Commit 89cf9e1f authored by Martin Storsjö's avatar Martin Storsjö

checkasm: arm: Make the indentation consistent with other files

This makes it easier to share code with e.g. the dav1d implementation
of checkasm.
Signed-off-by: 's avatarMartin Storsjö <martin@martin.st>
parent 07948f3d
......@@ -23,29 +23,29 @@
#include "libavutil/aarch64/asm.S"
const register_init, align=4
.quad 0x21f86d66c8ca00ce
.quad 0x75b6ba21077c48ad
.quad 0xed56bb2dcb3c7736
.quad 0x8bda43d3fd1a7e06
.quad 0xb64a9c9e5d318408
.quad 0xdf9a54b303f1d3a3
.quad 0x4a75479abd64e097
.quad 0x249214109d5d1c88
.quad 0x1a1b2550a612b48c
.quad 0x79445c159ce79064
.quad 0x2eed899d5a28ddcd
.quad 0x86b2536fcd8cf636
.quad 0xb0856806085e7943
.quad 0x3f2bf84fc0fcca4e
.quad 0xacbd382dcf5b8de2
.quad 0xd229e1f5b281303f
.quad 0x71aeaff20b095fd9
.quad 0xab63e2e11fa38ed9
.quad 0x21f86d66c8ca00ce
.quad 0x75b6ba21077c48ad
.quad 0xed56bb2dcb3c7736
.quad 0x8bda43d3fd1a7e06
.quad 0xb64a9c9e5d318408
.quad 0xdf9a54b303f1d3a3
.quad 0x4a75479abd64e097
.quad 0x249214109d5d1c88
.quad 0x1a1b2550a612b48c
.quad 0x79445c159ce79064
.quad 0x2eed899d5a28ddcd
.quad 0x86b2536fcd8cf636
.quad 0xb0856806085e7943
.quad 0x3f2bf84fc0fcca4e
.quad 0xacbd382dcf5b8de2
.quad 0xd229e1f5b281303f
.quad 0x71aeaff20b095fd9
.quad 0xab63e2e11fa38ed9
endconst
const error_message
.asciz "failed to preserve register"
.asciz "failed to preserve register"
endconst
......@@ -55,103 +55,103 @@ endconst
#define CLOBBER_STACK ((8*MAX_ARGS + 15) & ~15)
function checkasm_stack_clobber, export=1
mov x3, sp
mov x2, #CLOBBER_STACK
mov x3, sp
mov x2, #CLOBBER_STACK
1:
stp x0, x1, [sp, #-16]!
subs x2, x2, #16
b.gt 1b
mov sp, x3
ret
stp x0, x1, [sp, #-16]!
subs x2, x2, #16
b.gt 1b
mov sp, x3
ret
endfunc
#define ARG_STACK ((8*(MAX_ARGS - 8) + 15) & ~15)
function checkasm_checked_call, export=1
stp x29, x30, [sp, #-16]!
mov x29, sp
stp x19, x20, [sp, #-16]!
stp x21, x22, [sp, #-16]!
stp x23, x24, [sp, #-16]!
stp x25, x26, [sp, #-16]!
stp x27, x28, [sp, #-16]!
stp d8, d9, [sp, #-16]!
stp d10, d11, [sp, #-16]!
stp d12, d13, [sp, #-16]!
stp d14, d15, [sp, #-16]!
movrel x9, register_init
ldp d8, d9, [x9], #16
ldp d10, d11, [x9], #16
ldp d12, d13, [x9], #16
ldp d14, d15, [x9], #16
ldp x19, x20, [x9], #16
ldp x21, x22, [x9], #16
ldp x23, x24, [x9], #16
ldp x25, x26, [x9], #16
ldp x27, x28, [x9], #16
sub sp, sp, #ARG_STACK
stp x29, x30, [sp, #-16]!
mov x29, sp
stp x19, x20, [sp, #-16]!
stp x21, x22, [sp, #-16]!
stp x23, x24, [sp, #-16]!
stp x25, x26, [sp, #-16]!
stp x27, x28, [sp, #-16]!
stp d8, d9, [sp, #-16]!
stp d10, d11, [sp, #-16]!
stp d12, d13, [sp, #-16]!
stp d14, d15, [sp, #-16]!
