting from address ADDR might overlap an
already-clobbered argument area. This function is used to determine
if we should give up a sibcall. */
static bool
mem_might_overlap_already_clobbered_arg_p (rtx addr, poly_uint64 size)
poly_int64 i;
unsigned HOST_WIDE_INT start, end;
rtx val;
if (bitmap_empty_p (stored_args_map)
&& stored_args_watermark == HOST_WIDE_INT_M1U)
return false;
val = internal_arg_pointer_based_exp (addr, true);
if (val == NULL_RTX)
return false;
else if (!poly_int_rtx_p (val, &i))
return true;
if (known_eq (size, 0U))
return false;
i -= crtl->args.pretend_args_size;
i += crtl->args.pretend_args_size;
i = -i - size;
/* We can ignore any references to the function's pretend args,
which at this point would manifest as negative values of I. */
if (known_le (i, 0) && known_le (size, poly_uint64 (-i)))
return false;
start = maybe_lt (i, 0) ? 0 : constant_lower_bound (i);
if (!(i + size).is_constant (&end))
if (end > stored_args_watermark)
return true;
end = MIN (end, SBITMAP_SIZE (stored_args_map));
for (unsigned HOST_WIDE_INT k = start; k < end; ++k)
if (bitmap_bit_p (stored_args_map, k))
return true;
return false;
/* Do the register loads required for any wholly-register parms or any
parms which are passed both on the stack and in a register. Their
expressions were already evaluated.
Mark all register-parms as living through the call, putting these USE
insns in the CALL_INSN_FUNCTION_USAGE field.
When IS_SIBCALL, perform the check_sibcall_argument_overlap
checking, setting *SIBCALL_FAILURE if appropriate. */
static void
load_register_parameters (struct arg_data *args, int num_actuals,
rtx *call_fusage, int flags, int is_sibcall,
int *sibcall_failure)
int i, j;
for (i = 0; i < num_actuals; i++)
rtx reg = ((flags & ECF_SIBCALL)
? args[i].tail_call_reg : args[i].reg);
if (reg)
int partial = args[i].partial;
int nregs;
poly_int64 size = 0;
HOST_WIDE_INT const_size = 0;
rtx_insn *before_arg = get_last_insn ();
tree tree_value = args[i].tree_value;
tree type = TREE_TYPE (tree_value);
type = TREE_TYPE (first_field (type));
/* Set non-negative if we must move a word at a time, even if
just one word (e.g, partial == 4 && mode == DFmode). Set
to -1 if we just use a normal move insn. This value can be
zero if the argument is a zero size structure. */
nregs = -1;
if (GET_CODE (reg) == PARALLEL)
else if (partial)
gcc_assert (partial % UNITS_PER_WORD == 0);
nregs = partial / UNITS_PER_WORD;
else if (TYPE_MODE (type) == BLKmode)
/* Variable-sized parameters should be described by a
PARALLEL instead. */
const_size = int_size_in_bytes (type);
gcc_assert (const_size >= 0);
nregs = (const_size + (UNITS_PER_WORD - 1)) / UNITS_PER_WORD;
size = const_size;
size = GET_MODE_SIZE (args[i].mode);
/* Handle calls that pass values in multiple non-contiguous
locations. The Irix 6 ABI has examples of this. */
if (GET_CODE (reg) == PARALLEL)
emit_group_move (reg, args[i].parallel_value);
/* If simple case, just do move. If normal partial, store_one_arg
has already loaded the register for us. In all other cases,
load the register(s) from memory. */
else if (nregs == -1)
emit_move_insn (reg, args[i].value);
/* Handle case where we have a value that needs shifting
up to the msb. eg. a QImode value and we're padding
upward on a BYTES_BIG_ENDIAN machine. */
if (args[i].locate.where_pad
gcc_checking_assert (ordered_p (size, UNITS_PER_WORD));
if (maybe_lt (size, UNITS_PER_WORD))
rtx x;
poly_int64 shift
/* Assigning REG here rather than a temp makes
CALL_FUSAGE report the whole reg as used.
