stuff

Makefile

@@ -16,6 +16,12 @@ DOSK = $(shell find ka -name '*.k')
 vm/a.out: $(DOSK)
 	@cd ka && ../as/k-as.py dos.k 0x100000 ../vm/a.out
 
+.PHONY: clean
+
+clean:
+	@cd vm && make clean
+	@rm -f vm/a.out vm/k.exe vm/stdout.txt as/instrs.lst
+
 test: kas vm/a.out
 	@vm/k.exe vm/a.out > vm/stdout.txt
 	@rm -f vm/a.out

as/k-as.py

@@ -203,7 +203,7 @@ def parse():
 def parse_preproc(line):
     global pdata
 
-    tok = line.split(' ', 2)
+    tok = line.split(None, 2)
 
     # preprocessor
     if len(tok) > 1 and tok[1] == ':=':

ka/dos.k

@@ -16,10 +16,29 @@ _start:
     jmp .1
 
 ;
-; Essential definitions
+; Some definitions
 ;
+
+CHAR_MIN := 0x80
+SHRT_MIN := 0x8000
+INT_MIN  := 0x80000000
+LONG_MIN := 0x8000000000000000
+
+XCHAR_MIN := 0xFFFFFFFFFFFFFF80
+XSHRT_MIN := 0xFFFFFFFFFFFF8000
+XINT_MIN  := 0xFFFFFFFF80000000
+
 CHAR_MAX := 0x7F
-INT_MAX := 0x7FFF
+SHRT_MAX := 0x7FFF
+INT_MAX  := 0x7FFFFFFF
+LONG_MAX := 0x7FFFFFFFFFFFFFFF
+
+BYTE_MAX  := 0xFF
+WORD_MAX  := 0xFFFF
+LWORD_MAX := 0xFFFFFFFF
+QWORD_MAX := 0xFFFFFFFFFFFFFFFF
+
+STRLEN_MAX := 0xFFFFFFFF
 
 ;
 ; Include librairies

ka/fmt/itoa.k

@@ -30,14 +30,10 @@ _itoa:
     cmp ax2, 10     ; base 10
     jmp.nz .conv
 
-    cmp ax1, -9223372036854775808   ; LONG_MIN
-    jmp.z .min
-
     sgn rdx, ax1    ; extract ax1 sign
 
     cmp rdx, -1     ; negative?
-    not.z ax1       ; -x = ~x+1
+    neg.z ax1
-    inc.z ax1       ; need "neg" inst...
 
     ; main loop
 .conv:
@@ -50,7 +46,7 @@ _itoa:
     cmp rsx, 9      ; rsx > 9 ?
     jmp.a .nondec
 
-    add rsx, 48   ; '0'
+    add rsx, 48     ; '0'
     jmp .next
 
 .nondec:
@@ -90,13 +86,6 @@ _itoa:
     mov b[ax0+1], 0
     ret
 
-.min:
-    mov ax1, .min10
-    call strcpy
-    ret
-
-.min10 = "-9223372036854775808"
-
 ;
 ; wrappers
 ;

ka/main.k

@@ -5,6 +5,10 @@
 ; Main function
 ;
 main:
+    call itoa_test
+    ret
+
+stosb_test:
     cld
     mov rcx, 11
     mov rax, 33
@@ -44,7 +48,7 @@ movzx_test:
 
 itoa_test:
     mov ax0, .buf
-    mov ax1, -9223372036854775807
+    mov ax1, LONG_MIN
     mov ax2, 10
     call itoa
 
@@ -60,10 +64,6 @@ itoa_test:
     mov ax0, rax
     call print
 
-    mov rsi, 0x10
-    mov rdi, 8
-    lea rbi, b[rdi + rsi * 2 + 1]
-
     ret
 
 .buf = [32]
@@ -84,42 +84,55 @@ devtest:
 
 .buf = [32]
 
-strtest:
+str_test:
-    enter
-
     mov ax0, .msg
     call print
 
-    mov ax0, .buf
+    mov ax0, .buf1
     mov ax1, .msg
-    mov ax2, 5
+    call strcpy
+    
+    prn 10
+    mov ax0, .buf1
+    call print
+
+    mov ax0, .buf2
+    mov ax1, .msg
+    mov ax2, 8
     call strnzcpy
 
     prn 10
-    mov ax0, .buf
+    mov ax0, .buf2
-    mov ax1, 10
+    mov ax1, 12
     call print_n
 
-    mov ax0, .buf
+    mov ax0, .msg
+    mov ax1, .buf1
+    call strcmp
+    mov rbx, rax
+    
+    mov ax0, .buf2
     mov ax1, .msg
     call strcmp
+    mov rsx, rax
+
+    mov ax0, .msg
+    call strlen
 
-    leave
     ret
 
-.msg = "HelloWorld :)"
+.msg = "Hello World :)"
-.buf = [32]
+
+.buf1 = [32]
+.buf2 = "!!!!!!!!!!!!!"
 