movrel x9, register_init
ldp d8, d9, [x9], #16
ldp d10, d11, [x9], #16
ldp d12, d13, [x9], #16
ldp d14, d15, [x9], #16
ldp x19, x20, [x9], #16
ldp x21, x22, [x9], #16
ldp x23, x24, [x9], #16
ldp x25, x26, [x9], #16
ldp x27, x28, [x9], #16
sub sp, sp, #ARG_STACK
.equ pos, 0
.rept MAX_ARGS-8
// Skip the first 8 args, that are loaded into registers
ldr x9, [x29, #16 + 8*8 + pos]
str x9, [sp, #pos]
// Skip the first 8 args, that are loaded into registers
ldr x9, [x29, #16 + 8*8 + pos]
str x9, [sp, #pos]
.equ pos, pos + 8
.endr
mov x12, x0
ldp x0, x1, [x29, #16]
ldp x2, x3, [x29, #32]
ldp x4, x5, [x29, #48]
ldp x6, x7, [x29, #64]
blr x12
add sp, sp, #ARG_STACK
stp x0, x1, [sp, #-16]!
movrel x9, register_init
movi v3.8h, #0
mov x12, x0
ldp x0, x1, [x29, #16]
ldp x2, x3, [x29, #32]
ldp x4, x5, [x29, #48]
ldp x6, x7, [x29, #64]
blr x12
add sp, sp, #ARG_STACK
stp x0, x1, [sp, #-16]!
movrel x9, register_init
movi v3.8h, #0
.macro check_reg_neon reg1, reg2
ldr q1, [x9], #16
uzp1 v2.2d, v\reg1\().2d, v\reg2\().2d
eor v1.16b, v1.16b, v2.16b
orr v3.16b, v3.16b, v1.16b
ldr q1, [x9], #16
uzp1 v2.2d, v\reg1\().2d, v\reg2\().2d
eor v1.16b, v1.16b, v2.16b
orr v3.16b, v3.16b, v1.16b
.endm
check_reg_neon 8, 9
check_reg_neon 10, 11
check_reg_neon 12, 13
check_reg_neon 14, 15
uqxtn v3.8b, v3.8h
umov x3, v3.d[0]
check_reg_neon 8, 9
check_reg_neon 10, 11
check_reg_neon 12, 13
check_reg_neon 14, 15
uqxtn v3.8b, v3.8h
umov x3, v3.d[0]
.macro check_reg reg1, reg2
ldp x0, x1, [x9], #16
eor x0, x0, \reg1
eor x1, x1, \reg2
orr x3, x3, x0
orr x3, x3, x1
ldp x0, x1, [x9], #16
eor x0, x0, \reg1
eor x1, x1, \reg2
orr x3, x3, x0
orr x3, x3, x1
.endm
check_reg x19, x20
check_reg x21, x22
check_reg x23, x24
check_reg x25, x26
check_reg x27, x28
check_reg x19, x20
check_reg x21, x22
check_reg x23, x24
check_reg x25, x26
check_reg x27, x28
cbz x3, 0f
cbz x3, 0f
movrel x0, error_message
bl X(checkasm_fail_func)
movrel x0, error_message
bl X(checkasm_fail_func)
0:
ldp x0, x1, [sp], #16
ldp d14, d15, [sp], #16
ldp d12, d13, [sp], #16
ldp d10, d11, [sp], #16
ldp d8, d9, [sp], #16
ldp x27, x28, [sp], #16
ldp x25, x26, [sp], #16
ldp x23, x24, [sp], #16
ldp x21, x22, [sp], #16
ldp x19, x20, [sp], #16
ldp x29, x30, [sp], #16
ret
ldp x0, x1, [sp], #16
ldp d14, d15, [sp], #16
ldp d12, d13, [sp], #16
ldp d10, d11, [sp], #16
ldp d8, d9, [sp], #16
ldp x27, x28, [sp], #16
ldp x25, x26, [sp], #16
ldp x23, x24, [sp], #16
ldp x21, x22, [sp], #16
ldp x19, x20, [sp], #16
ldp x29, x30, [sp], #16
ret
endfunc
......@@ -29,22 +29,22 @@ ELF .eabi_attribute 10, 0 @ suppress Tag_FP_arch
#endif
const register_init, align=3
.quad 0x21f86d66c8ca00ce
.quad 0x75b6ba21077c48ad
.quad 0xed56bb2dcb3c7736
.quad 0x8bda43d3fd1a7e06
.quad 0xb64a9c9e5d318408
.quad 0xdf9a54b303f1d3a3
.quad 0x4a75479abd64e097
.quad 0x249214109d5d1c88
.quad 0x21f86d66c8ca00ce
.quad 0x75b6ba21077c48ad
.quad 0xed56bb2dcb3c7736
.quad 0x8bda43d3fd1a7e06
.quad 0xb64a9c9e5d318408
.quad 0xdf9a54b303f1d3a3
.quad 0x4a75479abd64e097
.quad 0x249214109d5d1c88
endconst
const error_message_fpscr
.asciz "failed to preserve register FPSCR, changed bits: %x"
.asciz "failed to preserve register FPSCR, changed bits: %x"
error_message_gpr:
.asciz "failed to preserve register r%d"
.asciz "failed to preserve register r%d"
error_message_vfp:
.asciz "failed to preserve register d%d"
.asciz "failed to preserve register d%d"
endconst
@ max number of args used by any asm function.
