129 replies to this topic

[the bigg]

WeiDU on Linux turns all file system calls to lowercase; hence, a single upper case character in the given path will fail. I don't think that dots will cause issues, however.

Edited by the bigg, 30 January 2011 - 08:49 AM.

[Turambar]

I've read that the BGT instalation on linux requires running tis2bg2 over wine.
I've seen, though, on the iesdp, that the .tis format is the same for all IE games: what does tis2bg2 really do? Couldn't the .tis files just be --get-biff_ed from bg1, and --make-biffed into bg2 just afterwards?

edit: added attachments for asc64 ONLY
new tp2 (test 4 last version):
new unix script: ; thanks to Anomaly for his help
debug from last reinstallation, if you need it:
Italian tra file:

Edited by Turambar, 04 January 2012 - 03:18 AM.

[the bigg]

IIRC, there's some difference in how overlays are handled between the engines; without tis2bg2, water becomes green when you enable 3d acceleration. After five minutes on Google, I found the source code to tis2bg2. From that archive, you should simply edit tis2bg2.cpp until you can get it to compile and have the same output on some sample files as running tis2bg2.exe via wine.

[Turambar]

IIRC, there's some difference in how overlays are handled between the engines; without tis2bg2, water becomes green when you enable 3d acceleration. After five minutes on Google, I found the source code to tis2bg2. From that archive, you should simply edit tis2bg2.cpp until you can get it to compile and have the same output on some sample files as running tis2bg2.exe via wine.

Thanks. Unfortunately, I don't know C, I don't nderstand what has to be included, what libs I need, ...
Is there anyone who can tell me how to compile it, or compile it on linux and attach it if it's possible?

[the bigg]

Done. You might want to try it on one file (or ten) and check that its output is the same as the Windows version via Wine before running a BGT setup (since I just toyed a bit with the #ifdefs, chances are that it was compiled as 64 bit or to use the wrong endianness).

Attached Files

•  tis2bg2_linux.zip   14.37K   416 downloads

[Turambar]

It fails due to buffer overflow:

Spoiler

*** buffer overflow detected ***: ./tis2bg2 terminated
======= Backtrace: =========
/lib32/libc.so.6(__fortify_fail+0x50)[0xf75aaa80]
/lib32/libc.so.6(+0xe39ba)[0xf75a99ba]
/lib32/libc.so.6(__strcpy_chk+0x44)[0xf75a8d34]
./tis2bg2[0x804a22e]
/lib32/libc.so.6(__libc_start_main+0xe7)[0xf74dcce7]
./tis2bg2[0x8048a41]
======= Memory map: ========
08048000-0804c000 r-xp 00000000 08:15 2104417 /home/francesco/Giochi/tistry/tis2bg2
0804c000-0804d000 r--p 00003000 08:15 2104417 /home/francesco/Giochi/tistry/tis2bg2
0804d000-0804e000 rw-p 00004000 08:15 2104417 /home/francesco/Giochi/tistry/tis2bg2
0951b000-0953c000 rw-p 00000000 00:00 0 [heap]
f74c4000-f74c6000 rw-p 00000000 00:00 0
f74c6000-f761a000 r-xp 00000000 08:15 657165 /lib32/libc-2.12.1.so
f761a000-f761b000 ---p 00154000 08:15 657165 /lib32/libc-2.12.1.so
f761b000-f761d000 r--p 00154000 08:15 657165 /lib32/libc-2.12.1.so
f761d000-f761e000 rw-p 00156000 08:15 657165 /lib32/libc-2.12.1.so
f761e000-f7621000 rw-p 00000000 00:00 0
f7621000-f763b000 r-xp 00000000 08:15 530922 /usr/lib32/libgcc_s.so.1
f763b000-f763c000 r--p 00019000 08:15 530922 /usr/lib32/libgcc_s.so.1
f763c000-f763d000 rw-p 0001a000 08:15 530922 /usr/lib32/libgcc_s.so.1
f763d000-f763e000 rw-p 00000000 00:00 0
f763e000-f7662000 r-xp 00000000 08:15 659079 /lib32/libm-2.12.1.so
f7662000-f7663000 r--p 00023000 08:15 659079 /lib32/libm-2.12.1.so
f7663000-f7664000 rw-p 00024000 08:15 659079 /lib32/libm-2.12.1.so
f7664000-f7743000 r-xp 00000000 08:15 530541 /usr/lib32/libstdc++.so.6.0.14
f7743000-f7747000 r--p 000de000 08:15 530541 /usr/lib32/libstdc++.so.6.0.14
f7747000-f7748000 rw-p 000e2000 08:15 530541 /usr/lib32/libstdc++.so.6.0.14
f7748000-f774f000 rw-p 00000000 00:00 0
f7772000-f7774000 rw-p 00000000 00:00 0
f7774000-f7775000 r-xp 00000000 00:00 0 [vdso]
f7775000-f7791000 r-xp 00000000 08:15 659366 /lib32/ld-2.12.1.so
f7791000-f7792000 r--p 0001b000 08:15 659366 /lib32/ld-2.12.1.so
f7792000-f7793000 rw-p 0001c000 08:15 659366 /lib32/ld-2.12.1.so
ff8e5000-ff906000 rw-p 00000000 00:00 0 [stack]
Annullato

