COMP3301 Assignment 1

OpenBSD Zones “Episode 3: Return of the Sys(call)”

Due: 3pm Monday in Week 5 (19th of August)

1 Academic Integrity

All assessments are individual. You should feel free to discuss aspects of C programming and assessment specifications with fellow students and discuss the related APIs in general terms.

You should not actively help (or seek help from) other students with the actual coding of your assessment. It is cheating to look at another student’s code, and it is cheating to allow your code to be seen or shared in printed or electronic form. You should note that all submitted code will be subject to automated checks for plagiarism and collusion. If we detect plagiarism or collusion (outside of the base code given to everyone), formal misconduct proceedings will be initiated against you.

If you’re having trouble, seek help from a teaching staff member. Do not be tempted to copy another student’s code. You should read and understand the statements on student misconduct in the course profile and on the school website: https://eecs.uq.edu.au/current-students/guidelines-and-policies-students/student-conduct.

1.1 Use of AI Tools

All assessment tasks evaluate students’ abilities, skills and knowledge without the aid of gen-erative Artificial Intelligence (AI) or Machine Translation (MT). Students are advised that the use of AI technologies to develop responses (e.g. code generation) is strictly prohibited and may constitute student misconduct under the Student Code of Conduct.

2 Introduction

This assignment will extend a basic implementation of “zones” in the OpenBSD kernel. The main area of improvement will be separating group and user permissions on zone operations.

You will be provided with a diff that adds the basic zones functionality to OpenBSD. You will need to make changes and improvements on top of this diff.

The purpose of this assignment is for you to demonstrate an understanding of the role of an operating system kernel and how it supports processes making system calls, as well as your skills in reading, understanding, and modifying existing code.

2.1 Background

Zones extend the isolation of processes beyond what is traditionally provided by UNIX and UNIX-like systems, including OpenBSD. Traditionally, all processes running on an OpenBSD are visible to all other processes. This can be demonstrated by running commands like top(1), ps(1), and pgrep(1)/pkill(1), which can show all processes running in a system:

$ ps - ax

PID TT STAT TIME COMMAND

1 ?? I 0:01.01 / sbin / init

35862 ?? Ip 0:00.01 / sbin / slaacd

9544 ?? Ip 0:00.01 slaacd : engine ( slaacd )

330** ?? IpU 0:00.01 slaacd : frontend ( slaacd )

96644 ?? IU 0:00.01 / sbin / dhcpleased

82639 ?? Ip 0:00.01 dhcpleased : engine ( dhcpleased )

68436 ?? IpU 0:00.01 dhcpleased : frontend ( dhcpleased )

6881 ?? IpU 0:00.01 / sbin / resolvd

69588 ?? IpU 0:00.03 syslogd : [ priv ] ( syslogd )

54598 ?? Spc 0:00.03 / usr / sbin / syslogd

14516 ?? IU 0:00.01 pflogd : [ priv ] ( pflogd )

15079 ?? Spc 0:00.12 pflogd : [ running ] -s 160 -i pflog0 -f / var / log /

pflog

94692 ?? S < pc 0:00.12 ntpd : ntp engine ( ntpd )

37809 ?? Sp 0:00.26 ntpd : dns engine ( ntpd )

1816 ?? I < pU 0:00.00 / usr / sbin / ntpd

63841 ?? I 0:00.01 sshd : / usr / sbin / sshd [ listener ] 0 of 10 -100

startups

83125 ?? Ip 0:00.02 / usr / sbin / smtpd

58972 ?? Ipc 0:00.02 smtpd : crypto ( smtpd )

99695 ?? Ipc 0:00.02 smtpd : control ( smtpd )

5777 ?? Ip 0:00.02 smtpd : lookup ( smtpd )

45996 ?? Ipc 0:00.04 smtpd : dispatcher ( smtpd )

37682 ?? Ipc 0:00.02 smtpd : queue ( smtpd )

97246 ?? Ipc 0:00.02 smtpd : scheduler ( smtpd )

48848 ?? IpU 0:00.00 sndiod : helper ( sndiod )

