GCVE-1988-2026-0574
Vulnerability from gna-1988 – Published: 2026-11-17 19:38 – Updated: 2026-11-17 19:38| Vendor | Product | Version | CPE status | |
|---|---|---|---|---|
| unknown | SAPBleed Pre-Auth Memory |
Affected:
unknown
|
guessed |
- related CVE-2026-44756
- related CVE-2026-76964
- related CVE-2026-89126
{
"containers": {
"cna": {
"affected": [
{
"product": "SAPBleed Pre-Auth Memory",
"vendor": "unknown",
"versions": [
{
"status": "affected",
"version": "unknown"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "finder",
"value": "Raschin Tavakoli via Fulldisclosure"
}
],
"descriptions": [
{
"lang": "en",
"value": "nullFaktor Security Advisory \u003c 2026-11-12 \u003e\n===========================================================\nTitle: SAPBleed: Pre-Auth Process Memory Leak and RCE\n in SAP ICM and Web Dispatcher, including the\n internal Web Dispatcher in SAP HANA XS\n\nVulnerability: Heap Buffer Overflow / Memory Leak / RCE\n\nCVE (ICM): CVE-2026-76964\nCVE (Web Disp): CVE-2026-89126\nImpact: Critical\nCVSS 3.0 Score: 10.0 (both CVEs)\nCVSS Vector: CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H\n\nProducts: SAP NetWeaver AS ABAP (ICM); SAP Web Dispatcher;\n SAP HANA XS (internal Web Dispatcher)\nAffected Versions: See SAP Security Notes 3778530 and 3778556\nVendor Homepage: https://www.sap.com\n\nSAP Note (ICM): 3778530\nSAP Note (Web Disp): 3778556\nDate: 2026-11-12\nAuthor: Raschin Tavakoli | nullFaktor GmbH\n https://www.nullfaktor.com\n office () nullfaktor com\n https://x.com/r4yzor9\n https://x.com/_nullfaktor\nBlog: https://nullfaktor.com/\n blog/cve-2026-76964-cve-2026-89126-sapbleed\n\nShort Summary\n=============\n\nSAPBleed is a pre-authentication memory-disclosure and RCE\nvulnerability affecting SAP ICM, SAP Web Dispatcher, and SAP HANA XS.\nAn Internet-facing SAP Web Dispatcher, acting as a reverse proxy,\ncan forward the attack to an ICM in a protected backend network\nsegment. The WebSocket connection becomes a path for leaking\nmemory from that internal component over the Internet. Attackers can\nrepeatedly read selected ICM memory, potentially exposing SSO tickets\nand private TLS keys. ICM-accessible shared memory also contains code\naddresses from separate ABAP dialog work processes (disp+work), even\nthough those processes are not directly exposed to the Internet.\n\nThe ICM can remain running during the memory disclosure. Traffic can\ntravel over TLS-protected WebSocket connections, concealing the leaked\npayload from network sensors without TLS visibility. Pre-authentication\ncode execution has also been demonstrated in both ICM and Web\nDispatcher.\n\nVendor description\n==================\nSAP SE is an enterprise software company headquartered in Walldorf,\nGermany. Its products include enterprise resource planning (ERP)\nsoftware and business applications.\n\nVulnerability\n=============\nSAPBleed is a pre-authentication heap buffer overflow in the WebSocket\nprocessing of SAP Internet Communication Manager (ICM) and SAP Web\nDispatcher, including Web Dispatcher deployments used with SAP HANA XS.\nThe flaw allows an unauthenticated attacker to corrupt WebSocket\nconnection state. Depending on how the corruption is exploited, the\naffected process can disclose memory or be used for remote code\nexecution. The attack is carried over a WebSocket connection and can\nreach affected services over HTTP or HTTPS, including TLS-encrypted\nconnections.