CVE-2014-3158 is an integer overflow in the options-file parser of pppd, the PPP daemon in routers, cellular modems and VPN gateways, in ppp before 2.4.7. A word longer than 2^31 characters wraps a length counter, and pppd writes past a heap buffer, which can expose privileged options. NVD scores it 7.5 with CVSS v2 only; ppp 2.4.7 fixes it.
Vulnerability summary
CVSS base score
7.5
High severity
EPSS score (FIRST)
3.5%
Probability of exploitation in the wild in the next 30 days
CISA KEV catalog
No
Not listed
Published
Vulnerability ID: CVE-2014-3158 · Source: NVD
Affected components
| Component | CPE | Version range |
|---|---|---|
| ppp (pppd) options parser | cpe:2.3:a:point-to-point_protocol_project:point-to-point_protocol:*:*:*:*:*:*:*:* | Up to and including 2.4.6 (NVD); fixed by commit 7658e825, first released in ppp 2.4.7 (2014-08-10) |
| Debian ppp package | pkg:deb/debian/ppp | squeeze fixed in 2.4.5-4+deb6u1 (DLA-74-1); wheezy fixed in 2.4.5-5.1+deb7u1 (DSA-3079-1); unstable fixed in 2.4.6-3; bookworm, trixie, forky and sid fixed (Debian security tracker, as of 2026-11-20) |
| Buildroot package/pppd | — | 2.4.6 in Buildroot 2014.05; security bump to 2.4.7 (commit 86b830d7) from Buildroot 2014.08 |
| Yocto Project poky meta/recipes-connectivity/ppp | — | dunfell 2.4.7, kirkstone 2.4.9, scarthgap 2.5.0: all include the fix (as of 2026-11-20) |
What does this mean for affected devices?
NVD describes CVE-2014-3158 as an "Integer overflow in the getword function in options.c in pppd in Paul's PPP Package (ppp) before 2.4.7" that lets attackers "access privileged options" via a long word in an options file, triggering a heap-based buffer overflow. The record was published on 2014-11-15 (NVD, CVE record). pppd is the daemon behind PPPoE WAN links, cellular modem data sessions, serial dial-up and PPTP/L2TP tunnels, so it sits in a large share of router and gateway firmware.
What does the fix change?
Paul Mackerras' commit 7658e825 of 2014-08-01 ("pppd: Eliminate potential integer overflow in option parsing", reported by Lee Campbell of Google) explains the bug. getword() counts the length of a word read from an options file in an int. Past MAXWORDLEN - 1 (1023) it stops storing characters but keeps incrementing the counter, which wraps to -2147483648 after 2147483647; the len < MAXWORDLEN-1 test then passes again and pppd "will start writing characters to memory again". The commit says this "may enable an attacker to overwrite the heap and thereby corrupt security-relevant variables". The fix stops incrementing len once it reaches MAXWORDLEN. Mackerras announced ppp 2.4.7 on 2014-08-10, saying the main reason for the release was this CVE and that he was "not aware of any complete working exploit".
How severe is CVE-2014-3158?
As of 2026-11-20 NVD scores this record with CVSS v2 only: the Primary metric is 7.5 (AV:N/AC:L/Au:N/C:P/I:P/A:P), and the record has no CVSS v3 metric. The CVE record, assigned by Google's Chrome CNA, carries no CNA score. NVD lists the weakness as CWE-119, and its CPE covers ppp up to and including 2.4.6. The network vector overstates the reach: the attacker has to get a word of more than 2 GB into an options file that pppd reads. Debian's DSA-3079-1 describes a local attacker who could gain root privileges, and the Debian security tracker notes a potential local privilege escalation for users in the "dip" group, where pppd is installed setuid root.
Which firmware builds carry it?
Any image whose pppd is older than 2.4.7 and has no backport. On most embedded devices pppd runs as root from the configuration system, so the risk depends on whether untrusted input can reach an options or peers file. The sources linked here name no device vendor products for this CVE as of 2026-11-20. Downstream status:
- Debian fixed squeeze in 2.4.5-4+deb6u1 (DLA-74-1), wheezy in 2.4.5-5.1+deb7u1 (DSA-3079-1) and unstable in 2.4.6-3; bookworm, trixie and later ship fixed packages.
- Buildroot shipped pppd 2.4.6 in 2014.05. Gustavo Zacarias' security bump to 2.4.7 ("Fixes CVE-2014-3158") is in Buildroot 2014.08 and later.
- Yocto Project poky builds ppp 2.4.7 on dunfell, 2.4.9 on kirkstone and 2.5.0 on scarthgap, all past the fix.
- OpenWrt builds 2.4.7 on the openwrt-18.06 branch and newer upstream snapshots on later branches.
How do you fix CVE-2014-3158?
Update to ppp 2.4.7 or later, or backport commit 7658e825 to an older tree. Since every pppd release from 2.4.7 has this fix, a current pppd also picks up later fixes such as CVE-2015-3310 in the RADIUS plugin and the EAP overflow CVE-2020-8597. Our pppd vulnerabilities in embedded firmware guide shows how to read the pppd version from an image and ranks the pppd CVEs; the memory corruption fix guide covers this bug class. Another overflow triggered by parsing a stored file is CVE-2021-25217 in ISC DHCP's lease file code.
