[SECURITY] HS EventMon uses forged event payload identity to trigger processor reset
维护者通常 1 天内回复
还没有人认领这个 Issue。
评估
- 难度
- 5/5
- 预计耗时
- 一周以上
- 新手友好度
- 35/100
- Issue 类型
- 缺陷
- 描述清晰度
- 基本清楚
- 活跃度
- 冷清
- 技术栈
- c
- 领域
- embedded-iot, security
调研方向
从 src/apps/hs/fsw/src/hs_monitors.c 中的决策点开始,然后跟踪事件经由 hs_app.c 和 hs_cmds.c 的进入路径,以确定是否有可供使用的可信发送方元数据。在 native_std 部署中复现伪造的 CFE_EVS_LONG_EVENT_MSG_MID 场景;当仅凭 payload 声明的身份已无法触发特权 EventMon 操作时,即表示完成。
由索引模型根据 Issue 内容生成。
描述
Checklist (Please check before submitting)
- I reviewed the Contributing Guide.
- I performed a cursory search to see if the bug report is relevant, not redundant, nor in conflict with other tickets.
Describe the bug
HS EventMon trusts the AppName and EventID fields embedded in incoming CFE_EVS_LONG_EVENT_MSG_MID packets, but it does not verify that the software-bus message actually originated from that named application. When a matching EventMon rule is configured, HS will execute privileged actions such as processor reset, app restart, or app deletion based only on spoofable payload identity.
In the reproduced native_std sample deployment, this is not a purely theoretical code issue. I reproduced an end-to-end chain where:
- attacker-controlled data was delivered through the normal UDP command-ingest path,
- CF/CFDP retained an attacker-controlled
HS.EventMon_Tblfile on the target runtime filesystem, - the normal
CFE_TBLmanagement path loaded and activated that table, and - a forged
CFE_EVS_LONG_EVENT_MSG_MIDpacket causedHSto executeCFE_ES_ResetCFE()and reset the processor.
To Reproduce
- Run the standard
native_stdsample deployment and startcore-cpu1. - Deliver a crafted
hs_emt_visa22.tblover the normalUDP -> CI_LAB -> CF/CFDPpath so that CF retains it as/cf/hs_emt_visa22.tbl. - Send normal
CFE_TBLcommands to:- abort any stale load state,
- load
/cf/hs_emt_visa22.tbl, - validate the inactive buffer, and
- activate
HS.EventMon_Tbl.
- Send the normal
HS EnableEventMoncommand. - Inject a forged
CFE_EVS_LONG_EVENT_MSG_MIDpacket whose payload sets:AppName = CFE_TIMEEventID = 0x2222
- Observe that
HSmatches the spoofed payload fields and performs a processor reset.
Observed runtime evidence from reproduction:
CF R2(4:8756): successfully retained file as /cf/hs_emt_visa22.tbl
HS 52: EventMon verify results: good = 1, bad = 0, unused = 15
CFE_TBL 36: HS validation successful for inactive 'HS.EventMon_Tbl'
CFE_TBL 37: HS Successfully Updated 'HS.EventMon_Tbl'
HS 27: Event Monitoring Enabled
HS 45: Event Monitor: APP:(CFE_TIME) EID:(8738): Action: Processor Reset
CFE_ES_ResetCFE: POWER ON RESET due to max proc resets (Commanded).
CFE_PSP: Exiting cFE with POWERON Reset status.
Expected behavior
HS should not make EventMon authorization decisions based solely on payload-declared identity. Before executing privileged actions, it should verify a trusted sender identity provided by the framework or reject externally supplied event traffic that lacks authenticated origin metadata.
Code snips
Primary decision point:
Event ingress path:
Relevant matching logic:
if ((ActionType != HS_EMTActType_NOACT)
&& (HS_AppData.EMTablePtr[TableIndex].EventID == EventPtr->Payload.PacketID.EventID))
{
if (strncmp(HS_AppData.EMTablePtr[TableIndex].AppName,
EventPtr->Payload.PacketID.AppName,
OS_MAX_API_NAME) == 0)
{
switch (ActionType)
{
case HS_EMTActType_PROC_RESET:
CFE_ES_ResetCFE(CFE_PSP_RST_TYPE_PROCESSOR);
Normal remote file-ingress prerequisite used in reproduction:
System observed on:
- Hardware: containerized native sample deployment
- OS: Linux (containerized native target)
- Versions: cFS sample deployment built from this repository using
build-native_std; cFE/OSAL/PSP sample stack as shipped in-tree
Additional context
Important scope note: this reproduced exploitability depends on deployment-specific reachability. In my reproduction, the sample deployment exposed a normal UDP ingest path through CI_LAB, and CF/CFDP could retain a remotely supplied table file to writable storage. If a mission removes or protects those paths, the attack surface changes.