\n\nThe memory disclosure occurs when corrupted connection state causes\nthe process to return heap contents beyond the data intended for its\nresponse. A Web Dispatcher can be targeted directly, exposing memory\nfrom its own process. Where affected traffic is forwarded to an ICM,\nthe ICM can also be targeted through the Web Dispatcher. Leaked process\nmemory may contain SSO tickets, cryptographic key material and process\naddresses that enable a full ASLR bypass.\n\nIn our tested laboratory deployment, the research PoCs demonstrated:\n\n - Direct disclosure of Web Dispatcher process memory, including its\n own address information, enabling an ASLR bypass for that process.\n\n - Repeatable reads of arbitrary ICM process memory, including when\n the ICM was reached through a Web Dispatcher.\n\n - Disclosure of ICM process addresses, enabling an ASLR bypass in\n that process.\n\n - Cross-process disclosure of code addresses belonging to separate\n ABAP dialog work processes (DIA/disp+work), defeating their own\n ASLR despite their separate process address spaces.\n\n - Remote identification of the SAP kernel release, build and patch\n level without SAP credentials.\n\n - Extraction of the server\u0027s private TLS key from ICM memory.\n\n - Extraction of a valid MYSAPSSO2 single sign-on ticket from active\n authenticated traffic present in ICM memory.\n\nThe impact extends beyond the ICM: SAP shared memory visible to the ICM\ncontains code addresses from separate DIA work processes. Our PoC\nrecovered their module base addresses over the network before\nauthentication. This removes ASLR as a barrier for address-dependent\nRCE flaws in those work processes as well.\n\nA leaked SSO ticket may allow impersonation of its user while the\nticket remains valid. Exposure of a TLS private key compromises the\nserver\u0027s key material; the effect on past traffic depends on the TLS\nconfiguration. Other secrets held in ICM memory may also be exposed,\ndepending on the system\u0027s workload and the data present at the time of\nan attack.\n\nSAPBleed is particularly critical because it exposes process memory\nover the Internet before authentication. The disclosed process\naddresses defeat ASLR in the affected processes. This can make other\nmemory-corruption flaws in those same processes exploitable when their\nremote code execution path requires an otherwise unknown address.\n\nOur SAP Overpass advisory (CVE-2026-44756), published September 10,\n2026, illustrates this: its RCE proof required an independently\nobtained module base address. SAPBleed can supply that address where\nboth issues affect the same ICM and can expose the corresponding\naddress for a DIA work process.\nSee:\nhttps://nullfaktor.com/blog/cve-2026-44756-sap-netweaver-pre-auth-rce\n\nOur research also established pre-authentication remote code execution\n(RCE) in both ICM and Web Dispatcher. We engineered the ICM\nmemory-extraction path to restore a consistent execution state after\neach read. In repeated laboratory runs, it extracted data reliably\nwithout crashing the ICM or dialog work processes, including when\ntraffic passed through a Web Dispatcher. The services remained\navailable during those runs.\n\nFurther exploitation details are withheld from this version.\n\nTested Environment\n==================\nThe issue was reproduced in a laboratory deployment of SAP NetWeaver AS\nABAP and SAP Web Dispatcher. Full build details are omitted from this\nversion.\n\nAttack Scenario\n===============\nAny affected SAP system that exposes the vulnerable WebSocket path to\nthe Internet can be attacked before authentication. Placing a Web\nDispatcher in front of the ICM does not shield that backend: our lab\ndemonstrated that attack traffic can pass through the Internet-facing\nproxy and disclose ICM process memory even when the ICM is not directly\nreachable from the Internet. The Web Dispatcher can also be attacked\ndirectly. No SAP credentials, prior session, or user interaction are\nrequired.