VYPR
Unrated severityNVD Advisory· Published Dec 15, 2020· Updated Aug 4, 2024

Unchecked buffer overrun in enc_untrusted_read

CVE-2020-8942

Description

An arbitrary memory read vulnerability in Asylo versions up to 0.6.0 allows an untrusted attacker to make a call to enc_untrusted_read whose return size was not validated against the requrested size. The parameter size is unchecked allowing the attacker to read memory locations outside of the intended buffer size including memory addresses within the secure enclave. We recommend upgrading past commit b1d120a2c7d7446d2cc58d517e20a1b184b82200

AI Insight

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

Asylo versions up to 0.6.0 contain an unchecked return size in enc_untrusted_read, allowing an attacker to read arbitrary memory including enclave data.

Vulnerability

In Asylo versions up to 0.6.0, the function enc_untrusted_read does not validate that the return size from the underlying read system call does not exceed the requested count [1]. The parameter size is passed directly to count and the return value is not checked against the buffer size, enabling an out-of-bounds memory read. The vulnerable code path is triggered when an untrusted caller invokes enc_untrusted_read via the host call interface. The commit b1d120a2c7d7446d2cc58d517e20a1b184b82200 fixes this by adding an abort if ret > count [1].

Exploitation

An untrusted attacker can call enc_untrusted_read with a small requested count while the underlying read returns more bytes than requested. Because the return size is not checked against count, the attacker can read memory locations beyond the intended buffer, including memory within the secure enclave. No authentication is required, and the attacker only needs the ability to make host calls from untrusted code.

Impact

Successful exploitation allows an attacker to read arbitrary memory outside the intended buffer, including sensitive data inside the secure enclave. This bypasses the confidentiality guarantees of the enclave, leading to information disclosure of protected data.

Mitigation

Users should upgrade Asylo to a version that includes commit b1d120a2c7d7446d2cc58d517e20a1b184b82200 [1]. No workaround is available for unpatched versions. The repository has been archived and is now read-only, so no further patches are expected.

AI Insight generated on May 27, 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: unspecified

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.