Function calling conventions and other conventions regarding the use of
machine registers and stack slots.
Require Import Coqlib.
Require Import Decidableplus.
Require Import AST.
Require Import Events.
Require Import Locations.
Require Archi.
Classification of machine registers
Machine registers (type
mreg in module
Locations) are divided in
the following groups:
-
Callee-save registers, whose value is preserved across a function call.
-
Caller-save registers that can be modified during a function call.
We follow the PowerPC/EABI application binary interface (ABI) in our choice
of callee- and caller-save registers.
Definition is_callee_save (
r:
mreg):
bool :=
match r with
|
R3 |
R4 |
R5 |
R6 |
R7 |
R8 |
R9 |
R10 |
R11 |
R12 =>
false
|
R14 |
R15 |
R16 |
R17 |
R18 |
R19 |
R20 |
R21 |
R22 |
R23 |
R24
|
R25 |
R26 |
R27 |
R28 |
R29 |
R30 |
R31 =>
true
|
F0 |
F1 |
F2 |
F3 |
F4 |
F5 |
F6 |
F7
|
F8 |
F9 |
F10 |
F11 |
F12 |
F13 =>
false
|
F14 |
F15 |
F16 |
F17 |
F18 |
F19 |
F20 |
F21 |
F22 |
F23
|
F24 |
F25 |
F26 |
F27 |
F28 |
F29 |
F30 |
F31 =>
true
end.
Definition destroyed_at_call :=
List.filter (
fun r =>
negb (
is_callee_save r))
all_mregs.
The following definitions are used by the register allocator.
When a PPC64 processor is used with the PPC32 ABI, the high 32 bits
of integer callee-save registers may not be preserved. So,
declare all integer registers as having size 32 bits for the purpose
of determining which variables can go in callee-save registers.
Definition callee_save_type (
r:
mreg):
typ :=
match r with
|
R3 |
R4 |
R5 |
R6 |
R7 |
R8 |
R9 |
R10 |
R11 |
R12
|
R14 |
R15 |
R16 |
R17 |
R18 |
R19 |
R20 |
R21 |
R22 |
R23 |
R24
|
R25 |
R26 |
R27 |
R28 |
R29 |
R30 |
R31 =>
Tany32
|
F0 |
F1 |
F2 |
F3 |
F4 |
F5 |
F6 |
F7
|
F8 |
F9 |
F10 |
F11 |
F12 |
F13 |
F14 |
F15
|
F16 |
F17 |
F18 |
F19 |
F20 |
F21 |
F22 |
F23
|
F24 |
F25 |
F26 |
F27 |
F28 |
F29 |
F30 |
F31 =>
Tany64
end.
Definition is_float_reg (
r:
mreg):
bool :=
match r with
|
R3 |
R4 |
R5 |
R6 |
R7 |
R8 |
R9 |
R10 |
R11 |
R12
|
R14 |
R15 |
R16 |
R17 |
R18 |
R19 |
R20 |
R21 |
R22 |
R23 |
R24
|
R25 |
R26 |
R27 |
R28 |
R29 |
R30 |
R31 =>
false
|
F0 |
F1 |
F2 |
F3 |
F4 |
F5 |
F6 |
F7
|
F8 |
F9 |
F10 |
F11 |
F12 |
F13
|
F14 |
F15 |
F16 |
F17 |
F18 |
F19 |
F20 |
F21 |
F22 |
F23
|
F24 |
F25 |
F26 |
F27 |
F28 |
F29 |
F30 |
F31 =>
true
end.
Definition int_caller_save_regs :=
R3 ::
R4 ::
R5 ::
R6 ::
R7 ::
R8 ::
R9 ::
R10 ::
R11 ::
R12 ::
nil.
Definition float_caller_save_regs :=
F0 ::
F1 ::
F2 ::
F3 ::
F4 ::
F5 ::
F6 ::
F7 ::
F8 ::
F9 ::
F10 ::
F11 ::
F12 ::
F13 ::
nil.
Definition int_callee_save_regs :=
R31 ::
R30 ::
R29 ::
R28 ::
R27 ::
R26 ::
R25 ::
R24 ::
R23 ::
R22 ::
R21 ::
R20 ::
R19 ::
R18 ::
R17 ::
R16 ::
R15 ::
R14 ::
nil.
