Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-84748 Overview
CVE-2026-84748 is an improper authorization vulnerability [CWE-285] in Apache bRPC versions prior to 1.18.0. The flaw allows remote attackers to access builtin services through the public port, bypassing the ServerOptions.internal_port gating mechanism. Builtin services such as brpc.vars and hotspots expose runtime metrics, flags, and profiler data that operators intend to restrict to the internal port. Any client reachable on the public port can invoke these services over protobuf-based protocols and disclose sensitive operational information.
Critical Impact
Remote attackers on the public port can invoke bRPC builtin services and extract runtime metrics, configuration flags, and profiler data without authentication.
Affected Products
- Apache bRPC versions prior to 1.18.0
- Deployments configured with ServerOptions.internal_port relying on it to gate builtin or Tabbed services
- Servers exposing any protobuf-based protocol (baidu_std, hulu_pbrpc, sofa_pbrpc, public_pbrpc, nshead_mcpack, nova_pbrpc, ubrpc2pb) on an untrusted network
Discovery Timeline
- 2026-10-21 - CVE-2026-84748 published to NVD
- 2026-10-22 - Last updated in NVD database
Technical Details for CVE-2026-84748
Vulnerability Analysis
Apache bRPC provides builtin services for runtime introspection, including variable exports (brpc.vars), hotspot profiling, and flag inspection. Operators scope these services to an internal management port by configuring ServerOptions.internal_port. The HTTP and HTTP/2 dispatch paths correctly honor this restriction by checking the security_mode flag before invoking a resolved method.
The protobuf-based dispatch paths deviate from this behavior. They compute security_mode during request handling but never consult it between method resolution and invocation. Because builtin services share the method map with user-defined services, a remote client on the public port can name a builtin method directly over a pb protocol and reach it.
The resulting disclosure includes performance counters, server configuration flags, and CPU or heap profiler output. This data assists adversaries in fingerprinting the deployment, locating other services, and identifying weak configuration prior to further exploitation.
Root Cause
The root cause is a missing authorization check [CWE-285] in the protobuf protocol dispatch pipeline. The security_mode flag that distinguishes public and internal callers is computed but not enforced for baidu_std, hulu_pbrpc, sofa_pbrpc, and the nshead-based protocols. Only the HTTP and h2 paths enforced the restriction, producing an inconsistent trust boundary across protocols sharing the same method registry.
Attack Vector
An attacker who can reach the public port sends a protobuf-formatted request naming a builtin method such as brpc.vars.default_method over any affected pb protocol. The server resolves the method against the shared method map and invokes it without checking that the caller originated from the internal port. The response returns the builtin service output directly to the attacker.
Exploitation requires no authentication and no user interaction. See the Apache Mailing List Thread for upstream discussion.
Detect Identity Attacks the Moment Authentication Fails
Singularity™ Identity surfaces credential misuse, lateral movement, and privilege escalation in real time, so you can contain the breach before it spreads.
Stop Identity AttacksDetection Methods for CVE-2026-84748
Indicators of Compromise
- Requests to builtin service method names (for example brpc.vars, hotspots, flags) arriving on the public bRPC port rather than the configured internal_port.
- Unexpected protobuf traffic over baidu_std, hulu_pbrpc, sofa_pbrpc, public_pbrpc, nshead_mcpack, nova_pbrpc, or ubrpc2pb referencing builtin method identifiers.
- Outbound responses from bRPC servers on public ports carrying profiler payloads or bvar metric dumps.
Detection Strategies
- Inspect bRPC access logs for builtin method invocations and correlate the source port against the configured internal_port.
- Instrument the server to log the computed security_mode value alongside each resolved method for post-hoc auditing.
- Monitor network boundaries for pb-protocol traffic reaching bRPC public ports from untrusted sources.
Monitoring Recommendations
- Alert on any invocation of brpc.vars, hotspots, or similar builtin services sourced from non-management network segments.
- Baseline normal pb-protocol method call distribution and flag deviations that include builtin method names.
- Capture packet metadata on public bRPC ports to retain evidence of enumeration attempts.
How to Mitigate CVE-2026-84748
Immediate Actions Required
- Upgrade Apache bRPC to version 1.18.0 or later across all server deployments.
- If an immediate upgrade is not possible, apply the upstream fix from apache/brpc pull request #3511 and rebuild.
- Restrict network access to the public bRPC port so that only trusted clients can reach it until patching completes.
Patch Information
Apache bRPC 1.18.0 enforces the security_mode check for all protobuf-based protocol dispatch paths, aligning them with the HTTP and h2 behavior. The patch is available in apache/brpc pull request #3511 for operators who need to backport the fix to earlier releases.
Workarounds
- Disable all protobuf-based protocols on the public port and route internal introspection traffic exclusively through the internal_port.
- Enforce network segmentation or firewall rules that block external access to builtin service method names.
- Remove or disable builtin and Tabbed services on servers that cannot be patched promptly.
# Configuration example: restrict the public bRPC port via host firewall
# Allow only the management subnet to reach the public port until upgrade to 1.18.0
iptables -A INPUT -p tcp --dport 8000 -s 10.0.0.0/24 -j ACCEPT
iptables -A INPUT -p tcp --dport 8000 -j DROP
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

