Messages in this thread |  | | Date | Wed, 20 Nov 1996 12:47:00 -0500 | From | "Theodore Y. Ts'o" <> | Subject | Re: real POSIX.1b semaphores |
| |
From: Ulrich Drepper <drepper@myware.rz.uni-karlsruhe.de> Date: 20 Nov 1996 02:23:42 +0100
> Err.... be careful about the security issues. It should not be possible > for a process which doesn't have access to a semaphore to just go into > the shared page and much with the sempahores. This pretty much tosses > out any idea where you use a single global page for all sempahores.....
Unfortunately there are not many security consideration in the POSIX semaphores. Either they are process local or they are available for any process. I read the section about semaphores twice and found no sign for protection.
Gallmeister's book speaks about protection for named semaphores (and it sounds reasonable). Nothing is specified for memory semaphores. This really seems to be a whole in the standard.
I had read the earlier proposal to mean that named semaphores would also go in globally accessible memory, which would have been a real problem.
Because of the security problems, I don't believe global unnamed semaphores are useful. Process local semaphores are useful for places where you need speed, and named semaphores for global access. But if you have global unnamed semaphores with no security, they really are pointless....
- Ted
|  |