My machine is 64bit, might that be the problem? It should have most 32-bit-compatibility libs, though. I might try on my laptop (which is 32bit) later.
PS: Perhaps I should start a new topic for this? where?

Edited by Turambar, 02 February 2011 - 01:03 PM.

[the bigg]

Can you upload the file it was failing on?

[Turambar]

Can you upload the file it was failing on?

It was the original AR2000.tis from bg1.
I tested it with wine and the windows tis2bg2 and it appears to work: converted file

[the bigg]

Done, I think (on AR2000 it has the same output whether run from Wine or natively). It works in an ext3 file system, but not in a folder mounted via VirtualBox's protocol - something about MMAP permissions.

Attached Files

•  tis2bg2.zip   41.5K   440 downloads

[Turambar]

Done, I think (on AR2000 it has the same output whether run from Wine or natively). It works in an ext3 file system, but not in a folder mounted via VirtualBox's protocol - something about MMAP permissions.

It still fails, with "Invalid input area name" or "Invalid input area directory":

Spoiler

./tis2bg2_linux -l ./ar2000.tis linux/ar2000.tis

[-----------------------------------------------]
[--- BG1->BG2 Area Fileset Converter ---]
[--- ---]
[--- ©KDiamond, Feb 2005, version 1.0 ---]
[-----------------------------------------------]

Usage : tis2bg2 [options] input_area_name output_area_name


Options : -s silent mode
-h save TIS header (for \override\ use)
-d delete input area tileset after conversion
-l write log file (tis2bg2.log)


Examples : tis2bg2 -h AR3300 myfolder1\AR6700
tis2bg2 -l myfolder1\AR3300 myfolder2\AR6700 -s -h -d


This program handles area's .wed, .tis, .mos and .bmp's.
For a complete area conversion .are and .bcs should be provided separately.


./
ar2000.tiar2000.tis
./ar2000.tiar2000.tis.wed

Error : Invalid input area name

I'm using it in my home folder, in my normal file system.
Moreover, the .exe you attached does not run through wine

Spoiler

wine ./tis2bg2.exe ar2000.tis wine2/ar2000.tis
err:service:validate_service_config Service L"Macmpms" is Win32 but has no image path set
err:service:scmdatabase_load_services Invalid configuration of service L"Macmpms" - skipping
err:module:import_dll Library cygwin1.dll (which is needed by L"Z:\\home\\francesco\\Giochi\\tistry\\tis2bg2.exe") not found
err:module:LdrInitializeThunk Main exe initialization for L"Z:\\home\\francesco\\Giochi\\tistry\\tis2bg2.exe" failed, status c0000135

Edited by Turambar, 03 February 2011 - 10:31 AM.

