A GPU kernel can read sensitive data from another GPU kernel (even from another user or app) through an optimized GPU memory region called _local memory_ on various architectures.

Insufficient validation of SPI flash addresses in the ASP (AMD Secure Processor) bootloader may allow an attacker to read data in memory mapped beyond SPI flash resulting in a potential loss of availability and integrity.

Insufficient input validation in CpmDisplayFeatureSmm may allow an attacker to corrupt SMM memory by overwriting an arbitrary bit in an attacker-controlled pointer potentially leading to arbitrary code execution in SMM.

When SMT is enabled, certain AMD processors may speculatively execute instructions using a target from the sibling thread after an SMT mode switch potentially resulting in information disclosure.

IBPB may not prevent return branch predictions from being specified by pre-IBPB branch targets leading to a potential information disclosure.

Execution unit scheduler contention may lead to a side channel vulnerability found on AMD CPU microarchitectures codenamed “Zen 1”, “Zen 2” and “Zen 3” that use simultaneous multithreading (SMT). By measuring the contention level on scheduler queues an attacker may potentially leak sensitive information.

Aliases in the branch predictor may cause some AMD processors to predict the wrong branch type potentially leading to information disclosure.

Mis-trained branch predictions for return instructions may allow arbitrary speculative code execution under certain microarchitecture-dependent conditions.

A potential vulnerability in some AMD processors using frequency scaling may allow an authenticated attacker to execute a timing attack to potentially enable information disclosure.