Amd Instinct Mi300X
Monthly
Attacker-controlled pointer dereference in AMD's ROCm Communication Collectives Library (RCCL) can escalate to remote code execution when a compromised peer rank - or a network-adjacent attacker on the collective fabric - participates in a multi-GPU or multi-node distributed workload. The defect (CWE-822, untrusted pointer dereference) affects RCCL running on AMD Instinct MI200/MI300/MI350-class accelerators and carries a CVSS 4.0 base score of 7.7 (vector AV:N/AC:H/AT:N/PR:L/UI:N, High confidentiality/integrity/availability), mapping to an assessed CVSS 3.1 of AV:N/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H. Exploitation is not confirmed as active (CVE is not in CISA KEV) and no public exploit code has been identified at time of analysis; the attack requires a pre-existing foothold as a peer rank inside the collective group rather than anonymous internet access, so this is a moderate-priority issue for distributed AMD GPU deployments rather than an internet-wide emergency.
Remote unauthenticated manipulation of AMD Instinct GPU configuration is possible in the AMD Device Metrics Exporter (DME), a ROCm-ecosystem monitoring component, because the service binds to an unrestricted IP address and exposes its control surface network-wide. Any attacker able to reach the listening port can push unauthorized GPU configuration changes, degrading or denying availability of the affected AMD Instinct MI-series accelerator (MI210/MI250/MI250X/MI300A/MI300X/MI308X/MI325X/MI350X/MI355X). No public exploit is identified at time of analysis, and EPSS is low (0.20%), but the CVSS 4.0 base score of 9.2 reflects trivial network reachability with high availability impact.
An out of bounds read in the remote management firmware could allow a privileged attacker read a limited section of memory outside of established bounds potentially resulting in loss of. Rated low severity (CVSS 1.8). No vendor patch available.
Heap-based buffer overflow in AMD's MxGPU-Virtualization (GIM) driver affects a range of AMD Instinct and Radeon PRO datacenter GPUs, allowing a local low-privileged attacker to trigger denial-of-service by exploiting a race condition on the global variable amdgv_cmd in an unsynchronized ioctl handler. AMD has issued Security Bulletin AMD-SB-6027 confirming the issue and releasing GIM Driver 8.4 for affected Instinct-series products; no public exploit code exists and the vulnerability is not listed in CISA KEV. With EPSS at 0.01% (2nd percentile) and CVSS 4.0 score of 2.0, this represents a very low real-world exploitation priority despite the heap overflow classification.
Improper isolation of VCN-JPEG HW register space could allow a malicious Guest Virtual Machine (VM) or a process to perform unauthorized access to the register space of the JPEG cores assigned a. Rated high severity (CVSS 8.8), this vulnerability is low attack complexity. No vendor patch available.
Improper cleanup of shared GPU firmware registers in AMD Instinct and Radeon Pro accelerators allows admin-privileged attackers within guest virtual machines to access registers allocated to other guest VMs, potentially compromising confidentiality, integrity, or availability across isolated workloads. The vulnerability requires local admin privileges within a guest VM and affects multiple GPU product lines used in data center and HPC environments.
Buffer overflow in AMD GPU driver IOCTL handler enables local privilege escalation to root on Linux systems running AMD Instinct or Radeon Pro GPUs. Authenticated local users with low privileges can exploit an out-of-bounds write vulnerability in the AMDGV_CMD_GET_DIAG_DATA IOCTL to achieve arbitrary kernel code execution. EPSS data not available; no public exploit or CISA KEV listing identified at time of analysis, suggesting limited active exploitation despite high CVSS 8.5 severity.
Insufficient parameter sanitization in the AMD TEE SOC Driver's DRV_SOC_CMD_ID_SRIOV_CHECK_TA_COMPAT command handler allows high-privileged local attackers to trigger incorrect shared memory mapping via malformed parameters, potentially disclosing sensitive information. The vulnerability affects AMD Radeon RX 6000/7000, Radeon Pro W6000/W7000, and Instinct MI-series GPUs. CVSS 1.8 reflects high-privilege requirement and local-only attack vector with severe attack complexity, but the information disclosure impact and active vendor acknowledgment indicate targeted risk to privileged processes.
Insufficient parameter sanitization in TEE SOC Driver on AMD Instinct MI300A, MI300X, MI308X, and MI325X accelerators allows local high-privileged attackers to issue malformed DRV_SOC_CMD_ID_SRIOV_COPY_VF_CHIPLET_REGS commands that write invalid data to remote Dies, resulting in availability impact through unexpected behavior. The vulnerability requires local access and high privileges, with a CVSS score of 1.8 indicating minimal real-world risk despite potential data integrity concerns.
