Hi everybody,
after hours I ve found a bug in TYA's opcode processing
causing some heavy headache here....
The bugfix will appear in TYA's next release I am working on:
to run both with jdk1.1.x and jdk1.2 (will be TYA v1.3).
The bug produces floating point unit overflows in certain
situations - OTOH FP modulo is not so often used.
(I've found in calculation of textures dealing the
new Java2D API.)
Unfortunately I've not the last TYA release 1.2v4 (@gonzalez)
at hand, so there's not yet a regular patch.
If you won't like to wait until releae date, please
edit tyarecode.c and replace the ``case 0x72:'' and ``case 0x73:''
sections ( part of the big outer switch statement )
with next lines following:
#define FXCH 0xC9D9 /* fxch */
#define FSTP_ST0 0xD8DD /* fstp st(0) */
case 0x72: // frem
CB(PUSHAX); // FIXAB
CW(FLDW_MSP);CB(SP_ADDR_BYTE3);
CB(POPAX);
CW(FLDW_MSP);CB(SP_ADDR_BYTE3);
CW(FPREM);
CW(FXCH);
CW(FSTP_ST0);
CW(FSTPW_MSP);CB(SP_ADDR_BYTE3);
CB(POPAX); // FIXAB
break;
case 0x73: // drem
CB(PUSHAX); // FIXAB
CW(FLDQW_MSP);CB(SP_ADDR_BYTE3);
CB(POPAX);
CB(POPAX);
CW(FLDQW_MSP);CB(SP_ADDR_BYTE3);
CW(FPREM);
CW(FXCH);
CW(FSTP_ST0);
CW(FSTPQW_MSP);CB(SP_ADDR_BYTE3);
CB(POPAX); // FIXAB
break;
Cheers,
Albrecht
----------------------------------------------------------------------
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
