July 31, 2002 at 8:12 pm
New security bulletin about an MDAC vulnerability affecting SQL Server 7.0 and 2000:
http://www.microsoft.com/technet/treeview/default.asp?url=/technet/security/bulletin/MS02-040.asp
This is another buffer overflow discovered by David Litchfield. An attacker must have the ability to execute a query on the server. This isn't as hard as it sounds, especially if an application leaves a server vulnerable to a SQL Injection attack.
A patch is available for MDAC versions 2.5, 2.6, and 2.7.
K. Brian Kelley
http://www.sqlservercentral.com/columnists/bkelley/
K. Brian Kelley
@kbriankelley
July 31, 2002 at 8:16 pm
I saw this today too - looks like 3 different patches? They couldnt do an all in one?!
Andy
August 1, 2002 at 3:25 am
What's this "DisallowAdhocAccess option" the faq claims is set using "the advanced sp_serveroption" to stop openrowset useage?
I've only got sql7 at work, but can't find it.
August 1, 2002 at 4:55 am
I'm not seeing anything in Books Online about it, though I've been looking, too. I've found how to do it through EM. When you set the provider options, the options you pick get applied to all connections using that provider. So with respect to sources connecting via OLE DB, here's how to set it up:
Start a new linked server connection:
Select the OLE DB Provider (even though we would normally select SQL Server when connecting to another SQL Server, it uses OLE DB):
We want to modify the Provider Options:
The last option disallows adhoc access:
BTW, this work around is also applicable to Microsoft Security Bulletin 01-032:
http://www.microsoft.com/technet/treeview/default.asp?url=/technet/security/bulletin/MS01-032.asp
K. Brian Kelley
http://www.sqlservercentral.com/columnists/bkelley/
K. Brian Kelley
@kbriankelley
August 1, 2002 at 10:42 am
Thanks, I've fixed that now.
I can't believe that openrowset is available to ALL users by default, and that it isn't restricted to the DSN's setup on the host server...
August 1, 2002 at 11:33 am
It is frightening. Some of the recent vulnerabilities wouldn't be a big deal except permissions default to the public role. Great googlymoogly!
Of course, someone's figured out that if they can get a buffer overflow attack to work, they can deploy a patch which will make SQL Server think everyone is a sysadmin. This was a side point of the weak hashes on the SQL Server login passwords from Litchfield.
I'm thinking neither Microsoft nor Oracle like this guy very much any more, but he's earnestly going after security holes and giving them a chance to respond.
K. Brian Kelley
http://www.sqlservercentral.com/columnists/bkelley/
K. Brian Kelley
@kbriankelley
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply