UID > 60,000

UID > 60,000

Post by nea » Fri, 22 Apr 1994 12:02:31



 Does anyone see anyproblem in making the MAXUID on the limit.h
to 200000.
 
 
 

UID > 60,000

Post by Mike Elliso » Fri, 22 Apr 1994 17:37:34


|>  Does anyone see anyproblem in making the MAXUID on the limit.h
|> to 200000.
|>

You could have problems with accounting etc as some unix systems only hold
a 2 byte user identifier giving a maximum of 2**16 (65536) possible uids,
before you get overlap. It seems on solaris that the uid has been increased
to a 4 byte value in accounting records but thta won't help interoperatability
with non solaris systems.

 
 
 

UID > 60,000

Post by Joerg Czerans » Sat, 23 Apr 1994 06:43:18



>  Does anyone see anyproblem in making the MAXUID on the limit.h
> to 200000.

Hm, you don't say, which OS.
You can't do this on SunOS 4.1.3.

To quote from /usr/include/sys/types.h:
typedef unsigned short uid_t;

However, on SunOS 5.x, the definition is changed to 'long'.
There may be problems with NFS and other programs that might
still rely on the fact that uids were stored in 'short's.

joerg

--


D38678 Clausthal-Zellerfeld    Voice (at work)  +49-5323-72-3896
Germany                        Voice (at home)  +49-5323-78858

 
 
 

UID > 60,000

Post by Peter Van E » Sat, 23 Apr 1994 08:06:35



> Does anyone see anyproblem in making the MAXUID on the limit.h
>to 200000.

        Many machines use an int (i.e. 16 bits) for the UID and thus the
maximum possible UID is 65535.
        Many machines (NeXTs for instance, and some versions of SunOS) have
trouble with UIDs greater than 32767 (since they use a signed int as the uid).
Various programs on a NeXT won't work for a user with a UID greater than
32767.
        I believe that Posix mandates a 32 bit UID, and that at least AIX
supports 32 bit UIDs.

Peter Van Epp / Operations and Technical Support
Simon Fraser University, Burnaby, B.C. Canada

 
 
 

UID > 60,000

Post by Casper H.S. D » Sat, 23 Apr 1994 17:06:36



>To quote from /usr/include/sys/types.h:
>typedef unsigned short uid_t;
>However, on SunOS 5.x, the definition is changed to 'long'.
>There may be problems with NFS and other programs that might
>still rely on the fact that uids were stored in 'short's.

But still, all places in the kernel that matter hardwite
the value 60000 as max.

Casper

 
 
 

UID > 60,000

Post by Paul Flinde » Sun, 24 Apr 1994 00:48:25



> Does anyone see anyproblem in making the MAXUID on the limit.h
>to 200000.

Well if your uid_t is 16 bits you might have problems
eg on an SGI running Irix 4.0.5f

smee$ grep uid_t /usr/include/sys/types.h
/usr/include/sys/types.h:typedef        u_short uid_t;

--

Delcam International PLC.       Tel:  +44 21 766 5544   Fax: +44 21 766 5511

 
 
 

UID > 60,000

Post by nea » Mon, 25 Apr 1994 11:58:41


: smee$ grep uid_t /usr/include/sys/types.h
: /usr/include/sys/types.h:typedef        u_short uid_t;

The following is from the HP 700 and 800 that I am working on
types.h:     typedef long uid_t;                /* For user IDs */

 
 
 

1. UID > 60,000

In Unix System V Release 4 for UID is defined as 60000 yet is defined
as a long. I would like to bump up the UID to about 200000 from what
I've been able to see in the .h file it does not seem to be a problem.
Any percieved problems with changing the UID to 200000. All networked
systems will be Unix System V Release V

2. 640mb MO drive on solaris x86 2.5.1?

3. User Ids (uid) > 65k or 32 bits uids

4. Jail - Some general questions

5. HELP need root->other->root uid

6. A program to save running process?

7. <><><> MOUNTING EXTENDED PARTITION <><><>

8. Meddling with a Packet in IP stack

9. Wanted: <><><> Unix Specialist <><><>

10. LILO help <><><><><><>

11. >>---> Software Jobs! >>--->

12. UID / effective UID problem

13. effective UID vs. Real UID with su - problem