CVE-2024-45005 in Linux
Summary
by MITRE • 09/04/2024
In the Linux kernel, the following vulnerability has been resolved:
KVM: s390: fix validity interception issue when gisa is switched off
We might run into a SIE validity if gisa has been disabled either via using kernel parameter "kvm.use_gisa=0" or by setting the related sysfs attribute to N (echo N >/sys/module/kvm/parameters/use_gisa).
The validity is caused by an invalid value in the SIE control block's gisa designation. That happens because we pass the uninitialized gisa origin to virt_to_phys() before writing it to the gisa designation.
To fix this we return 0 in kvm_s390_get_gisa_desc() if the origin is 0. kvm_s390_get_gisa_desc() is used to determine which gisa designation to set in the SIE control block. A value of 0 in the gisa designation disables gisa usage.
The issue surfaces in the host kernel with the following kernel message as soon a new kvm guest start is attemted.
kvm: unhandled validity intercept 0x1011 WARNING: CPU: 0 PID: 781237 at arch/s390/kvm/intercept.c:101 kvm_handle_sie_intercept+0x42e/0x4d0 [kvm]
Modules linked in: vhost_net tap tun xt_CHECKSUM xt_MASQUERADE xt_conntrack ipt_REJECT xt_tcpudp nft_compat x_tables nf_nat_tftp nf_conntrack_tftp vfio_pci_core irqbypass vhost_vsock vmw_vsock_virtio_transport_common vsock vhost vhost_iotlb kvm nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 ip_set nf_tables sunrpc mlx5_ib ib_uverbs ib_core mlx5_core uvdevice s390_trng eadm_sch vfio_ccw zcrypt_cex4 mdev vfio_iommu_type1 vfio sch_fq_codel drm i2c_core loop drm_panel_orientation_quirks configfs nfnetlink lcs ctcm fsm dm_service_time ghash_s390 prng chacha_s390 libchacha aes_s390 des_s390 libdes sha3_512_s390 sha3_256_s390 sha512_s390 sha256_s390 sha1_s390 sha_common dm_mirror dm_region_hash dm_log zfcp scsi_transport_fc scsi_dh_rdac scsi_dh_emc scsi_dh_alua pkey zcrypt dm_multipath rng_core autofs4 [last unloaded: vfio_pci]
CPU: 0 PID: 781237 Comm: CPU 0/KVM Not tainted 6.10.0-08682-gcad9f11498ea #6 Hardware name: IBM 3931 A01 701 (LPAR) Krnl PSW : 0704c00180000000 000003d93deb0122 (kvm_handle_sie_intercept+0x432/0x4d0 [kvm])
R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:0 PM:0 RI:0 EA:3 Krnl GPRS: 000003d900000027 000003d900000023 0000000000000028 000002cd00000000 000002d063a00900 00000359c6daf708 00000000000bebb5 0000000000001eff 000002cfd82e9000 000002cfd80bc000 0000000000001011 000003d93deda412 000003ff8962df98 000003d93de77ce0 000003d93deb011e 00000359c6daf960 Krnl Code: 000003d93deb0112: c020fffe7259 larl %r2,000003d93de7e5c4 000003d93deb0118: c0e53fa8beac brasl %r14,000003d9bd3c7e70 #000003d93deb011e: af000000 mc 0,0 >000003d93deb0122: a728ffea lhi %r2,-22 000003d93deb0126: a7f4fe24 brc 15,000003d93deafd6e 000003d93deb012a: 9101f0b0 tm 176(%r15),1 000003d93deb012e: a774fe48 brc 7,000003d93deafdbe 000003d93deb0132: 40a0f0ae sth %r10,174(%r15) Call Trace: [] kvm_handle_sie_intercept+0x432/0x4d0 [kvm]
([] kvm_handle_sie_intercept+0x42e/0x4d0 [kvm])
[] vcpu_post_run+0x1d0/0x3b0 [kvm]
[] __vcpu_run+0xea/0x2d0 [kvm]
[] kvm_arch_vcpu_ioctl_run+0x16a/0x430 [kvm]
[] kvm_vcpu_ioctl+0x190/0x7c0 [kvm]
[] vfs_ioctl+0x2e/0x70
[] __s390x_sys_ioctl+0xc2/0xd0
[] __do_syscall+0x1f2/0x2e0
[] system_call+0x70/0x98
Last Breaking-Event-Address: [] __warn_printk+0xe8/0xf0
Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.