Attacker-controlled pointer dereference in AMD's ROCm Communication Collectives Library (RCCL) can escalate to remote code execution when a compromised peer rank - or a network-adjacent attacker on the collective fabric - participates in a multi-GPU or multi-node distributed workload. The defect (CWE-822, untrusted pointer dereference) affects RCCL running on AMD Instinct MI200/MI300/MI350-class accelerators and carries a CVSS 4.0 base score of 7.7 (vector AV:N/AC:H/AT:N/PR:L/UI:N, High confidentiality/integrity/availability), mapping to an assessed CVSS 3.1 of AV:N/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H. Exploitation is not confirmed as active (CVE is not in CISA KEV) and no public exploit code has been identified at time of analysis; the attack requires a pre-existing foothold as a peer rank inside the collective group rather than anonymous internet access, so this is a moderate-priority issue for distributed AMD GPU deployments rather than an internet-wide emergency.
Remote unauthenticated manipulation of AMD Instinct GPU configuration is possible in the AMD Device Metrics Exporter (DME), a ROCm-ecosystem monitoring component, because the service binds to an unrestricted IP address and exposes its control surface network-wide. Any attacker able to reach the listening port can push unauthorized GPU configuration changes, degrading or denying availability of the affected AMD Instinct MI-series accelerator (MI210/MI250/MI250X/MI300A/MI300X/MI308X/MI325X/MI350X/MI355X). No public exploit is identified at time of analysis, and EPSS is low (0.20%), but the CVSS 4.0 base score of 9.2 reflects trivial network reachability with high availability impact.
An out of bounds read in the remote management firmware could allow a privileged attacker read a limited section of memory outside of established bounds potentially resulting in loss of. Rated low severity (CVSS 1.8). No vendor patch available.
Heap-based buffer overflow in AMD's MxGPU-Virtualization (GIM) driver affects a range of AMD Instinct and Radeon PRO datacenter GPUs, allowing a local low-privileged attacker to trigger denial-of-service by exploiting a race condition on the global variable amdgv_cmd in an unsynchronized ioctl handler. AMD has issued Security Bulletin AMD-SB-6027 confirming the issue and releasing GIM Driver 8.4 for affected Instinct-series products; no public exploit code exists and the vulnerability is not listed in CISA KEV. With EPSS at 0.01% (2nd percentile) and CVSS 4.0 score of 2.0, this represents a very low real-world exploitation priority despite the heap overflow classification.
Improper isolation of VCN-JPEG HW register space could allow a malicious Guest Virtual Machine (VM) or a process to perform unauthorized access to the register space of the JPEG cores assigned a. Rated high severity (CVSS 8.8), this vulnerability is low attack complexity. No vendor patch available.
Improper cleanup of shared GPU firmware registers in AMD Instinct and Radeon Pro accelerators allows admin-privileged attackers within guest virtual machines to access registers allocated to other guest VMs, potentially compromising confidentiality, integrity, or availability across isolated workloads. The vulnerability requires local admin privileges within a guest VM and affects multiple GPU product lines used in data center and HPC environments.
Buffer overflow in AMD GPU driver IOCTL handler enables local privilege escalation to root on Linux systems running AMD Instinct or Radeon Pro GPUs. Authenticated local users with low privileges can exploit an out-of-bounds write vulnerability in the AMDGV_CMD_GET_DIAG_DATA IOCTL to achieve arbitrary kernel code execution. EPSS data not available; no public exploit or CISA KEV listing identified at time of analysis, suggesting limited active exploitation despite high CVSS 8.5 severity.
Insufficient parameter sanitization in the AMD TEE SOC Driver's DRV_SOC_CMD_ID_SRIOV_CHECK_TA_COMPAT command handler allows high-privileged local attackers to trigger incorrect shared memory mapping via malformed parameters, potentially disclosing sensitive information. The vulnerability affects AMD Radeon RX 6000/7000, Radeon Pro W6000/W7000, and Instinct MI-series GPUs. CVSS 1.8 reflects high-privilege requirement and local-only attack vector with severe attack complexity, but the information disclosure impact and active vendor acknowledgment indicate targeted risk to privileged processes.
Insufficient parameter sanitization in TEE SOC Driver on AMD Instinct MI300A, MI300X, MI308X, and MI325X accelerators allows local high-privileged attackers to issue malformed DRV_SOC_CMD_ID_SRIOV_COPY_VF_CHIPLET_REGS commands that write invalid data to remote Dies, resulting in availability impact through unexpected behavior. The vulnerability requires local access and high privileges, with a CVSS score of 1.8 indicating minimal real-world risk despite potential data integrity concerns.