CVE-2026-93164
Estado: RecibidaSin puntuar—
In the Linux kernel, the following vulnerability has been resolved:
uprobes/x86: Move optimized uprobe from nop5 to nop10
Andrii reported an issue with optimized uprobes [1] that can clobber redzone area with call instruction storing return address on stack where user code may keep temporary data without adjusting rsp.
Fixing this by moving the optimized uprobes on top of 10-bytes nop instruction, so we can squeeze another instruction to escape the redzone area before doing the call, like:
Note the lea instruction is used to adjust the rsp register without changing the flags.
We use nop10 and following transformation to optimized instructions above and back as suggested by Peterz [2].
Leer descripción completaMostrar menos
Optimize path (int3_update_optimize):
Unoptimize path (int3_update_unoptimize):
Note as explained in [2] we need to use following nop10: PF1 PF2 ESC NOPL MOD SIB DISP32 NOP10: 0x66, 0x2e, 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 -- cs nopw 0x00000000(%rax,%rax,1)
which means we need to allow 0x2e prefix which maps to INAT_PFX_CS attribute in is_prefix_bad function.
Also changing the uprobe syscall error when called out of uprobe trampoline to -EPROTO, so we are able to detect the fixed kernel.
The optimized uprobe performance stays the same:
[1] https://lore.kernel.org/bpf/20260509003146.976844-1-andrii@kernel.org/ [2] https://lore.kernel.org/bpf/20260518104306.GU3102624@noisy.programming.kicks-ass.net/#t
Detalles técnicos trazas, registros y código del informe original
lea -0x80(%rsp), %rsp
call tramp
1) Initial state after set_swbp() installed the uprobe:
cc 2e 0f 1f 84 00 00 00 00 00
From offset 0 this is INT3 followed by the tail of the original
10-byte NOP.
After a previous unoptimization bytes 5..9 may still contain the
old call instruction, which remains valid for threads already there.
2) Rewrite the LEA tail and call displacement:
cc [8d 64 24 80 e8 d0 d1 d2 d3]
From offset 0 this traps on the uprobe INT3. Bytes 1..9 are not
executable entry points while byte 0 is trapped.
3) Publish the first LEA byte:
[48] 8d 64 24 80 e8 d0 d1 d2 d3
From offset 0 this is:
lea -0x80(%rsp), %rsp
call <uprobe-trampoline>
1) Initial optimized state:
48 8d 64 24 80 e8 d0 d1 d2 d3
Same as 3) above.
2) Trap new entries before restoring the NOP bytes:
[cc] 8d 64 24 80 e8 d0 d1 d2 d3
From offset 0 this traps. A thread that had already executed the
LEA can still reach the intact CALL at offset 5.
3) Restore bytes 1..4 of the original NOP while keeping byte 0 trapped
and byte 5 as CALL.
cc [2e 0f 1f 84] e8 d0 d1 d2 d3
From offset 0 this still traps. Offset 5 is still the CALL for any
thread that was already past the first LEA byte.
4) Publish the first byte of the original NOP:
[66] 2e 0f 1f 84 e8 d0 d1 d2 d3
From offset 0 this is the restored 10-byte NOP; the CALL opcode and
displacement are now only NOP operands. Offset 5 still decodes as
CALL for a thread that was already there.
Tthere is only a single target uprobe-trampoline for the given nop10
instruction address, so the CALL instruction will not be changed across
unoptimization/optimization cycles.
Therefore, any task that is preempted at the CALL instruction is guaranteed
to observe that CALL and not anything else.
uprobe-nop : 3.129 ± 0.013M/s
uprobe-push : 3.045 ± 0.006M/s
uprobe-ret : 1.095 ± 0.004M/s
--> uprobe-nop10 : 7.170 ± 0.020M/s
uretprobe-nop : 2.143 ± 0.021M/s
uretprobe-push : 2.090 ± 0.000M/s
uretprobe-ret : 0.942 ± 0.000M/s
--> uretprobe-nop10: 3.381 ± 0.003M/s
usdt-nop : 3.245 ± 0.004M/s
--> usdt-nop10 : 7.256 ± 0.023M/sCVSS
NVD no ha asignado puntuación CVSS a esta CVE (habitual desde el cambio de política de abril de 2026).
