diff options
Diffstat (limited to 'tinyc/tests/asmtest.S')
| -rw-r--r-- | tinyc/tests/asmtest.S | 455 |
1 files changed, 436 insertions, 19 deletions
diff --git a/tinyc/tests/asmtest.S b/tinyc/tests/asmtest.S index 358a8239f..55787054a 100644 --- a/tinyc/tests/asmtest.S +++ b/tinyc/tests/asmtest.S @@ -1,3 +1,4 @@ +# gas comment with ``gnu'' style quotes /* some directive tests */ @@ -9,14 +10,23 @@ .int 1, 2, 3 .align 8 .byte 1 - .align 16, 0x90 +/* .align 16, 0x90 gas is too clever for us with 0x90 fill */ + .balign 4, 0x92 + .align 16, 0x91 /* 0x91 tests the non-clever behaviour */ .skip 3 .skip 15, 0x90 .string "hello\0world" - +/* Macro expansion should work like with C, the #n shouldn't be parsed + as asm line comment */ +#define __stringify(n) #n +#define stringify(n) __stringify(n) + .skip 8,0x90 + .asciz stringify(BLA) + .skip 8,0x90 + +# 28 "asmtest.S" # a line directive (and a line comment) + movl %eax, %ebx # some more asm comment /* some label tests */ - - movl %eax, %ebx L1: movl %eax, %ebx mov 0x10000, %eax @@ -36,7 +46,7 @@ mov %al, 0x10000 mov $1, %edx mov $1, %dx -mov $1, %dl +mov $1, %cl movb $2, 0x100(%ebx,%edx,2) movw $2, 0x100(%ebx,%edx,2) movl $2, 0x100(%ebx,%edx,2) @@ -44,17 +54,55 @@ movl %eax, 0x100(%ebx,%edx,2) movl 0x100(%ebx,%edx,2), %edx movw %ax, 0x100(%ebx,%edx,2) +movw $0x1122,%si +movl $0x112233,%edx +movl $0x80000000, %esi +movl $-0x7fffffff, %edi +#ifdef __x86_64__ +mov $0x11223344,%rbx +movq $0x11223344,%rbx +mov $0x1122334455,%rbx +movq $0x1122334455,%rbx +movl $0x11334455,(%rbx) +#endif + mov %eax, 0x12(,%edx,2) +#ifdef __i386__ mov %cr3, %edx mov %ecx, %cr3 movl %cr3, %eax movl %tr3, %eax movl %db3, %ebx movl %dr6, %eax +#else +mov %cr3, %rdx +mov %rcx, %cr3 +movq %cr3, %rax +movq %db3, %rbx +movq %dr6, %rax +mov %cr8, %rsi +mov %rdi, %cr8 +#endif movl %fs, %ecx movl %ebx, %fs +#ifdef __x86_64__ +movq %r8, %r9 +movq %r10, %r11 +movq %r12, %r13 +movq %r14, %r15 +movq %rax, %r9 +movq %r15, %rsi +inc %r9b +dec %r10w +not %r11d +negq %r12 +decb %r13b +incw %r14w +notl %r15d +#endif + movsbl 0x1000, %eax movsbw 0x1000, %ax movswl 0x1000, %eax @@ -66,19 +114,47 @@ movl %ebx, %fs movzb 0x1000, %eax movzb 0x1000, %ax + mov $0x12345678,%eax + +#ifdef __x86_64__ + movzb 0x1000, %rax + movzbq 0x1000, %rbx + movsbq 0x1000, %rdx + movzwq 0x1000, %rdi + movswq 0x1000, %rdx + movslq %eax, %rcx + mov $0x12345678,%rax + mov $0x12345678,%rdx + mov $0x12345678,%r10 + mov $0x123456789abcdef0,%rax + mov $0x123456789abcdef0,%rcx + mov $0x123456789abcdef0,%r11 +#endif +#ifdef __i386__ pushl %eax - pushw %ax push %eax push %cs +#else + pushq %rax + push %rax +#endif + pushw %ax push %gs push $1 push $100 + push 0x42(%eax) + pop 0x43(%esi) +#ifdef __i386__ popl %eax - popw %ax pop %eax pop %ds +#else + popq %rax + pop %rax +#endif + popw %ax pop %fs xchg %eax, %ecx @@ -109,22 +185,61 @@ movl %ebx, %fs leal 0x1000(%ebx), %ecx lea 0x1000(%ebx), %ecx +#ifdef __i386__ les 0x2000, %eax lds 0x2000, %ebx + lss 0x2000, %edx +#endif lfs 0x2000, %ecx lgs 0x2000, %edx - lss 0x2000, %edx addl $0x123, %eax add $0x123, %ebx +add $-16, %ecx +add $-0x123, %esi +add $1, %bx +add $1, %ebx +add $-1, %bx +add $-1, %ebx +add $127, %bx +addl $127, %ebx +addl $-128, %ebx +addl $-128, %ebx +addl $-129, %ebx +addl $128, %ebx +addl $255, %ebx +addl $256, %ebx +andb $0xf, %ah +andb $-15, %cl +xorb $127, %dh +cmpb $42, (%eax) addl $0x123, 0x100 addl $0x123, 0x100(%ebx) addl $0x123, 0x100(%ebx,%edx,2) addl $0x123, 0x100(%esp) +addl $0x123, (3*8)(%esp) addl $0x123, (%ebp) addl $0x123, (%esp) cmpl $0x123, (%esp) +#ifdef __x86_64__ +xor %bl,%ah +xor %bl,%r8b +xor %r9b,%bl +xor %sil,%cl +add %eax,(%r8d) +add %ebx,(%r9) +add %edx,(%r10d,%r11d) +add %ecx,(%r12,%r13) +add %esi,(%r14,%r15,4) +add %edi,0x1000(%rbx,%r12,8) +add %r11,0x1000(%ebp,%r9d,8) +movb $12, %ah +movb $13, %bpl +movb $14, %dil +movb $15, %r12b +#endif + add %eax, (%ebx) add (%ebx), %eax @@ -176,6 +291,8 @@ add (%ebx), %dl div %bl div %ecx, %eax +and $15,%bx +and $-20,%edx shl %edx shl $10, %edx @@ -192,21 +309,53 @@ shrd %eax, %edx L4: call 0x1000 call L4 +#ifdef __i386__ call *%eax +#else +call *%rax +#endif call *0x1000 call func1 +.global L5,L6 + +L5: +L6: + +#ifdef __i386__ lcall $0x100, $0x1000 +#else +lcall *0x100 +lcall *(%rax) +#endif jmp 0x1000 +jmp *(%edi) +#ifdef __i386__ jmp *%eax +#else +jmp *%rax +#endif jmp *0x1000 +#ifdef __i386__ ljmp $0x100, $0x1000 +#else +ljmp *0x100 +ljmp *(%rdi) +ljmpl *(%esi) +ljmpw *(%esi) +#endif ret - ret $10 +#ifdef __i386__ +retl +retl $10 +#else +retq +retq $10 +#endif lret @@ -234,14 +383,20 @@ L3: seto %al + setc %al + setcb %al setnp 0x1000 setl 0xaaaa setg %dl fadd fadd %st(1), %st + fadd %st(0), %st(1) fadd %st(3) + fmul %st(0),%st(0) + fmul %st(0),%st(1) + faddp %st(5) faddp faddp %st(1), %st @@ -341,9 +496,9 @@ L3: fnstcw 0x1002 fstcw 0x1002 fnstsw 0x1004 - fnstsw %eax + fnstsw (%eax) fstsw 0x1004 - fstsw %eax + fstsw (%eax) fnclex fclex fnstenv 0x1000 @@ -387,6 +542,7 @@ L3: fwait bswap %edx +bswapl %ecx xadd %ecx, %edx xaddb %dl, 0x1000 xaddw %ax, 0x1000 @@ -397,6 +553,10 @@ cmpxchgw %ax, 0x1000 cmpxchgl %eax, 0x1000 invlpg 0x1000 cmpxchg8b 0x1002 +#ifdef __x86_64__ +cmpxchg16b (%rax) +cmpxchg16b (%r10,%r11) +#endif fcmovb %st(5), %st fcmove %st(5), %st @@ -416,32 +576,48 @@ fucomip %st(5), %st cmovo 0x1000, %eax cmovs 