 ;
 ; Exit function
 ;
 exit:
-    enter
-
     mov ax0, .msg
     call print
 
-    leave
     ret
 
 .msg = "Goodbye World :(\n"

ka/str/strcmp.k

@@ -9,44 +9,24 @@
 ;    1 if the first character that does not match has a greater value in str1 than in str2
 ;   -1 if the first character that does not match has a lower value in str1 than in str2
 ;
-
 strcmp:
-    cmp b[ax0], b[ax1]
+    mov ax2, STRLEN_MAX
-    jmp.nz .1
+    jmp strncmp
-
-    test b[ax0], b[ax0]
-    jmp.z .1
-
-    inc ax0
-    inc ax1
-    jmp strcmp
-
-.1:
-    mov rax, b[ax0]
-    sub rax, b[ax1]
-    sgn rax, rax
-    ret
 
 ;
 ; int strncmp(const char *str1, const char *str2, int maxn)
 ;
 strncmp:
     mov rcx, ax2
+    mov.cxz rax, 0
+    ret.cxz
 
-.0:
+    cld
-    cmp b[ax0], b[ax1]
+    rep.e cmpzsb ax0, ax1
-    jmp.nz .1
 
-    test b[ax0], b[ax0]
-    jmp.z .1
-
-    inc ax0
-    inc ax1
-    loop .0
-
-.1:
     mov rax, b[ax0]
     sub rax, b[ax1]
     sgn rax, rax
+
     ret
 

ka/str/strcpy.k

@@ -5,34 +5,19 @@
 ; void strcpy(char *, const char *)
 ;
 strcpy:
-    mov b[ax0], b[ax1]
+    mov ax2, STRLEN_MAX
-
+    jmp strncpy
-    test b[ax1], b[ax1]
-    ret.z
-
-    inc ax0
-    inc ax1
-    jmp strcpy
 
 ;
 ; void strncpy(char *, const char *, int)
 ;
 strncpy:
     mov rcx, ax2
-    j.cxz .2
+    ret.cxz
-    dec rcx
 
-.1:
+    cld
-    mov b[ax0], b[ax1]
+    rep.nz movsb ax0, ax1
 
-    test b[ax1], b[ax1]
-    ret.z
-
-    inc ax0
-    inc ax1
-    loop .1
-
-.2:
     ret
 
 ;
@@ -40,20 +25,15 @@ strncpy:
 ;
 strnzcpy:
     mov rcx, ax2
-    j.cxz .2
+    ret.cxz
+
     dec rcx
+    jmp.cxz .1
+
+    cld
+    rep.nz movsb ax0, ax1
 
 .1:
-    mov b[ax0], b[ax1]
-
-    test b[ax1], b[ax1]
-    ret.z
-
-    inc ax0
-    inc ax1
-    loop .1
-
-.2:
     mov b[ax0], 0
     ret
 

ka/str/strlen.k

@@ -1,37 +1,22 @@
 ; The OS/K Team licenses this file to you under the MIT license.
 ; See the LICENSE file in the project root for more information.
 
-;
-; int strlen(char *)
-;
-strlen:
-    xor rax, rax
-
-.1:
-    test b[ax0], b[ax0]
-    ret.z
-
-    inc rax
-    inc ax0
-    jmp .1
-
 ;
 ; int strnlen(char *, int)
 ;
 strnlen:
-    xor rax, rax
+    cld
-
     mov rcx, ax1
-    j.cxz .2
+    rep.nz scasb ax0, 0
-    dec rcx
+    
-
+    mov rax, ax1
-.1:
+    sub rax, rcx
-    test b[ax0], b[ax0]
-    ret.z
-
-    inc rax
-    inc ax0
-    loop .1
-
-.2:
     ret
+
+;
+; int strlen(char *)
+;
+strlen:
+    mov ax1, STRLEN_MAX
+    jmp strnlen
+

vm/in/INSTRS

@@ -408,6 +408,9 @@ stosb r rim
 #           %str = %str - sizeof(x)
 #       FI
 #
+# Preserves CF, OF and SF
+# Sets ZF according to the loaded value
+#
 # When no parameters are given, %dest = RAX and %str = RSI
 # When one parameter is given, %dest = $1 and %str = RSI
 # When two parameters are given, %dest = $1 and %str = $2
@@ -460,6 +463,19 @@ cmpsb
 cmpsb r
 cmpsb r r
 
