check if carry flag is set
Using inline assembler [gcc, intel, c], how to check if the carry flag is set after an o开发者_开发百科peration?
sbb %eax,%eax
will store -1 in eax if the carry flag is set, 0 if it is clear. There's no need to pre-clear eax to 0; subtracting eax from itself does that for you. This technique can be very powerful since you can use the result as a bitmask to modify the results of computations in place of using conditional jumps.
You should be aware that it is only valid to test the carry flag if it was set by arithmetic performed INSIDE the inline asm block. You can't test carry of a computation that was performed in C code because there are all sorts of ways the compiler could optimize/reorder things that would clobber the carry flag.
With conditional jumps jc
(jump if carry) or jnc
(jump if not carry).
Or you can store the carry flag,
;; Intel syntax
mov eax, 0
adc eax, 0 ; add with carry
However the x86 assembler hes dedicated fast ALU flag test instructions named SETcc where the cc is desired ALU flag. So you can write:
setc AL //will set AL register to 1 or clear to 0 depend on carry flag
or
setc byte ptr [edx] //will set memory byte on location edx depend on carry flag
or even
setc byte ptr [CarryFlagTestByte] //will set memory variable on location CarryFlagTestByte depend on carry flag
With SETcc instruction you can test flags like carry, zero, sign, overflow or parity, some SETcc instructions allow to test two flags at once.
EDIT: Added simple test made in Delphi to disappear a doubt about term fast
procedure TfrmTest.ButtonTestClick(Sender: TObject);
function GetCPUTimeStamp: int64;
asm
rdtsc
end;
var
ii, i: int64;
begin
i := GetCPUTimeStamp;
asm
mov ecx, 1000000
@repeat:
mov al, 0
adc al, 0
mov al, 0
adc al, 0
mov al, 0
adc al, 0
mov al, 0
adc al, 0
loop @repeat
end;
i := GetCPUTimeStamp - i;
ii := GetCPUTimeStamp;
asm
mov ecx, 1000000
@repeat:
setc al
setc al
setc al
setc al
loop @repeat
end;
ii := GetCPUTimeStamp - ii;
caption := IntToStr(i) + ' ' + IntToStr(ii));
end;
The loop (1M iterations) wich using instruction setc is more than 5 times faster than loop with adc instriuction.
EDIT: Added second test which test result stored in register AL comulative in register CL to be more realistic case.
procedure TfrmTestOtlContainers.Button1Click(Sender: TObject);
function GetCPUTimeStamp: int64;
asm
rdtsc
end;
var
ii, i: int64;
begin
i := GetCPUTimeStamp;
asm
xor ecx, ecx
mov edx, $AAAAAAAA
shl edx, 1
mov al, 0
adc al, 0
add cl, al
shl edx, 1
mov al, 0
adc al, 0
add cl, al
shl edx, 1
mov al, 0
adc al, 0
add cl, al
shl edx, 1
mov al, 0
adc al, 0
add cl, al
shl edx, 1
mov al, 0
adc al, 0
add cl, al
shl edx, 1
mov al, 0
adc al, 0
add cl, al
shl edx, 1
mov al, 0
adc al, 0
add cl, al
shl edx, 1
mov al, 0
adc al, 0
add cl, al
end;
i := GetCPUTimeStamp - i;
ii := GetCPUTimeStamp;
asm
xor ecx, ecx
mov edx, $AAAAAAAA
shl edx, 1
setc al
add cl, al
shl edx, 1
setc al
add cl, al
shl edx, 1
setc al
add cl, al
shl edx, 1
setc al
add cl, al
shl edx, 1
setc al
add cl, al
shl edx, 1
setc al
add cl, al
shl edx, 1
setc al
add cl, al
shl edx, 1
setc al
add cl, al
end;
ii := GetCPUTimeStamp - ii;
caption := IntToStr(i) + ' ' + IntToStr(ii);
end;
Rutine part with SETcc instruction is still faster for about 20%.
The first function performs unsigned addition and then tests for overflow using the carry flag (CF). The volatile's must remain. Otherwise the optimizer will rearrange instructions, which pretty much ensures an incorrect result. I've seen the optimizer change the jnc
to a jae
(which is also based on CF).