[the bigg]

Parameter passing is a fugly mess (but what do you expect out of the BiG congregation). ./tis2bg2_linux -l ./ar2000 linux/ar2000 should work (everything before the last slash is the path, everything after is the SOURCE_RES of this particular area). I also do not know or care why a given tool refuses to work under Wine if I'm working on a native equivalent.

[Turambar]

Parameter passing is a fugly mess (but what do you expect out of the BiG congregation). ./tis2bg2_linux -l ./ar2000 linux/ar2000 should work (everything before the last slash is the path, everything after is the SOURCE_RES of this particular area). I also do not know or care why a given tool refuses to work under Wine if I'm working on a native equivalent.

Ok it worked now. I diffed the output from wine and linux. The tis are the same, but the wed files differ (but the .wed treated with wine was obtained with an older version of tis2bg2, the one included in BGT).
The two files, opened with IETME, appear the same and correct (whereas the original file from bg1 had green water); NI can open both, and I can't see where they actually differ.

[the bigg]

I hope that the contents are identical, except that the version made with the native Linux client has some padding appended to the bottom; if so, it's because memory mapping works differently between Linux and Windows. If not, I have more hacking to do.

EIDT: you can use weidu --cmp-from linux.wed --cmp-to wine.wed to get a crude diff if you don't know of other binary diff tools.

Edited by the bigg, 04 February 2011 - 02:25 AM.

[Turambar]

I checked them with WeiDU and there were only 3 bytes to change:
To turn the linux version into wine

COPY ~linux\ar2000.wed~ ~wine\ar2000.wed~
// patches to turn [linux\ar2000.wed] into [wine\ar2000.wed]
	WRITE_BYTE 0x24 65 // 0x41 == A
	WRITE_BYTE 0x25 82 // 0x52 == R
	WRITE_BYTE 0x2b 52 // 0x34 == 4

Vice versa, to turn wine into linux

COPY ~wine\ar2000.wed~ ~linux\ar2000.wed~
// patches to turn [wine\ar2000.wed] into [linux\ar2000.wed]
	WRITE_BYTE 0x24 97 // 0x61 == a
	WRITE_BYTE 0x25 114 // 0x72 == r
	WRITE_BYTE 0x2b 0 // 0x00

Those offsets are all in the same field: name of the tileset. They are the 1st, 2nd and last character.
I understand the first two: the letters only change case (the old version, through wine, used capital letters; the new, linux one uses lower case).
The last byte is correctly 0 in linux (no character).
I don't understand why the last character is 0x34 (ascii for 4) in the wine version.
Moreover, and that's most astonishing, NI sees the two files as having the same 8 bytes at 0x25: their values (decimal) are 065 082 050 048 048 048 000 000 (hex 0x41 0x52 0x32 0x30 0x30 0x30 0x00 0x00); this is different from what WeiDU says for either file (NI says upper case as the wine version, but the last byte is 0 as in linux)

[the bigg]

The probable cause for the 0x00 vs. 0x34 are differences between how mmap() works in Linux and in Windows (or Wine); probably not worth obsessing over (most likely the Windows/Wine version preserves whatever data was in the memory, while the Linux version sets the whole file to 0x00 before proceeding). As per why it shows as 0x00 in Near Infinity, C strings are delimited by a NUL (0x00) character (and the convention is preserved by Java), thus NI reads the string as "ar2000<end>" and pads the rest of the 8 bytes with NUL characters.

[thom_raindog]

Can I throw a question in here as well? I managed to get bgt-linux to a point where it will at least start the installation, but it coughs up an error about not being able to find CHITIN.KEY right away, which is weird as it is right there in the same directory as the game. This might be a simple case of me being dense, but can someone point me to a solution?

I have NOT registered the games, but I would assume that makes no difference here.

[the bigg]

Did you run tolower?

[thom_raindog]

I swear I did tun tolower.. maybe at the wrong time? In any case, installation is running as we speak (after symlinking my BG1 dir to a strictly lowercase one ). Keep fingers crossed and thank you!

[thom_raindog]

That did it! Thanks so much.