+#
+# Safe compare bytes in strings (CMPZSx)
+#
+# Behaves precisely like CMPSx, except in the following case:
+#   - If both [%str1] and [%str2] are zero, clears ZF (indicating NOT EQUAL)
+#
+# This prevents 'rep.e cmpsb' from looping infinitely when both strings
+# have the exact same content; this allows for short strcmp's
+#
+cmpzsb
+cmpzsb r
+cmpzsb r r
+
 #
 # Move value from string to string (MOVSx)
 #
@@ -472,6 +488,9 @@ cmpsb r r
 #           %str2 = %str2 - sizeof(x)
 #       FI
 #
+# Preserves CF, OF and SF
+# Sets ZF according to the moved value
+#
 # When no parameters are given, %str1 = RDI and %str2 = RSI
 # When one parameter is given, %str1 = RDI and %str2 = $1
 # When two parameters are given, %str1 = $1 and %str2 = $2

vm/in/arith.c

@@ -7,7 +7,8 @@
 
 IMPL_START_2(sgn)
 {
-    v1 = (long)v2 < 0 ? (ulong)-1L : 1;
+    v1 = (v2 == 0 ? 0 : 
+            ((long)v2 < 0 ? (ulong)-1L : 1));
 }
 IMPL_OUT;
 

vm/in/string.c

@@ -85,6 +85,10 @@ void lods_impl(ctx_t *ctx, acc_t *p1, acc_t *p2, uint len)
     }
 
     R(reg1) = readmem(ctx, R(reg2), len);
+    
+    flg = (R(reg1) == 0 ? flg|ZF : flg&~ZF);
+
+    STR_MOVE(reg2, len);
 }
 
 IMPL_START_0(lodsb)
@@ -113,22 +117,219 @@ IMPL_END;
 
 //----------------------------------------------------------------------------//
 
+void scas_impl(ctx_t *ctx, acc_t *p1, acc_t *p2, uint len)
+{
+    ulong reg, val;
+
+    if (p2) {
+        DECV(v2, p2);
+        reg = p1->reg;
+        val = v2;
+    }
+
+    else if (p1) {
+        DECV(v1, p1);
+        reg = RDI;
+        val = v1;
+    }
+
+    else {
+        reg = RDI;
+        val = rax;
+    }
+
+    ulong x = readmem(ctx, R(reg), len);
+    COMPARE(x, val);
+
+    STR_MOVE(reg, len);
+}
+
 IMPL_START_0(scasb)
 {
+    scas_impl(ctx, p1, p2, 1);
+}
+IMPL_END;
+
+IMPL_START_0(scasw)
+{
+    scas_impl(ctx, p1, p2, 2);
+}
+IMPL_END;
+
+IMPL_START_0(scasl)
+{
+    scas_impl(ctx, p1, p2, 4);
+}
+IMPL_END;
+
+IMPL_START_0(scasq)
+{
+    scas_impl(ctx, p1, p2, 8);
 }
 IMPL_END;
 
 //----------------------------------------------------------------------------//
 
+void cmps_impl(ctx_t *ctx, acc_t *p1, acc_t *p2, uint len)
+{
+    ulong reg1, reg2;
+
+    if (p2) {
+        reg1 = p1->reg;
+        reg2 = p2->reg;
+    }
+
+    else if (p1) {
+        reg1 = RDI;
+        reg2 = p1->reg;
+    }
+
+    else {
+        reg1 = RDI;
+        reg2 = RSI;
+    }
+
+    ulong x1 = readmem(ctx, R(reg1), len);
+    ulong x2 = readmem(ctx, R(reg2), len);
+
+    COMPARE(x1, x2);
+
+    STR_MOVE(reg1, len);
+    STR_MOVE(reg2, len);
+}
+
 IMPL_START_0(cmpsb)
 {
+    cmps_impl(ctx, p1, p2, 1);
+}
+IMPL_END;
+
+IMPL_START_0(cmpsw)
+{
+    cmps_impl(ctx, p1, p2, 2);
+}
+IMPL_END;
+
+IMPL_START_0(cmpsl)
+{
+    cmps_impl(ctx, p1, p2, 4);
+}
+IMPL_END;
+
+IMPL_START_0(cmpsq)
+{
+    cmps_impl(ctx, p1, p2, 8);
 }
 IMPL_END;
 
