CVE-2020-12352, named BadChoice, is an information leak in the Linux kernel's Bluetooth A2MP code: a nearby attacker who knows a device's Bluetooth address can read uninitialized kernel stack memory, which reveals pointers that defeat KASLR. NVD scores it 6.5, and Google says it affects Linux 3.6 and later, a wider range than NVD's.
Vulnerability summary
CVSS base score
6.5
Medium severity
EPSS score (FIRST)
5.8%
Probability of exploitation in the wild in the next 30 days
CISA KEV catalog
No
Not listed
Published
Vulnerability ID: CVE-2020-12352 · Source: NVD
Affected components
| Component | CPE | Version range |
|---|---|---|
| Linux kernel Bluetooth subsystem (net/bluetooth/a2mp.c, CONFIG_BT_HS) | cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:* | 3.6 and later per Google; NVD: 5.4 to before 5.4.72, 5.8 to before 5.8.16, 5.9 through 5.9.13; fixed by mainline commit eddb773 |
| Dell Hybrid Client | — | 1.0 and 1.1; remediated by a Security Addon (Dell DSA-2021-003) |
What does this mean for affected devices?
CVE-2020-12352 is the information-leak half of BleedingTooth. Google's advisory GHSA-7mh3-gq28-gfrq describes "a stack-based information leak affecting Linux kernel 3.6 and higher" in net/bluetooth/a2mp.c: an attacker in short distance who knows the victim's Bluetooth address can "retrieve kernel stack information containing various pointers that can be used to predict the memory layout and to defeat KASLR", and the leak "may contain other valuable information such as the encryption keys". The BleedingTooth write-up uses it to make the BadKarma flaw, CVE-2020-12351, reliably exploitable.
NVD scores it 6.5 (medium) with vector CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N, weakness CWE-909 (missing initialization of a resource) (NVD). NVD's configuration, though, only covers 5.4 to before 5.4.72, 5.8 to before 5.8.16 and 5.9 through 5.9.13. A scanner that relies on NVD ranges can therefore miss the flaw on 4.9, 4.14 or 4.19 kernels, which are common in older BSPs and fall inside Google's "3.6 and higher". The fix is mainline commit eddb773, "Bluetooth: A2MP: Fix not initializing all members", dated 2020-08-06, and the Debian tracker lists it in stretch 4.9.240-1 (DLA-2420-1) and buster 4.19.152-1 (DSA-4774-1).
A kernel option narrows the exposure. A2MP serves Bluetooth High Speed, and the same fix series included "Bluetooth: Disable High Speed by default", which notes that High Speed hardware "is very uncommon nowadays"; a later commit, "Bluetooth: Remove BT_HS", removed the code because the feature left the Bluetooth Core specification in 5.3. A kernel built without CONFIG_BT_HS does not carry the A2MP channel.
For devices, Dell's DSA-2021-003 lists this CVE with CVE-2020-12351 for Dell Hybrid Client 1.0 and 1.1, fixed by a Security Addon. OpenEmbedded-Core's bluez5 recipe marks it cpe-incorrect because "This issue has kernel fixes rather than bluez fixes", so Yocto builds track it on the kernel, not on BlueZ.
The CVE is not in the CISA KEV catalog as of 2026-11-16 (catalog version 2026.11.13), and FIRST EPSS gives it 0.0576, the 93rd percentile, in the 2026-11-15 dataset (CISA KEV, FIRST EPSS).
Device makers should check CONFIG_BT_HS first, then update or patch the kernel with eddb773 regardless of what an NVD range match says. Our BlueZ and Linux Bluetooth vulnerabilities in embedded devices guide explains the NVD range gap and the kernel checks, and our CSAF and VEX for device makers guide shows how to state "not affected" for a kernel built without Bluetooth.
Evidence
- Stack-based information leak in net/bluetooth/a2mp.c affecting Linux 3.6 and higher; leaks pointers that defeat KASLR, possibly encryption keys. — Google Security Research GHSA-7mh3-gq28-gfrq
- NVD CVSS v3.1 6.5 (CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N), CWE-909; ranges 5.4 to before 5.4.72, 5.8 to before 5.8.16, 5.9 through 5.9.13. — NVD
- Fix commit eddb773 (2020-08-06): A2MP: Fix not initializing all members. — Linux kernel commit eddb773
- Commit b176dd0 disables Bluetooth High Speed by default; commit e7b0229 removes BT_HS. — Linux kernel commits b176dd0 and e7b0229
- Dell Hybrid Client 1.0 and 1.1 affected; Security Addon remediation. — Dell DSA-2021-003
- OpenEmbedded-Core bluez5 recipe: CVE_STATUS cpe-incorrect, kernel fixes rather than bluez fixes. — OpenEmbedded-Core bluez5 recipe
- Debian fixed stretch in 4.9.240-1 (DLA-2420-1) and buster in 4.19.152-1 (DSA-4774-1). — Debian Security Tracker
- Not in the CISA KEV catalog as of 2026-11-16 (catalog version 2026.11.13). — CISA KEV
- EPSS 0.0576 (93rd percentile), dataset 2026-11-15, read 2026-11-16. — FIRST EPSS

Related advisories
Sources
- NVD
- Google Security Research GHSA-7mh3-gq28-gfrq (BadChoice)
- Google Security Research: BleedingTooth write-up
- Linux commit eddb773: A2MP: Fix not initializing all members
- Linux commit b176dd0: Disable High Speed by default
- Linux commit e7b0229: Remove BT_HS
- Dell DSA-2021-003: Dell Hybrid Client, Intel BlueZ vulnerabilities
- OpenEmbedded-Core: bluez5 recipe
- Debian Security Tracker
- CISA KEV
- FIRST EPSS
Frequently asked questions
Why can an NVD-based scan miss CVE-2020-12352 on older kernels?
NVD's configuration starts at Linux 5.4, while Google's advisory says the leak affects Linux 3.6 and higher. A 4.14 or 4.19 BSP kernel without commit eddb773 can be affected even though it matches no NVD range (per NVD and Google Security Research).
Does turning off CONFIG_BT_HS mitigate CVE-2020-12352?
It removes the exposure. The leak is in the A2MP code that serves Bluetooth High Speed; the fix series disabled High Speed by default and a later commit removed BT_HS (per Linux commits b176dd0 and e7b0229). Patch the kernel as well.
Which devices does a vendor list as affected by CVE-2020-12352?
Dell's DSA-2021-003 lists Dell Hybrid Client 1.0 and 1.1 with a Security Addon fix (per Dell).
Is CVE-2020-12352 in the CISA KEV catalog?
No, not as of 2026-11-16 (catalog version 2026.11.13). FIRST EPSS gives it 0.0576, the 93rd percentile, in the 2026-11-15 dataset (per CISA KEV and FIRST EPSS).
Questions about this page?