**188 ?? I < pc 0:00.00 / usr / bin / sndiod

96369 ?? Ip 0:00.02 / usr / sbin / cron

45067 ?? I 0:00.07 sshd : dlg [ priv ] ( sshd )

**638 ?? S 0:00.03 sshd : dlg@ttyp0 ( sshd )

1**0 p0 Sp 0:00.02 - ksh ( ksh )

169** p0 R + pU /2 0:00.00 ps - ax

33428 00 I + pU 0:00.01 / usr / libexec / getty std .9600 tty00

$

‘‘‘

While all processes are visible to each other, they are restricted from interacting with each other based on the user that each process is running as. A non-root user can only signal their own processes. Attempts to signal processes running as another user fails:

$ whoami

dlg

$ ps -U _sndio

PID TT STAT TIME COMMAND

**188 ?? I < pc 0:00.00 / usr / bin / sndiod

$ kill **188

ksh : kill : **188: Operation not permitted

$

However, the root user is allowed to signal any process:

$ doas kill **188

doas ( dlg@comp3301 . eait . uq . edu . au ) password :

$ ps -U _sndio

PID TT STAT TIME COMMAND

$

3 Zones Implementation

Zones are implemented for this assignment to add further isolation of processes. Processes running within a zone can only see and interact with processes running within the same zone, regardless of which user within the zone is running the commands. This implementation is loosely modelled on the design of Solaris Zones as described in PSARC/2002/174.

The exception to this enhanced isolation is for processes running in the ”global” zone, which is the default zone that is created and exists on boot. Processes running in the global zone can see all other processes in the system, including those running in other (non-global) zones, and the root user in the global zone can signal any of these processes too. However, non-root users in the global zone cannot signal processes in other zones, even if they are running as the same user.

The provided diff implements changes to the kernel and several userland utilities and adds a zone(8) command and man page. The zone(8) command provides several sub-commands that expose the functionality of the kernel zone subsystem.

3.1 Provided Zone Syscalls

zone_create()

zoneid_t zone_create ( const char * zonename ) ;

zone_create() creates a new zone id for use in the system, with a unique name specified by zonename.

zone_destroy()

int zone_destroy ( zoneid_t z ) ;

zone_destroy() deletes the specified zone instance. The zone must have no running processes inside it for the request to succeed.

zone_enter()

int zone_enter ( zoneid_t z ) ;

zone_enter() moves the current process into the specified zone.

zone_list()

int zone_list ( zoneid_t * zs , size_t * nzs ) ;

In the global zone zone_list() provides the list of zones in the running system as an array of zoneid ts. If run in a non-global zone, the list will only contain the current zone.

zone_name()

int zone_name ( zoneid_t z , char * name , size_t namelen ) ;

The zone_name() syscall provides the name of the zone identified by the z argument. If run in a non-global zone the z id must be the identifier for the current zone. In the global zone it can be any zone identifier.

zone_id()

zoneid_t zone_id ( const char * name ) ;

zone_id() provides the id associated with the name zone. If run in a non-global zone, only the current zone name may be specified. If name is a NULL pointer the zone id calling process is running in is returned.

zone_stats()

int zone_stats ( zoneid_t z , struct zstats * zstats ) ;

zone_stats() provides an assortment of operating system statistics resulting from processes in the zone associated with the id z.

zone_rename()

int zone_rename ( zoneid_t z , char * newname ) ;

zone_rename() alters the name of the zone identified by the z argument. The new name will be the name provided in the newname argument. zone_rename() handles the necessary tree updates on the zone names tree.

This syscall will be necessary for you to implement the zone rename subcommand.