 //----------------------------------------------------------------------------//
 
+void cmpzs_impl(ctx_t *ctx, acc_t *p1, acc_t *p2, uint len)
+{
+    ulong reg1, reg2;
+
+    if (p2) {
+        reg1 = p1->reg;
+        reg2 = p2->reg;
+    }
+
+    else if (p1) {
+        reg1 = RDI;
+        reg2 = p1->reg;
+    }
+
+    else {
+        reg1 = RDI;
+        reg2 = RSI;
+    }
+
+    ulong x1 = readmem(ctx, R(reg1), len);
+    ulong x2 = readmem(ctx, R(reg2), len);
+
+    COMPARE(x1, x2);
+
+    if (!x1 && !x2)
+        flg &= ~ZF;
+
+    STR_MOVE(reg1, len);
+    STR_MOVE(reg2, len);
+}
+
+IMPL_START_0(cmpzsb)
+{
+    cmpzs_impl(ctx, p1, p2, 1);
+}
+IMPL_END;
+
+IMPL_START_0(cmzpsw)
+{
+    cmpzs_impl(ctx, p1, p2, 2);
+}
+IMPL_END;
+
+IMPL_START_0(cmpzsl)
+{
+    cmpzs_impl(ctx, p1, p2, 4);
+}
+IMPL_END;
+
+IMPL_START_0(cmpzsq)
+{
+    cmpzs_impl(ctx, p1, p2, 8);
+}
+IMPL_END;
+
+//----------------------------------------------------------------------------//
+
+void movs_impl(ctx_t *ctx, acc_t *p1, acc_t *p2, uint len)
+{
+    ulong reg1, reg2;
+
+    if (p2) {
+        reg1 = p1->reg;
+        reg2 = p2->reg;
+    }
+
+    else if (p1) {
+        reg1 = RDI;
+        reg2 = p1->reg;
+    }
+
+    else {
+        reg1 = RDI;
+        reg2 = RSI;
+    }
+
+    ulong x = readmem(ctx, R(reg2), len);
+    writemem(ctx, x, R(reg1), len);
+
+    flg = (x == 0 ? flg|ZF : flg&~ZF);
+
+    STR_MOVE(reg1, len);
+    STR_MOVE(reg2, len);
+}
+
 IMPL_START_0(movsb)
 {
+    movs_impl(ctx, p1, p2, 1);
+}
+IMPL_END;
+
+IMPL_START_0(movsw)
+{
+    movs_impl(ctx, p1, p2, 2);
+}
+IMPL_END;
+
+IMPL_START_0(movsl)
+{
+    movs_impl(ctx, p1, p2, 4);
+}
+IMPL_END;
+
+IMPL_START_0(movsq)
+{
+    movs_impl(ctx, p1, p2, 8);
 }
 IMPL_END;
 

vm/pc/decd.c

@@ -7,7 +7,7 @@
 // Imperatively read the "DECD" file before reading this code
 //
 
-static void check_param_type(ctx_t *ctx, uint prm, uchar fmt)
+static void check_param_type(ctx_t *ctx, instr_t *in, uint prm, uchar fmt)
 {
     bool ok;
 
@@ -22,8 +22,8 @@ static void check_param_type(ctx_t *ctx, uint prm, uchar fmt)
 
     if (!ok)
         _except(ctx, E_ILL,
-            "FT1 or FT2 not matching INSTR's expected parameter types: "
+            "FT1 or FT2 not matching %s's expected parameter types: "
-            "fmt=0x%x prm=0x%x", fmt, prm);
+            "fmt=0x%x prm=0x%x", in->full, fmt, prm);
 }
 
 void decode(ctx_t *ctx)
@@ -102,7 +102,7 @@ skip_w2:
         return;
     }
 
-    check_param_type(ctx, in->prm1, f1);
+    check_param_type(ctx, in, in->prm1, f1);
     extract_param(ctx, &p1, f1);
 
     //
@@ -121,7 +121,7 @@ skip_w2:
         return;
     }
 
-    check_param_type(ctx, in->prm2, f2);
+    check_param_type(ctx, in, in->prm2, f2);
     extract_param(ctx, &p2, f2);
 
     exec_instr(ctx, in, &p1, &p2, lock, rep, cond, pc);

vm/pc/dump.c

@@ -76,10 +76,10 @@ void dump_instr(ctx_t *ctx,
     log("0x%lX: ", pc);
 
     if (lock)
-        log("lock ");
+        log("lock");
 
     if (rep)
-        log("rep ");
+        log(" rep");
 
     if (cond)
     {
@@ -95,8 +95,12 @@ void dump_instr(ctx_t *ctx,
 
         log("%s", cond_suffixes[cond]);
     }
+    
+    if (!rep)
+        log("\t\t");
 
-    log("\t");
+    else
+        log("\t");
 
     log("%s\t", in->name);