A while ago I asked:
>Does the sticky bit do anything for executables? The chmod(2) man page says:
> t Save text permission.
> In earlier versions of the UNIX system, setting this permission bit
> caused the text segment of a program to remain in virtual memory after
> its first use. The system thus avoided having to transfer the program
> code of frequently accessed programs into the paging area. A t appears
> in the execute position of the all others option to indicate that the
> file has this bit (the sticky bit) set.
>Does OSF/1 respect this bit? I can't even think of a good way to test it.
>Maybe someone could look at the source. Is there a better way to keep a
>process in memory? Thanks.
berrigan_at_kent.wednet.edu (Bob Berrigan) wrote:
>Back in February you sent out a message on keeping executables in memory
>(use sticky bit or some other means). Did you get any worthwhile info. you
>could pass along?
So, I figured I'd pollute the list with the answers as well. Here they are:
haymanR_at_icefog.uacn.alaska.edu:
The sticky bit under OSF/1 is ignored except when it is on directories, then
only the owner (and root) may delete the owner's files under that directory.
To lock something in memory, one would either let the UBC hold it in memory due
to the many accesses it has, one could use the MFS (memory file system) for the
files, or one could wire the pages into memory via the mlock, munlock, mlockall,
and munlockall calls - see the DEC OSF/1 Guide to Realtime Programming.
Randy M. Hayman
haymanr_at_icefog.alaska.edu
Knut.Hellebo_at_nho.hydro.com:
I think nowadays the stickybit is used only on directories. If the sticky
bit is on then only the owner of files in that directory can remove them.
As far as I know the stickybit otherwise is ignored.
-drew
-------
USMail: OTA Limited Partnership E-mail: drew_at_ox.com
1 Manhattanville Road Phone: +1 914 694 5800
Purchase, New York 10577 FAX: +1 914 694 5831
"Welcome, to the machine..."
Received on Wed Mar 29 1995 - 17:48:10 NZST