Analysis
by VulDB Data Team • 10/15/2024
The vulnerability described in CVE-2024-45005 affects the Linux kernel's KVM implementation on IBM s390 architecture systems. This issue manifests when the General Interrupt Subsystem Architecture (GISA) is disabled through either kernel parameter configuration or sysfs attribute manipulation. The core problem lies in the improper handling of uninitialized memory values within the SIE control block, specifically in the gisa designation field. When GISA is switched off, the kernel attempts to pass an uninitialized value from the gisa origin to the virt_to_phys() function, resulting in an invalid memory address being written into the SIE control block. This invalid value triggers a SIE validity intercept, which is a hardware-generated exception indicating an invalid control block configuration. The resulting system crash occurs during KVM guest startup operations, as evidenced by the kernel log message "kvm: unhandled validity intercept 0x1011" and subsequent stack trace information pointing to the kvm_handle_sie_intercept function in the intercept.c file. The vulnerability is categorized under CWE-252, representing an unchecked return value, and aligns with ATT&CK technique T1059.3.005 related to command and scripting interpreter usage in kernel contexts.
The technical flaw stems from a race condition in memory initialization within the kvm_s390_get_gisa_desc() function. This function is responsible for determining the appropriate gisa designation to set in the SIE control block, which controls whether GISA functionality is enabled or disabled for virtual machines. When GISA is disabled, the function should return a value of zero to effectively disable GISA usage in the control block. However, due to the uninitialized gisa origin value being passed through virt_to_phys() before proper initialization, an invalid physical address is generated and written to the control block. The fix implemented addresses this by adding a conditional check in kvm_s390_get_gisa_desc() that returns zero when the origin is zero, ensuring that GISA is properly disabled when requested. This modification prevents the invalid memory address from being written into the SIE control block, thereby eliminating the validity intercept that would otherwise occur during guest initialization.
The operational impact of this vulnerability is significant for virtualized environments running on IBM s390 systems, particularly in data center and mainframe virtualization scenarios. When exploited, the vulnerability causes immediate system instability during KVM guest creation, leading to kernel oops and system crashes that can result in service interruptions and potential data loss. The vulnerability affects systems where administrators have explicitly disabled GISA functionality either through kernel boot parameters or runtime sysfs configuration, which is a common configuration in certain virtualization deployments. The issue is particularly concerning because it represents a denial-of-service condition that can be triggered by any attempt to create a new KVM guest when GISA is disabled, making it a critical issue for production environments. The vulnerability also demonstrates a failure in proper initialization sequences within kernel subsystems, which could potentially be exploited to gain unauthorized access or escalate privileges, although the direct attack vector is primarily focused on system stability.
Mitigation strategies for this vulnerability include immediate patch application to the Linux kernel, which implements the fix by ensuring proper initialization of gisa origin values before they are passed to virt_to_phys(). Administrators should also verify that GISA configuration settings are properly initialized and validated when changing kernel parameters or sysfs attributes. Systems running vulnerable kernel versions should be updated to the latest stable releases containing the fix. Additionally, monitoring for kernel messages indicating validity intercepts or related warnings should be implemented as part of routine system maintenance. For environments where immediate patching is not feasible, administrators can consider disabling KVM functionality entirely or using alternative virtualization approaches. The fix addresses the root cause by ensuring that when GISA is disabled, the SIE control block is properly configured with a zero value, which effectively disables GISA usage without causing system instability. This approach aligns with the principle of least privilege and proper resource management in kernel space operations, preventing invalid memory references that could compromise system integrity.