freebsd-dev/sys/libkern/arm/divsi3.S
Ian Lepore 25166187e3 Fix unwind-info errors in our hand-written arm assembler code.
We have functions nested within functions, and places where we start a
function then never end it, we just jump to the middle of something else.
We tried to express this with nested ENTRY()/END() macros (which result
in .fnstart and .fnend directives), but it turns out there's no way to
express that nesting in ARM EHABI unwind info, and newer tools treat
multiple .fnstart directives without an intervening .fnend as an error.

These changes introduce two new macros, EENTRY() and EEND().  EENTRY()
creates a global label you can call/jump to just like ENTRY(), but it
doesn't emit a .fnstart.  EEND() is a no-op that just documents the
conceptual endpoint that matches up with the same-named EENTRY().

This is based on patches submitted by Stepan Dyatkovskiy, but I made some
changes and added the EEND() stuff, so blame any problems on me.

Submitted by:	Stepan Dyatkovskiy <stpworld@narod.ru>
2014-08-01 18:24:44 +00:00

409 lines
8.5 KiB
ArmAsm

/* $NetBSD: divsi3.S,v 1.4 2003/04/05 23:27:15 bjh21 Exp $ */
/*-
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <machine/asm.h>
__FBSDID("$FreeBSD$");
/*
* stack is aligned as there's a possibility of branching to L_overflow
* which makes a C call
*/
ENTRY_NP(__umodsi3)
stmfd sp!, {lr}
sub sp, sp, #4 /* align stack */
bl .L_udivide
add sp, sp, #4 /* unalign stack */
mov r0, r1
ldmfd sp!, {pc}
END(__umodsi3)
ENTRY_NP(__modsi3)
stmfd sp!, {lr}
sub sp, sp, #4 /* align stack */
bl .L_divide
add sp, sp, #4 /* unalign stack */
mov r0, r1
ldmfd sp!, {pc}
.L_overflow:
#if !defined(_KERNEL) && !defined(_STANDALONE)
mov r0, #8 /* SIGFPE */
bl PIC_SYM(_C_LABEL(raise), PLT) /* raise it */
mov r0, #0
#else
/* XXX should cause a fatal error */
mvn r0, #0
#endif
RET
END(__modsi3)
ENTRY_NP(__udivsi3)
#ifdef __ARM_EABI__
EENTRY_NP(__aeabi_uidiv)
EENTRY_NP(__aeabi_uidivmod)
#endif
.L_udivide: /* r0 = r0 / r1; r1 = r0 % r1 */
eor r0, r1, r0
eor r1, r0, r1
eor r0, r1, r0
/* r0 = r1 / r0; r1 = r1 % r0 */
cmp r0, #1
bcc .L_overflow
beq .L_divide_l0
mov ip, #0
movs r1, r1
bpl .L_divide_l1
orr ip, ip, #0x20000000 /* ip bit 0x20000000 = -ve r1 */
movs r1, r1, lsr #1
orrcs ip, ip, #0x10000000 /* ip bit 0x10000000 = bit 0 of r1 */
b .L_divide_l1
.L_divide_l0: /* r0 == 1 */
mov r0, r1
mov r1, #0
RET
#ifdef __ARM_EABI__
EEND(__aeabi_uidiv)
EEND(__aeabi_uidivmod)
#endif
END(__udivsi3)
ENTRY_NP(__divsi3)
#ifdef __ARM_EABI__
EENTRY_NP(__aeabi_idiv)
EENTRY_NP(__aeabi_idivmod)
#endif
.L_divide: /* r0 = r0 / r1; r1 = r0 % r1 */
eor r0, r1, r0
eor r1, r0, r1
eor r0, r1, r0
/* r0 = r1 / r0; r1 = r1 % r0 */
cmp r0, #1
bcc .L_overflow
beq .L_divide_l0
ands ip, r0, #0x80000000
rsbmi r0, r0, #0
ands r2, r1, #0x80000000
eor ip, ip, r2
rsbmi r1, r1, #0
orr ip, r2, ip, lsr #1 /* ip bit 0x40000000 = -ve division */