Definition float_callee_save_regs :=
F31 ::
F30 ::
F29 ::
F28 ::
F27 ::
F26 ::
F25 ::
F24 ::
F23 ::
F22 ::
F21 ::
F20 ::
F19 ::
F18 ::
F17 ::
F16 ::
F15 ::
F14 ::
nil.
Definition dummy_int_reg :=
R3.
(* Used in Coloring. *)
Definition dummy_float_reg :=
F0.
(* Used in Coloring. *)
How to use registers for register allocation.
We favor the use of caller-save registers, using callee-save registers
only when no caller-save is available.
Record alloc_regs :=
mk_alloc_regs {
preferred_int_regs:
list mreg;
remaining_int_regs:
list mreg;
preferred_float_regs:
list mreg;
remaining_float_regs:
list mreg
}.
Definition allocatable_registers (
_:
unit) :=
{|
preferred_int_regs :=
int_caller_save_regs;
remaining_int_regs :=
int_callee_save_regs;
preferred_float_regs :=
float_caller_save_regs;
remaining_float_regs :=
float_callee_save_regs |}.
Function calling conventions
The functions in this section determine the locations (machine registers
and stack slots) used to communicate arguments and results between the
caller and the callee during function calls. These locations are functions
of the signature of the function and of the call instruction.
Agreement between the caller and the callee on the locations to use
is guaranteed by our dynamic semantics for Cminor and RTL, which demand
that the signature of the call instruction is identical to that of the
called function.
Calling conventions are largely arbitrary: they must respect the properties
proved in this section (such as no overlapping between the locations
of function arguments), but this leaves much liberty in choosing actual
locations. To ensure binary interoperability of code generated by our
compiler with libraries compiled by another PowerPC compiler, we
implement the standard conventions defined in the PowerPC/EABI
application binary interface.
Location of function result
The result value of a function is passed back to the caller in
registers R3 or F1 or R3, R4, depending on the type of the returned value.
We treat a function without result as a function with one integer result.
Definition loc_result_32 (
s:
signature) :
rpair mreg :=
match proj_sig_res s with
|
Tint |
Tany32 =>
One R3
|
Tfloat |
Tsingle |
Tany64 =>
One F1
|
Tlong =>
Twolong R3 R4
end.
Definition loc_result_64 (
s:
signature) :
rpair mreg :=
match proj_sig_res s with
|
Tint |
Tlong |
Tany32 |
Tany64 =>
One R3
|
Tfloat |
Tsingle =>
One F1
end.
Definition loc_result :=
if Archi.ptr64 then loc_result_64 else loc_result_32.
The result registers have types compatible with that given in the signature.
Lemma loc_result_type:
forall sig,
subtype (
proj_sig_res sig) (
typ_rpair mreg_type (
loc_result sig)) =
true.
Proof.
The result locations are caller-save registers
Lemma loc_result_caller_save:
forall (
s:
signature),
forall_rpair (
fun r =>
is_callee_save r =
false) (
loc_result s).
Proof.
If the result is in a pair of registers, those registers are distinct and have type Tint at least.
Lemma loc_result_pair:
forall sg,
match loc_result sg with
|
One _ =>
True
|
Twolong r1 r2 =>
r1 <>
r2 /\
proj_sig_res sg =
Tlong
/\
subtype Tint (
mreg_type r1) =
true /\
subtype Tint (
mreg_type r2) =
true
/\
Archi.ptr64 =
false
end.
Proof.
The location of the result depends only on the result part of the signature
Lemma loc_result_exten:
forall s1 s2,
s1.(
sig_res) =
s2.(
sig_res) ->
loc_result s1 =
loc_result s2.
Proof.
Location of function arguments
The PowerPC EABI states the following convention for passing arguments
to a function:
-
The first 8 integer arguments are passed in registers R3 to R10.
-
The first 8 float arguments are passed in registers F1 to F8.
-
The first 4 long integer arguments are passed in register pairs R3,R4 ... R9,R10.
-
Extra arguments are passed on the stack, in Outgoing slots, consecutively
assigned (1 word for an integer argument, 2 words for a float),
starting at word offset 0.