Strictly speaking, the call only uses SIZE
bytes at the msb end, but it doesn't seem worth
generating rtl to say that. */
reg = gen_rtx_REG (word_mode, REGNO (reg));
x = expand_shift (LSHIFT_EXPR, word_mode,
reg, shift, reg, 1);
if (x != reg)
emit_move_insn (reg, x);
/* If we have pre-computed the values to put in the registers in
the case of non-aligned structures, copy them in now. */
else if (args[i].n_aligned_regs != 0)
for (j = 0; j < args[i].n_aligned_regs; j++)
emit_move_insn (gen_rtx_REG (word_mode, REGNO (reg) + j),
/* If we need a single register and the source is a constant
VAR_DECL with a simple constructor, expand that constructor
via a pseudo rather than read from (possibly misaligned)
memory. PR middle-end/95126. */
else if (nregs == 1
&& partial == 0
&& !args[i].pass_on_stack
&& VAR_P (tree_value)
&& TREE_READONLY (tree_value)
&& !TREE_SIDE_EFFECTS (tree_value)
&& immediate_const_ctor_p (DECL_INITIAL (tree_value)))
rtx target = gen_reg_rtx (word_mode);
store_constructor (DECL_INITIAL (tree_value), target, 0,
int_expr_size (DECL_INITIAL (tree_value)),
reg = gen_rtx_REG (word_mode, REGNO (reg));
emit_move_insn (reg, target);
else if (partial == 0 || args[i].pass_on_stack)
/* SIZE and CONST_SIZE are 0 for partial arguments and
the size of a BLKmode type otherwise. */
gcc_checking_assert (known_eq (size, const_size));
rtx mem = validize_mem (copy_rtx (args[i].value));
/* Check for overlap with already clobbered argument area,
providing that this has non-zero size. */
if (is_sibcall
&& const_size != 0
&& (mem_might_overlap_already_clobbered_arg_p
(XEXP (args[i].value, 0), const_size)))
*sibcall_failure = 1;
if (const_size % UNITS_PER_WORD == 0
|| MEM_ALIGN (mem) % BITS_PER_WORD == 0)
move_block_to_reg (REGNO (reg), mem, nregs, args[i].mode);
if (nregs > 1)
move_block_to_reg (REGNO (reg), mem, nregs - 1,
rtx dest = gen_rtx_REG (word_mode, REGNO (reg) + nregs - 1);
unsigned int bitoff = (nregs - 1) * BITS_PER_WORD;
unsigned int bitsize = const_size * BITS_PER_UNIT - bitoff;
rtx x = extract_bit_field (mem, bitsize, bitoff, 1, dest,
word_mode, word_mode, false,
x = expand_shift (LSHIFT_EXPR, word_mode, x,
BITS_PER_WORD - bitsize, dest, 1);
if (x != dest)
emit_move_insn (dest, x);
/* Handle a BLKmode that needs shifting. */
if (nregs == 1 && const_size < UNITS_PER_WORD
&& args[i].locate.where_pad == PAD_DOWNWARD
rtx dest = gen_rtx_REG (word_mode, REGNO (reg));
int shift = (UNITS_PER_WORD - const_size) * BITS_PER_UNIT;
enum tree_code dir = (BYTES_BIG_ENDIAN
rtx x;
x = expand_shift (dir, word_mode, dest, shift, dest, 1);
if (x != dest)
emit_move_insn (dest, x);
/* When a parameter is a block, and perhaps in other cases, it is
possible that it did a load from an argument slot that was
already clobbered. */
if (is_sibcall
&& check_sibcall_argument_overlap (before_arg, &args[i], 0))
*sibcall_failure = 1;
/* Handle calls that pass values in multiple non-contiguous
locations. The Irix 6 ABI has examples of this. */
if (GET_CODE (reg) == PARALLEL)
use_group_regs (call_fusage, reg);
else if (nregs == -1)
use_reg_mode (call_fusage, reg, TYPE_MODE (type));
else if (nregs > 0)
use_regs (call_fusage, REGNO (reg), nregs);
/* We need to pop PENDING_STACK_ADJUST bytes. But, if the arguments
wouldn't fill up an even multiple of PREFERRED_UNIT_STACK_BOUNDARY
bytes, then we would need to push some additional bytes to pad the
arguments. So, we try to compute an adjust to the stack pointer for an
amount that will leave the stack under-aligned by UNADJUSTED_ARGS_SIZE
bytes. Then, when the arguments are pushed the stack will be perfectly
Return true if this optimization is possible, storing the adjustment
in ADJUSTMENT_OUT and setting ARGS_SIZE->CONSTANT to the number of
bytes that should be popped after the call. */
static bool
combine_pending_stack_adjustment_and_call (poly_int64_pod *adjustment_out,
poly_int64 unadjusted_args_size,
struct args_size *args_size,
unsigned int preferred_unit_stack_boundary)
/* The number of bytes to pop so that the stack will be
under-aligned by UNADJUSTED_ARGS_SIZE bytes. */
poly_int64 adjustment;
/* The alignment of the stack after the arguments are pushed, if we
just pushed the arguments without adjust the stack here. */
unsigned HOST_WIDE_INT unadjusted_alignment;
if (!known_misalignment (stack_pointer_delta + unadjusted_args_size,
return false;
/* We want to get rid of as many of the PENDING_STACK_ADJUST bytes
