{"id":"CVE-2026-11809","aliases":["EUVD-2026-55843","GHSA-6r86-hvv2-h6g4"],"status":"published","title":"UpdateHub probe: uninitialized-heap out-of-bounds read of network-supplied metadata","description":"The UpdateHub OTA client in subsys/mgmt/updatehub/updatehub.c contains an out-of-bounds / uninitialized-memory read in z_impl_updatehub_probe(). The probe response from the UpdateHub server is copied into a heap buffer (metadata) that is correctly NUL-terminated, but a second buffer (metadata_copy) is allocated with k_malloc (unzeroed) and filled with memcpy(metadata_copy, metadata, strlen(metadata)), which omits the terminating NUL. Everything after the copied content remains uninitialized heap.\n\nWhen the first json_obj_parse() over the array descriptor fails, the code falls back to json_obj_parse(metadata_copy, strlen(metadata_copy), ...). The strlen() call scans past the copied bytes through uninitialized heap and, if no zero byte is found before the end of the allocation, reads beyond the buffer; the resulting over-long length is then parsed as JSON. The probe payload is fully controlled by the (malicious, compromised, or — without the optional CONFIG_UPDATEHUB_DTLS — on-path) UpdateHub server, which can craft a large payload that fails the first parse to drive this path.\n\nThe consequence is a read of uninitialized heap, with a worst case of an out-of-bounds read past the metadata_copy allocation that can fault and crash the update thread/device, producing a network-triggerable denial of service. The over-read data is consumed only internally to evaluate the update and is not returned to the attacker, so there is no direct information disclosure and no out-of-bounds write.\n\nThe fix zeroes metadata_copy with memset before the copy, guaranteeing NUL termination and bounding strlen() within the allocation.","severity":{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:L","level":"low","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-10","score":0.00353,"percentile":0.29311,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":false},"cwes":["CWE-125","CWE-908"],"published":"2026-08-10T20:36:33Z","modified":"2026-11-04T04:59:16Z","products":[{"vendor":"zephyrproject","product":"zephyr"},{"vendor":"zephyrproject-rtos","product":"zephyr"}],"references":[{"url":"https://github.com/zephyrproject-rtos/zephyr/commit/69bbed5e601495a2076f25d822adbf991cae5c01","tags":["patch","fix"]},{"url":"https://github.com/zephyrproject-rtos/zephyr/security/advisories/GHSA-6r86-hvv2-h6g4","tags":["advisory"]},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/11xxx/CVE-2026-11809.json","tags":["advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-11809","tags":["advisory"]},{"url":"https://github.com/zephyrproject-rtos/zephyr","tags":["package"]}],"sources":{"cve":[{"id":"CVE-2026-11809","title":"UpdateHub probe: uninitialized-heap out-of-bounds read of network-supplied metadata","description":"The UpdateHub OTA client in subsys/mgmt/updatehub/updatehub.c contains an out-of-bounds / uninitialized-memory read in z_impl_updatehub_probe(). The probe response from the UpdateHub server is copied into a heap buffer (metadata) that is correctly NUL-terminated, but a second buffer (metadata_copy) is allocated with k_malloc (unzeroed) and filled with memcpy(metadata_copy, metadata, strlen(metadata)), which omits the terminating NUL. Everything after the copied content remains uninitialized heap.\n\nWhen the first json_obj_parse() over the array descriptor fails, the code falls back to json_obj_parse(metadata_copy, strlen(metadata_copy), ...). The strlen() call scans past the copied bytes through uninitialized heap and, if no zero byte is found before the end of the allocation, reads beyond the buffer; the resulting over-long length is then parsed as JSON. The probe payload is fully controlled by the (malicious, compromised, or — without the optional CONFIG_UPDATEHUB_DTLS — on-path) UpdateHub server, which can craft a large payload that fails the first parse to drive this path.\n\nThe consequence is a read of uninitialized heap, with a worst case of an out-of-bounds read past the metadata_copy allocation that can fault and crash the update thread/device, producing a network-triggerable denial of service. The over-read data is consumed only internally to evaluate the update and is not returned to the attacker, so there is no direct information disclosure and no out-of-bounds write.\n\nThe fix zeroes metadata_copy with memset before the copy, guaranteeing NUL termination and bounding strlen() within the allocation.","status":"published","published":"2026-08-10T20:36:33.022Z","modified":"2026-11-04T04:59:16.61Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:L","level":"low","assessor":"cna"}],"cwes":["CWE-125","CWE-908"],"updated_at":"2026-11-04T05:21:48.223566Z"}],"euvd":[{"id":"EUVD-2026-55843","description":"The UpdateHub OTA client in subsys/mgmt/updatehub/updatehub.c contains an out-of-bounds / uninitialized-memory read in z_impl_updatehub_probe(). The probe response from the UpdateHub server is copied into a heap buffer (metadata) that is correctly NUL-terminated, but a second buffer (metadata_copy) is allocated with k_malloc (unzeroed) and filled with memcpy(metadata_copy, metadata, strlen(metadata)), which omits the terminating NUL. Everything after the copied content remains uninitialized heap.