[Solved] TC 16bit and riva128 or TNT driver in 3.11 = errors

[dallasl]

*** quick answer: those issues were caused by the nvidia riva128/tnt (agp or pci) driver for wfw3.11 ***

Hello!

I started using the 16bit version in my old computers running win3.11 (yeah wfw3.11 , don't ask why I keep them ) but at some random times I get this message:

WARNING: The TOTALCMD executable file is corrupted, possible VIRUS!
Totalcmd will close. Please run a virus scanner as soon as possible!

I understand this is to prevent certain actions but:
1. I use my registered key (so no alterations to exe for evil reasons , the registration email is the same gmail one I use here)
2. All files going to those old pcs pass through my main system which is protected with Kasprersky AntiVirus 2012. No viruses here...

It can take 1min or it can take 5 hours but the corruption message will appear. And this happens in 2 seperate pcs , I could see that one has bad ram or ide cable or something like that but that's not the case (one dual boots to win98SE and I never had any corruption messages or crashes...)

I have tried versions 6.58 and 6.57 and they behave the same. Also TC is the only program that I have issues in my 3.11 setups.
I just re-start TC after it closes and continue working, it didn't mess my filesystem or anything btw.

I must add that I have installed 3.11 on a FAT32 partition (with a special patch to remove win3.x dos6.22 check) could it be that TC is expecting a FAT16 partition or something similar and the detection routine fails
?
Also before anyone asks, no - there is no hdd corruption or errors or anything at all with this setup ever.

I hope there is an easy solution to this, thanks for listening!

[gmb]

Is Totalcmd.exe packed with UPX?
Maybe there's alarm because of UPX.

[dallasl]

I have no idea, I use the 16bit version downloaded from ghisler-com ... didn't mess with the .exe in any way

[ghisler(Author)]

No, it has nothing to do with UPX - TC reads its EXE and calculates a checksum to make sure that the EXE wasn't modified. If you get this with an unmodified EXE, it may be due to abad RAM chip. Try to create a large ZIP archive - if you get a CRC error, it's bad RAM.

[dallasl]

I just made a 676MB zip three times in a row (p200mmx/128MB ram) , did not encounter any problems at all...
I am using this particular machine for some years now and I never encountered any ram-related problems or corruption in archives, I don't see any reason to believe I have some faulty ram...?

[ghisler(Author)]

Can you post the Md5 checksum of the EXE, please? You can create it with TC itself (Files - Create checksums).

[dallasl]

sure I can, I get this:

for version 6.58: 9d7aee9b3240af4642421d458ce7db2c *tcmd16.exe
for version 6.57: 1b87f9636d161d8889ad3378e94ceb63 *tcmd16.exe

haven't tried older versions, both display the same error I posted above

edit:
I have currently 6.58 installed and I started randomly creating MD5s of the EXE... in about an hour I did ~15 tests, 2 times I got a different MD5 on the exe ... in these 2 times I made another md5 immediately after and I got the same I get all the time (same as above)
the last different 6.58 MD5 I saved was: bbd359abffe6a1f9d3d0e844fc95fa88 *tcmd16.exe



edit2:
got bored and made a low-quality video with my phone
see it youtu.be/odRkqg3IZgg
there are 5mins of constant sfv checking, and 3-4 fails in files (mainly the exe that is big) and see after the 5:00 I open a bunch of tc instances and they start failing....

[Flint]

dallasl
Are you sure Win3.1 can work with FAT32 natively? I always thought it couldn't; the check you turned off so bravely was not added there just for fun. TC behaviour should not depend on the filesystem used, but the operating system's may, and this would affect TC which relies on API provided.

You can try following: make a bunch of copies of the same file tcmd16.exe and compare them by contents afterwards, whether they were copied correctly and all have the same contents. Maybe, the FAT driver sometimes slips off and returns wrong contents of some sector…

[dallasl]

Yes, 3.1 can work with FAT32, they don't really care.... they never had a problem with FAT32/LBA , their "problem" was ms implementing stuff in later windows to lock 3.x out... nowdays they just stop supporting the OS (no more updates, fixes, etc) and they achieve their goal, back then noone would really care IMHO (except companies that would move on to NT or similar)
The check I removed was a DOS version check. Back then MS wanted 3.1x to run only on their DOS6.22 version and nothing else (to remove competition from DR-DOS mainly).
I have thought the possibility of fat32 corruption or something but the only thing that displays this behavior is TC... (also I am not using any of the 32bit file/disk access drivers just to be sure)

[ghisler(Author)]

I just checked the MD5 of TC 6.58, 9d7aee9b3240af4642421d458ce7db2c is correct. Therefore the error must be happening when the checksum is tested, so it is really most probably a RAM problem...

[sztihamer]

If I were you I'd rather run a Memtest86 or Memtest86+ test on the machine, and see if any errors would come up. If that's the case it definitely means that the RAM is bad. If you have multiple RAM modules, try removing the others and just leave on in your system and then run the tool. It can take a while to finish all the and you don't see any red errors it means that your memory is working fine.

Memtest86 or Memtest86+ will run even on 386 and 486 processors, so if you have one of those you can still use this tool. You will find a lot of images on Google that shows i7 and other Athlon processors, but it will run fine even on a Pentium-MMX or a Pentium II just fine.
For example heres' one running on an old machine: http://img183.imageshack.us/img183/3954/asustxp4x20tu6.jpg

If you see something like this it means your have memory problems, and the only way to fix it is by replacing the bad module. http://farm1.static.flickr.com/227/504300235_1f83c25ddb.jpg

[dallasl]

I already checked with memtest 86+ the 5x86 machine... takes ages to complete some cycles but nothing was wrong...
Haven't run memtest on the pentium yet, I'll do it now and report some hours later

[petermad]

WARNING: The TOTALCMD executable file is corrupted, possible VIRUS!
Totalcmd will close. Please run a virus scanner as soon as possible!

This error message is a bit odd since the name of the 16 bit Total Commander version 6.5x executables are TCMD16.EXE - not TOTALCMD.EXE - are you by accident trying to run the 32 bit version.

Earlier 16 bit versions used the filename WINCMD16.EXE, and before that WINCMD.EXE.

IIRC TOTALCMD.EXE was introduced as filename for 32bit version 5.50.

EDIT: I just tried with a manually corrupted TCMD16.EXE ( 6.58 ) - and I get the same message as dallasl - so Ghisler apparantly just used the same mesage for 32 and 16 bit version.

[dallasl]

@petermad: haven't payed attention to the name, before I post here I googled about the problem and just copied/pasted the error message here... next time I see it I will report

@thread
In the meantime after ~5 hours of memtest and 4 complete passes on the ram check there are no errors whatsoever. And it makes sense because like I said I use this computer for 98SE/DOS apps and never encountered a freeze or corruption due to ram being bad...

Image: http://i41.tinypic.com/3468bqh.jpg

[petermad]

2dallasl

I edited my previous message:

I just tried with a manually corrupted TCMD16.EXE ( 6.58 ) - and I get the same message as dallasl - so Ghisler apparantly just used the same mesage for 32 and 16 bit version.