Skip to content

Windows defender finds a virus in current master branch #554

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
rmn20 opened this issue Mar 27, 2023 · 3 comments
Closed

Windows defender finds a virus in current master branch #554

rmn20 opened this issue Mar 27, 2023 · 3 comments
Labels
wontfix This will not be worked on

Comments

@rmn20
Copy link

rmn20 commented Mar 27, 2023

I'm using Windows 10 LTSC
At the moment, on the state of this commit, windows defender finds a virus in the master branch.

image

@anzz1
Copy link
Contributor

anzz1 commented Mar 27, 2023

Breaking News
AV ""product"" sucks ass, water is wet.

To be clear, I'm not calling you stupid, I'm calling your AV software stupid. Users are not to blame, the companies should make passable products instead of falsely advertising them as such.

I've wasted so much time sending legitimate software to various AV companies to be whitelisted that I can't even count; all the while the AV companies do nothing to stop the spread of actually malicious stuff.

I'm done doing that though. If someone else wants to send the binaries to AV companies to be whitelisted so they don't hit the "unknown file, must be a virus" mark, feel free to do that.

I cannot fully express my opinion towards AV companies within the GitHub etiquette but I'm sure you can read between the lines.

@gjmulder gjmulder added the wontfix This will not be worked on label Mar 27, 2023
@h-h-h-h
Copy link
Contributor

h-h-h-h commented Mar 27, 2023

On virustotal.com, MaxSecure thinks llama-master-34c1072-bin-win-avx2-x64.zip from the release page is "Trojan.Malware.300983.susgen".

When I compile the codebase as "x64-Release" with Visual Studio 2022, I get the same 6 .exe files as in the ready-made release (by their names). Of those, only test-quantize.exe is flagged on virustotal.com, again as "Trojan.Malware.300983.susgen".

In my understanding, test-quantize.exe is made out of tests/test-quantize.c. I could only find the method calls ggml_quantize_q4_0() and ggml_quantize_q4_1() there that call into the rest of the codebase. And those methods are also called in other code that I assume is used to create the other .exe files which aren't flagged.

So, flagging test-quantize.exe in my build seems irrational. But the app also worked without it for me; so you can just delete it.

What .exe files are flagged exactly how is different for the initially mentioned official download, though.

@anzz1
Copy link
Contributor

anzz1 commented Mar 28, 2023

So, flagging test-quantize.exe in my build seems irrational.

You are now confusing antivirus software with something that is rational.

I can't fathom how they are so bad at detecting viruses while simultaneously flagging everything that isn't, even having all the resources in the world and being made by billion dollar companies. There is simply no excuse. There would be some very simple ways of improving their heuristics by a leap, lot of it involving following the codepaths to what actual system libraries are accessed and what are the actual syscalls being made. However, they dont do that and basically resort to the same CTRL+F style ""heuristics"" that has been the modus operandi since the 1990s. It simply doesn't and can't work properly and is a completely irrational approach. Maybe it did once make a little bit of sense, when there were like 10 different viruses total in existence and everything was shared in diskettes and floppies. It however, does not make any sense today.

If your security plan is a blindfolded monkey with a shotgun, you can hardly blame the result to be illogical.

@gjmulder gjmulder closed this as not planned Won't fix, can't repro, duplicate, stale Mar 28, 2023
AAbushady pushed a commit to AAbushady/llama.cpp that referenced this issue Jan 27, 2024
* .sh script V1

* koboldcpp.sh polish

* koboldcpp.sh dist generator

* Include html's in dist

* RWKV in Linux Dist

* Lower dependency requirements

* Eliminate wget dependency

* More distinct binary name

I know its technically amd64, but I don't want to cause confusion among nvidia users.

* Use System OpenCL

Unsure how this will behave in the pyinstaller build, but pocl ended up CPU only. With a bit of luck the pyinstaller uses the one from the actual system if compiled in a system without opencl, while conda now includes it for that specific system.

* Add cblas dependency

Missing this causes compile failures on some system's

* ICD workaround

Ideally we find a better solution, but conda forces ICD and needs this for the successful compile. However, pyinstaller then embeds the ICD causing it to be limited to the system it was compiled for. By temporarily removing the ICD pyinstaller can't find it and everything remains functional. Ideally we do this on a pyinstaller level, but I could not find any good options to do so yet.

---------

Co-authored-by: root <root@DESKTOP-DQ1QRAG>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
wontfix This will not be worked on
Projects
None yet
Development

No branches or pull requests

4 participants