Re: Code Sign verification on Leopard

2009-10-15 Thread Clark Cox
On Thu, Oct 15, 2009 at 8:51 AM, Charles Srstka wrote: > On Oct 15, 2009, at 9:57 AM, Clark S. Cox III wrote: > >> The malicious code could just move the entire original bundle wholesale. >>  Code signature check still sees the original bundle. >> >> Sent from my iPhone > > Presumably, this would

Re: Code Sign verification on Leopard

2009-10-15 Thread David Duncan
On Oct 15, 2009, at 8:51 AM, Charles Srstka wrote: Presumably, this would be more noticeable to the user than simply copying a binary file inside an opaque app bundle that most users never look inside. The malicious code could also break the code sign checking APIs to always return 'true

Re: Code Sign verification on Leopard

2009-10-15 Thread Charles Srstka
On Oct 15, 2009, at 9:57 AM, Clark S. Cox III wrote: The malicious code could just move the entire original bundle wholesale. Code signature check still sees the original bundle. Sent from my iPhone Presumably, this would be more noticeable to the user than simply copying a binary file i

Re: Code Sign verification on Leopard

2009-10-15 Thread Clark S. Cox III
The malicious code could just move the entire original bundle wholesale. Code signature check still sees the original bundle. Sent from my iPhone On Oct 14, 2009, at 21:42, Charles Srstka wrote: On Oct 14, 2009, at 11:40 PM, Jeff Laing wrote: Actually, heck, you wouldn't even need that

Re: Code Sign verification on Leopard

2009-10-14 Thread Kyle Sluder
On Wed, Oct 14, 2009 at 9:35 PM, Charles Srstka wrote: > Unless, of course, the app checked its code signature. Put malicious library in temp dir. Set DYLD_LIBRARY_PATH. Game over. --Kyle Sluder ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)

Re: Code Sign verification on Leopard

2009-10-14 Thread Charles Srstka
On Oct 14, 2009, at 11:40 PM, Jeff Laing wrote: Actually, heck, you wouldn't even need that. All a virus would have to do would be to move the binary somewhere else and put a binary in its place that does something malicious and then launches the real binary, and the user would never tell t

RE: Code Sign verification on Leopard

2009-10-14 Thread Jeff Laing
> Actually, heck, you wouldn't even need that. All a virus would have to > do would be to move the binary somewhere else and put a binary in its > place that does something malicious and then launches the real binary, > and the user would never tell the difference. > > Unless, of course, the app c

Re: Code Sign verification on Leopard

2009-10-14 Thread Charles Srstka
On Oct 14, 2009, at 10:50 AM, Clark Cox wrote: You say that as if "getting infected by" and "being specifically targeted by" a virus are different things. Either a virus has targeted your application, and as such knows enough about it to modify it's code (and therefore replace/defeat any signatu

Re: Code Sign verification on Leopard

2009-10-14 Thread Clark Cox
On Wed, Oct 14, 2009 at 12:37 AM, Charles Srstka wrote: > On Oct 13, 2009, at 10:41 AM, Jens Alfke wrote: > >> This code sample seems to be designed to verify the binary that it's >> compiled into. That's sort of useless for security purposes, like yelling >> downstairs "are you a burglar?" If you

Re: Code Sign verification on Leopard

2009-10-14 Thread Charles Srstka
On Oct 13, 2009, at 10:41 AM, Jens Alfke wrote: This code sample seems to be designed to verify the binary that it's compiled into. That's sort of useless for security purposes, like yelling downstairs "are you a burglar?" If your own code's already been modified, it's easy enough for the h

Re: Code Sign verification on Leopard

2009-10-13 Thread jonat...@mugginsoft.com
On 13 Oct 2009, at 19:28, Jens Alfke wrote: On Oct 13, 2009, at 9:25 AM, jonat...@mugginsoft.com wrote: But it's not useless in the sense that it provides feedback that the code IS signed. The code merely allows me to detect if I have screwed up my build settings and managed to break the

Re: Code Sign verification on Leopard

2009-10-13 Thread Jens Alfke
On Oct 13, 2009, at 9:25 AM, jonat...@mugginsoft.com wrote: But it's not useless in the sense that it provides feedback that the code IS signed. The code merely allows me to detect if I have screwed up my build settings and managed to break the code signing. Sure, it's useful for that. Bu

Re: Code Sign verification on Leopard

2009-10-13 Thread jonat...@mugginsoft.com
On 13 Oct 2009, at 16:41, Jens Alfke wrote: That's sort of useless for security purposes, like yelling downstairs "are you a burglar?" But it's not useless in the sense that it provides feedback that the code IS signed. The code merely allows me to detect if I have screwed up my build se

Re: Code Sign verification on Leopard

2009-10-13 Thread Jens Alfke
This code sample seems to be designed to verify the binary that it's compiled into. That's sort of useless for security purposes, like yelling downstairs "are you a burglar?" If your own code's already been modified, it's easy enough for the hacker to disable the code that does the checking

Re: Code Sign verification on Leopard

2009-10-13 Thread jonat...@mugginsoft.com
On 13 Oct 2009, at 08:34, Jakub Bednar wrote: Hi list, I have read CodeSigningGuide and CodeSigningRef from Apple. In the CodeSigningRef every method has a note, that it is available in 10.6 and later. So I just want to make sure. On Leopard, there is now Cocoa or other API for verifying