CVE-2026-98104 is a vulnerability in the Linux kernel's net/sched cls_u32 module where duplicate handles can be created when the node ID pool is exhausted. This occurs because the function gen_new_kid() returns a fallback handle value instead of reporting an error when ID ranges are full, causing handle uniqueness to break. Similarly, gen_new_htid() can return 0 when the handle pool is full, leading to aliasing of root tables. This flaw can be triggered by adding a large number of u32 filters or proto entries, causing collisions and potential cross-protocol node manipulation.
Rapid7 Bulk Export MCP versions 0.2.5 through 0.6.1 suffer from a GraphQL query injection issue in the export-status component (`get_export_status` in `src/export_manager.py`), whereby the `export_id` value — an unvalidated MCP tool argument reaching the function via the `check_rapid7_export_status` and `download_rapid7_export` tools — is interpolated directly into the GraphQL query string. A crafted `export_id` containing quote and brace characters can terminate the intended `export(id: "...")` selection early and append attacker-controlled root-level selections (for example, schema introspection), producing a single well-formed GraphQL document that is then sent to the Rapid7 export API under the operator's own API key.
Notably, this issue does not grant an existing actor any access they do not already have: every injected query executes within the operator's own already-authenticated API scope, using the operator's own valid API key, and cannot cross a tenant or account boundary. A directly-malicious operator gains nothing they could not already do by calling the API directly; the realistic exposure is limited to a compromised or careless upstream MCP client, or indirect prompt injection forwarding an unvalidated identifier. This is fixed in version 0.6.2, which passes `export_id` as a parameterized GraphQL variable (`$exportId: ID!`).
A vulnerability in the Linux kernel's ASoC Intel avs component involves out-of-bounds access due to improper indexing of the init_configs array. The flaw arises when an ID larger or equal to the number of init_configs is used, leading to potential memory access issues. The issue has been addressed by refactoring the code to avoid storing IDs and instead using a list of pointers to actual config entries, reusing existing verification code.
A vulnerability in the Linux kernel's BPF subsystem involves improper handling of scalar register IDs during linked register collection, which could lead to incorrect register state synchronization. The issue was fixed by changing the order of linked register collection to prevent resurrection of dropped scalar IDs. This fix addresses potential inaccuracies in register state tracking within BPF programs.
CVE-2026-98023 is a vulnerability in the Linux kernel's VXLAN driver where dynamic forwarding database (FDB) entries referencing a nexthop ID were not properly rejected. This flaw could lead to list corruption and use-after-free bugs due to race conditions when multiple VXLAN devices share the same FDB nexthop list. The issue has been fixed by rejecting such invalid dynamic FDB entries and ensuring proper locking during list mutations.
CVE-2026-98002 is a high-severity vulnerability in the Linux kernel's AMD IOMMU nested domain allocation code. The flaw arises from an ineffective error check when allocating a nested domain ID, where a negative error code is incorrectly stored in an unsigned variable, causing the error condition to be missed. This leads to the nested domain being set up with an invalid host domain ID instead of failing properly. The issue has been fixed by correcting the error handling logic to properly check the allocation result before storing it.
A vulnerability in the Linux kernel's netfilter nfnetlink_log subsystem allows a race condition during concurrent instance destruction. This can cause a general protection fault (kernel crash) due to improper handling of reference counts and unlinking from hash tables when multiple sockets interact with the same port ID. The issue has been fixed in specific Linux kernel commits.
A NULL pointer dereference vulnerability was identified and fixed in the Linux kernel's s390/debug subsystem. The issue was introduced by a commit that removed a null-id check from the debug_set_level() function, potentially causing a NULL pointer dereference when debug_register() results were unchecked. The fix restores the null check in the external debug_set_level() wrapper to prevent this condition.
CVE-2026-97598 addresses a vulnerability in the Linux kernel's IPv4 routing table ID allocation mechanism. The automatic allocation of table IDs for IPv4 rules previously searched the entire range up to RT_TABLE_MAX (0xFFFFFFFF), potentially causing unbounded delays while holding the RTNL lock. The fix bounds this search to 4096, limiting the lock hold duration and changing user-visible behavior by causing rule addition to fail with ENOBUFS if the range is exhausted. This automatic allocation path is legacy, IPv4-only, undocumented, and unused by major components like NetworkManager and systemd.
CVE-2026-97580 is a high-severity vulnerability in the Linux kernel's media subsystem, specifically in the rkvdec HEVC decoder. The flaw involves unbounded loops and array indexing based on untrusted picture parameter set (PPS) data, which could lead to out-of-bounds writes in hardware tables. The issue has been fixed by bounding tile loops and validating PPS IDs before indexing arrays.
CVE-2026-97570 is a vulnerability in the Linux kernel's bnxt_en network driver where software TPA IDs are not properly bounded, leading to out-of-bounds memory access. This can cause kernel crashes due to invalid pointer dereferences when the firmware advertises fewer concurrent TPAs than the software expects. The issue is fixed by allocating the rx_tpa array with a size rounded up to the next power of two and masking the firmware TPA ID accordingly.
A vulnerability in the Linux kernel's net: usb: pegasus driver involves incorrect pointer arithmetic on the usb_device_id table when dynamic IDs are used. This flaw causes out-of-bound indexing due to the usb_device_id parameter residing on the heap. The issue has been addressed by changing the handling of dynamic IDs to use the driver_info field and removing problematic pointer arithmetic.
