Talk:Oni engine patches (Mac PPC)

Just tested the upgrade for 1024x1024 and Oni crashed. First of all OniSplit works fine with a 1024x1024 targa image, I can import and export.

When I imported a 1024x1024 image (delorean.tga) I got the level 4 splash screen, but then I got a black screen, I had to force quit Oni. So no crash report. Next I did the same image at 512x512, same result, except this time I got Oni's crash report. Here it is:

Process: Oni (1.0 (Mac OS X) (v1.36) (01/28/03))
exception = 0x00000001, code = 0x00063328;0x00000001;0x88428882;

5 threads:
Thread 0:
srr0: 0x9000b348  srr1: 0x0000d030   cr: 0x24022024  xer: 0x00000000  lr: 0x9000b29c ctr: 0x9000b340
r0: 0xffffffe1  r8: 0x00000000  r16: 0x00000000  r24: 0xf0182840
r1: 0xf0182770  r9: 0x00000000  r17: 0x00000000  r25: 0x00000450
r2: 0xa1b1c1d3  r10: 0x91444fa8  r18: 0x00006803  r26: 0x00006603
r3: 0xf0182840  r11: 0xa0006a28  r19: 0x00000000  r27: 0x00000000
r4: 0x03000006  r12: 0x9000b340  r20: 0x05c8046a  r28: 0x00000000
r5: 0x00000000  r13: 0x00000000  r21: 0x177348d1  r29: 0x03000006
r6: 0x00000450  r14: 0x00000000  r22: 0x0063e7b8  r30: 0x03000006
r7: 0x00006603  r15: 0x00000001  r23: 0x00000000  r31: 0x907de670
   0 -- 0x9000b348 -- _mach_msg_trap
   1 -- 0x9000b29c -- _mach_msg
   2 -- 0x907de998 -- ___CFRunLoopRun
   3 -- 0x907de29c -- _CFRunLoopRunSpecific
   4 -- 0x91458524 -- __ZN10HALRunLoop9OwnThreadEPv
   5 -- 0x914582c4 -- __ZN9CAPThread5EntryEPS_
   6 -- 0x9002bd08 -- __pthread_body

Thread 1:
srr0: 0x00063328  srr1: 0x0000f030   cr: 0x84000248  xer: 0x00000004  lr: 0x0012aec8 ctr: 0x0b127490
r0: 0x0012aec8  r8: 0x00186418  r16: 0x00000000  r24: 0x00000000
r1: 0xbffff9e0  r9: 0x88428842  r17: 0x00000000  r25: 0x00195f6d
r2: 0x0b207498  r10: 0x0b127498  r18: 0x00000000  r26: 0x00000000
r3: 0x00000000  r11: 0x0187de20  r19: 0x00000000  r27: 0x00000000
r4: 0x000003fc  r12: 0x00063300  r20: 0x00000000  r28: 0xbffffb28
r5: 0x000003fc  r13: 0x00000000  r21: 0x00000000  r29: 0x001910bc
r6: 0x0019624c  r14: 0x00000000  r22: 0x00000000  r30: 0x00218c3c
r7: 0x43300000  r15: 0x00000000  r23: 0x00000000  r31: 0x0012afb4
   0 -- 0x00063328 -- _ONrGameState_GetEnvironment
   1 -- 0x0012aec8 -- _P3iDisplayDecals
   2 -- 0x0012afc4 -- _P3rDisplayStaticDecals
   3 -- 0x00062b38 -- _ONiGameState_Display_Reflectable
   4 -- 0x00062d20 -- _ONrGameState_Display
   5 -- 0x00003d48 -- _ONiRunGame
   6 -- 0x00004610 -- _ONiMain
   7 -- 0x0000470c -- _main
   8 -- 0x00002b40 -- __start
   9 -- 0x00002970 -- start

