Short: Safely?! remove sytem patches (sample code) Author: llucius@millcomm.com (L.Lucius) Uploader: llucius millcomm com (L Lucius) Type: dev/src Version: 1.2 Architecture: m68k-amigaos Safe?! Patch Removal By Leland Lucius v1.2 IMPORTANT!!!! ============= The code contained in this archive is only an example. It is not complete and has never been executed (it does compile though). Additional error checking should be added and customized to fit an individual's needs. WHAT IS THIS? ============= If you frequent the c.s.a.p newsgroup, you have probably noticed that the subject of safely removing AND freeing memory occupied by a system patch comes up quite a bit. And that there are some that believe it to be impossible, while others (like me) keep thinking that there just has to be a way to do it. Well, I THINK I've found THAT way (actually two). The first, found in files "remove1.c" and "patch1.asm", is an implementation that is suitable for frequently called routines since it does not use any additional CPU to track callers. It however CAN'T call any other routines outside of the patch (see code). The second, found in files "remove2.c" and "patch2.asm", is only slightly different than the first. It keeps track of the callers of the routine and thus uses slightly more CPU, but it CAN call any other routines. Both implementations SHOULD be callable from interrupts. Both implementations allow for memory deallocation. Both implementations are position independent. NOTICE ====== This code is released to the public domain. Hell, I'm not even sure if it'll work! B-) DISCUSSION? =========== If you find that the code isn't safe (or is), let's discuss it on c.s.a.p so that everyone interested can remark. Besides another thread about patch removal is inevitable anyway, right? B-) Leland CHANGES ======= Corrected the vector checks. Changed to just free the patch if there are no current users. Corrected patch exit during removal.