3.2 zone(8)

usage : zone create zonename 1

zone destroy zonename 2

zone exec zonename command ... 3

zone list 4

zone id [ zonename ] 5

zone name [ zid ] 6

zone stats [ - H ] [ - o property [ ,...] zone [...] 7

The zone(8) program uses the zone syscalls to allow systems administrators or operators to use the zone subsystem in the kernel.

zone create

zone create uses the zone_create() syscall to create a zone with the specified name.

zone destroy

zone destroy uses the zone_destroy() syscall to create a zone with the specified name. If a zone with the specified name does not exist, zone(8) will attempt to interpret the argument as a numeric zone identifier.

zone exec

zone exec uses the zone_enter() syscall to move itself into the specified zone, and then executes the program. If a zone with the specified name does not exist, zone(8) will attempt to interpret the argument as a numeric zone identifier.

zone list

zone list uses the zone_list() syscall to fetch a list of ids for the currently running zones, and iterates over it calling the zone_name() syscall to print out the list of zone ids and names.

zone name / zone id

zone name and zone id use their associated syscalls zone_name() and zone_id() to return the name of a zone given its id, or the id of a zone given its name.

zone stats

zone stats uses the zone_stat() syscall to obtain and print out to the user a series of statis-tics from processes running in the current zone. See the manual page in zone(8) for more information.

3.3 Your Tasks

You will be adding additional functionality to a series of zone(8) sub-commands, adding three new zone(8) sub-commands, and implementing any necessary changes to the kernel zones system to support them.

Your additional functionality centers around zone permissions. Files have an associated “user” and “group”, and this user or group may have permission to operate on the file. Your task is to associate zones with a particular owner and group, and allow the owner of the zone and users who are in that group to perform. operations on the zone (regardless of whether they are the owner of the zone).

In short, where zones are now only controllable by root, your changes will allow the owner of a zone and a different group of users to control a zone.

The additional sub-commands you will be implementing are: zone rename, which will change the name of a zone; zone chown, which will change the owner of a zone in a manner similar to the existing chown(8); and zone chgrp, which will change the group of a zone in a manner similar to the existing chgrp(8).

4 Instructions

To complete the assignment, you will need to do the following.

4.1 Apply the diff

- Fetch https :// stluc . manta . uqcloud . net / comp3301 / public /2024/ a1 - zones - base . patch 1

- Create an a1 branch 2

- ‘ git checkout -b a1 openbsd-7.5‘ 3

- Apply the base patch to the a1 branch 4

- ‘ git am / path / to / a1 - zones - base . patch ‘ in / usr / src 5

- Build the kernel 6

- ‘ cd / usr / src / sys / arch / amd64 / compile / GENERIC . MP ‘ 7

- ‘ make obj ‘ 8

- ‘ make config ‘ 9

- ‘ make -j 5 ‘ 10

- ‘ doas make install ‘ 11

- Reboot into the kernel 12

- ‘ doas reboot ‘ 13

- ‘ make obj ‘ in / usr / src 14

- ‘ doas make includes ‘ in / usr / src / include 15

- Verify the zones syscalls are in / usr / include / sys / syscall . h 16

- Verify / usr / include / sys / zones . h exists 17

- Make and install libc 18

- ‘ cd / usr / src / lib / libc ‘ 19

- ‘ make -j 5 ‘ 20

- ‘ doas make install ‘ 21

- Optional : make ps , and pkill / pgrep 22

- make zone (8) 23

- ‘ cd / usr / src / usr . sbin / zone ‘ 24

- ‘ make ‘ 25

- ‘ doas make install ‘ 26

- Verify ‘ zone (8) ‘ and the zones subsystem works : 27

$ zone list 28

ID NAME 29

0 global 30

$ zone create 31

usage : zone create zonename **

$ zone create test 33

zone : create : Operation not permitted 34

$ doas zone create test 35

doas ( dlg@comp3301 . eait . uq . edu . au ) password : 36

$ zone list 37

ID NAME 38

0 global 39

42101 test 40

$ zone id 41

0 42

$ zone id test 43

42101 44

$ zone exec test ps - aux 45

zone : enter : Operation not permitted 46

$ doas zone exec test ps - aux **

USER PID % CPU % MEM VSZ RSS TT STAT STARTED TIME COMMAND 48

root 41705 0.0 0.1 628 580 p0 R + pU /0 3:37 PM 0:00.14 ps - aux 49

$ doas zone exec test zone id 50

42101 51

$ doas zone exec test zone id global 52

zone : id : No such process 53

$ 54

As you add the functionality specified in the next sections, some of these steps will be repeated. eg, changing the kernel means rebuilding and installing the kernel. Adding a syscall means making the syscall stub as a function visible in the headers (make includes), and callable through libc.