/* ip bit 0x80000000 = -ve remainder */
.L_divide_l1:
mov r2, #1
mov r3, #0
/*
* If the highest bit of the dividend is set, we have to be
* careful when shifting the divisor. Test this.
*/
movs r1,r1
bpl .L_old_code
/*
* At this point, the highest bit of r1 is known to be set.
* We abuse this below in the tst instructions.
*/
tst r1, r0 /*, lsl #0 */
bmi .L_divide_b1
tst r1, r0, lsl #1
bmi .L_divide_b2
tst r1, r0, lsl #2
bmi .L_divide_b3
tst r1, r0, lsl #3
bmi .L_divide_b4
tst r1, r0, lsl #4
bmi .L_divide_b5
tst r1, r0, lsl #5
bmi .L_divide_b6
tst r1, r0, lsl #6
bmi .L_divide_b7
tst r1, r0, lsl #7
bmi .L_divide_b8
tst r1, r0, lsl #8
bmi .L_divide_b9
tst r1, r0, lsl #9
bmi .L_divide_b10
tst r1, r0, lsl #10
bmi .L_divide_b11
tst r1, r0, lsl #11
bmi .L_divide_b12
tst r1, r0, lsl #12
bmi .L_divide_b13
tst r1, r0, lsl #13
bmi .L_divide_b14
tst r1, r0, lsl #14
bmi .L_divide_b15
tst r1, r0, lsl #15
bmi .L_divide_b16
tst r1, r0, lsl #16
bmi .L_divide_b17
tst r1, r0, lsl #17
bmi .L_divide_b18
tst r1, r0, lsl #18
bmi .L_divide_b19
tst r1, r0, lsl #19
bmi .L_divide_b20
tst r1, r0, lsl #20
bmi .L_divide_b21
tst r1, r0, lsl #21
bmi .L_divide_b22
tst r1, r0, lsl #22
bmi .L_divide_b23
tst r1, r0, lsl #23
bmi .L_divide_b24
tst r1, r0, lsl #24
bmi .L_divide_b25
tst r1, r0, lsl #25
bmi .L_divide_b26
tst r1, r0, lsl #26
bmi .L_divide_b27
tst r1, r0, lsl #27
bmi .L_divide_b28
tst r1, r0, lsl #28
bmi .L_divide_b29
tst r1, r0, lsl #29
bmi .L_divide_b30
tst r1, r0, lsl #30
bmi .L_divide_b31
/*
* instead of:
* tst r1, r0, lsl #31
* bmi .L_divide_b32
*/
b .L_divide_b32
.L_old_code:
cmp r1, r0
bcc .L_divide_b0
cmp r1, r0, lsl #1
bcc .L_divide_b1
cmp r1, r0, lsl #2
bcc .L_divide_b2
cmp r1, r0, lsl #3
bcc .L_divide_b3
cmp r1, r0, lsl #4
bcc .L_divide_b4
cmp r1, r0, lsl #5
bcc .L_divide_b5
cmp r1, r0, lsl #6
bcc .L_divide_b6
cmp r1, r0, lsl #7
bcc .L_divide_b7
cmp r1, r0, lsl #8
bcc .L_divide_b8
cmp r1, r0, lsl #9
bcc .L_divide_b9
cmp r1, r0, lsl #10
bcc .L_divide_b10
cmp r1, r0, lsl #11
bcc .L_divide_b11
cmp r1, r0, lsl #12
bcc .L_divide_b12
cmp r1, r0, lsl #13
bcc .L_divide_b13
cmp r1, r0, lsl #14
bcc .L_divide_b14
cmp r1, r0, lsl #15
bcc .L_divide_b15
cmp r1, r0, lsl #16
bcc .L_divide_b16
cmp r1, r0, lsl #17
bcc .L_divide_b17
cmp r1, r0, lsl #18
bcc .L_divide_b18
cmp r1, r0, lsl #19
bcc .L_divide_b19
cmp r1, r0, lsl #20
bcc .L_divide_b20
cmp r1, r0, lsl #21
bcc .L_divide_b21
cmp r1, r0, lsl #22
bcc .L_divide_b22
cmp r1, r0, lsl #23
bcc .L_divide_b23
cmp r1, r0, lsl #24
bcc .L_divide_b24
cmp r1, r0, lsl #25
bcc .L_divide_b25
cmp r1, r0, lsl #26
bcc .L_divide_b26
cmp r1, r0, lsl #27
bcc .L_divide_b27
cmp r1, r0, lsl #28
bcc .L_divide_b28
cmp r1, r0, lsl #29
bcc .L_divide_b29
cmp r1, r0, lsl #30
bcc .L_divide_b30
.L_divide_b32:
cmp r1, r0, lsl #31
subhs r1, r1,r0, lsl #31
addhs r3, r3,r2, lsl #31
.L_divide_b31:
cmp r1, r0, lsl #30
subhs r1, r1,r0, lsl #30
addhs r3, r3,r2, lsl #30
.L_divide_b30:
