flawopen.com/Teardowns/cve-2023-4211-arm-mali-gpu-kernel-race-condition

● CVE-2023-4211 · CVSS 5.5 · Moyenne
Recherche · FlawOpen

CVE-2023-4211: Arm Mali GPU Kernel Memory Race Condition Teardown

Analyse technique détaillée du code source et mesures de durcissement pour vulnerabilidade : How an unlocked page table teardown in drivers/gpu/arm/midgard/mali_kbase_mem.c enabled commercial spyware to bypass Android sandboxes.

💡 Explication en Langage Simple (ELI5)

Analogie concrète : Imagine a hotel where guests return room keys to the front desk. A dishonest guest hands in their key, and while the clerk is marking the room empty in the computer, the guest's accomplice runs into the room, replaces the door lock with their own, and starts renting out the room privately without the hotel knowing.

Concepts Clés et Termes

Mali kbase Driver
The Linux kernel module interfacing Arm Mali GPU cores with user-space rendering libraries.
Page Table Mapping (mmap)
Mapping GPU hardware physical memory pages into user-space process virtual memory addresses.
Race Condition
A timing bug where two execution threads attempt to access and modify shared resources simultaneously without adequate synchronization.
Privilege Escalation
Gaining kernel execution privileges from an unprivileged sandbox application.

Analyse de Cause Racine

La cause fondamentale provient de paramètres de limites non validés dans les systèmes open source, permettant une désynchronisation d'état et le contournement des contrôles de sécurité.

Déroulement de l'Attaque Étape par Étape

Step 1

Étape d'attaque : Multi-Threaded Memory Allocation

Mécanisme technique d'exploitation : The attacker launches two threads mapping GPU memory chunks via mmap().

Step 2

Étape d'attaque : Race Unmap Against Access

Mécanisme technique d'exploitation : Thread A invokes munmap() while Thread B simultaneously queues GPU drawing commands on the same region.

Step 3

Étape d'attaque : Free Memory Access

Mécanisme technique d'exploitation : The kernel frees the memory pages, but Thread B's GPU commands execute against the freed pages.

Step 4

Étape d'attaque : Kernel Control Hijack

Mécanisme technique d'exploitation : The attacker sprays page tables into the slot, rewriting GPU descriptors to access all physical device RAM.

Code Source : Vulnérable vs Sécurisé

IMPLÉMENTATION VULNÉRABLE
// VULNERABLE: drivers/gpu/arm/midgard/mali_kbase_mem.c
int kbase_mem_free(struct kbase_context *kctx, u64 gpu_addr) {
    struct kbase_va_region *reg = kbase_region_tracker_find(kctx, gpu_addr);
    
    // ROOT CAUSE:
    // Memory region is unmapped and freed WITHOUT holding the context's page lock!
    // Concurrent GPU commands can still dereference 'reg' while it is being destroyed!
    kbase_gpu_vm_lock(kctx);
    kbase_mem_free_region(kctx, reg); // Frees underlying pages
    kbase_gpu_vm_unlock(kctx);
    return 0;
}
PATCH SÉCURISÉ ET ROBUSTE
// SECURE: drivers/gpu/arm/midgard/mali_kbase_mem.c patch
int kbase_mem_free(struct kbase_context *kctx, u64 gpu_addr) {
    // 1. Acquire global context lock BEFORE looking up region
    kbase_gpu_vm_lock(kctx);
    
    struct kbase_va_region *reg = kbase_region_tracker_find(kctx, gpu_addr);
    if (!reg) {
        kbase_gpu_vm_unlock(kctx);
        return -EINVAL;
    }
    
    // 2. Wait for all in-flight GPU job chains to complete before deallocating
    kbase_wait_for_in_flight_jobs(kctx, reg);
    
    kbase_mem_free_region(kctx, reg);
    kbase_gpu_vm_unlock(kctx);
    return 0;
}

Liste de Contrôle de Sécurité pour l'Ingénierie

Sources