\n\nMitigation\n==========\nInstall the corrections in SAP Security Notes 3778530 (ICM) and 3778556\n(Web Dispatcher) at the fixed patch levels for each installed release.\nAs a temporary measure, disable WebSocket HTTP upgrades in the relevant\nICM and Web Dispatcher profiles with\nicm/HTTP/support_websocket_upgrade=FALSE and restart the components.\nThis can disrupt functions that require WebSocket. SAP FAQ Note 3778788\nprovides further guidance on temporary workarounds. Until updates are\napplied, restrict network access to affected systems.\n\nProof of Concept\n================\n\nThe following excerpts are from our laboratory tests. All\nattacks reach the target before authentication. SIDs and OS\nuser names have been redacted, and line wrapping has been\nadjusted for readability.\n\n\nSAPBleed Integrated Tool\n-------------------------\n\nThe SAPBleed tool automates the full chain from initial memory\ndisclosure through ASLR bypass to secret extraction without attacker\ncredentials or a prior attacker session. The SSO test below uses\nconcurrent authenticated traffic from a separate lab client.\n\nTLS private-key extraction (ICM PIE leak, kernel fingerprint,\nTLS key recovery in one session):\n\n$\u003e python3 sapbleed.py secure-bank 50001 tls\n\nSAPBleed Lab Demo\nTarget: secure-bank:50001\nAuthentication: none (pre-auth)\nOperations: ICM PIE -\u003e Kernel fingerprint -\u003e TLS key\n\n[CHECK] Connecting to ICM ... OK 00:01\n\n[1/3] Leaking the ICM PIE base\n[OK] ICM PIE: 0x*** 00:08\n\n[2/3] Identifying the kernel build and patch level\n[OK] ICM kernel 793, patch level 101\n (read from the kernel patch table) 00:25\n\n[3/3] Extracting the TLS private key\n[OK] TLS private key extracted 00:28\n Saved to: ~/.sapbleed/loot/tls_private_key.pem\n\nRESULT: 3/3 operations succeeded\nTOTAL TIME: 01:04\n\nSSO ticket extraction (ICM PIE leak, kernel fingerprint, and\nrecovery of a MYSAPSSO2 ticket from concurrent victim traffic):\n\nIn one terminal, a victim sends authenticated HTTPS requests:\n\n$\u003e python3 victim_sim.py secure-bank 50001\n\nVictim simulator: secure-bank:50001 every 0.3s\nTicket: ~/.sapbleed/victim_ticket.b64 (1164 characters)\nPress Ctrl+C to stop.\n\n[1] HTTP 200 (14259 bytes) AUTHENTICATED (ticket valid)\n[2] HTTP 200 (14259 bytes) AUTHENTICATED (ticket valid)\n[3] HTTP 200 (14259 bytes) AUTHENTICATED (ticket valid)\n\nWith that traffic running, the unauthenticated client extracts the\nvictim\u0027s sso ticket in another terminal:\n\n$\u003e python3 sapbleed.py secure-bank 50001 sso\n\nSAPBleed Lab Demo\nTarget: secure-bank:50001\nAuthentication: none (pre-auth)\nOperations: ICM PIE -\u003e Kernel fingerprint -\u003e SSO ticket\nLoot: ~/.sapbleed/loot\nSSO traffic: on non-productive systems, use victim_sim.py\n to simulate SSO traffic\n\n[CHECK] Connecting to ICM ... OK 00:01\n\n[1/3] Leaking the ICM PIE base\n[OK] ICM PIE: 0x*** 00:20\n\n[2/3] Identifying the kernel build and patch level (icman)\n[OK] ICM kernel 793, patch level 101\n (read from the kernel patch table) 00:13\n\n[3/3] Extracting an SSO ticket\n[OK] SSO ticket extracted 00:29\n Saved to: ~/.sapbleed/loot/mysapsso2_ticket.b64\n\nRESULT: 3/3 operations succeeded\nTOTAL TIME: 01:04\n\nArtifacts:\n ~/.sapbleed/loot/mysapsso2_ticket.b64\n\nBenchmark: ~/.sapbleed/loot/benchmark.json\n\nCross-process ASLR bypass (ICM PIE leak extended to DIA\nwork-process addresses via SAP shared memory):\n\n$\u003e python3 sapbleed.py secure-bank 50001 dia\n\nSAPBleed Lab Demo\nTarget: