An attacker could cooperatively pass data from one secure GPU process to another secure GPU process through shared secure memory allocations in the kernel module. Additionally, an attacker could disrupt the operation of another secure GPU process leading to image corruption / GPU hardware recovery. Sharing secure memory allocations among various GPU secure processes allows an attacker to corrupt shared resource affecting other users.