cmp r1, r0, lsl #29
subhs r1, r1,r0, lsl #29
addhs r3, r3,r2, lsl #29
.L_divide_b29:
cmp r1, r0, lsl #28
subhs r1, r1,r0, lsl #28
addhs r3, r3,r2, lsl #28
.L_divide_b28:
cmp r1, r0, lsl #27
subhs r1, r1,r0, lsl #27
addhs r3, r3,r2, lsl #27
.L_divide_b27:
cmp r1, r0, lsl #26
subhs r1, r1,r0, lsl #26
addhs r3, r3,r2, lsl #26
.L_divide_b26:
cmp r1, r0, lsl #25
subhs r1, r1,r0, lsl #25
addhs r3, r3,r2, lsl #25
.L_divide_b25:
cmp r1, r0, lsl #24
subhs r1, r1,r0, lsl #24
addhs r3, r3,r2, lsl #24
.L_divide_b24:
cmp r1, r0, lsl #23
subhs r1, r1,r0, lsl #23
addhs r3, r3,r2, lsl #23
.L_divide_b23:
cmp r1, r0, lsl #22
subhs r1, r1,r0, lsl #22
addhs r3, r3,r2, lsl #22
.L_divide_b22:
cmp r1, r0, lsl #21
subhs r1, r1,r0, lsl #21
addhs r3, r3,r2, lsl #21
.L_divide_b21:
cmp r1, r0, lsl #20
subhs r1, r1,r0, lsl #20
addhs r3, r3,r2, lsl #20
.L_divide_b20:
cmp r1, r0, lsl #19
subhs r1, r1,r0, lsl #19
addhs r3, r3,r2, lsl #19
.L_divide_b19:
cmp r1, r0, lsl #18
subhs r1, r1,r0, lsl #18
addhs r3, r3,r2, lsl #18
.L_divide_b18:
cmp r1, r0, lsl #17
subhs r1, r1,r0, lsl #17
addhs r3, r3,r2, lsl #17
.L_divide_b17:
cmp r1, r0, lsl #16
subhs r1, r1,r0, lsl #16
addhs r3, r3,r2, lsl #16
.L_divide_b16:
cmp r1, r0, lsl #15
subhs r1, r1,r0, lsl #15
addhs r3, r3,r2, lsl #15
.L_divide_b15:
cmp r1, r0, lsl #14
subhs r1, r1,r0, lsl #14
addhs r3, r3,r2, lsl #14
.L_divide_b14:
cmp r1, r0, lsl #13
subhs r1, r1,r0, lsl #13
addhs r3, r3,r2, lsl #13
.L_divide_b13:
cmp r1, r0, lsl #12
subhs r1, r1,r0, lsl #12
addhs r3, r3,r2, lsl #12
.L_divide_b12:
cmp r1, r0, lsl #11
subhs r1, r1,r0, lsl #11
addhs r3, r3,r2, lsl #11
.L_divide_b11:
cmp r1, r0, lsl #10
subhs r1, r1,r0, lsl #10
addhs r3, r3,r2, lsl #10
.L_divide_b10:
cmp r1, r0, lsl #9
subhs r1, r1,r0, lsl #9
addhs r3, r3,r2, lsl #9
.L_divide_b9:
cmp r1, r0, lsl #8
subhs r1, r1,r0, lsl #8
addhs r3, r3,r2, lsl #8
.L_divide_b8:
cmp r1, r0, lsl #7
subhs r1, r1,r0, lsl #7
addhs r3, r3,r2, lsl #7
.L_divide_b7:
cmp r1, r0, lsl #6
subhs r1, r1,r0, lsl #6
addhs r3, r3,r2, lsl #6
.L_divide_b6:
cmp r1, r0, lsl #5
subhs r1, r1,r0, lsl #5
addhs r3, r3,r2, lsl #5
.L_divide_b5:
cmp r1, r0, lsl #4
subhs r1, r1,r0, lsl #4
addhs r3, r3,r2, lsl #4
.L_divide_b4:
cmp r1, r0, lsl #3
subhs r1, r1,r0, lsl #3
addhs r3, r3,r2, lsl #3
.L_divide_b3:
cmp r1, r0, lsl #2
subhs r1, r1,r0, lsl #2
addhs r3, r3,r2, lsl #2
.L_divide_b2:
cmp r1, r0, lsl #1
subhs r1, r1,r0, lsl #1
addhs r3, r3,r2, lsl #1
.L_divide_b1:
cmp r1, r0
subhs r1, r1, r0
addhs r3, r3, r2
.L_divide_b0:
tst ip, #0x20000000
bne .L_udivide_l1
mov r0, r3
cmp ip, #0
rsbmi r1, r1, #0
movs ip, ip, lsl #1
bicmi r0, r0, #0x80000000 /* Fix incase we divided 0x80000000 */
rsbmi r0, r0, #0
RET
.L_udivide_l1:
tst ip, #0x10000000
mov r1, r1, lsl #1
orrne r1, r1, #1
mov r3, r3, lsl #1
cmp r1, r0
subhs r1, r1, r0
addhs r3, r3, r2
mov r0, r3
RET
#ifdef __ARM_EABI__
EEND(__aeabi_idiv)
EEND(__aeabi_idivmod)
#endif
END(__divsi3)