A vulnerability in the Linux kernel's xusbatm USB driver was fixed to prevent reliance on unsafe pointer arithmetic involving dynamic USB device ID tables. The flaw caused out-of-bounds pointer arithmetic when dynamic IDs were used, which the driver could not properly support. The fix involved avoiding this pointer arithmetic and disabling dynamic ID support for xusbatm.
CVE-2026-97531 is a high-severity vulnerability in the Linux kernel's qla2xxx SCSI driver affecting version 4.11. It involves improper handling of virtual ports (vports) during report ID acquisition, where a vport marked for deletion is not properly skipped, leading to use-after-free conditions. This can cause the kernel to dereference freed memory, potentially resulting in system crashes or other impacts.
The cisco_firesight_manager_ACL_rule_export module in misp-modules generates a shell script (.sh) that authenticates to and calls the Cisco fireSIGHT Manager API. The module interpolates configuration values (IP address, login, password, domain ID, policy ID) and MISP attribute values (destination IPs, URLs, event info comments) directly into single-quoted shell string assignments without any escaping or sanitization. Because the values are placed inside single-quoted shell strings, any value containing a single-quote character (e.g., a crafted ip-dst or url attribute value submitted to MISP) breaks out of the quoting context, allowing an attacker to inject arbitrary shell commands into the exported script. A security analyst who subsequently executes the generated .sh file unmodified would run the injected commands with their own privileges, potentially exposing fireSIGHT Manager credentials, modifying ACL rules, or compromising the analyst workstation. Additionally, the module contained a secondary defect where the variable 'config' was only assigned inside a conditional block but referenced unconditionally afterward, causing a NameError (denial of service) when the request payload lacked a 'config' key. The vulnerability requires the attacker to have the ability to submit MISP events or attributes containing a single-quote character and the victim to execute the exported script. No authentication bypass is required beyond standard MISP event-submission privileges.
The Knit Pay – Cashfree, Instamojo, Razorpay, PayPal and more plugin for WordPress is vulnerable to Privilege Escalation in all versions up to, and including, 9.6.1.0. This is due to the `maybe_update_user_role()` function reading the target role directly from an attacker-controlled Gravity Forms entry field — configured via the feed's `user_role_field_id` — and passing it to `WP_User::set_role()` without validating the supplied value against an allowlist of permitted roles. This makes it possible for authenticated attackers, with Subscriber-level access and above, to elevate their privileges to administrator by tampering with the hidden role field value at form submission time. Exploitation is further enabled by the fact that $0 orders are synchronously marked as SUCCESS during form submission without requiring a real payment, and when no GF User Registration user can be resolved, the role assignment target falls back to `$lead['created_by']` — the currently authenticated submitter's own user ID — making any authenticated form submitter an eligible exploitation target.
Uncontrolled Resource Consumption (CWE-400 / CWE-407) in the ANTLR 3 search query parser (QueryParser / Query.g) in Gerrit Code Review versions 2.0.19 through 3.12.9, 3.13.0 through 3.13.8, and 3.14.0 through 3.14.2 allows an unauthenticated remote attacker (or an authenticated user if anonymous read access is disabled) to cause a persistent denial of service (CPU exhaustion and HTTP worker thread pool starvation requiring a server restart) via crafted search queries containing deeply nested parentheses sent to query evaluation endpoints (/changes/?q=, /accounts/?q=, /groups/?query=, /projects/?query=, /Documentation/?q=, /changes/{id}/query?expression=, or SSH gerrit query). Because syntactic predicates in conditionOr and conditionAnd recurse via conditionBase without memoization prior to capability or visibility checks and worker threads do not abort when the client disconnects, a small number of requests (such as 25 requests matching default httpd.maxThreads) can permanently pin all HTTP worker threads. This issue is fixed in Gerrit Code Review versions 3.12.10, 3.13.9, and 3.14.3.
CVE-2026-97324 is an improper authorization vulnerability in the ruoyi-vue-pro product by YunaiV, specifically affecting version 2026.08. The flaw exists in the updateDemoOrderPaid function of the Demo-order Payment Callback Handler component, where manipulation of the ID argument allows unauthorized actions. The vulnerability can be exploited remotely without authentication. Public exploit code is available, but no vendor response or patch has been provided.
The EmbedVideo Extension is a MediaWiki extension which adds a parser function called #ev and various parser tags for embedding video clips from various video sharing services. Prior to 4.1.0, with $wgEmbedVideoRequireConsent disabled (not the default), the urls for videos are passed into an iframe src attribute without sanitization. When given a malformed url or id, the src attribute can be escaped via double quotes, allowing for html/javascript injection. Version 4.1.0 contains a patch.
A vulnerability in the Linux kernel's drm/amdkfd component allows cross-process information leakage via SMI events. The flaw occurs because the function kfd_smi_ev_enabled() skips a superuser privilege check when the process ID (pid) is zero. This causes PROCESS_START, PROCESS_END, and VMFAULT events to be emitted with pid=0 but containing another process's PID and command name, enabling any user in the render group with access to /dev/kfd to monitor all GPU workloads.