as possible -- leaving just enough left to cancel out the
UNADJUSTED_ALIGNMENT. In other words, we want to ensure that the
PENDING_STACK_ADJUST is non-negative, and congruent to
/* Begin by trying to pop all the bytes. */
unsigned HOST_WIDE_INT tmp_misalignment;
if (!known_misalignment (pending_stack_adjust,
return false;
unadjusted_alignment -= tmp_misalignment;
adjustment = pending_stack_adjust;
/* Push enough additional bytes that the stack will be aligned
after the arguments are pushed. */
if (preferred_unit_stack_boundary > 1 && unadjusted_alignment)
adjustment -= preferred_unit_stack_boundary - unadjusted_alignment;
/* We need to know whether the adjusted argument size
(UNADJUSTED_ARGS_SIZE - ADJUSTMENT) constitutes an allocation
or a deallocation. */
if (!ordered_p (adjustment, unadjusted_args_size))
return false;
/* Now, sets ARGS_SIZE->CONSTANT so that we pop the right number of
bytes after the call. The right number is the entire
PENDING_STACK_ADJUST less our ADJUSTMENT plus the amount required
by the arguments in the first place. */
= pending_stack_adjust - adjustment + unadjusted_args_size;
*adjustment_out = adjustment;
return true;
/* Scan X expression if it does not dereference any argument slots
we already clobbered by tail call arguments (as noted in stored_args_map
Return nonzero if X expression dereferences such argument slots,
zero otherwise. */
static int
check_sibcall_argument_overlap_1 (rtx x)
RTX_CODE code;
int i, j;
const char *fmt;
if (x == NULL_RTX)
return 0;
code = GET_CODE (x);
/* We need not check the operands of the CALL expression itself. */
if (code == CALL)
return 0;
if (code == MEM)
return (mem_might_overlap_already_clobbered_arg_p
(XEXP (x, 0), GET_MODE_SIZE (GET_MODE (x))));
/* Scan all subexpressions. */
fmt = GET_RTX_FORMAT (code);
for (i = 0; i < GET_RTX_LENGTH (code); i++, fmt++)
if (*fmt == 'e')
if (check_sibcall_argument_overlap_1 (XEXP (x, i)))
return 1;
else if (*fmt == 'E')
for (j = 0; j < XVECLEN (x, i); j++)
if (check_sibcall_argument_overlap_1 (XVECEXP (x, i, j)))
return 1;
return 0;
/* Scan sequence after INSN if it does not dereference any argument slots
we already clobbered by tail call arguments (as noted in stored_args_map
bitmap). If MARK_STORED_ARGS_MAP, add stack slots for ARG to
stored_args_map bitmap afterwards (when ARG is a register MARK_STORED_ARGS_MAP
should be 0). Return nonzero if sequence after INSN dereferences such argument
slots, zero otherwise. */
static int
check_sibcall_argument_overlap (rtx_insn *insn, struct arg_data *arg,
int mark_stored_args_map)
poly_uint64 low, high;
unsigned HOST_WIDE_INT const_low, const_high;
if (insn == NULL_RTX)
insn = get_insns ();
insn = NEXT_INSN (insn);
for (; insn; insn = NEXT_INSN (insn))
if (INSN_P (insn)
&& check_sibcall_argument_overlap_1 (PATTERN (insn)))
if (mark_stored_args_map)
low = -arg->locate.slot_offset.constant - arg->locate.size.constant;
low = arg->locate.slot_offset.constant;
high = low + arg->locate.size.constant;
const_low = constant_lower_bound (low);
if (high.is_constant (&const_high))
for (unsigned HOST_WIDE_INT i = const_low; i < const_high; ++i)
bitmap_set_bit (stored_args_map, i);
stored_args_watermark = MIN (stored_args_watermark, const_low);
return insn != NULL_RTX;
/* Given that a function returns a value of mode MODE at the most
significant end of hard register VALUE, shift VALUE left or right
as specified by LEFT_P. Return true if some action was needed. */
shift_return_value (machine_mode mode, bool left_p, rtx value)
gcc_assert (REG_P (value) && HARD_REGISTER_P (value));
machine_mode value_mode = GET_MODE (value);
poly_int64 shift = GET_MODE_BITSIZE (value_mode) - GET_MODE_BITSIZE (mode);
if (known_eq (shift, 0))
return false;
/* Use ashr rather than lshr for right shifts. This is for the benefit
of the MIPS port, which requires SImode values to be sign-extended
when stored in 64-bit registers. */
if (!force_expand_binop (value_mode, left_p ? ashl_optab : ashr_optab,
value, gen_int_shift_amount (value_mode, shift),
value, 1, OPTAB_WIDEN))
gcc_unreachable ();
return true;
/* If X is a likely-spilled register value, copy it to a pseudo
register and return that register. Return X otherwise. */
static rtx
avoid_likely_spilled_reg (rtx x)
rtx new_rtx;
if (REG_P (x)
&& targetm.class_likely_spilled_p (REGNO_REG_CLASS (REGNO (x))))
/* Make sure that we generate a REG rather than a CONCAT.