A note on errors

We have over-specified the errors you should return from your syscalls - if you do not require an error code (for example, never returning ENOMEM on memory failures because you never allocate any memory) then you do not have to use it. The reverse is also true - if you find an error case that is not listed, choose an appropriate error from errno(2). We will not explicitly test all errors, but during your code interview, we will expect you to be able to explain the suitability of the error codes you use.

4.2 Zone Rename

The zone(8) commands should be extended to enable renaming of zones. Zones should only be able to be renamed by the owner, root, or members of the zone’s group. Additionally, the global zone cannot be renamed, and zone names must be unique.

$ zone 1

usage : zone create zonename 2

zone destroy zonename 3

zone exec zonename command ... 4

zone list 5

zone name [ id ] 6

zone id [ zonename ] 7

zone rename zone newname 8

$ doas zone create foo 9

$ zone list 10

ID NAME 11

0 global 12

289 foo 13

$ doas zone rename 298 bar 14

$ zone list 15

ID NAME 16

0 global 17

289 bar 18

$ doas zone rename 0 something 19

zone : name : Permission denied 20

$ doas zone rename 289 global 21

zone : name : File exists 22

4.3 Modifications to Existing Syscalls

zone_create() syscall

The zone_create() syscall should now ensure that the created zone is associated with the group of the user that created it, as well as the user themself. Additionally, this will mean ensuring that non-root users can create zones. The definition of zone_create() should not change - it should still take a single char *zonename as its argument.

All other syscalls

The full suite of zone_* syscalls should permit users with matching credentials (owner or group) to perform. zone operations on them, not only the root user.

4.4 Zone name and zone list

zone_name() syscall

The zone_name() syscall should be renamed to zone_info(). Subsequently, it should return not only the name and namelen, but a struct, containing the id of the user and the id of the group that has permission to control the zone. The zone(8) userland sub-command for zone name should also be modified in line with these changes - the name should be changed to zone info and the additional information should be provided to the user.

zone list

The zone list subcommand should now take flags: -o and -g. If the -o flag is provided, the owner of the zone should be printed, and if the -g flag is provided, the zone’s group should be printed. If both flags are provided, print both. The extra fields should be printed as extra columns in the current table format.

4.5 Zone chown and chgrp

The zone(8) commands and the kernel zones system should be extended to enable changing the owner and group of a zone. Zone owners and groups should only be able to be changed by the owner, root, or members of the zone’s group. Additionally, the owner of the global zone cannot be changed.

$ zone 1

usage : zone create zonename 2

zone destroy zonename 3

zone exec zonename command ... 4

zone list 5

zone name [zoneid] 6

zone id [ zonename ] 7

zone chown zone user 8

zone chgrp zone group 9

The two subcommands you are adding are zone chown and zone chgrp. zone chown takes the name of a zone and uses the zone_chown() syscall to change its owner to the user with the specified name. If a zone with the name zonename does not exist, zone(8) will attempt to interpret the argument as a numeric zone identifier. zone chgrp behaves similarly, but instead it uses the zone_chgrp() syscall to change the zone’s group to the specified group name.

To support these subcommands, you will need to implement the following system calls:

zone_chown() syscall

int zone_chown ( zoneid_t z , uid_t user ) ;

The zone_chown() syscall alters the owner of the zone identified by the z argument. The new owner should be the owner identified by the user argument. If called from a non-global zone then the z id must be the identifier for the current zone, but in the global zone it can be any zone identifier.

Potential Errors:

 EPERM - the user does not have permission to alter the zone z

 ESRCH - the zone identified by z does not exist

 ENOMEM - the system was not able to allocate memory

 EINVAL - the zone to alter was the global zone

zone_chgrp() syscall

int zone_chgrp ( zoneid_t z , gid_t group ) ;

The zone_chgrp() syscall alters the owner of the zone identified by the z argument. The new owner should be the group identified by the group argument. If called from a non-global zone then the z id must be the identifier for the current zone, but in the global zone it can be any zone identifier.

