GHSA-P69H-P269-MRVC
Vulnerability from github – Published: 2026-09-11 21:31 – Updated: 2026-09-13 09:32In the Linux kernel, the following vulnerability has been resolved:
apparmor: fix cred UAF caused by begin_current_label_crit_section()
AppArmor's begin_current_label_crit_section() is a scary function called from lots of LSM hooks (in particular VFS/socket-related ones) that checks if the label referenced by the current creds is marked FLAG_STALE, and if so, attempts to use aa_replace_current_label() to replace the creds with an updated version that uses a new label.
The first problem with this is that it would directly lead to UAF of
struct cred if anything in the kernel takes a pointer to the current
creds and accesses these past a security hook invocation that replaces
creds, like so:
const struct cred *cred = current_cred();
alloc_file_pseudo(...);
uid_t uid = cred->euid;
I don't know if anything in the kernel actually does this, but I think it is very surprising that this pattern could lead to UAF.
The second problem is that things go wrong when aa_replace_current_label()
runs with overridden credentials. aa_replace_current_label() bails out if
current_cred() != current_real_cred() (mirroring the check in
proc_pid_attr_write()), but this check can't actually reliably detect
overridden credentials because the overridden creds can be the same as the
objective creds.
So in approximately the following scenario, things go wrong:
- task begins with (as both objective and subjective creds), with refcount=2
- task grabs an extra reference on for overriding
- task calls override_creds(), which returns a pointer to the old subjective creds ()
- task enters AppArmor LSM hook
- AppArmor checks that objective/subjective creds are equal
- AppArmor replaces both cred pointers with and drops 2 refs on
- task leaves AppArmor LSM hook
- task calls revert_creds()
- now task->cred is while task->real_cred is , but the task_struct logically holds two references to
- another task drops the extra reference on that was used for overriding, refcount drops to 0
- now task->real_cred points to freed creds
At this point, any access to current_cred() will be UAF.
I have a test case where I run aa-disable on a profile while a process using that profile is blocked on splice() from a FUSE passthrough file into a full pipe; after the profile update, the pipe becomes empty, splice() resumes, the credentials go out of sync, and a subsequent getuid() syscall results in a KASAN UAF splat.
To fix this, instead of directly replacing creds, do it via task_work that will run at the end of the current syscall. (The point in time at which the cred replacement happens should have no correctness impact; it is just a performance optimization to avoid unnecessarily touching the refcount of the new label.)
Note that AppArmor still performs direct cred replacements in the sb_pivotroot LSM hook after this change, and that direct cred replacements can still happen in VFS ->write() callbacks via proc_pid_attr_write().
There are two options for what to do with aa_dup_task_ctx(): Either explicitly reset new->label_replacement_pending after the entire aa_task_ctx has been copied, or switch to manually copying members over. I am switching to manually copying members over because that should make bugs more obvious.
{
"affected": [],
"aliases": [
"CVE-2026-89762"
],
"database_specific": {
"cwe_ids": [],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-09-11T20:20:07Z",
"severity": "HIGH"
},
"details": "In the Linux kernel, the following vulnerability has been resolved:\n\napparmor: fix cred UAF caused by begin_current_label_crit_section()\n\nAppArmor\u0027s begin_current_label_crit_section() is a scary function called\nfrom lots of LSM hooks (in particular VFS/socket-related ones) that checks\nif the label referenced by the current creds is marked FLAG_STALE, and if\nso, attempts to use aa_replace_current_label() to replace the creds with an\nupdated version that uses a new label.\n\nThe first problem with this is that it would directly lead to UAF of\n`struct cred` if anything in the kernel takes a pointer to the current\ncreds and accesses these past a security hook invocation that replaces\ncreds, like so:\n```\nconst struct cred *cred = current_cred();\nalloc_file_pseudo(...);\nuid_t uid = cred-\u003eeuid;\n```\nI don\u0027t know if anything in the kernel actually does this, but I think it\nis very surprising that this pattern could lead to UAF.\n\nThe second problem is that things go wrong when aa_replace_current_label()\nruns with overridden credentials. aa_replace_current_label() bails out if\n`current_cred() != current_real_cred()` (mirroring the check in\nproc_pid_attr_write()), but this check can\u0027t actually reliably detect\noverridden credentials because the overridden creds can be the same as the\nobjective creds.\n\nSo in approximately the following scenario, things go wrong:\n\n1. task begins with \u003ccreds A\u003e (as both objective and subjective creds),\n with refcount=2\n2. task grabs an extra reference on \u003ccreds A\u003e for overriding\n3. task calls override_creds(\u003ccreds A\u003e), which returns a pointer to the old\n subjective creds (\u003ccreds A\u003e)\n4. task enters AppArmor LSM hook\n5. AppArmor checks that objective/subjective creds are equal\n6. AppArmor replaces both cred pointers with \u003ccreds B\u003e and drops 2 refs on\n \u003ccreds A\u003e\n7. task leaves AppArmor LSM hook\n8. task calls revert_creds(\u003ccreds A\u003e)\n9. now task-\u003ecred is \u003ccreds A\u003e while task-\u003ereal_cred is \u003ccreds B\u003e, but the\n task_struct logically holds two references to \u003ccreds B\u003e\n10. another task drops the extra reference on \u003ccreds A\u003e that was used for\n overriding, refcount drops to 0\n11. now task-\u003ereal_cred points to freed creds\n\nAt this point, any access to current_cred() will be UAF.\n\nI have a test case where I run aa-disable on a profile while a process\nusing that profile is blocked on splice() from a FUSE passthrough file into\na full pipe; after the profile update, the pipe becomes empty, splice()\nresumes, the credentials go out of sync, and a subsequent getuid() syscall\nresults in a KASAN UAF splat.\n\nTo fix this, instead of directly replacing creds, do it via task_work that\nwill run at the end of the current syscall. (The point in time at which the\ncred replacement happens should have no correctness impact; it is just a\nperformance optimization to avoid unnecessarily touching the refcount of\nthe new label.)\n\nNote that AppArmor still performs direct cred replacements in the\nsb_pivotroot LSM hook after this change, and that direct cred replacements\ncan still happen in VFS -\u003ewrite() callbacks via proc_pid_attr_write().\n\nThere are two options for what to do with aa_dup_task_ctx(): Either\nexplicitly reset new-\u003elabel_replacement_pending after the entire\naa_task_ctx has been copied, or switch to manually copying members over.\nI am switching to manually copying members over because that should make\nbugs more obvious.",
"id": "GHSA-p69h-p269-mrvc",
"modified": "2026-09-13T09:32:30Z",
"published": "2026-09-11T21:31:39Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-89762"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/361488984d668235438faac28635f3632351407f"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/3f4ae5fab613dca01d6a2a8210dd832e009fcf47"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/580f777d6d9fd07fc034bd1aeba5c30fd48871a2"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/587a6a92b93ec314c583bbf747413af170d42540"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.