Moves into CONCATs can need nontrivial instructions,
and the whole point of this function is to avoid
using the hard register directly in such a situation. */
generating_concat_p = 0;
new_rtx = gen_reg_rtx (GET_MODE (x));
generating_concat_p = 1;
emit_move_insn (new_rtx, x);
return new_rtx;
return x;
/* Helper function for expand_call.
Return false is EXP is not implementable as a sibling call. */
static bool
can_implement_as_sibling_call_p (tree exp,
rtx structure_value_addr,
tree funtype,
tree fndecl,
int flags,
tree addr,
const args_size &args_size)
if (!targetm.have_sibcall_epilogue ())
"machine description does not have"
" a sibcall_epilogue instruction pattern");
return false;
/* Doing sibling call optimization needs some work, since
structure_value_addr can be allocated on the stack.
It does not seem worth the effort since few optimizable
sibling calls will return a structure. */
if (structure_value_addr != NULL_RTX)
maybe_complain_about_tail_call (exp, "callee returns a structure");
return false;
/* Check whether the target is able to optimize the call
into a sibcall. */
if (!targetm.function_ok_for_sibcall (fndecl, exp))
maybe_complain_about_tail_call (exp,
"target is not able to optimize the"
" call into a sibling call");
return false;
/* Functions that do not return exactly once may not be sibcall
optimized. */
if (flags & ECF_RETURNS_TWICE)
maybe_complain_about_tail_call (exp, "callee returns twice");
return false;
if (flags & ECF_NORETURN)
maybe_complain_about_tail_call (exp, "callee does not return");
return false;
maybe_complain_about_tail_call (exp, "volatile function type");
return false;
/* If the called function is nested in the current one, it might access
some of the caller's arguments, but could clobber them beforehand if
the argument areas are shared. */
if (fndecl && decl_function_context (fndecl) == current_function_decl)
maybe_complain_about_tail_call (exp, "nested function");
return false;
/* If this function requires more stack slots than the current
function, we cannot change it into a sibling call.
crtl->args.pretend_args_size is not part of the
stack allocated by our caller. */
if (maybe_gt (args_size.constant,
crtl->args.size - crtl->args.pretend_args_size))
maybe_complain_about_tail_call (exp,
"callee required more stack slots"
" than the caller");
return false;
/* If the callee pops its own arguments, then it must pop exactly
the same number of arguments as the current function. */
if (maybe_ne (targetm.calls.return_pops_args (fndecl, funtype,
targetm.calls.return_pops_args (current_function_decl,
maybe_complain_about_tail_call (exp,
"inconsistent number of"
" popped arguments");
return false;
if (!lang_hooks.decls.ok_for_sibcall (fndecl))
maybe_complain_about_tail_call (exp, "frontend does not support"
" sibling call");
return false;
/* All checks passed. */
return true;
/* Update stack alignment when the parameter is passed in the stack
since the outgoing parameter requires extra alignment on the calling
function side. */
static void
update_stack_alignment_for_call (struct locate_and_pad_arg_data *locate)
if (crtl->stack_alignment_needed < locate->boundary)
crtl->stack_alignment_needed = locate->boundary;
if (crtl->preferred_stack_boundary < locate->boundary)
crtl->preferred_stack_boundary = locate->boundary;
/* Generate all the code for a CALL_EXPR exp
and return an rtx for its value.
Store the value in TARGET (specified as an rtx) if convenient.
If the value is stored in TARGET then TARGET is returned.
If IGNORE is nonzero, then we ignore the value of the function call. */
expand_call (tree exp, rtx target, int ignore)
/* Nonzero if we are currently expanding a call. */
static int currently_expanding_call = 0;
/* RTX for the function to be called. */
rtx funexp;
/* Sequence of insns to perform a normal "call". */
rtx_insn *normal_call_insns = NULL;
/* Sequence of insns to perform a tail "call". */
rtx_insn *tail_call_insns = NULL;
/* Data type of the function. */
tree funtype;
tree type_arg_types;
tree rettype;
/* Declaration of the function being called,
or 0 if the function is computed (not known by name). */
tree fndecl = 0;
/* The type of the function being called. */
tree fntype;
bool try_tail_call = CALL_EXPR_TAILCALL (exp);
bool must_tail_call = CALL_EXPR_MUST_TAIL_CALL (exp);
int pass;
/* Register in which non-BLKmode value will be returned,
or 0 if no value or if value is BLKmode. */
rtx valreg;
/* Address where we should return a BLKmode value;
0 if value not BLKmode. */
rtx structure_value_addr = 0;
/* Nonzero if that address is being passed by treating it as
an extra, implicit first parameter. Otherwise,
it is passed by being copied directly into struct_value_rtx. */
int structure_value_addr_parm = 0;
/* Holds the value of implicit argument for the struct value. */
tree structure_value_addr_value = NULL_TREE;
/* Size of aggregate value wanted, or zero if none wanted
or if we are using the non-reentrant PCC calling convention
or expecting the value in registers. */
poly_int64 struct_value_size = 0;
/* Nonzero if called function returns an aggregate in memory PCC style,
by returning the address of where to find it. */
int pcc_struct_value = 0;
rtx struct_value = 0;
/* Number of actual parameters in this call, including struct value addr. */
int num_actuals;
/* Number of named args. Args after this are anonymous ones
and they must all go on the stack. */
int n_named_args;
/* Number of complex actual arguments that need to be split. */
int num_complex_actuals = 0;
/* Vector of information about each argument.
