Emulator Cheats

FYI, Aaron forgot to add external submissions in u4. hence both the new searching feature and Pugsy's enhancements are expected to be included in u5

I have added the function to look for byte, word, dword and qword signed and unsigned.

http://www.megaupload.com/?d=JKTBDQ57

Great stuff, works a treat

WOW ! I was momentanely out of the scene ! GREAT NEW ! Is there someone here able to explain how to activate/use it under the mame debug ? Thank you !!

Start mame with -debug command (or change debug to 1 in mame.ini), start to play, when you want initialize the cheat go in debugger and write "ci" (for more information write "help cheatinit"), now all memory of first CPU is initialized, continue to play, when want do a search go in debugger and write "cn <condition>" (for more information write "help cheatnext"). For view the current result write "cl" or "cl <filename>" if you want save the result to a file.
For other information write "help cheats" for view all available options.

Thank you very much ronco ! Ma sei italiano ?

Do you think it can be implemented inside the mame ingame menu (TAB) ?

Of course, I am Italian.

I don't believe is possible to make a menu like the old version, because this function is inside the debugger and if MAME is loaded without the debugger this function is not available.

ENG - So why don't add a suitable menu in mame debugger insted of using the command line ?

ITA - Allora perche'non aggiungere un comodo menu direttamente nel mame debugger invece di utilizzare la linea di comando ?

In the past I used to think a menu driven cheat facility was essential but now I'm leaning to just the current debugger system. If you use it enough you will find that it's actually quicker than using a MAME menu based cheat search facility. Just write down the 'short' commands on a piece of paper in front of your monitor/keyboard and you will quickly get used to it.

Anyway, there would be problems getting any cheat finding/editing functions into MAME as the cheating search/editing menus were always considered a hack of the UI system. There are still features of the old cheat engine missing like being able to easily 'watch' any cheat results and adding the results to the cheat menus automatically + other stuff but for now I'm actually really happy with what we have.

On a side node any chance of adding some macroed cheatinit areas so for instance in the gba driver in MESS, 'cheatinit gba'/'cheatinit XX gba' would automatically cheat init the two RAM areas of 02000000-0203FFFF & 03000000-03007FFF because at the moment 'cheatinit' can't initialise alone (too big) so you have to do a specific cheatinit followed by a cheatrange. The macro capability could be used for MAME stuff too...so in the PSX based games 'ci psx' would just search the 80xxxxxx range rather than the mirrors as well.

@asper
I think that Pugsy has reason and however would not be approved from the mamedevs.

@Pugsy
You are right, it needs to look a system for remove the mirrors from the search. For the gba problem the fast solution is make a file with the desired command (cheatinit + cheatrange) saving the file with name 'gba' and load with the command "source gba".

s.ronco wrote:@Pugsy
You are right, it needs to look a system for remove the mirrors from the search. For the gba problem the fast solution is make a file with the desired command (cheatinit + cheatrange) saving the file with name 'gba' and load with the command "source gba".

Thank you that works a charm, the 'source' command slipped completely under my radar ...I used used something similar in the past but it only ran the scripts as the debugger loaded the first time.

This is a diff for remove the mirrors from the search, before submit to mamedev, I think is better make some test, I have tested only on gba and psx.

http://www.megaupload.com/?d=EERQ79X3

Excellent, the benefits outweigh the disadvantages as ci is a LOT faster for games with massive memory maps.

Only run into one problem so far.

'ci' in mace now gives:-

No writeable bytes found in this area

Probably the problem is not in the cheat system but is in the address translation of the mips CPU.

Can you see the RAM (0x00000000 - 0x007fffff) of mace in the memory window?

The same problem exists in mk/mk2/mk3.

Yes I can see the asterisks in the memory edit/view window for those addresses. There have been issues in the past with how the cheat engine accesses the memory map so I'm not that surprised it's raised it's head again.


Possible fix, If a plain 'ci' returns "No writeable bytes found in this area" can you put a check in the code so that rather than display the message it effectively does a ci BOTTOM,TOP instead, that way we get best of both worlds.



EDIT: I don't know where your post went - it had gone by the time I had hit the submit button, hope it wasn't me?!!

Don't worry I have tried to edit and have accidentally delete it. (I need to sleep)
However I think to have resolved, the problem is the address translation inside the debugger, not in the CPU, now all seem work.

http://www.megaupload.com/?d=GV08BVV3

After some test I have noted that the result in seattle driver is not the logical address, but is the physical address this is a problem because it needs to use the raw patch.
For now the best way for search cheats on seattle.c is force the correct range ("ci xx, 0x80000000, 0x7fffff"). I have submitted this change to Aaron and I have asked for some clarification of this situation, I hope can help me.


http://www.megaupload.com/?d=AVT61Q09

Thanks, the search in the mk series now works well.

I really appreciate your fantastic work, you made a polished cheat search engine in next to no time when other mamedevs have appeared to ignore the lack of one for well over a year. BRAVO!

Aaron has fixed the problem making all cheats search in physical addresses and adding the prefixes 'p' (physical) and 'l' (logical) to specify the type of address. I think this is not mandatory but only a specification for remove some ambiguous situations.

Example:

ppb@1000 = physical program space byte at address 0x1000
ldw@2000 = logical data space word at address 0x2000

I hope it is non-mandatory...otherwise people will complain when every single cheat stops working It wouldn't be that hard for me to change my conversion util to fix..but would would still require a fair bit of manual work for all hard coded xml cheats I have in my cheat file. Progress is good though and if I do have to change things in the cheat files so be it.

s.ronco wrote:Aaron has fixed the problem making all cheats search in physical addresses and adding the prefixes 'p' (physical) and 'l' (logical) to specify the type of address. I think this is not mandatory but only a specification for remove some ambiguous situations.

Example:

ppb@1000 = physical program space byte at address 0x1000
ldw@2000 = logical data space word at address 0x2000

it's not mandatory. if no l/p is provided, MAME uses the default value, i.e. 'l'

no need to change all the cheats