\n\nWhen the first json_obj_parse() over the array descriptor fails, the code falls back to json_obj_parse(metadata_copy, strlen(metadata_copy), ...). The strlen() call scans past the copied bytes through uninitialized heap and, if no zero byte is found before the end of the allocation, reads beyond the buffer; the resulting over-long length is then parsed as JSON. The probe payload is fully controlled by the (malicious, compromised, or — without the optional CONFIG_UPDATEHUB_DTLS — on-path) UpdateHub server, which can craft a large payload that fails the first parse to drive this path.\n\nThe consequence is a read of uninitialized heap, with a worst case of an out-of-bounds read past the metadata_copy allocation that can fault and crash the update thread/device, producing a network-triggerable denial of service. The over-read data is consumed only internally to evaluate the update and is not returned to the attacker, so there is no direct information disclosure and no out-of-bounds write.\n\nThe fix zeroes metadata_copy with memset before the copy, guaranteeing NUL termination and bounding strlen() within the allocation.","status":"published","published":"2026-08-10T20:36:33Z","modified":"2026-11-04T04:59:16Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:L","level":"low"}],"updated_at":"2026-11-04T07:00:56.720215Z"}],"nvd":[{"id":"CVE-2026-11809","description":"The UpdateHub OTA client in subsys/mgmt/updatehub/updatehub.c contains an out-of-bounds / uninitialized-memory read in z_impl_updatehub_probe(). The probe response from the UpdateHub server is copied into a heap buffer (metadata) that is correctly NUL-terminated, but a second buffer (metadata_copy) is allocated with k_malloc (unzeroed) and filled with memcpy(metadata_copy, metadata, strlen(metadata)), which omits the terminating NUL. Everything after the copied content remains uninitialized heap.\n\nWhen the first json_obj_parse() over the array descriptor fails, the code falls back to json_obj_parse(metadata_copy, strlen(metadata_copy), ...). The strlen() call scans past the copied bytes through uninitialized heap and, if no zero byte is found before the end of the allocation, reads beyond the buffer; the resulting over-long length is then parsed as JSON. The probe payload is fully controlled by the (malicious, compromised, or — without the optional CONFIG_UPDATEHUB_DTLS — on-path) UpdateHub server, which can craft a large payload that fails the first parse to drive this path.\n\nThe consequence is a read of uninitialized heap, with a worst case of an out-of-bounds read past the metadata_copy allocation that can fault and crash the update thread/device, producing a network-triggerable denial of service. The over-read data is consumed only internally to evaluate the update and is not returned to the attacker, so there is no direct information disclosure and no out-of-bounds write.\n\nThe fix zeroes metadata_copy with memset before the copy, guaranteeing NUL termination and bounding strlen() within the allocation.","status":"published","published":"2026-08-10T21:17:18.19Z","modified":"2026-11-04T05:18:29.98Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:L","level":"low","assessor":"secondary"}],"cwes":["CWE-125","CWE-908"],"updated_at":"2026-11-04T06:53:18.101176Z"}],"osv":[{"id":"CVE-2026-11809","title":"UpdateHub probe: uninitialized-heap out-of-bounds read of network-supplied metadata","description":"The UpdateHub OTA client in subsys/mgmt/updatehub/updatehub.c contains an out-of-bounds / uninitialized-memory read in z_impl_updatehub_probe(). The probe response from the UpdateHub server is copied into a heap buffer (metadata) that is correctly NUL-terminated, but a second buffer (metadata_copy) is allocated with k_malloc (unzeroed) and filled with memcpy(metadata_copy, metadata, strlen(metadata)), which omits the terminating NUL. Everything after the copied content remains uninitialized heap.\n\nWhen the first json_obj_parse() over the array descriptor fails, the code falls back to json_obj_parse(metadata_copy, strlen(metadata_copy), ...). The strlen() call scans past the copied bytes through uninitialized heap and, if no zero byte is found before the end of the allocation, reads beyond the buffer; the resulting over-long length is then parsed as JSON. The probe payload is fully controlled by the (malicious, compromised, or — without the optional CONFIG_UPDATEHUB_DTLS — on-path) UpdateHub server, which can craft a large payload that fails the first parse to drive this path.\n\nThe consequence is a read of uninitialized heap, with a worst case of an out-of-bounds read past the metadata_copy allocation that can fault and crash the update thread/device, producing a network-triggerable denial of service. The over-read data is consumed only internally to evaluate the update and is not returned to the attacker, so there is no direct information disclosure and no out-of-bounds write.\n\nThe fix zeroes metadata_copy with memset before the copy, guaranteeing NUL termination and bounding strlen() within the allocation.","status":"published","published":"2026-08-10T20:36:33.022Z","modified":"2026-11-05T08:04:56.823988548Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:L","level":"low"}],"cwes":["CWE-125","CWE-908"],"updated_at":"2026-11-05T09:08:14.330889Z"}]}}