Arguments are numbered in the order they will be pushed,
not the order they are written. */
struct arg_data *args;
/* Total size in bytes of all the stack-parms scanned so far. */
struct args_size args_size;
struct args_size adjusted_args_size;
/* Size of arguments before any adjustments (such as rounding). */
poly_int64 unadjusted_args_size;
/* Data on reg parms scanned so far. */
CUMULATIVE_ARGS args_so_far_v;
cumulative_args_t args_so_far;
/* Nonzero if a reg parm has been scanned. */
int reg_parm_seen;
/* Nonzero if this is an indirect function call. */
/* Nonzero if we must avoid push-insns in the args for this call.
If stack space is allocated for register parameters, but not by the
caller, then it is preallocated in the fixed part of the stack frame.
So the entire argument block must then be preallocated (i.e., we
ignore PUSH_ROUNDING in that case). */
int must_preallocate = !targetm.calls.push_argument (0);
/* Size of the stack reserved for parameter registers. */
int reg_parm_stack_space = 0;
/* Address of space preallocated for stack parms
(on machines that lack push insns), or 0 if space not preallocated. */
rtx argblock = 0;
/* Mask of ECF_ and ERF_ flags. */
int flags = 0;
int return_flags = 0;
/* Define the boundary of the register parm stack space that needs to be
saved, if any. */
int low_to_save, high_to_save;
rtx save_area = 0; /* Place that it is saved */
unsigned int initial_highest_arg_in_use = highest_outgoing_arg_in_use;
char *initial_stack_usage_map = stack_usage_map;
unsigned HOST_WIDE_INT initial_stack_usage_watermark = stack_usage_watermark;
char *stack_usage_map_buf = NULL;
poly_int64 old_stack_allocated;
/* State variables to track stack modifications. */
rtx old_stack_level = 0;
int old_stack_arg_under_construction = 0;
poly_int64 old_pending_adj = 0;
int old_inhibit_defer_pop = inhibit_defer_pop;
/* Some stack pointer alterations we make are performed via
allocate_dynamic_stack_space. This modifies the stack_pointer_delta,
which we then also need to save/restore along the way. */
poly_int64 old_stack_pointer_delta = 0;
rtx call_fusage;
tree addr = CALL_EXPR_FN (exp);
int i;
/* The alignment of the stack, in bits. */
unsigned HOST_WIDE_INT preferred_stack_boundary;
/* The alignment of the stack, in bytes. */
unsigned HOST_WIDE_INT preferred_unit_stack_boundary;
/* The static chain value to use for this call. */
rtx static_chain_value;
/* See if this is "nothrow" function call. */
if (TREE_NOTHROW (exp))
flags |= ECF_NOTHROW;
/* See if we can find a DECL-node for the actual function, and get the
function attributes (flags) from the function decl or type node. */
fndecl = get_callee_fndecl (exp);
if (fndecl)
fntype = TREE_TYPE (fndecl);
flags |= flags_from_decl_or_type (fndecl);
return_flags |= decl_return_flags (fndecl);
fntype = TREE_TYPE (TREE_TYPE (addr));
flags |= flags_from_decl_or_type (fntype);
rettype = TREE_TYPE (exp);
struct_value = targetm.calls.struct_value_rtx (fntype, 0);
/* Warn if this value is an aggregate type,
regardless of which calling convention we are using for it. */
if (AGGREGATE_TYPE_P (rettype))
warning (OPT_Waggregate_return, "function call has aggregate value");
/* If the result of a non looping pure or const function call is
ignored (or void), and none of its arguments are volatile, we can
avoid expanding the call and just evaluate the arguments for
side-effects. */
if ((flags & (ECF_CONST | ECF_PURE))
&& (flags & ECF_NOTHROW)
&& (ignore || target == const0_rtx
|| TYPE_MODE (rettype) == VOIDmode))
bool volatilep = false;
tree arg;
call_expr_arg_iterator iter;
FOR_EACH_CALL_EXPR_ARG (arg, iter, exp)
volatilep = true;
if (! volatilep)
FOR_EACH_CALL_EXPR_ARG (arg, iter, exp)
expand_expr (arg, const0_rtx, VOIDmode, EXPAND_NORMAL);
return const0_rtx;
reg_parm_stack_space = REG_PARM_STACK_SPACE (!fndecl ? fntype : fndecl);
if (! OUTGOING_REG_PARM_STACK_SPACE ((!fndecl ? fntype : TREE_TYPE (fndecl)))
&& reg_parm_stack_space > 0 && targetm.calls.push_argument (0))
must_preallocate = 1;
/* Set up a place to return a structure. */
/* Cater to broken compilers. */
if (aggregate_value_p (exp, fntype))
/* This call returns a big structure. */
pcc_struct_value = 1;
if (!poly_int_tree_p (TYPE_SIZE_UNIT (rettype), &struct_value_size))
struct_value_size = -1;
/* Even if it is semantically safe to use the target as the return
slot, it may be not sufficiently aligned for the return type. */
&& target
&& MEM_P (target)
/* If rettype is addressable, we may not create a temporary.