/* Performs r = a + b, returns 1 if the result is safe (no overflow), 0 otherwise */
int add_u32(uint32_t a, uint32_t b, uint32_t* r)
{
volatile int no_carry = 1;
volatile uint32_t result = a + b;
asm volatile
(
"jnc 1f ;"
"movl $0, %[xc] ;"
"1: ;"
: [xc] "=m" (no_carry)
);
if(r)
*r = result;
return no_carry;
}
The next function is for the signed ints. Same use of volatile applies. Note that signed integer math jumps on OF flag via jno
. I've seen the optimizer change this to a jnb
(which is also based on OF).
/* Performs r = a + b, returns 1 if the result is safe (no overflow), 0 otherwise */
int add_i32(int32_t a, int32_t b, int32_t* r)
{
volatile int no_overflow = 1;
volatile int32_t result = a + b;
asm volatile
(
"jno 1f ;"
"movl $0, %[xo] ;"
"1: ;"
: [xo] "=m" (no_overflow)
);
if(r)
*r = result;
return no_overflow;
}
In the big picture, you might use the functions as follows. In the same big picture, many folks will probably reject the extra work and aesthetic non-beauty until pwn'd by an overflow/wrap/underflow
int r, a, b;
...
if(!add_i32(a, b, &r))
abort(); // Integer overflow!!!
...
The inline GCC assembly is available in GCC 3.1 and above. See Assembler Instructions with C Expression Operands, or search for 'GCC Extended Assembly'.
Finally, the same in Visual Studio would be as follows (not much difference in code generation), but syntax is much easier since MASM allows you to jump to a C label:
/* Performs r = a + b, returns 1 if the result is safe (no overflow), 0 otherwise */
int add_i32(__int32 a, __int32 b, __int32* r)
{
volatile int no_overflow = 1;
volatile __int32 result = a + b;
__asm
{
jno NO_OVERFLOW;
mov no_overflow, 0;
NO_OVERFLOW:
}
if(r)
*r = result;
return no_overflow;
}
On the bad side, the above MASM code is only applicable for x86 assembly. For x64 assembly, there is no inlining so you will have to code it up in assembly (in a separate file) and use use MASM64 to compile.
This may give an idea or solution if it's correct. I struggled with testing for wrap around until I found out about in-line assembly. I tried to test with various edge values and seems to work correctly. Program takes input from cmdln and converts it to integer and outputs hex and binary values.
gcc version 11.2.1
$> gcc -Wall -std=c99 -O2 -o uilt uilt.c
snippet:
size_t i = 0;
int mul = 10;
uint128_t sum = 0;
int int_array[48] = {0};
// fill arr. with ea. str val in argv[1] str. converted to int vals.
while (i < strlen(argv[1])) {
// chk they are digit chars, if not, skip iter
if (isdigit(argv[1][i]) == 0) {
i++;
continue;
}
int_array[i] = (argv[1][i] - 48);
sum = int_array[i] + (sum * mul);
/* check carry flag */
__asm__ goto("jc %l0"
: /* no outputs */
: /* no inputs */
: /* no clobbers */
: carry);
/* no carry */
goto its_good;
carry:
system("clear");
printf("\n\n\tERROR!!!\
\n\n\t!!!!!!! uilt has ABORTED !!!!!!\
\n\tCmdln arg exceeds 2^127 bit limit\
\n\twhen converted from string to 127\
\n\tbit unsigned __int128.\n\n");
exit(1);
its_good:
i++;
}
some output:
[jim@nitroII uiltDev]$ ./uilt 1
Dec: 1
Hex: 0x0001
Bin: 0x0001
[jim@nitroII uiltDev]$ ./uilt 255
Dec: 255
Hex: 0x00ff
Bin: 0x0000 1111 1111
[jim@nitroII uiltDev]$ ./uilt 18446744073709551616
Dec: 18446744073709551616
Hex: 0x0001 0000 0000 0000 0000
Bin: 0x0001 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
[jim@nitroII uiltDev]$ ./uilt 340282366920938463463374607431768211455
Dec: 340282366920938463463374607431768211455
Hex: 0x0000 ffff ffff ffff ffff ffff ffff ffff ffff
Bin: 0x0000 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111
Dec: 340282366920938463463374607431768211456
ERROR!!!
!!!!!!! uilt has ABORTED !!!!!!
Cmdln arg exceeds 2^127 bit limit
when converted from string to 127
bit unsigned __int128.
精彩评论