Is CVE-2014-3158 exploited?
It is not in the CISA KEV catalog as of 2026-11-20 (catalog version 2026.11.18, 1,786 entries). FIRST EPSS gives it 0.03546, the 89.5th percentile, in the 2026-11-19 dataset (CISA KEV, FIRST EPSS).
Evidence
- NVD (read 2026-11-20): status Modified; published 2014-11-15; NVD Primary CVSS v2 7.5 AV:N/AC:L/Au:N/C:P/I:P/A:P; no CVSS v3 metric; CWE-119 (NVD); CPE point-to-point_protocol_project:point-to-point_protocol up to and including 2.4.6; references commit 7658e825, DSA-3079 and the linux-ppp 2.4.7 release post. — NVD
- CVE record (checked 2026-11-20): state PUBLISHED, CNA Chrome (Google), published 2014-11-15; no CNA CVSS metric. — CVE.org
- ppp commit 7658e825 (Paul Mackerras, 2014-08-01, reported by Lee Campbell of Google), read 2026-11-20: getword() kept incrementing int len past MAXWORDLEN-1, which wraps and re-enables writes, possibly corrupting security-relevant heap variables; fix stops incrementing at MAXWORDLEN. Contained in tag ppp-2.4.7, not in ppp-2.4.6. — ppp-project/ppp
- linux-ppp post 'ppp-2.4.7 released' (Paul Mackerras, 2014-08-10), read 2026-11-20: main reason for the release is CVE-2014-3158, which may let an unprivileged attacker access privileged options; author not aware of a complete working exploit. — linux-ppp mailing list (MARC)
- Debian security tracker and DSA-3079-1 (2014-11-28), read 2026-11-20: local attacker could gain root; squeeze 2.4.5-4+deb6u1 (DLA-74-1), wheezy 2.4.5-5.1+deb7u1 (DSA-3079-1), unstable 2.4.6-3; bookworm 2.4.9-1+1.1, trixie 2.5.2-1+1, forky/sid 2.5.5-1+1 fixed; tracker note on potential local privilege escalation for users in the dip group. — Debian
- Buildroot (read 2026-11-20): package/pppd 2.4.6 in tag 2014.05; commit 86b830d7 'pppd: security bump to version 2.4.7' (Gustavo Zacarias, 2014-08-13, 'Fixes CVE-2014-3158'); 2.4.7 in tag 2014.08. — Buildroot
- Yocto poky (read 2026-11-20): meta/recipes-connectivity/ppp has ppp_2.4.7.bb on dunfell, ppp_2.4.9.bb on kirkstone, ppp_2.5.0.bb on scarthgap. — Yocto Project poky
- OpenWrt (read 2026-11-20): package/network/services/ppp PKG_VERSION 2.4.7 on openwrt-18.06; later branches build upstream git snapshots (2.4.7 2019-05-25 on 19.07, 2.4.8 2020-10-03 on 21.02, 2.4.9 on 22.03). — OpenWrt
- Not in the CISA KEV catalog, version 2026.11.18 (1,786 entries), checked 2026-11-20. — CISA KEV
- EPSS 0.03546, percentile 0.89542, dataset 2026-11-19, read 2026-11-20. — FIRST EPSS

Related advisories
Sources
- NVD
- CVE.org record CVE-2014-3158 (CNA: Chrome)
- ppp commit 7658e825: pppd: Eliminate potential integer overflow in option parsing
- linux-ppp: ppp-2.4.7 released (Paul Mackerras, 2014-08-10)
- Debian security tracker: CVE-2014-3158
- Debian Security Advisory DSA-3079-1: ppp security update
- Buildroot commit 86b830d7: pppd: security bump to version 2.4.7
- Yocto Project poky (dunfell): meta/recipes-connectivity/ppp
- OpenWrt openwrt-18.06: ppp package Makefile
- CISA KEV
- FIRST EPSS
Frequently asked questions
Which ppp version fixes CVE-2014-3158?
ppp 2.4.7, released on 2014-08-10, is the first release with the fix (commit 7658e825). Debian backported it to 2.4.5 for squeeze (2.4.5-4+deb6u1) and wheezy (2.4.5-5.1+deb7u1).
Who can exploit CVE-2014-3158 on a device?
Someone who can place a word longer than 2^31 characters in an options file pppd reads. Debian describes it as a local route to root on systems where pppd is setuid root. On embedded devices pppd usually runs as root from the configuration system, so the question is whether untrusted input can reach its options or peers files.
Why does NVD show only a CVSS v2 score?
As of 2026-11-20 the NVD record for CVE-2014-3158 has a CVSS v2 Primary metric (7.5, AV:N/AC:L/Au:N/C:P/I:P/A:P) and no CVSS v3 metric, and the CNA record has no score. Tools that sort by CVSS v3 alone will show it as unscored.
Is CVE-2014-3158 in CISA KEV, and what is its EPSS score?
It is not in the CISA KEV catalog as of 2026-11-20 (catalog version 2026.11.18). FIRST EPSS gives it 0.03546, the 89.5th percentile, in the 2026-11-19 dataset.
Questions about this page?