MCUboot's Zephyr serial-recovery upload handler bs_upload() in boot/boot_serial/src/boot_serial.c decodes the host-supplied image field of an MCUmgr image-upload request into img_num (a full 32-bit value) and latches it on the off == 0 packet without any range check. It then passes img_num straight to flash_area_id_from_multi_image_slot(img_num, 0), which on the Zephyr port resolves to __flash_area_ids_for_slot() in boot/zephyr/include/sysflash/sysflash.h and indexes the static const int all_slots[] table as all_slots[img_num 2]. That table holds only MCUBOOT_IMAGE_NUMBER 2 entries, so any img_num >= MCUBOOT_IMAGE_NUMBER is an out-of-bounds read.
The path is reachable by anyone with access to the serial / SMP recovery console, which is unauthenticated by design, and the bad value is supplied before any image signature is verified. The defect is specific to the default non-direct upload path (MCUBOOT_SERIAL_DIRECT_IMAGE_UPLOAD disabled) on the Zephyr port with the array-backed slot layout (not CONFIG_SINGLE_APPLICATION_SLOT). It is not gated on a multi-image configuration: MCUBOOT_IMAGE_NUMBER defaults to 1, so even a default single-image build is affected for any image >= 1. The MCUBOOT_SERIAL_DIRECT_IMAGE_UPLOAD path is safe (its flash_area_id_from_direct_image() switch returns -EINVAL for unknown indices).
flash_area_id_from_multi_image_slot() takes a signed int image_index while img_num is uint32_t, so the attacker controls the sign of the index as well as its magnitude: an image of 0xFFFFFFFE converts to -2 and reads all_slots[-4], off the front of the table, while large positive values read off the end (CWE-192). (0xFFFFFFFF itself is the "no image given" sentinel and selects image 0.) Either way the index multiply img * 2 is performed in int, so a wild offset yields a bus/hard fault, giving a reliable denial of service of the bootloader during recovery. For moderately out-of-range indices the read returns adjacent .rodata as a garbage flash-area id that is handed to flash_area_open(); this normally fails with MGMT_ERR_EINVAL, but if the garbage id happens to match a valid non-image partition, the subsequent erase/write can target a partition outside the image set (CWE-787). The attacker-supplied value is never echoed back, so there is no information disclosure. The fix is to reject img_num >= BOOT_IMAGE_NUMBER on the off == 0 packet, before the slot lookup. The comparison must be made on the unsigned img_num: an equivalent guard applied after the conversion to int would not reject the negative case above. __flash_area_ids_for_slot() should bounds-check its own arguments as defence in depth.
This vulnerability was reported independently twice: by NVIDIA Project Vanessa on 2026-06-22, and by Kai Phan (@kaimandalic) on 2026-07-19 as GHSA-8v33-88p6-jffj, which is closed as a duplicate of this advisory. Both original reports are reproduced in full below.
A denial of service vulnerability exists in XZ Utils due to a buffer overflow in index decoding. This flaw affects Red Hat Enterprise Linux 10.0 Extended Update Support across multiple architectures. The vulnerability is rated as moderate severity by Red Hat Product Security. A security update addressing this issue is available.
pgEdge RAG Server versions 1.0.0 and earlier have an unbounded corpus read vulnerability at the query endpoint, allowing unauthenticated callers to exhaust database and server resources. The server rebuilds an in-memory BM25 index from the entire corpus on every request without limiting the read size, causing resource consumption to scale with corpus size. Hybrid search is enabled by default and cannot be disabled per request, increasing exposure. This can lead to a denial-of-service condition. The issue is fixed in version 2.0.0 by introducing a configurable limit on documents read and allowing disabling the expensive keyword search per query.
BACnet Stack is a BACnet open source protocol stack C library for embedded systems. Prior to versions 1.4.5, 1.5.1, and 1.6.0, in the official apps/epics client tool on v1.6.0-rc1 (not in the generic BACnet RPM ACK decoder alone) a malicious BACnet/IP endpoint can reply to the EPICS tool’s legitimate ReadPropertyMultiple request with a syntactically valid ReadPropertyMultiple-ACK that contains more heading properties than the tool’s fixed local storage can hold. The response is decoded successfully and later consumed by ProcessRPMData(), which writes property values into a fixed global array using an unbounded linear index. This results in a reproducible global-buffer-overflow in apps/epics/main.c. This is a network-reachable memory corruption in the official epics tool’s response-processing path. The root cause is that apps/epics treats a remotely supplied variable-length RPM property list as if it were a trusted, fixed-size local heading-property set. This issue has been patched in versions 1.4.5, 1.5.1, and 1.6.0.
pgEdge RAG Server versions 1.0.0 and earlier have a concurrency flaw causing shared keyword index leakage between unauthenticated concurrent requests. This allows documents from one caller to be returned to another, defeating tenant or customer scoping filters without requiring injection or malformed input. The issue is fixed in version 2.0.0 by building the index per request instead of sharing it. Workarounds include disabling the hybrid keyword index feature, but upgrading is strongly recommended.