If target is properly aligned at runtime and the compiler
just doesn't know about it, it will work fine, otherwise it
will be UB. */
&& (TREE_ADDRESSABLE (rettype)
|| !(MEM_ALIGN (target) < TYPE_ALIGN (rettype)
&& targetm.slow_unaligned_access (TYPE_MODE (rettype),
MEM_ALIGN (target)))))
structure_value_addr = XEXP (target, 0);
/* For variable-sized objects, we must be called with a target
specified. If we were to allocate space on the stack here,
we would have no way of knowing when to free it. */
rtx d = assign_temp (rettype, 1, 1);
structure_value_addr = XEXP (d, 0);
target = 0;
/* Figure out the amount to which the stack should be aligned. */
preferred_stack_boundary = PREFERRED_STACK_BOUNDARY;
if (fndecl)
struct cgraph_rtl_info *i = cgraph_node::rtl_info (fndecl);
/* Without automatic stack alignment, we can't increase preferred
stack boundary. With automatic stack alignment, it is
unnecessary since unless we can guarantee that all callers will
align the outgoing stack properly, callee has to align its
stack anyway. */
if (i
&& i->preferred_incoming_stack_boundary
&& i->preferred_incoming_stack_boundary < preferred_stack_boundary)
preferred_stack_boundary = i->preferred_incoming_stack_boundary;
/* Operand 0 is a pointer-to-function; get the type of the function. */
funtype = TREE_TYPE (addr);
gcc_assert (POINTER_TYPE_P (funtype));
funtype = TREE_TYPE (funtype);
/* Count whether there are actual complex arguments that need to be split
into their real and imaginary parts. Munge the type_arg_types
appropriately here as well. */
if (targetm.calls.split_complex_arg)
call_expr_arg_iterator iter;
tree arg;
FOR_EACH_CALL_EXPR_ARG (arg, iter, exp)
tree type = TREE_TYPE (arg);
if (type && TREE_CODE (type) == COMPLEX_TYPE
&& targetm.calls.split_complex_arg (type))
type_arg_types = split_complex_types (TYPE_ARG_TYPES (funtype));
type_arg_types = TYPE_ARG_TYPES (funtype);
if (flags & ECF_MAY_BE_ALLOCA)
cfun->calls_alloca = 1;
/* If struct_value_rtx is 0, it means pass the address
as if it were an extra parameter. Put the argument expression
in structure_value_addr_value. */
if (structure_value_addr && struct_value == 0)
/* If structure_value_addr is a REG other than
virtual_outgoing_args_rtx, we can use always use it. If it
is not a REG, we must always copy it into a register.
If it is virtual_outgoing_args_rtx, we must copy it to another
register in some cases. */
rtx temp = (!REG_P (structure_value_addr)
&& stack_arg_under_construction
&& structure_value_addr == virtual_outgoing_args_rtx)
? copy_addr_to_reg (convert_memory_address
(Pmode, structure_value_addr))
: structure_value_addr);
structure_value_addr_value =
make_tree (build_pointer_type (TREE_TYPE (funtype)), temp);
structure_value_addr_parm = 1;
/* Count the arguments and set NUM_ACTUALS. */
num_actuals =
call_expr_nargs (exp) + num_complex_actuals + structure_value_addr_parm;
/* Compute number of named args.
First, do a raw count of the args for INIT_CUMULATIVE_ARGS. */
if (type_arg_types != 0)
= (list_length (type_arg_types)
/* Count the struct value address, if it is passed as a parm. */
+ structure_value_addr_parm);
/* If we know nothing, treat all args as named. */
n_named_args = num_actuals;
/* Start updating where the next arg would go.
On some machines (such as the PA) indirect calls have a different
calling convention than normal calls. The fourth argument in
INIT_CUMULATIVE_ARGS tells the backend if this is an indirect call
or not. */
INIT_CUMULATIVE_ARGS (args_so_far_v, funtype, NULL_RTX, fndecl, n_named_args);
args_so_far = pack_cumulative_args (&args_so_far_v);
/* Now possibly adjust the number of named args.