This reproduction did not rely on manually placing files inside the target runtime filesystem and did not call internal APIs directly. The malicious EMT file was delivered through the system's normal ingress path, activated through the normal CFE_TBL management path, and the forged event was injected through the same sample deployment's normal message-ingest surface.
Another scope note: in the reproduced setup, the packet sender and core-cpu1 were run within the same container/network namespace and traffic was sent to 127.0.0.1:1234. Therefore, this demonstrates that the shipped sample deployment's normal network-ingest surface can be abused to trigger privileged HS actions, but it should not be overstated as proof that every real mission exposes the same surface directly to an external network.
Suggested fix:
- bind
HSEventMon decisions to authenticated sender identity, not payload-declared app name; - reject or quarantine externally supplied event packets that are not emitted by trusted EVS-originated paths;
- if EVS-origin metadata exists or can be added, require
HSto compare against that trusted sender identity before executing privileged actions.
Proposed severity view:
- This is best described as a real authorization/trust-boundary failure with deployment-conditional network reachability.
- In the reproduced sample deployment, it results in a visible processor reset with no need to compromise the named victim app whose identity is spoofed.
- A CVSS v3.1 base score around
8.8is defensible for the reproduced sample deployment (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:H) if the report text keeps the deployment-prerequisite caveat. - In my view this is CVE-worthy because the vulnerable behavior is not simply “HS can reset on events”; the issue is that
HSauthorizes those privileged actions using spoofable payload identity instead of trusted sender identity, and that weakness is dynamically exploitable in the shipped sample deployment.
Reporter Info
Prepared draft based on local reproduction and source review.
- 主要语言
- C
- 星标
- 1.6k
- 派生
- 421
- 平均合并
- 1 天 42 分钟
- 30 天内合并 PR
- 12
环境准备
- 没有 Dockerfile 或 Docker Compose 文件
- 有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue,再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库,在一个分支上完成修改。
- 提交 Pull Request,并在描述里引用这个 Issue 编号。
nasa/cFS 的其他 Issue
-
[SECURITY] Off-by-one bounds check in SBN HKPeerCmd/HKNetCmd allows access to invalid net/peer slots可能已有人在做 @0x000NULL 于 38 天前认领。 未关闭
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
-
难度 1/5 1 小时以内 新手友好度 92/100
nasa/cFS#1039 · 1 个 reaction ·
维护者通常 1 天内回复
-
continuous-integration
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
-
[SECURITY] Missing length validation for CFE_EVS_SEND_HK_MID in EVS ProcessCommandPacket可能已有人在做 @linerfan5114 于 204 天前认领。 未关闭bug
难度 2/5 1-3 小时 新手友好度 76/100
维护者通常 1 天内回复
-
[SECURITY] Missing length validation for CFE_ES_SEND_HK_MID in ES TaskPipe可能已有人在做 @linerfan5114 于 204 天前认领。 未关闭bug
难度 2/5 1-3 小时 新手友好度 78/100
维护者通常 1 天内回复
相似的 Issue
-
难度 2/5 1-3 小时 新手友好度 80/100
ESPresense/ESPresense#2680 · 2 条评论 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 78/100
OpenPrinting/cups#1828 · 4 条评论 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 74/100
ImageMagick/ImageMagick#9134 ·
维护者通常 1 天内回复
-
Builder: a structure opened without calculation data shows the Diffraction (fibre) view by default未关闭bug
难度 2/5 1-3 小时 新手友好度 65/100
FriendsofECCE/ECCE#1245 · 1 条评论 ·
-
Arch: ARM/AArch64 Doc
难度 1/5 1 小时以内 新手友好度 90/100
RT-Thread/rt-thread#12082 · 1 条评论 ·
维护者通常 2 天内回复