0x1000, %eax cmovns %edx, %edi + cmovne %ax, %si +#ifdef __x86_64__ + bswapq %rsi + bswapq %r10 + cmovz %rdi,%rbx +#endif int $3 int $0x10 +#ifdef __i386__ pusha popa - clc - cld +#endif + clc # another comment + cld # a comment with embedded ' tick cli clts cmc lahf sahf +#ifdef __i386__ pushfl popfl +#else + pushfq + popfq +#endif pushf popf stc std sti +#ifdef __i386__ aaa aas daa das aad aam + into +#endif cbw cwd cwde @@ -452,7 +628,6 @@ int $0x10 cltd leave int3 - into iret rsm hlt @@ -470,6 +645,24 @@ int $0x10 repz repne repnz + nop + + lock ;negl (%eax) + wait ;pushf + rep ;stosb + repe ;lodsb + repz ;cmpsb + repne;movsb + repnz;outsb + + /* handle one-line prefix + ops */ + lock negl (%eax) + wait pushf + rep stosb + repe lodsb + repz cmpsb + repne movsb + repnz outsb invd wbinvd @@ -479,7 +672,34 @@ int $0x10 rdmsr rdpmc ud2 +#ifdef __x86_64__ + syscall + sysret + sysretq + lfence + mfence + sfence + prefetchnta 0x18(%rdx) + prefetcht0 (%rcx) + prefetcht1 (%rsi) + prefetcht2 (%rdi) + prefetchw (%rdi) + clflush 0x1000(%rax,%rcx) + fxsaveq (%rdx) + fxsaveq (%r11) + fxrstorq (%rcx) + fxrstorq (%r10) +#endif + + lar %ax,%dx + lar %eax,%dx + lar %ax,%edx + lar %eax,%edx +#ifdef __x86_64__ + lar %ax,%rdx + lar %eax,%rdx +#endif emms movd %edx, %mm3 movd 0x1000, %mm2 @@ -524,35 +744,232 @@ int $0x10 +#ifdef __i386__ boundl %edx, 0x10000 boundw %bx, 0x1000 - + arpl %bx, 0x1000 +#endif lar 0x1000, %eax lgdt 0x1000 lidt 0x1000 lldt 0x1000 - lmsw 0x1000 - lsl 0x1000, %ecx - ltr 0x1000 - sgdt 0x1000 sidt 0x1000 sldt 0x1000 +#ifdef __x86_64__ + lgdtq 0x1000 + lidtq 0x1000 + sgdtq 0x1000 + sidtq 0x1000 + + swapgs + + str %rdx + str %r9 +#endif + + lmsw 0x1000 + lsl 0x1000, %ecx + ltr 0x1000 + ltr %si smsw 0x1000 str 0x1000 + str %ecx + str %dx verr 0x1000 verw 0x1000 +#ifdef __i386__ push %ds pushw %ds pushl %ds pop %ds popw %ds popl %ds +#endif fxsave 1(%ebx) fxrstor 1(%ecx) +#ifdef __i386__ pushl $1 +#else + pushq $1 +#endif pushw $1 push $1 + +#ifdef __ASSEMBLER__ // should be defined, for S files + inc %eax +#endif + +#ifndef _WIN32 +ft1: ft2: ft3: ft4: ft5: ft6: ft7: ft8: ft9: + xor %eax, %eax + ret + +.type ft1,STT_FUNC +.type ft2,@STT_FUNC +.type ft3,%STT_FUNC +.type ft4,"STT_FUNC" +.type ft5,function +.type ft6,@function +.type ft7,%function +.type ft8,"function" +#endif + + pause +.rept 6 + nop +.endr +.fill 4,1,0x90 + +.section .text.one,"ax" +nop +.previous +.pushsection .text.one,"ax" +nop +.pushsection .text.two,"ax" +nop +.popsection +.popsection + +1: ud2 +.pushsection __bug_table,"a" +.align 8 +2: .long 1b - 2b + .long 0x600000 - 2b + .long 1b + 42 + .long 43 + 1b + .long 2b + 144 + .long 145 + 2b + .word 164, 0 + .org 2b+32 +#ifdef __x86_64__ + .quad 1b +#else + .long 1b +#endif +.popsection +3: mov %eax,%ecx +4: +.pushsection .text.three, "ax" +nop +.skip (-((4b-3b) > 0) * 2) , 0x90 +.popsection + +.globl overrideme +.weak overrideme + nop +.globl notimplemented +notimplemented: + ret +.set overrideme, notimplemented +overrideme = notimplemented +overrideme: + ret + + movd %esi, %mm1 + movd %edi, %xmm2 + movd (%ebx), %mm3 + movd (%ebx), %xmm3 + movd %mm1, %esi + movd %xmm2, %edi + movd %mm3, (%edx) + movd %xmm3, (%edx) +#ifdef __x86_64__ + movd %rsi, %mm1 + movd %rdi, %xmm2 + movd (%rbx), %mm3 + movd (%rbx), %xmm3 + movd %mm1, %r12 + movd %xmm2, %rdi + movd %mm3, (%r8) + movd %xmm3, (%r13) +#endif + + movq (%ebp), %mm1 + movq %mm2, (%edi) + movq (%edi), %xmm3 + movq %mm4, %mm5 +#ifdef __x86_64__ + movq %rcx, %mm1 + movq %rdx, %xmm2 + movq %r13, %xmm3 + /* movq mem64->xmm is encoded as f30f7e by GAS, but as + 660f6e by tcc (which really is a movd and would need + a REX.W prefix to be movq). */ + movq (%rsi), %xmm3 + movq %mm1, %rdx + movq %xmm3, %rcx + movq %xmm4, (%rsi) +#endif + +#define TEST_MMX_SSE(insn) \ + insn %mm1, %mm2; \ + insn %xmm2, %xmm3; \ + insn (%ebx), %xmm3; +#define TEST_MMX_SSE_I8(insn) \ + TEST_MMX_SSE(insn) \ + insn $0x42, %mm4; \ + insn $0x42, %xmm4; + + TEST_MMX_SSE(packssdw) + TEST_MMX_SSE(packsswb) + TEST_MMX_SSE(packuswb) + TEST_MMX_SSE(paddb) + TEST_MMX_SSE(paddw) + TEST_MMX_SSE(paddd) + TEST_MMX_SSE(paddsb) + TEST_MMX_SSE(paddsw) + TEST_MMX_SSE(paddusb) + TEST_MMX_SSE(paddusw) + TEST_MMX_SSE(pand) + TEST_MMX_SSE(pandn) + TEST_MMX_SSE(pcmpeqb) + TEST_MMX_SSE(pcmpeqw) + TEST_MMX_SSE(pcmpeqd) + TEST_MMX_SSE(pcmpgtb) + TEST_MMX_SSE(pcmpgtw) + TEST_MMX_SSE(pcmpgtd) + TEST_MMX_SSE(pmaddwd) + TEST_MMX_SSE(pmulhw) + TEST_MMX_SSE(pmullw) + TEST_MMX_SSE(por) + TEST_MMX_SSE(psllw) +TEST_MMX_SSE_I8(psllw) + TEST_MMX_SSE(pslld) +TEST_MMX_SSE_I8(pslld) + TEST_MMX_SSE(psllq) +TEST_MMX_SSE_I8(psllq) + TEST_MMX_SSE(psraw) +TEST_MMX_SSE_I8(psraw) + TEST_MMX_SSE(psrad) +TEST_MMX_SSE_I8(psrad) + TEST_MMX_SSE(psrlw) +TEST_MMX_SSE_I8(psrlw) + TEST_MMX_SSE(psrld) +TEST_MMX_SSE_I8(psrld) + TEST_MMX_SSE(psrlq) +TEST_MMX_SSE_I8(psrlq) + TEST_MMX_SSE(psubb) + TEST_MMX_SSE(psubw) + TEST_MMX_SSE(psubd) + TEST_MMX_SSE(psubsb) + TEST_MMX_SSE(psubsw) + TEST_MMX_SSE(psubusb) + TEST_MMX_SSE(psubusw) + TEST_MMX_SSE(punpckhbw) + TEST_MMX_SSE(punpckhwd) + TEST_MMX_SSE(punpckhdq) + TEST_MMX_SSE(punpcklbw) + TEST_MMX_SSE(punpcklwd) + TEST_MMX_SSE(punpckldq) + TEST_MMX_SSE(pxor) + + cvtpi2ps %mm1, %xmm2 + cvtpi2ps (%ebx), %xmm2 + TEST_MMX_SSE(pmaxsw) + TEST_MMX_SSE(pmaxub) + TEST_MMX_SSE(pminsw) + TEST_MMX_SSE(pminub) |