Login Name Checker

[Mr.Slime 142]

Hi guys.

The game core sends different login result to the server for password wrong and user id not found.

Just one example it's possibile to test with sending packets 9 accounts per second.

 

Affects: Every game core

 

Fix:

Search for this in input_db.cpp:

case HEADER_DG_LOGIN_NOT_EXIST:
		LoginFailure(DESC_MANAGER:.instance().FindByHandle(m_dwHandle), "NOID");
		break;

case HEADER_DG_LOGIN_WRONG_PASSWD:
		LoginFailure(DESC_MANAGER:.instance().FindByHandle(m_dwHandle),	"WRONGPWD");
		break;

and replace the WRONGPWD with NOID.

 

So you blocked the way to detect if an account exists.

 

Best Regards,

Mr.Slime

[Metin2 Dev 12287]

I think that better way is change the lines in locale_game.txt

[Mr.Slime 142]

C++ better of .txt ahahha

[Galet 592]

35 minutes ago, Mr.Slime said:

C++ better of .txt ahahha

The result is just the same, no matter which one is better, the locale_game.txt method is a bit better because you don't have to change anything in sources and moreover the result will be exactly the same

[Vanilla 1538]

Okay you can change locale_game.txt but what if a user changes it back?

c++ > txt files

With his fix the server won't respond with an accurate message. Next thing is that people could write a tool to connect with the server and start the bruteforce - this way they won't even need the locale_game.txt and just fetch the result packet the server sends them. Telling people to consider changing a simple text file on a local client instead of a serverside source isn't really that secure^^' It's easier, yes. But comfortability sacrifices security.

 

Therefore: Thanks, I for myself didn't see it. Should be in the next vanilla core release with an option to change it (of course on by default so make sure it's secure)

[Galet 592]

I never told .txt was more secure than c++ but it's really more usefull to create a new error message with packet than using the NOID statement, or at least change the text in locale_game.txt

However, if the client is securized a bit, there won't be any problem for locale_game, but indeed you can trace the packet to fetch the result, you're right !

So yeah, c++ method is way better than only a locale_game.txt method, but I think using both could be better (new packet, new error, new message)

[Mr.Slime 142]

Thanks @Vanilla

[RealReznov 15]

Thanks, it's useful

[Mr.Slime 142]

Thanks @RealReznov

[RealReznov 15]

No problem

But i tested it, and it still says the text of LOGIN_FAILURE_WRONG_PASSWORD after compile. Did you tested it?

[Mr.Slime 142]

Sure

[RealReznov 15]

Then I don't know why doesn't works for me

(input_db had compiled)

I've rewrited login_failure_wrong_password in locale_game to check it, and it says it, that i rewrited when i type wrong pw.

[ragem0re 171]

It's just a copy from

DevChuckNorris

@ VIP-Area