......@@ -58,107 +58,107 @@ endconst
.macro clobbercheck variant
.equ pushed, 4*9
function checkasm_checked_call_\variant, export=1
push {r4-r11, lr}
push {r4-r11, lr}
.ifc \variant, vfp
vpush {d8-d15}
fmrx r4, FPSCR
push {r4}
vpush {d8-d15}
fmrx r4, FPSCR
push {r4}
.equ pushed, pushed + 16*4 + 4
.endif
movrel r12, register_init
movrel r12, register_init
.ifc \variant, vfp
vldm r12, {d8-d15}
vldm r12, {d8-d15}
.endif
ldm r12, {r4-r11}
ldm r12, {r4-r11}
sub sp, sp, #ARG_STACK_A
sub sp, sp, #ARG_STACK_A
.equ pos, 0
.rept MAX_ARGS-4
ldr r12, [sp, #ARG_STACK_A + pushed + 8 + pos]
str r12, [sp, #pos]
ldr r12, [sp, #ARG_STACK_A + pushed + 8 + pos]
str r12, [sp, #pos]
.equ pos, pos + 4
.endr
mov r12, r0
mov r0, r2
mov r1, r3
ldrd r2, r3, [sp, #ARG_STACK_A + pushed]
blx r12
add sp, sp, #ARG_STACK_A
mov r12, r0
mov r0, r2
mov r1, r3
ldrd r2, r3, [sp, #ARG_STACK_A + pushed]
blx r12
add sp, sp, #ARG_STACK_A
push {r0, r1}
movrel r12, register_init
push {r0, r1}
movrel r12, register_init
.ifc \variant, vfp
.macro check_reg_vfp, dreg, offset
ldrd r2, r3, [r12, #8 * (\offset)]
vmov r0, lr, \dreg
eor r2, r2, r0
eor r3, r3, lr
orrs r2, r2, r3
bne 4f
ldrd r2, r3, [r12, #8 * (\offset)]
vmov r0, lr, \dreg
eor r2, r2, r0
eor r3, r3, lr
orrs r2, r2, r3
bne 4f
.endm
.irp n, 8, 9, 10, 11, 12, 13, 14, 15
@ keep track of the checked double/SIMD register
mov r1, #\n
check_reg_vfp d\n, \n-8
@ keep track of the checked double/SIMD register
mov r1, #\n
check_reg_vfp d\n, \n-8
.endr
.purgem check_reg_vfp
fmrx r1, FPSCR
ldr r3, [sp, #8]
eor r1, r1, r3
@ Ignore changes in bits 0-4 and 7
bic r1, r1, #0x9f
@ Ignore changes in the topmost 5 bits
bics r1, r1, #0xf8000000
bne 3f
fmrx r1, FPSCR
ldr r3, [sp, #8]
eor r1, r1, r3
@ Ignore changes in bits 0-4 and 7
bic r1, r1, #0x9f
@ Ignore changes in the topmost 5 bits
bics r1, r1, #0xf8000000
bne 3f
.endif
@ keep track of the checked GPR
mov r1, #4
@ keep track of the checked GPR
mov r1, #4
.macro check_reg reg1, reg2=
ldrd r2, r3, [r12], #8
eors r2, r2, \reg1
bne 2f
add r1, r1, #1
ldrd r2, r3, [r12], #8
eors r2, r2, \reg1
bne 2f
add r1, r1, #1
.ifnb \reg2
eors r3, r3, \reg2
bne 2f
eors r3, r3, \reg2
bne 2f
.endif
add r1, r1, #1
add r1, r1, #1
.endm
check_reg r4, r5
check_reg r6, r7
check_reg r4, r5
check_reg r6, r7
@ r9 is a volatile register in the ios ABI
#ifdef __APPLE__
check_reg r8
check_reg r8
#else
check_reg r8, r9
check_reg r8, r9
#endif
check_reg r10, r11
check_reg r10, r11
.purgem check_reg
b 0f
b 0f
4:
movrel r0, error_message_vfp
b 1f
movrel r0, error_message_vfp
b 1f
3:
movrel r0, error_message_fpscr
b 1f
movrel r0, error_message_fpscr
b 1f
2:
movrel r0, error_message_gpr
movrel r0, error_message_gpr
1:
blx X(checkasm_fail_func)
blx X(checkasm_fail_func)
0:
pop {r0, r1}
pop {r0, r1}
.ifc \variant, vfp
pop {r2}
fmxr FPSCR, r2
vpop {d8-d15}
pop {r2}
fmxr FPSCR, r2
vpop {d8-d15}
.endif
pop {r4-r11, pc}
pop {r4-r11, pc}
endfunc
.endm
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment