CVE-2026-74702 (GCVE-0-2026-74702)
Vulnerability from cvelistv5
Published
2026-08-22 15:33
Modified
2026-08-25 05:41
Severity ?
VLAI Severity ?
EPSS score ?
Summary
In the Linux kernel, the following vulnerability has been resolved:
vhost-scsi: reject feature changes after endpoint
vhost_scsi_setup_vq_cmds() runs from VHOST_SCSI_SET_ENDPOINT and allocates
each command's protection scatterlist array (prot_sgl) according to the
acknowledged VIRTIO_SCSI_F_T10_PI bit. The command pools are not rebuilt
when VHOST_SET_FEATURES changes that bit later.
Although virtio feature bits must not change after feature negotiation,
vhost_scsi_set_features() currently accepts such a request after the
endpoint is active and updates acked_features. Enabling T10-PI after
endpoint setup therefore leaves prot_sgl NULL while the I/O path follows
the new feature bit.
For a 129-page protection payload, vhost_scsi_mapal() passes the missing
first chunk to sg_alloc_table_chained():
sg_alloc_table_chained(table, 129, first_chunk=NULL,
nents_first_chunk=inline_sg_cnt)
sg_pool_index() then hits:
BUG_ON(nents > SG_CHUNK_SIZE); /* 129 > 128 */
The kernel reported the following call trace and register state:
Call Trace:
<TASK>
? __sg_alloc_table+0x1d8/0x250
? __pfx_vhost_run_work_list+0x10/0x10 [vhost]
sg_alloc_table_chained+0x59/0xf0
? __pfx_sg_pool_alloc+0x10/0x10
? vhost_scsi_calc_sgls.constprop.0+0x43/0x60 [vhost_scsi]
vhost_scsi_handle_vq+0xf02/0x1700 [vhost_scsi]
? __pfx_vhost_scsi_handle_vq+0x10/0x10 [vhost_scsi]
vhost_scsi_handle_kick+0x37/0x50 [vhost_scsi]
vhost_run_work_list+0x8e/0xd0 [vhost]
vhost_task_fn+0xe1/0x210
ret_from_fork+0x348/0x540
</TASK>
RIP: 0010:0x4
CR2 = 0x4
RSP: 0018:ffffc90000dbf940 EFLAGS: 00010202
RAX: ffffffff82396810 RBX: ffff88811dc28b80 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000820 RDI: 0000000000000081
VHOST_F_LOG_ALL is a vhost-specific runtime feature and remains the only
exception.
Reject changes to any feature other than VHOST_F_LOG_ALL while the
endpoint is active. This preserves the existing runtime log toggle while
preventing feature-dependent command resources and data-path state from
becoming inconsistent. Userspace must clear the endpoint before changing
any other negotiated feature and set the endpoint up again afterward.
References
Impacted products
{
"containers": {
"cna": {
"affected": [
{
"defaultStatus": "unaffected",
"product": "Linux",
"programFiles": [
"drivers/vhost/scsi.c"
],
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"versions": [
{
"lessThan": "a06e4611d45518896fbff4f45d9581578b107e91",
"status": "affected",
"version": "bf2d650391be508dd8b5e188b65ed32300cf3489",
"versionType": "git"
},
{
"lessThan": "9a3eb77a612f9d158e4d27df43677a014e9cfa55",
"status": "affected",
"version": "bf2d650391be508dd8b5e188b65ed32300cf3489",
"versionType": "git"
},
{
"lessThan": "42bc45df5905e2b7dccb72adaf7730f66cfbe03f",
"status": "affected",
"version": "bf2d650391be508dd8b5e188b65ed32300cf3489",
"versionType": "git"
}
]
},
{
"defaultStatus": "affected",
"product": "Linux",
"programFiles": [
"drivers/vhost/scsi.c"
],
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"versions": [
{
"status": "affected",
"version": "6.15"
},
{
"lessThan": "6.15",
"status": "unaffected",
"version": "0",
"versionType": "semver"
},
{
"lessThanOrEqual": "6.18.*",
"status": "unaffected",
"version": "6.18.45",
"versionType": "semver"
},
{
"lessThanOrEqual": "7.1.*",
"status": "unaffected",
"version": "7.1.9",
"versionType": "semver"
},
{
"lessThanOrEqual": "*",
"status": "unaffected",
"version": "7.2",
"versionType": "original_commit_for_fix"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
"versionEndExcluding": "6.18.45",
"versionStartIncluding": "6.15",
"vulnerable": true
},
{
"criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
"versionEndExcluding": "7.1.9",
"versionStartIncluding": "6.15",
"vulnerable": true
},
{
"criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
"versionEndExcluding": "7.2",
"versionStartIncluding": "6.15",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"descriptions": [
{
"lang": "en",
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\nvhost-scsi: reject feature changes after endpoint\n\nvhost_scsi_setup_vq_cmds() runs from VHOST_SCSI_SET_ENDPOINT and allocates\neach command\u0027s protection scatterlist array (prot_sgl) according to the\nacknowledged VIRTIO_SCSI_F_T10_PI bit. The command pools are not rebuilt\nwhen VHOST_SET_FEATURES changes that bit later.\n\nAlthough virtio feature bits must not change after feature negotiation,\nvhost_scsi_set_features() currently accepts such a request after the\nendpoint is active and updates acked_features. Enabling T10-PI after\nendpoint setup therefore leaves prot_sgl NULL while the I/O path follows\nthe new feature bit.\n\nFor a 129-page protection payload, vhost_scsi_mapal() passes the missing\nfirst chunk to sg_alloc_table_chained():\n\n sg_alloc_table_chained(table, 129, first_chunk=NULL,\n nents_first_chunk=inline_sg_cnt)\n\nsg_pool_index() then hits:\n\n BUG_ON(nents \u003e SG_CHUNK_SIZE); /* 129 \u003e 128 */\n\nThe kernel reported the following call trace and register state:\n\n Call Trace:\n \u003cTASK\u003e\n ? __sg_alloc_table+0x1d8/0x250\n ? __pfx_vhost_run_work_list+0x10/0x10 [vhost]\n sg_alloc_table_chained+0x59/0xf0\n ? __pfx_sg_pool_alloc+0x10/0x10\n ? vhost_scsi_calc_sgls.constprop.0+0x43/0x60 [vhost_scsi]\n vhost_scsi_handle_vq+0xf02/0x1700 [vhost_scsi]\n ? __pfx_vhost_scsi_handle_vq+0x10/0x10 [vhost_scsi]\n vhost_scsi_handle_kick+0x37/0x50 [vhost_scsi]\n vhost_run_work_list+0x8e/0xd0 [vhost]\n vhost_task_fn+0xe1/0x210\n ret_from_fork+0x348/0x540\n \u003c/TASK\u003e\n\n RIP: 0010:0x4\n CR2 = 0x4\n RSP: 0018:ffffc90000dbf940 EFLAGS: 00010202\n RAX: ffffffff82396810 RBX: ffff88811dc28b80 RCX: 0000000000000000\n RDX: 0000000000000000 RSI: 0000000000000820 RDI: 0000000000000081\n\nVHOST_F_LOG_ALL is a vhost-specific runtime feature and remains the only\nexception.\n\nReject changes to any feature other than VHOST_F_LOG_ALL while the\nendpoint is active. This preserves the existing runtime log toggle while\npreventing feature-dependent command resources and data-path state from\nbecoming inconsistent. Userspace must clear the endpoint before changing\nany other negotiated feature and set the endpoint up again afterward."
}
],
"metrics": [
{
"cvssV3_1": {
"baseScore": 8.8,
"baseSeverity": "HIGH",
"vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H",
"version": "3.1"
},
"scenarios": [
{
"lang": "en",
"value": "AV:L - vhost-scsi is only reachable through virtio-scsi virtqueue kicks and VHOST_* ioctls on /dev/vhost-scsi by the VMM or a VM guest; it is not exposed via any network protocol or remote packet path.\nAC:L - Once VIRTIO_SCSI_F_T10_PI is toggled after VHOST_SCSI_SET_ENDPOINT, a virtio-scsi request with protection data spanning 129+ pages deterministically drives vhost_scsi_mapal() into sg_alloc_table_chained() with NULL prot_sgl and triggers BUG_ON(nents \u003e SG_CHUNK_SIZE); no races or victim-specific memory layout are required.\nPR:L - No host root or CAP_SYS_ADMIN is needed: an unprivileged cloud VM tenant can negotiate virtio-scsi T10-PI (QEMU propagates VHOST_SET_FEATURES) and submit the oversized PI I/O, and vhost ioctls have no capability gate for any process holding the vhost-scsi fd.\nUI:N - Exploitation requires only attacker-controlled virtio-scsi I/O and/or VMM feature updates during normal VM operation; no administrator or end-user action beyond routine VM use is needed.\nS:C - On KVM hosts using virtio-scsi with the vhost backend, a malicious guest reaches host-kernel vhost-scsi worker code and can panic the hypervisor, crossing the guest-to-host virtual machine security boundary.\nC:H - Enabling T10-PI after endpoint setup leaves prot_sgl NULL while the data path allocates and maps protection scatterlists from the shared sg pool; the reporter hit a fault at address 0x4, and smaller requests can complete with mismatched inline/pool metadata that corrupts kernel heap bookkeeping on free.\nI:H - Attacker-controlled protection iterators are mapped into kernel scatterlists passed to the LIO/TCM target stack under inconsistent feature state, yielding heap writes and unreliable sg_free_table_chained() teardown that can be leveraged for control-flow hijack beyond the immediate BUG_ON.\nA:H - The demonstrated path hits BUG_ON(nents \u003e SG_CHUNK_SIZE) in sg_pool_index() from the vhost worker during vhost_scsi_handle_vq(), producing a host kernel oops/panic that terminates the hypervisor and all co-resident VMs."
}
]
}
],
"providerMetadata": {
"dateUpdated": "2026-08-25T05:41:45.057Z",
"orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"shortName": "Linux"
},
"references": [
{
"url": "https://git.kernel.org/stable/c/a06e4611d45518896fbff4f45d9581578b107e91"
},
{
"url": "https://git.kernel.org/stable/c/9a3eb77a612f9d158e4d27df43677a014e9cfa55"
},
{
"url": "https://git.kernel.org/stable/c/42bc45df5905e2b7dccb72adaf7730f66cfbe03f"
}
],
"title": "vhost-scsi: reject feature changes after endpoint",
"x_generator": {
"engine": "bippy-1.2.0"
}
}
},
"cveMetadata": {
"assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"assignerShortName": "Linux",
"cveId": "CVE-2026-74702",
"datePublished": "2026-08-22T15:33:02.223Z",
"dateReserved": "2026-08-15T05:44:03.927Z",
"dateUpdated": "2026-08-25T05:41:45.057Z",
"state": "PUBLISHED"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2"
}
Loading…
Loading…
Sightings
| Author | Source | Type | Date |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or seen somewhere by the user.
- Confirmed: The vulnerability is confirmed from an analyst perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: This vulnerability was exploited and seen by the user reporting the sighting.
- Patched: This vulnerability was successfully patched by the user reporting the sighting.
- Not exploited: This vulnerability was not exploited or seen by the user reporting the sighting.
- Not confirmed: The user expresses doubt about the veracity of the vulnerability.
- Not patched: This vulnerability was not successfully patched by the user reporting the sighting.
Loading…
Loading…