Thread 2:
srr0: 0x90054388  srr1: 0x0200f030   cr: 0x24008244  xer: 0x00000000  lr: 0x90070be8 ctr: 0x90054380
r0: 0xffffffd9  r8: 0x91468918  r16: 0x00000000  r24: 0x00000000
r1: 0xf0284b00  r9: 0xa0001fac  r17: 0x00000000  r25: 0xa00009dc
r2: 0x00000001  r10: 0x00acc0d9  r18: 0x00000000  r26: 0x0063ff94
r3: 0x00000031  r11: 0xa0006be0  r19: 0x00000000  r27: 0x0063ffc0
r4: 0x00002f03  r12: 0x90054380  r20: 0x00000000  r28: 0xf0284bb0
r5: 0x00000000  r13: 0x00000000  r21: 0x00000000  r29: 0xa0001fac
r6: 0x00acc0d9  r14: 0x00000000  r22: 0x00000000  r30: 0xa0001fac
r7: 0xf0284d58  r15: 0x00000000  r23: 0x00000000  r31: 0x900709dc
   0 -- 0x90054388 -- _semaphore_timedwait_signal_trap
   1 -- 0x90070be8 -- _pthread_cond_timedwait_relative_np
   2 -- 0x914696ac -- __ZN7CAGuard7WaitForEy
   3 -- 0x914695bc -- __ZN7CAGuard9WaitUntilEy
   4 -- 0x91467800 -- __ZN11HP_IOThread8WorkLoopEv
   5 -- 0x91467498 -- __ZN11HP_IOThread11ThreadEntryEPS_
   6 -- 0x914582c4 -- __ZN9CAPThread5EntryEPS_
   7 -- 0x9002bd08 -- __pthread_body

Thread 3:
srr0: 0x9003288c  srr1: 0x0000d030   cr: 0x82000002  xer: 0x00000000  lr: 0x33332814 ctr: 0x90032880
r0: 0x00000007  r8: 0x00000000  r16: 0x00000000  r24: 0x00000000
r1: 0xf00807d0  r9: 0xa0010204  r17: 0x00000000  r25: 0x00000000
r2: 0x000000d9  r10: 0x90032824  r18: 0x00000000  r26: 0xf0080bec
r3: 0x00005997  r11: 0x42000008  r19: 0x00000000  r27: 0x00063328
r4: 0x00000000  r12: 0x90032880  r20: 0x00000000  r28: 0x00000001
r5: 0x00000000  r13: 0x00000000  r21: 0x00000000  r29: 0x00005997
r6: 0x00000000  r14: 0x00000000  r22: 0x00000000  r30: 0x00000002
r7: 0x00000000  r15: 0x00000000  r23: 0x00000000  r31: 0x333325c8
   0 -- 0x9003288c -- _wait4
   1 -- 0x33332814 -- _OCCHandleException
   2 -- 0x33331e34 -- _OCCExc_catch_exception_raise_state_identity
   3 -- 0x333329cc -- __Xexception_raise_state_identity
   4 -- 0x33332adc -- _OCCExc_server
   5 -- 0x33331ee0 -- +[OCCCrashCatcher(MachPrivate) _handleExceptions]
   6 -- 0x92bf6118 -- _forkThreadForFunction
   7 -- 0x9002bd08 -- __pthread_body

Thread 4:
srr0: 0x9002c3c8  srr1: 0x0000f030   cr: 0x24000084  xer: 0x00000000  lr: 0x90030eac ctr: 0x9002c3c0
r0: 0xffffffdb  r8: 0xf0101a00  r16: 0x00000000  r24: 0x00000000
r1: 0xf0101c80  r9: 0xa0001fac  r17: 0x00000000  r25: 0x00000000
r2: 0x00000001  r10: 0x90a3f628  r18: 0x00000000  r26: 0xa0000cdc
r3: 0x00003003  r11: 0xa0006bf4  r19: 0x00000000  r27: 0x00627168
r4: 0x00002d03  r12: 0x9002c3c0  r20: 0x00000000  r28: 0xa0001fac
r5: 0x000003e8  r13: 0x00000000  r21: 0x00000000  r29: 0x00627194
r6: 0xffffffff  r14: 0x00000000  r22: 0x00000000  r30: 0xa0001fac
r7: 0x000000ff  r15: 0x00000000  r23: 0x00000000  r31: 0x90030cdc
   0 -- 0x9002c3c8 -- _semaphore_wait_signal_trap
   1 -- 0x90030eac -- _pthread_cond_wait
   2 -- 0x92bfd284 -- -[NSConditionLock lockWhenCondition:]
   3 -- 0x001548a4 -- -[SoundChannelProcessor(Private) _processQueue]
   4 -- 0x0015483c -- -[SoundChannelProcessor processQueueForever:]
   5 -- 0x92bf6118 -- _forkThreadForFunction
   6 -- 0x9002bd08 -- __pthread_body