Normally, don't include the last named arg if anonymous args follow.
We do include the last named arg if
targetm.calls.strict_argument_naming() returns nonzero.
(If no anonymous args follow, the result of list_length is actually
one too large. This is harmless.)
If targetm.calls.pretend_outgoing_varargs_named() returns
nonzero, and targetm.calls.strict_argument_naming() returns zero,
this machine will be able to place unnamed args that were passed
in registers into the stack. So treat all args as named. This
allows the insns emitting for a specific argument list to be
independent of the function declaration.
If targetm.calls.pretend_outgoing_varargs_named() returns zero,
we do not have any reliable way to pass unnamed args in
registers, so we must force them into memory. */
if (type_arg_types != 0
&& targetm.calls.strict_argument_naming (args_so_far))
else if (type_arg_types != 0
&& ! targetm.calls.pretend_outgoing_varargs_named (args_so_far))
/* Don't include the last named arg. */
/* Treat all args as named. */
n_named_args = num_actuals;
/* Make a vector to hold all the information about each arg. */
args = XCNEWVEC (struct arg_data, num_actuals);
/* Build up entries in the ARGS array, compute the size of the
arguments into ARGS_SIZE, etc. */
initialize_argument_information (num_actuals, args, &args_size,
n_named_args, exp,
structure_value_addr_value, fndecl, fntype,
args_so_far, reg_parm_stack_space,
&old_stack_level, &old_pending_adj,
&must_preallocate, &flags,
&try_tail_call, CALL_FROM_THUNK_P (exp));
if (args_size.var)
must_preallocate = 1;
/* Now make final decision about preallocating stack space. */
must_preallocate = finalize_must_preallocate (must_preallocate,
num_actuals, args,
/* If the structure value address will reference the stack pointer, we
must stabilize it. We don't need to do this if we know that we are
not going to adjust the stack pointer in processing this call. */
if (structure_value_addr
&& (reg_mentioned_p (virtual_stack_dynamic_rtx, structure_value_addr)
|| reg_mentioned_p (virtual_outgoing_args_rtx,
&& (args_size.var
&& maybe_ne (args_size.constant, 0))))
structure_value_addr = copy_to_reg (structure_value_addr);
/* Tail calls can make things harder to debug, and we've traditionally
pushed these optimizations into -O2. Don't try if we're already
expanding a call, as that means we're an argument. Don't try if
there's cleanups, as we know there's code to follow the call. */
if (currently_expanding_call++ != 0
|| (!flag_optimize_sibling_calls && !CALL_FROM_THUNK_P (exp))
|| args_size.var
|| dbg_cnt (tail_call) == false)
try_tail_call = 0;
/* Workaround buggy C/C++ wrappers around Fortran routines with
character(len=constant) arguments if the hidden string length arguments
are passed on the stack; if the callers forget to pass those arguments,
attempting to tail call in such routines leads to stack corruption.
Avoid tail calls in functions where at least one such hidden string
length argument is passed (partially or fully) on the stack in the
caller and the callee needs to pass any arguments on the stack.
See PR90329. */
if (try_tail_call && maybe_ne (args_size.constant, 0))
for (tree arg = DECL_ARGUMENTS (current_function_decl);
arg; arg = DECL_CHAIN (arg))
subrtx_iterator::array_type array;
if (MEM_P (*iter))
try_tail_call = 0;
/* If the user has marked the function as requiring tail-call
optimization, attempt it. */
if (must_tail_call)
try_tail_call = 1;
/* Rest of purposes for tail call optimizations to fail. */
if (try_tail_call)
try_tail_call = can_implement_as_sibling_call_p (exp,
flags, addr, args_size);
/* Check if caller and callee disagree in promotion of function
return value. */
if (try_tail_call)
machine_mode caller_mode, caller_promoted_mode;
machine_mode callee_mode, callee_promoted_mode;
int caller_unsignedp, callee_unsignedp;
tree caller_res = DECL_RESULT (current_function_decl);
caller_unsignedp = TYPE_UNSIGNED (TREE_TYPE (caller_res));
caller_mode = DECL_MODE (caller_res);
callee_unsignedp = TYPE_UNSIGNED (TREE_TYPE (funtype));
callee_mode = TYPE_MODE (TREE_TYPE (funtype));
= promote_function_mode (TREE_TYPE (caller_res), caller_mode,
TREE_TYPE (current_function_decl), 1);
= promote_function_mode (TREE_TYPE (funtype), callee_mode,
funtype, 1);
if (caller_mode != VOIDmode
&& (caller_promoted_mode != callee_promoted_mode
|| ((caller_mode != caller_promoted_mode
|| callee_mode != callee_promoted_mode)
&& (caller_unsignedp != callee_unsignedp