-
No stack space is reserved for the arguments that are passed in registers.
Definition int_param_regs :=
R3 ::
R4 ::
R5 ::
R6 ::
R7 ::
R8 ::
R9 ::
R10 ::
nil.
Definition float_param_regs :=
F1 ::
F2 ::
F3 ::
F4 ::
F5 ::
F6 ::
F7 ::
F8 ::
nil.
Fixpoint loc_arguments_rec
(
tyl:
list typ) (
ir fr ofs:
Z) {
struct tyl} :
list (
rpair loc) :=
match tyl with
|
nil =>
nil
| (
Tint |
Tany32)
as ty ::
tys =>
match list_nth_z int_param_regs ir with
|
None =>
One (
S Outgoing ofs ty) ::
loc_arguments_rec tys ir fr (
ofs + 1)
|
Some ireg =>
One (
R ireg) ::
loc_arguments_rec tys (
ir + 1)
fr ofs
end
|
Tsingle as ty ::
tys =>
match list_nth_z float_param_regs fr with
|
None =>
let ty :=
if Archi.single_passed_as_single then Tsingle else Tany64 in
let ofs :=
align ofs (
typesize ty)
in
One (
S Outgoing ofs ty) ::
loc_arguments_rec tys ir fr (
ofs + (
typesize ty))
|
Some freg =>
One (
R freg) ::
loc_arguments_rec tys ir (
fr + 1)
ofs
end
| (
Tfloat |
Tany64)
as ty ::
tys =>
match list_nth_z float_param_regs fr with
|
None =>
let ofs :=
align ofs 2
in
One (
S Outgoing ofs ty) ::
loc_arguments_rec tys ir fr (
ofs + 2)
|
Some freg =>
One (
R freg) ::
loc_arguments_rec tys ir (
fr + 1)
ofs
end
|
Tlong ::
tys =>
let ir :=
align ir 2
in
match list_nth_z int_param_regs ir,
list_nth_z int_param_regs (
ir + 1)
with
|
Some r1,
Some r2 =>
Twolong (
R r1) (
R r2) ::
loc_arguments_rec tys (
ir + 2)
fr ofs
|
_,
_ =>
let ofs :=
align ofs 2
in
(
if Archi.ptr64
then One (
S Outgoing ofs Tlong)
else Twolong (
S Outgoing ofs Tint) (
S Outgoing (
ofs + 1)
Tint)) ::
loc_arguments_rec tys ir fr (
ofs + 2)
end
end.
loc_arguments s returns the list of locations where to store arguments
when calling a function with signature s.
Definition loc_arguments (
s:
signature) :
list (
rpair loc) :=
loc_arguments_rec s.(
sig_args) 0 0 0.
Argument locations are either caller-save registers or Outgoing
stack slots at nonnegative offsets.
Definition loc_argument_acceptable (
l:
loc) :
Prop :=
match l with
|
R r =>
is_callee_save r =
false
|
S Outgoing ofs ty =>
ofs >= 0 /\ (
typealign ty |
ofs)
|
_ =>
False
end.
Definition loc_argument_charact (
ofs:
Z) (
l:
loc) :
Prop :=
match l with
|
R r =>
In r int_param_regs \/
In r float_param_regs
|
S Outgoing ofs'
ty =>
ofs' >=
ofs /\ (
typealign ty |
ofs')
|
_ =>
False
end.
Remark loc_arguments_rec_charact:
forall tyl ir fr ofs p,
In p (
loc_arguments_rec tyl ir fr ofs) ->
forall_rpair (
loc_argument_charact ofs)
p.
Proof.
Lemma loc_arguments_acceptable:
forall (
s:
signature) (
p:
rpair loc),
In p (
loc_arguments s) ->
forall_rpair loc_argument_acceptable p.
Proof.
Global Hint Resolve loc_arguments_acceptable:
locs.
Lemma loc_arguments_main:
loc_arguments signature_main =
nil.
Proof.
reflexivity.
Qed.
Normalization of function results and parameters
No normalization needed.
Definition return_value_needs_normalization (
t:
rettype) :=
false.
Definition parameter_needs_normalization (
t:
rettype) :=
false.