Ed

Hm, I didn't notice you had tried 512x512 before and still got a crash. Oh well... "the truth is somewhere completely different". Sorry.

geyser 11:40, 13 August 2008 (CEST)

Maybe this will be a clue, when using the new hex of 10, I was able to start level 4. Oni only crashed when the delorean (Using the new 512x512 texture) came into view. Previously, with the hex at 40, Oni crashed right after the splash screen. EdT 15:49, 13 August 2008 (CEST)

I have tested some more on my side and it turned out that the mysterious crash for 1024x1024 textures was some kind of user error (like forgetting to save or something equally stupid). So 1024x1024 works just fine on PC now, and there is thus no similarity between our cases anymore. Also, the crash right after the splashscreen (which one, BTW? intro splashscreen of a level?) seems very suspicious. So, could you please: 1) double-check that you're changing stuff at the right address; 2) let us have a look at the engine after you patched it. --geyser 21:48, 13 August 2008 (CEST)

I tried the 512 patch on a fresh unpatched version of Oni. Same result. I can start level 4, move around, but when I turn to look at the Delorean, Oni crashes. I'm using the files you provided with the new M3GM and OBAN files. However, if I change TXMPdelorean to 256x256, no crash. When I used the previous patch of 40, then Oni would crash after the splash screen for level 4. Note: in both cases, I had imported the TXMPdelorean file into level4 not level0. Here is the patched engine: http://edt.oni2.net/files/Oni512.zip EdT 01:14, 14 August 2008 (CEST)

Sorry for the late reply. Let's run a clean experiment on your side while we double-check OMNI's engine:

1. Try and import another large texture than the DeLorean's (a skybox, a poster, Master Chief, anything) and test this other large texture without involving the DeLorean at all. Make sure that you rule out everything else that may make Oni crash (e.g., an old M3GM).
2. As for the DeLorean, make sure its M3GM was imported with a recent version of OniSplit: it could be that it suffers from the same issue as the one that used to plague early TRBS import (i.e., it may be missing extra storage space at the end of the TXCA).

Let us know if it still crashes for the "clean experiment" (just a texture replaced with a large one). And oh, if it crashes, send us the TXMP, just in case.

geyser 15:33, 21 August 2008 (CEST)

hmmm, time for a stupid question, what is the option in OniSplit to create a 512x512 TXMP? I just tried to import a 512x512 texture, but when I exported that file it was 256x256. I imported these files: http://edt.oni2.net/AE_Files/TXMPAIR_FLOOR002.tga and http://edt.oni2.net/AE_Files/TXMPPOSTER1.tga for the Airport lobby floor and posterEdT 23:52, 21 August 2008 (CEST)

So you mean that all this time you were in fact letting OniSplit downsample your textures to 256x256? Then I really don't understand what about the texture may have caused the crashes you reported. The option for disabling the auto-downsampling (and thus retaining the native size of your picture) is -large, e.g.: OniSplit.exe -create:txmp TXMPcreated -format:bgr32 -large TXMPPOSTER1.tga --geyser 00:24, 22 August 2008 (CEST)

No, that's not the case. The first account with the 1024x1024, I used the "-large" option (But today I forgot about that command) With the Delorean crash I used your TXMP .oni file which was 512x512. Anyways, now I used the -large option and imported the file TXMPAIR_FLOOR002.tga as brg555. As Oni was in the process of loading level 4, it crashed. I replaced that with a 256x256 version, no crash. Next I imported TXMPPOSTER1.tga as brg32, Oni crashed as it was loading level 4, replaced with 256x256, back to normal. EdT 01:01, 22 August 2008 (CEST)

---

Hmm, about the Shift key not being 'accessible'... I actually never noticed before that one cannot type capital letters on the command line, and if I did notice it I would have assumed the PC version was the same. So, one point and one question:

• It's not exactly true that the Shift key is not accessible or not recognized, because I can type important characters like the underscore and parentheses. It's merely that uppercase letters cannot be typed.
• Is the lack of access to uppercase letters an issue? This is total news to me, so naturally I am curious as to whether this has been a problem on the Mac side of things that I never knew about. --Iritscen 03:04, 7 October 2008 (CEST)

The PC command line has no major issues: uppercase letters can be entered both on PC and PC demo. There are occasional glitches with autocompletion and some inconsistencies with multithreading (i.e., "fork"ing may not work as expected when launched from the console). Also, the command line is not displayed properly or at all at large resolutions.