|| partial_subreg_p (caller_mode, callee_mode)))))
try_tail_call = 0;
maybe_complain_about_tail_call (exp,
"caller and callee disagree in"
" promotion of function"
" return value");
/* Ensure current function's preferred stack boundary is at least
what we need. Stack alignment may also increase preferred stack
boundary. */
for (i = 0; i < num_actuals; i++)
if (reg_parm_stack_space > 0
|| args[i].reg == 0
|| args[i].partial != 0
|| args[i].pass_on_stack)
update_stack_alignment_for_call (&args[i].locate);
if (crtl->preferred_stack_boundary < preferred_stack_boundary)
crtl->preferred_stack_boundary = preferred_stack_boundary;
preferred_stack_boundary = crtl->preferred_stack_boundary;
preferred_unit_stack_boundary = preferred_stack_boundary / BITS_PER_UNIT;
if (flag_callgraph_info)
record_final_call (fndecl, EXPR_LOCATION (exp));
/* We want to make two insn chains; one for a sibling call, the other
for a normal call. We will select one of the two chains after
initial RTL generation is complete. */
for (pass = try_tail_call ? 0 : 1; pass < 2; pass++)
int sibcall_failure = 0;
bool normal_failure = false;
/* We want to emit any pending stack adjustments before the tail
recursion "call". That way we know any adjustment after the tail
recursion call can be ignored if we indeed use the tail
call expansion. */
saved_pending_stack_adjust save;
rtx_insn *insns, *before_call, *after_args;
rtx next_arg_reg;
if (pass == 0)
/* State variables we need to save and restore between
iterations. */
save_pending_stack_adjust (&save);
if (pass)
flags &= ~ECF_SIBCALL;
flags |= ECF_SIBCALL;
/* Other state variables that we must reinitialize each time
through the loop (that are not initialized by the loop itself). */
argblock = 0;
call_fusage = 0;
/* Start a new sequence for the normal call case.
From this point on, if the sibling call fails, we want to set
sibcall_failure instead of continuing the loop. */
start_sequence ();
/* Don't let pending stack adjusts add up to too much.
Also, do all pending adjustments now if there is any chance
this might be a call to alloca or if we are expanding a sibling
call sequence.
Also do the adjustments before a throwing call, otherwise
exception handling can fail; PR 19225. */
if (maybe_ge (pending_stack_adjust, 32)
|| (maybe_ne (pending_stack_adjust, 0)
&& (flags & ECF_MAY_BE_ALLOCA))
|| (maybe_ne (pending_stack_adjust, 0)
&& flag_exceptions && !(flags & ECF_NOTHROW))
|| pass == 0)
do_pending_stack_adjust ();
/* Precompute any arguments as needed. */
if (pass)
precompute_arguments (num_actuals, args);
/* Now we are about to start emitting insns that can be deleted
if a libcall is deleted. */
if (pass && (flags & ECF_MALLOC))
start_sequence ();
/* Check the canary value for sibcall or function which doesn't
return and could throw. */
if ((pass == 0
|| ((flags & ECF_NORETURN) != 0 && tree_could_throw_p (exp)))
&& crtl->stack_protect_guard
&& targetm.stack_protect_runtime_enabled_p ())
stack_protect_epilogue ();
adjusted_args_size = args_size;
/* Compute the actual size of the argument block required. The variable
and constant sizes must be combined, the size may have to be rounded,
and there may be a minimum required size. When generating a sibcall
pattern, do not round up, since we'll be re-using whatever space our
caller provided. */
= compute_argument_block_size (reg_parm_stack_space,
fndecl, fntype,
(pass == 0 ? 0
: preferred_stack_boundary));
old_stack_allocated = stack_pointer_delta - pending_stack_adjust;
/* The argument block when performing a sibling call is the
incoming argument block. */
if (pass == 0)
argblock = crtl->args.internal_arg_pointer;
= plus_constant (Pmode, argblock, crtl->args.pretend_args_size);
= plus_constant (Pmode, argblock, -crtl->args.pretend_args_size);
HOST_WIDE_INT map_size = constant_lower_bound (args_size.constant);
stored_args_map = sbitmap_alloc (map_size);
bitmap_clear (stored_args_map);
stored_args_watermark = HOST_WIDE_INT_M1U;
/* If we have no actual push instructions, or shouldn't use them,
make space for all args right now. */
else if (adjusted_args_size.var != 0)
if (old_stack_level == 0)
emit_stack_save (SAVE_BLOCK, &old_stack_level);
old_stack_pointer_delta = stack_pointer_delta;
old_pending_adj = pending_stack_adjust;
pending_stack_adjust = 0;
/* stack_arg_under_construction says whether a stack arg is
being constructed at the old stack level. Pushing the stack
gets a clean outgoing argument block. */