www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - asm: ja and jae mixed up?

reply =?UTF-8?B?VGhvbWFzIEvDvGhuZQ==?= writes:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Instruction - Description
JA	- Jump if above (CF=0 and ZF=0).
JAE 	- Jump if above or equal (CF=0).

Am I missing something or should the values for test_jae be
(test_ja - 1)?

int test_ja(int i){
	asm{
		mov EAX, 0;

	loop_start:
		add EAX, 1;
		cmp i, EAX;
		ja loop_start;

		mov i, EAX;
	}
	return i;
}

int test_jae(int i){
	asm{
		mov EAX, 0;

	loop_start:
		add EAX, 1;
		cmp i, EAX;
		jae loop_start;

		mov i, EAX;
	}
	return i;
}

int main(){
	for(int i=0; i<5; i++){
		printf("[%i] ja:%i - jae:%i\n", i,
			test_ja(i), test_jae(i));
	}
	return 0;
}

Output:
[1] ja:1 - jae:2
[2] ja:2 - jae:3
[3] ja:3 - jae:4
[4] ja:4 - jae:5

Expected Output:
[1] ja:2 - jae:1
[2] ja:3 - jae:2
[3] ja:4 - jae:3
[4] ja:5 - jae:4

Thomas
-----BEGIN PGP SIGNATURE-----

iD8DBQFC/85z3w+/yD4P9tIRAhiUAKCJYfdnSFhabu94P9RxWVphK0mfbACfSeBT
LiPZDlCebBXii6MBAo771d0=
=NuUY
-----END PGP SIGNATURE-----
Aug 14 2005
parent reply Dave <Dave_member pathlink.com> writes:
In article <ddoimh$2m8n$1 digitaldaemon.com>, =?UTF-8?B?VGhvbWFzIEvDvGhuZQ==?=
says...

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Instruction - Description
JA	- Jump if above (CF=0 and ZF=0).
JAE 	- Jump if above or equal (CF=0).

Am I missing something or should the values for test_jae be
(test_ja - 1)?


It looks like it is working correctly, as:

cmp i, EAX;
ja loop_start; // Branch if i > EAX

cmp i, EAX;
jae loop_start; // Branch if i >= EAX

- Dave


int test_ja(int i){
	asm{
		mov EAX, 0;

	loop_start:
		add EAX, 1;
		cmp i, EAX;
		ja loop_start;

		mov i, EAX;
	}
	return i;
}

int test_jae(int i){
	asm{
		mov EAX, 0;

	loop_start:
		add EAX, 1;
		cmp i, EAX;
		jae loop_start;

		mov i, EAX;
	}
	return i;
}

int main(){
	for(int i=0; i<5; i++){
		printf("[%i] ja:%i - jae:%i\n", i,
			test_ja(i), test_jae(i));
	}
	return 0;
}

Output:
[1] ja:1 - jae:2
[2] ja:2 - jae:3
[3] ja:3 - jae:4
[4] ja:4 - jae:5

Expected Output:
[1] ja:2 - jae:1
[2] ja:3 - jae:2
[3] ja:4 - jae:3
[4] ja:5 - jae:4

Thomas
-----BEGIN PGP SIGNATURE-----

iD8DBQFC/85z3w+/yD4P9tIRAhiUAKCJYfdnSFhabu94P9RxWVphK0mfbACfSeBT
LiPZDlCebBXii6MBAo771d0=
=NuUY
-----END PGP SIGNATURE-----
Aug 14 2005
next sibling parent Dave <Dave_member pathlink.com> writes:
In article <ddp4gv$1gf$1 digitaldaemon.com>, Dave says...

In article <ddoimh$2m8n$1 digitaldaemon.com>, =?UTF-8?B?VGhvbWFzIEvDvGhuZQ==?=
says...

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Instruction - Description
JA	- Jump if above (CF=0 and ZF=0).
JAE 	- Jump if above or equal (CF=0).

Am I missing something or should the values for test_jae be
(test_ja - 1)?


It looks like it is working correctly, as:

cmp i, EAX;
ja loop_start; // Branch if i > EAX

cmp i, EAX;
jae loop_start; // Branch if i >= EAX

- Dave


Oh yea - jg or jge would be better to use because they are for signed operands
(ja, jae are for unsigned).
Aug 14 2005
prev sibling parent reply =?UTF-8?B?VGhvbWFzIEvDvGhuZQ==?= writes:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dave schrieb:

 In article <ddoimh$2m8n$1 digitaldaemon.com>, =?UTF-8?B?VGhvbWFzIEvDvGhuZQ==?=
 says...
 
 Instruction - Description
 JA	- Jump if above (CF=0 and ZF=0).
 JAE 	- Jump if above or equal (CF=0).
 
 Am I missing something or should the values for test_jae be
 (test_ja - 1)?
 
 
 

 It looks like it is working correctly, as:


 

 cmp i, EAX;
 ja loop_start; // Branch if i > EAX


 

 cmp i, EAX;
 jae loop_start; // Branch if i >= EAX




A rewrite of the test code:

=====
uint asm_ja(uint i){
	asm{
		mov EAX, 0;

	loop_start:
		add EAX, 1;
		cmp i, EAX;
		ja loop_start;

		mov i, EAX;
	}
	return i;
}

uint d1_ja(uint i){
	uint x=0;
	
	do{
		x+=1;
	}while(!(x>i))
	
	return x;
}

uint d2_ja(uint i){
	uint x=0;
	
	do{
		x+=1;
	}while(!(i>x))
	
	return x;
}

int main(){
	for(uint i=2; i<10; i++){
		printf("[%i]\td1:%i\td2:%i\tasm:%i\n",
			i, d1_ja(i), d2_ja(i), asm_ja(i));
	}
	return 0;
}
=====

Output:
[2]     d1:3    d2:1    asm:2
[3]     d1:4    d2:1    asm:3
[4]     d1:5    d2:1    asm:4
[5]     d1:6    d2:1    asm:5
[6]     d1:7    d2:1    asm:6
[7]     d1:8    d2:1    asm:7
[8]     d1:9    d2:1    asm:8
[9]     d1:10   d2:1    asm:9

Neither d1 - "do ... while(!(x>i))" - nor
d2 - "do ... while(!(i>x))" - yields the same results as asm's
"ja loop_start;".

Am I a bit dense?

Thomas
-----BEGIN PGP SIGNATURE-----

iD8DBQFDADUV3w+/yD4P9tIRAnBbAJ9eb/v76zOA/mq+M4pBCvMom2y1HgCgx6QR
ySH/zhWPSxTf1Arh3h4fWUA=
=EUxF
-----END PGP SIGNATURE-----
Aug 14 2005
parent reply Thomas Kühne <Thomas_member pathlink.com> writes:
In article <ddpcbj$7if$1 digitaldaemon.com>, =?UTF-8?B?VGhvbWFzIEvDvGhuZQ==?=
says...


*smashing head against the wall*

never mind 



 Aug 15 2005



  parent  "Walter" <newshound digitalmars.com>  writes:


"Thomas Kühne" <Thomas_member pathlink.com> wrote in message
news:ddpsop$net$1 digitaldaemon.com...

 In article <ddpcbj$7if$1 digitaldaemon.com>,


=?UTF-8?B?VGhvbWFzIEvDvGhuZQ==?=

 says...

 *smashing head against the wall*

 never mind


We all have days like that <g>

P.S. An easier way to see if the code gen is correct for inline assembler is
to run:
    obj2asm -o foo.obj
on it, and compare the opcode with the Intel manuals (which can be
downloaded from www.intel.com).



 Aug 15 2005