Grafana is an open-source platform for monitoring and observability.
When an Elasticsearch data source is configured with a time-based index pattern, Grafana
generated one index name for every interval in the requested query time range without any upper
limit. A user with permission to query the data source, which includes users holding only the
Viewer role in a default configuration, could submit a single small query with an extreme time
range and cause the server to allocate memory until it was terminated. The allocation occurred
before any request was sent to Elasticsearch, so the condition applied whether or not the
Elasticsearch cluster was reachable.
Data sources configured with a static index name are not affected.
A vulnerability in PostgreSQL's REINDEX CONCURRENTLY operation on a dropped index allows an object creator to execute arbitrary code as the operating system user running the database. This issue affects versions prior to PostgreSQL 18.7, 17.12, 16.16, 15.20, and 14.25. The flaw arises from an index expression that issues DROP INDEX, enabling code execution.
SQL injection in PostgreSQL reindexdb allows an object creator to execute arbitrary SQL as the database user running reindexdb, via a crafted index name embedded in the parallel catalog query for --jobs with --index. Within major versions 17-18, minor versions before PostgreSQL 18.7 and 17.12 are affected. Versions before PostgreSQL 17 are unaffected.
A cleartext storage flaw was found in the AAP Gateway Operator. When reconciling an AutomationController, the operator reads the protected resource-server Secret (Secret/<aap-name>-resource-server, key controller_resource_server_secret) and copies its plaintext value into the generated AutomationController custom resource under spec.extra_settings (RESOURCE_SERVER.SECRET_KEY). Although the operator uses no_log to keep the value out of its own logs, serializing it into the custom resource exposes it to any principal able to read the AutomationController CR without permission to read the backing Kubernetes Secret, breaking the expected OpenShift RBAC separation between the two resources. The value may also propagate into Kubernetes API/etcd storage, GitOps and Argo diffs and history, rendered manifests, support bundles, and audit data. The resource-server secret is intended for trusted Gateway-to-Controller service authentication, and its disclosure may allow a user to impersonate a trusted Controller service and reach privileged service-index APIs.
Out-of-bounds Read vulnerability in the Guththila XML parser wrapper of Apache Axis2/C. On some malformed XML input the Guththila parser returned a raw input byte in place of an event code, and the AXIOM reader wrapper used that value as an index into a ten-entry lookup table without checking its range, reading heap memory past the end of the table. An unauthenticated client could trigger this by sending a crafted XML request to a server built with Guththila, the default parser, causing the request to be parsed incorrectly or, in some cases, the process to crash. The memory read is not returned to the client. This issue affects Apache Axis2/C: through 2.0.0. The fix is available in commits 23110287924e79d05a5b4712298c355d1f05c606 and f01683ec10e8fddb87279ec9351b3239021223e6 and will be included in the next release.
A security vulnerability has been detected in Moonshot AI Kimi Code up to 2.5.2. This affects the function isOriginAllowed of the file apps/kimi-code/dist-web/assets/index-HU0LCM-X.js of the component Web UI. Such manipulation of the argument kimi_origin leads to open redirect. The attack can be executed remotely. The exploit has been disclosed publicly and may be used. The vendor was contacted early about this disclosure but did not respond in any way.
An authorization bypass vulnerability exists in website-index.php in Revive Adserver version 6.0.8 and earlier. This flaw allows an authenticated manager, who is supposed to be restricted to one agency, to view website listings of other agencies by modifying the agencyid request parameter. Although detailed website pages still enforce ownership checks, the disclosed information includes website names, identifiers, and timestamps. The issue is resolved in version 6.0.9 by deriving the agency identifier from the authenticated account rather than the request parameter.
Apache Roller versions 4.0 through 6.1.6 contain a vulnerability where an unauthenticated user can access blog entries that have been withdrawn to draft status via the AtomPub API. Although these entries are removed from the weblog, they remain in the search index and can be viewed through public search results. AtomPub is disabled by default. The issue is fixed in version 6.1.7, which restricts search results to published entries only.
Apache Roller versions 4.0 through 6.1.6 contain a vulnerability that allows unauthenticated visitors to retrieve email addresses of commenters who left approved comments. This occurs because the search index stores commenter email addresses and the public weblog search accepts queries on that field, including wildcard queries that can confirm email addresses character by character. The vulnerability is classified as CWE-200, Exposure of Sensitive Information to an Unauthorized Actor. Users are advised to upgrade to version 6.1.7 or later and rebuild the search index to mitigate the issue.