Hello Philipp,
Thursday, December 30, 2021, 4:31:50 PM, you wrote:
>> > Thursday, December 30, 2021, 2:44:26 PM, you wrote:
>> > Oh, i see! Missed that. Thank you again.
>> > PKK> On 30.12.21 12:41, Tony Pavlov via Sdcc-user wrote:
>>>> So, __banked forces __sdcccall(0), that's nice.
>> > PKK> For parameters, and caller vs. callee cleanup, essentially yes. But
>> > not for the return value.
>> > >
PKK> Stated with different words:
PKK> Calls to __banked functions are handled like calls to functions with
variable arguments.
There is a problem with returning uint8_t results from __banked functions. By
the new convention
it is returned in A, but the tail code of the trampoline function is very
likely spoil A, while
having result in E, DE, DEHL is perfect here.
the new trampoline looks like this:
___sdcc_bcall_ehl:: ; Performs a long call.
ldh a, (__current_bank)
push af ; Push the current bank onto the stack
ld a, e
ldh (__current_bank), a
ld (.MBC_ROM_PAGE), a ; Perform the switch
rst 0x20 ; call HL
pop bc ; Pop the old bank
ld c, a
ld a, b
ldh (__current_bank), a
ld (.MBC_ROM_PAGE), a
ld a, c
ret
while the old one was:
___sdcc_bcall_ehl:: ; Performs a long call.
ldh a, (__current_bank)
push af ; Push the current bank onto the stack
ld a, e
ldh (__current_bank), a
ld (.MBC_ROM_PAGE), a ; Perform the switch
rst 0x20 ; call HL
pop af ; Pop the old bank
ldh (__current_bank), a
ld (.MBC_ROM_PAGE), a
ret
which is shorter and faster.
__banked already force to pass parameters on stack. why not simply
imply __sdcccall(0) for __banked?
look at the trampoline from the native SDCC Z80 library:
___sdcc_bcall_ehl::
call get_bank
push af
inc sp
call ___sdcc_bjump_ehl
dec sp
pop bc
push af
ld a, b
call set_bank
pop af
ret
push af/pop af plus loads, what for?
another annoying thing is one byte hidden parameter for the bank number on the
Z80 target.
inc sp/dec sp everywhere, while on gbz80 two bytes are reserved - that is much
faster! why
not unify that? also, on systems like MSX you may want to save/restore more
than one page
and two bytes may be very useful here!
--
Best regards,
Tony mailto:unt...@mail.ru
_______________________________________________
Sdcc-user mailing list
Sdcc-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sdcc-user
