bluez: Whitelist CVE-2020-12351 and CVE-2020-12352
Quoted 32 lines · Robert Joslyn
According to the Intel security advisory [1], these CVEs are mitigated by the following kernel commits:
eddb7732119d53400f48a02536a84c509692faa8Bluetooth: A2MP: Fix not initializing all membersf19425641cb2572a33cb074d5e30283720bd4d22Bluetooth: L2CAP: Fix callingsk_filteron non-socket based channelb560a208cda0297fef6ff85bbfd58a8f0a52a543Bluetooth: MGMT: Fix not checking ifBT_HSis enableda2ec905d1e160a33b2e210e45ad30445ef26ce0eBluetooth: fix kernel oops instore_pending_adv_reportThe latest of these commits were backported from 5.10 to the stable kernel tree in the 5.8.16 and 5.4.72 releases. Since the kernels provied by OE-core contain these fixes, mark them as whitelisted.
Signed-off-by Robert Joslyn <[email protected]>---
meta/recipes-connectivity/bluez5/bluez5_5.55.bb | 2 ++1 file changed, 2 insertions(+)meta/recipes-connectivity/bluez5/bluez5_5.55.bb +0
-0@@ -3,6 +3,8 @@ require bluez5.incSRC_URI[md5sum] = "
94972b8bc7ade60c72b0ffa6ccff2c0a"
SRC_URI[sha256sum] = "8863717113c4897e2ad3271fc808ea245319e6fd95eed2e934fae8e0894e9b88"meta/recipes-connectivity/bluez5/bluez5_5.55.bb +2
-0+CVE_CHECK_WHITELIST += "CVE-2020-12351 CVE-2020-12352" +
Quoted 5 lines
# noinst programs in Makefile.tools that are conditional on READLINE # support NOINST_TOOLS_READLINE ?= " \Signature · 1 line
2.26.2
I’m not sure if this is the best solution, but figured I’d send the patch and see what others think. The CVEs call out bluez running on the Linux kernel in the vulnerable CPE names, but no versions are listed. It seems to me like the CPE names associated with the CVE should really just be against the kernel, with appropriate version numbers added. Is this something likely to be updated in the CVE database, or is whitelisting them here the best option? Does cve-check.bbclass handle the “bluez running on/with linux_kernel” case described in the CVE?
Thanks,
Robert