Thanks for the disambiguation about the Shift key. Ed didn't make it clear how exactly it didn't work. Do you confirm that the only characters that can't be entered are uppercase letters? Also, Ed, how long have your 3 "known issues" been around? Right from the moment when the Dev Mode was unlocked?

The original variables and functions (preset of scripted) are all lowercase, but team names are capitalized, and many AI names contain uppercase letters as well. So lowercase-only input can be frustrating. The obvious workaround is to use scripts more, so that all command line statements can be lowercase.

geyser 03:36, 7 October 2008 (CEST)

Just confirmed it to make sure; only a-z cannot be modified by the Shift key, i.e., become uppercase. All other characters that require Shift can be typed. And thanks for the answer on what uses uppercase, the team names and AI names slipped my mind. I tested that as well, hoping that the Mac would ignore case, but alas; the console indeed does not recognize any team names without the uppercase letters. --Iritscen 04:02, 7 October 2008 (CEST)

Iritscen, you beat me to it. The SHIFT key does not work in Dev Mode to enter uppercase letters (I should have been more specific, sorry). How long have I've known these issues? The SHIFT key problem was discovered when you, (geyser) asked me to enter some code in the console, I forgot when that happened. The Window mode issue was known as soon as it was figured out how to run Omni group's version in a window. I'm sure this was mentioned at some point, either here or at OCF. The problems with the Numpad key was known back in Jan 2007 and you suggested using the key_config.txt file EdT 04:11, 7 October 2008 (CEST)

---

Update on large textures

Sorry, but we still haven't got a clear idea of what's going on. Ed's patch is correct, yet the crash log indicates that texture data was written on top of other stuff (i.e. outside of its supposed buffer). This probably means there's another texture buffer somewhere, specific to the Mac build. And we haven't tried looking for it yet. --geyser 04:53, 9 October 2008 (CEST)

Update on projectile awareness

Sorry, but browsing and patching PowerPC ASM code is not pleasant. There are tricky pointers all over the place, which is why Neo is not too hot for fixing the messy awareness on the Mac (even though the nature of the mess is essentially known). --geyser 04:53, 9 October 2008 (CEST)

"Finally, some good news"

Here's a tentative set of in-place patches to enable chr_focus, chr_set_class and chr_location for the Omni engine. A bunch of pointers and strings to edit. I wrote the strings with a fixed-width font, ° standing for the null character, with the old and new strings above one another so you can see what goes where. Make sure you don't mess up with the 2 strings being shifted backwards. --geyser 04:53, 9 October 2008 (CEST)

making cm_barabus into chr_focus: fixing pointers

Offset	Old hex	New hex	Description
0x1B906C	00 16 F8 34	00 16 F8 34	pointer to BSL name; unchanged
0x1B9078	00 16 F8 40	00 16 F8 40	pointer to help string; unchanged
0x1B907C	00 16 F8 5C	00 16 F8 5C	pointer to signature string; unchanged
0x1B9080	00 03 58 BC	00 04 19 6C	pointer to function; now pointing to _SetMainCharacter

making chr_hold_key into chr_location: fixing pointers

Offset	Old hex	New hex	Description
0x1B918C	00 16 EE 24	00 16 EE 23	pointer to BSL name; shifting 1 char backwards
0x1B9198	00 16 FB E8	00 16 FB E8	pointer to help string; unchanged
0x1B919C	00 16 FC 20	00 16 FC 20	pointer to signature string; unchanged
0x1B91A0	00 03 45 A8	00 04 17 20	pointer to function; now pointing to _iSetAnyCharacterLocation