Potential Errors:

 EPERM - the user does not have permission to alter the zone z

 ESRCH - the zone identified by z does not exist

 ENOMEM - the system was not able to allocate memory

 EINVAL - the zone to alter was the global zone

5 Other Requirements & Suggestions

5.1 Code Style

Your code is to be written according to OpenBSD’s style. guide, as per the style(9) man page.

An automatic tool for checking for style. violations is available at:

https://stluc.manta.uqcloud.net/comp3301/public/2022/cstyle.pl

This tool will be used to calculate your style. marks for this assignment.

5.2 Compilation

Your code for this assignment is to be built on an amd64 OpenBSD 7.5 system identical to your course-provided VM.

The following steps must succeed:

 make obj; make config; make in src/sys/arch/amd64/compile/GENERIC.MP

 make obj; make includes in src

 make obj; make; make install in src/lib/libc

 make obj; make; make install in src/usr.sbin/zone

The existing Makefiles in the provided code are functional as-is, but may need modification as part of your work for this assignment. Note that the existing Makefile ensures the -Wall flag is passed to the compiler, as well as a few other warning and error-related flags.

5.3 Provided code

The provided code which forms the basis for this assignment can be downloaded as a single patch file at:

https://stluc.manta.uqcloud.net/comp3301/public/2024/a**zones-base.patch

You should create a new a1 branch in your repository based on the openbsd-7.5 tag using git checkout, and then apply this base patch using the git am command:

$ git checkout -b a1 openbsd -7.5 1

$ ftp https :// stluc . manta . uqcloud . net / comp3301 / public /2024/ a1 - zones - base . patch 2

$ git am < a1 - zones - base . patch 3

$ git push origin a1 4

5.4 Recommendations

The following order will likely be the most reasonable way to complete this assignment:

1. Download, build, and install the zones patch.

2. Add the zone rename subcommand to zone(8).

3. Minimally modify zone_create() to store credentials.

4. Rewrite zone_name() to zone_info().

This ensures you have a way to view the credentials of a zone.

5. Add the zone_chown() and zone_chgrp() syscalls.

6. Add the corresponding zone chown and zone chgrp commands to zone(8).

7. Fix up any tiny bugs and ensure it’s all working. But you did that as you were going... right?

Additionally, it is strongly recommended (and in some cases, required) that the following APIs be considered for use as part of your changes:

 ucred(9) - provides necessary handlers for dealing with user and group credentials
 copyin(9)/copyout(9) - provides the ability to copy data across the userspace boundary

 user_from_uid(3) - conversions from group/user name to id and back

 strtonum(3) - BSD style. safe string to int conversions

 Finally, you may wish to look at the header file sys/proc.h to see how user and group credentials are currently stored by threads.

6 Reflection

Provide a reflection on your implementation by briefly answering the following questions:

1. Describe the steps you took or draw a flowchart.

2. Describe an error that you encountered.

3. Describe how the error was debugged.

4. Describe how the bug was solved.

Upload both pdf and your answers it as a pdf to the Blackboard a1 reflection submission. Page length is a maximum 2 pages or less. Pdf name must be your STUDENT NUMBER - a1.pdf. Note this is your XXXXXXXX ID number and not sXXXXXXX login.

7 Submission

Submission must be made electronically by committing to your course-provided Git repository on source.eait.uq.edu.au. In order to mark your assignment the markers will check out the a1 branch from your repository. Code checked into any other branch in your repository will not be marked.

As per the source.eait.uq.edu.au usage guidelines, you should only commit source code and Makefiles.

Your a1 branch should consist of:

 The openbsd-7.5 base commit

 The A1 base patch commit

 Your commit(s) for adding the required functionality

7.1 Marking

Your submission will be marked by course tutors and staff, during an in-person demo with you, at your lab session during the due week. You must attend your session, in-person, otherwise your submission will not be marked. Online attendence, e.g. zoom, is not permitted.


請加QQ：99515681  郵箱：99515681@qq.com   WX：codinghelp