secure-bank:50001\nAuthentication: none (pre-auth)\nOperations: ICM PIE -\u003e Kernel fingerprint\n -\u003e disp+work PIE\n\n[CHECK] Connecting to ICM ... OK 00:01\n\n[1/3] Leaking the ICM PIE base\n[OK] ICM PIE: 0x*** 00:22\n\n[2/3] Identifying the kernel build and patch level\n[OK] ICM kernel 793, patch level 101\n (read from the kernel patch table) 00:25\n\n[3/3] Leaking the disp+work PIE base via shared memory\n[OK] disp+work PIE: 0x*** 00:23\n\nRESULT: 3/3 operations succeeded\nTOTAL TIME: 01:12\n\n\n1. Internet Communication Manager (ICM)\n=======================================\n\n1.1 Memory Disclosure (ICM Heap Leak)\n---------------------------------------\n\nThe standalone pre-authentication heap-leak PoC returns a 16 KiB\nWebSocket binary frame containing uninitialised ICM heap data. This\nis one response, not a limit on the total disclosure: the ICM PoC tool\nuses a ROP-based reader to read selected ICM memory\naddresses over subsequent WebSocket connections. The ICM survives;\nno PIE base or heap address is needed as input for the initial leak:\n\n$\u003e python3 poc-icm-websocket-heapleak.py secure-bank 50001\n\nicman pre-auth wire-only heap leak\ntarget: secure-bank:50001\n (WebSocket connection)\n[*] warming the allocator\n (churning freed pointer-bearing chunks)...\n[*] attempt 1/3\n[+] WebSocket upgraded\n[+] crafted frames altered connection state\n[+] received 16390 bytes from icman\n\nRESULT: OVER-READ FIRED - 16390 bytes of an\n uninitialised icman heap buffer leaked to\n the client (icman survives; pre-auth;\n wire-only; no PIE/heap base"
}
],
"providerMetadata": {
"dateUpdated": "2026-11-17T19:38:27Z",
"orgId": "4e2abfbf-4a2a-4b76-a4e0-d77c18ba156c",
"shortName": "VULNARCHIVE"
},
"references": [
{
"tags": [
"technical-description",
"exploit"
],
"url": "https://vuln.freearchive.org/archive/full-disclosure/2026/Nov/36"
},
{
"tags": [
"technical-description"
],
"url": "https://seclists.org/fulldisclosure/2026/Nov/36"
},
{
"url": "https://nmap.org/mailman/listinfo/fulldisclosure"
},
{
"url": "https://nullfaktor.com/"
},
{
"url": "https://nullfaktor.com/blog/cve-2026-44756-sap-netweaver-pre-auth-rce"
},
{
"url": "https://seclists.org/fulldisclosure/"
},
{
"url": "https://www.nullfaktor.com"
},
{
"url": "https://www.nullfaktor.com/#kontakt"
},
{
"url": "https://www.nullfaktor.com/blog"
},
{
"url": "https://www.sap.com"
},
{
"url": "https://x.com/_nullfaktor"
},
{
"url": "https://x.com/r4yzor9"
}
],
"source": {
"defect": [
"https://seclists.org/fulldisclosure/2026/Nov/36"
],
"discovery": "EXTERNAL"
},
"title": "SAPBleed: Pre-Auth Memory Disclosure and RCE in SAP NetWeaver, Web Dispatcher and HANA XS (CVE-2026-76964, CVE-2026-89126)",
"x_gcve": [
{
"recordType": "analysis",
"relationships": [
{
"destId": "CVE-2026-44756",
"type": "related"
},
{
"destId": "CVE-2026-76964",
"type": "related"
},
{
"destId": "CVE-2026-89126",
"type": "related"
}
],
"vulnId": "GCVE-1988-2026-0574",
"x_vulnarchive": {
"archiveUrl": "https://vuln.freearchive.org/archive/full-disclosure/2026/Nov/36",
"automated": true,
"contentSha256": "a7c1abe87dddabc5d6254f8d195f8567ebf60fb453f0a6c792f4fac32e5cfb90",
"evidenceScore": 10,
"messageId": "",
"originalUrl": "https://seclists.org/fulldisclosure/2026/Nov/36",
"policy": "vulnarchive-1",
"sourceFormat": "text/html",
"sourcePublishedAt": "2026-11-13T20:44:21Z"
}
}
]
}
},
"cveMetadata": {
"assignerOrgId": "4e2abfbf-4a2a-4b76-a4e0-d77c18ba156c",
"assignerShortName": "VULNARCHIVE",
"datePublished": "2026-11-17T19:38:27Z",
"dateUpdated": "2026-11-17T19:38:27Z",
"state": "PUBLISHED",
"vulnId": "GCVE-1988-2026-0574"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.