Bug 62481 - lastcomm uses 16-bit UIDs
Summary: lastcomm uses 16-bit UIDs
Keywords:
Status: CLOSED DUPLICATE of bug 131757
Alias: None
Product: Fedora
Classification: Fedora
Component: psacct
Version: 3
Hardware: i386
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Charlie Bennett
QA Contact:
URL:
Whiteboard:
: 56829 (view as bug list)
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2002-04-01 15:57 UTC by Andy Grimm
Modified: 2007-11-30 22:10 UTC (History)
3 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2006-02-21 18:48:39 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
Kernel patch to provide 32bit UID/GID support, and also fix highuid bug (1.30 KB, patch)
2002-04-01 20:52 UTC, Mike A. Harris
no flags Details | Diff

Description Andy Grimm 2002-04-01 15:57:47 UTC
From Bugzilla Helper:
User-Agent: Mozilla/4.76 [en] (X11; U; SunOS 5.8 sun4u)

Description of problem:
lastcomm uses 16-bits UIDs, while the rest of the system now supports 32-bit
UIDs.  This could be a security issue if the lowest 16 bits of a UID match the
UID of another user (for example, UID 65536 or 131072 would show up as root). 
If both users with "equivalent" UIDs are logged in, there is no way to tell who
ran which commands.

Version-Release number of selected component (if applicable):


How reproducible:
Always

Steps to Reproduce:
1. Create a user with a high UID (65536 or greater)
2. Make sure accounting is on
3. login as the newly created user
4. Run some commands
5. Become root
6. Run lastcomm
	

Actual Results:  Commands run by the user show up as having been run by (UID mod
65536).

Expected Results:  Commands should have been attributed to the correct user.

Additional info:

Comment 1 Mike A. Harris 2002-04-01 18:02:38 UTC
Believe it or not, this is actually not a bug.  It is really the lack of
the kernel process accounting code being 32bit UID aware.  In other words,
the kernel's accounting code is purely 16bit UID/GID, among other problems.

Fixing this, requires fixing it in the kernel first, and then in the
userland utilities.  Doing so however may break binary compatibility,
and so a full fix may have to wait for Linux 2.5.x development
cycle, which will become official in 2.6.0.

I will have a look into implementing support in a compatible way however
in the meantime.

Comment 2 Mike A. Harris 2002-04-01 20:52:44 UTC
Created attachment 51686 [details]
Kernel patch to provide 32bit UID/GID support, and also fix highuid bug

Comment 3 Mike A. Harris 2002-04-01 23:07:59 UTC
*** Bug 56829 has been marked as a duplicate of this bug. ***

Comment 4 Mike A. Harris 2002-04-01 23:09:13 UTC
Adding Cc's...

Working on userland support...

Comment 5 Mike A. Harris 2002-04-15 17:42:38 UTC
I've implemented support for the userland tools but have not put it into
our packaging yet, because I'd like to test it fully first.  Once that
is done, I will probably release it as an enhancment release.

Comment 6 Andy Grimm 2002-05-31 01:02:47 UTC
I finally got around to testing this code, after the patch appeared in the kernel 
which shipped with redhat 7.3.  I was disappointed however, to see that the
psacct package had not been patched.  I made a couple of fixes to lastcomm.c
and sa.c (changing ac_uid to ac_uid32 in a few places), then patched
/usr/include/linux/acct.h and /usr/include/sys/acct.h in the same way that you
patched the kernel source.  32-bit accounting now works wonderfully on my system.
I can submit the patches if you like.

Comment 7 Chris Ricker 2002-07-10 02:45:37 UTC
Still broken in the Limbo public beta

Comment 8 Mike A. Harris 2002-11-13 21:14:42 UTC
ag381597)  You can't just replace those variables like that or you will
break compatibility.  It will work with our kernel and nobody elses.



Comment 9 Charlie Bennett 2004-08-27 22:08:43 UTC
This was not fixable without upstream changes to the kernel.
The 2.6 kernel now has support for 32-bit uid/gid in the accounting code
which is enabled by setting:

CONFIG_BSD_PROCESS_ACCT=y
CONFIG_BSD_PROCESS_ACCT_V3=y

The userspace tools are being modified to support the V3 data format.

This bug will be used to track these changes against Fedora Core 3.

Comment 10 Charlie Bennett 2004-09-07 15:48:52 UTC

*** This bug has been marked as a duplicate of 131757 ***

Comment 11 Red Hat Bugzilla 2006-02-21 18:48:39 UTC
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.


Note You need to log in before you can comment on or make changes to this bug.