VYPR
Unrated severityNVD Advisory· Published Aug 12, 2020· Updated Sep 17, 2024

Confidential Information Disclosure vulnerability in Asylo

CVE-2020-8905

Description

A buffer length validation vulnerability in Asylo versions prior to 0.6.0 allows an attacker to read data they should not have access to. The 'enc_untrusted_recvfrom' function generates a return value which is deserialized by 'MessageReader', and copied into three different 'extents'. The length of the third 'extents' is controlled by the outside world, and not verified on copy, allowing the attacker to force Asylo to copy trusted memory data into an untrusted buffer of significantly small length.. We recommend updating Asylo to version 0.6.0 or later.

AI Insight

LLM-synthesized narrative grounded in this CVE's description and references.

A buffer length validation vulnerability in Asylo's enc_untrusted_recvfrom allows an attacker to read sensitive data by copying trusted memory into an undersized untrusted buffer.

Vulnerability

A buffer length validation vulnerability exists in Asylo versions prior to 0.6.0. The enc_untrusted_recvfrom function in asylo/platform/host_call/trusted deserializes a return value via MessageReader and copies it into three extents. The length of the third extent is controlled by the outside world and is not verified during the copy operation. This allows an attacker to supply a small actual data size but a large buffer length (len), causing memcpy to read beyond the received data buffer into adjacent trusted memory [1].

Exploitation

An attacker who can control the length field in a message sent to the enclave (e.g., from the untrusted host or over the network) can trigger the vulnerability. By providing a small actual data size for the third extent while specifying a large len parameter, the memcpy call copies len bytes from the received data buffer, which includes data beyond the actual received data, thus leaking trusted memory contents [1]. No authentication or user interaction is required beyond the ability to send messages to the enclave.

Impact

Successful exploitation allows an attacker to read trusted memory data that should not be accessible, resulting in information disclosure. The attacker gains access to sensitive data within the Asylo enclave's trusted memory, compromising confidentiality [1].

Mitigation

The vulnerability is fixed in Asylo version 0.6.0. The fix is implemented in commit 299f804acbb95a612ab7c504d25ab908aa59ae93, which changes memcpy(buf, buffer_received.data(), len) to memcpy(buf, buffer_received.data(), std::min(len, buffer_received.size())) [1]. Users should update to Asylo 0.6.0 or later. No workarounds are documented.

AI Insight generated on May 26, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.

Affected products

2
  • Google/asylollm-fuzzy
    Range: <0.6.0
  • Google LLC/Asylov5
    Range: stable

Patches

0

No patches discovered yet.

Vulnerability mechanics

AI mechanics synthesis has not run for this CVE yet.

References

1

News mentions

0

No linked articles in our index yet.