making ai2_panic into chr_set_class: fixing pointers

Offset	Old hex	New hex	Description
0x1B95DC	00 16 F2 EC	00 16 F2 EA	pointer to BSL name; shifting 2 chars backwards
0x1B95E8	00 17 09 48	00 17 09 48	pointer to help string; unchanged
0x1B95EC	00 17 09 68	00 17 09 68	pointer to signature string; unchanged
0x1B95F0	00 03 55 F8	00 04 06 14	pointer to function; now pointing to _iSetCharacterClass

making cm_barabus into chr_focus: fixing strings

Offset	0x16E834	0x16E840	0x16E85C
Old chars	cm_barabus°°	special camera for barabus°°	[ai_name:string | script_id:int] away:float up:float time:int°°°
New chars	chr_focus°°°	sets a new player character°	chr_index:int°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°
Description	replacing name in-place	replacing help string in-place	replacing signature string in-place

making chr_hold_key into chr_location: fixing strings

Offset	0x16DE23	0x16EBE8	0x16EC20
Old chars	°chr_holdkey°	forces a character to hold a key down for some frames°°°	[ai_name:string | script_id:int] key_name:string num_frames:int°
New chars	chr_location°	sets location of a character or prints current location°	chr_index:int [[loc_x:float loc_y:float loc_z:float] | ]°°°°°°°°
Description	replacing/shifting name	replacing help string in-place	replacing signature string in-place

making ai2_panic into chr_set_class: fixing strings

Offset	0x16E2EA	0x16F948	0x16F968
Old chars	°°ai2_panic°°°	makes an AI panic or not panic°°	[ai_name:string | script_id:int] timer:int°°
New chars	chr_set_class°	sets class of a given character°	chr_id:int [oncc_name:string | oncc_id:int]°
Description	replacing/shifting name	replacing help string in-place	replacing signature string in-place

Thank you for the patches! chr_set_class works!!! chr_focus works!!! chr_location crashes, I included the report on a separate page. Here is the modified engine: http://edt.oni2.net/files/OniTest.zip for your examination. EdT 07:26, 9 October 2008 (CEST)

Updated the pointer-to-function for chr_location, should work now. Will add weapon_reset later today. Thanks for the crash report (but you should have linked to it ;) ). And I have no idea how you managed to mess up chr_focus the way you did ^_^ --geyser 17:26, 9 October 2008 (CEST)

How did I mess up chr_focus? When I first tried it, I was able to focus on the new character, but couldn't move him, then I read ssg's script command page and realized I needed the command ai2_takecontrol 1, so once I tried the script on that page, I realized chr_focus worked. EdT 23:35, 9 October 2008 (CEST)

---

OK, Ed, here's a last one that should make your Oni 100% PC-OTA-compatible (I don't remember any other important command that you're missing, will check). Looking forward to massive OTA porting and mission scripting now ;) --geyser 22:14, 9 October 2008 (CEST)

I settled on ai2_skill_save even though it works OK, because we're running short of commands, so I'd rather use something that I know is a mere commodity than something potentially useful that we haven't figured out yet. --geyser 22:14, 9 October 2008 (CEST)

making ai2_skill_save into weapon_reset: fixing pointers

Offset	Old hex	New hex	Description
0x1B97A4	00 17 0F 10	00 17 0F 10	pointer to BSL name; unchanged
0x1B97B0	00 17 0F 20	00 17 0F 20	pointer to help string; unchanged
0x1B97B4	00 16 A1 48	00 16 A1 48	pointer to signature string; unchanged
0x1B97B8	00 03 0D 90	00 0A 56 38	pointer to function; now pointing to _WPiWeapons_Reset

making ai2_skill_save into weapon_reset: fixing strings

Offset	0x16FF10	0x16FF20	0x169148
Old chars	ai2_skill_save°°	saves the shooting skill being edited out as a text file°°°°	°°°°
New chars	weapon_reset°°°°	deletes unheld weapons and spawns initially present weapons°	°°°°
Description	replacing name in-place	replacing help string in-place	empty signature string; unchanged

geyser, weapon_reset works. I did a quick test of OTA and it runs fine, but its not 100% compatible, since I cannot access the menus, using the function keys. Going to Dev Mode and pressing F9 starts the recording, but the OTA menus do not appear, pressing F3 does not cycle arena. EdT 23:35, 9 October 2008 (CEST)

The Mac version doesn't have F2/F3 binding. Another patch, maybe? :) Gumby 00:15, 10 October 2008 (CEST)