Probabilidad de explotación (EPSS)
- Probabilidad de explotación en los próximos 30 días: 0.20%
- Percentil entre todas las CVEs puntuadas: 9
- Fecha de la puntuación: 6/10/2026
EPSS (Exploit Prediction Scoring System, de FIRST) estima la probabilidad de que una vulnerabilidad sea explotada en 30 días. Complementa a CVSS (impacto) y a CISA KEV (explotación confirmada).
Tecnologías afectadas (1)
⚠ Inferidas por IA a partir de la descripción — NVD aún no ha analizado esta CVE; no son CPE verificados.
Referencias
JSON original (NVD)
Mostrar
{
"id": "CVE-2026-93164",
"cveTags": [],
"metrics": {},
"affected": [
{
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"affectedData": [
{
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"product": "Linux",
"versions": [
{
"status": "affected",
"version": "ba2bfc97b4629b10bd8d02b36e04f3932a04cac4",
"lessThan": "1b3fecd09910040668b752f3377d7218ca5c59ac",
"versionType": "git"
},
{
"status": "affected",
"version": "ba2bfc97b4629b10bd8d02b36e04f3932a04cac4",
"lessThan": "554ba38456dad8053a1a80afe6ae6da9eff745cc",
"versionType": "git"
}
],
"programFiles": [
"arch/x86/kernel/uprobes.c"
],
"defaultStatus": "unaffected"
},
{
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"product": "Linux",
"versions": [
{
"status": "affected",
"version": "6.18"
},
{
"status": "unaffected",
"version": "0",
"lessThan": "6.18",
"versionType": "semver"
},
{
"status": "unaffected",
"version": "7.2.6",
"versionType": "semver",
"lessThanOrEqual": "7.2.*"
},
{
"status": "unaffected",
"version": "7.3-rc1",
"versionType": "original_commit_for_fix",
"lessThanOrEqual": "*"
}
],
"programFiles": [
"arch/x86/kernel/uprobes.c"
],
"defaultStatus": "affected"
}
]
}
],
"published": "2026-09-17T17:18:11.890",
"references": [
{
"url": "https://git.kernel.org/stable/c/1b3fecd09910040668b752f3377d7218ca5c59ac",
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
},
{
"url": "https://git.kernel.org/stable/c/554ba38456dad8053a1a80afe6ae6da9eff745cc",
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
}
],
"vulnStatus": "Received",
"descriptions": [
{
"lang": "en",
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\nuprobes/x86: Move optimized uprobe from nop5 to nop10\n\nAndrii reported an issue with optimized uprobes [1] that can clobber\nredzone area with call instruction storing return address on stack\nwhere user code may keep temporary data without adjusting rsp.\n\nFixing this by moving the optimized uprobes on top of 10-bytes nop\ninstruction, so we can squeeze another instruction to escape the\nredzone area before doing the call, like:\n\n lea -0x80(%rsp), %rsp\n call tramp\n\nNote the lea instruction is used to adjust the rsp register without\nchanging the flags.\n\nWe use nop10 and following transformation to optimized instructions\nabove and back as suggested by Peterz [2].\n\nOptimize path (int3_update_optimize):\n\n 1) Initial state after set_swbp() installed the uprobe:\n cc 2e 0f 1f 84 00 00 00 00 00\n\n From offset 0 this is INT3 followed by the tail of the original\n 10-byte NOP.\n\n After a previous unoptimization bytes 5..9 may still contain the\n old call instruction, which remains valid for threads already there.\n\n 2) Rewrite the LEA tail and call displacement:\n cc [8d 64 24 80 e8 d0 d1 d2 d3]\n\n From offset 0 this traps on the uprobe INT3. Bytes 1..9 are not\n executable entry points while byte 0 is trapped.\n\n 3) Publish the first LEA byte:\n [48] 8d 64 24 80 e8 d0 d1 d2 d3\n\n From offset 0 this is:\n lea -0x80(%rsp), %rsp\n call <uprobe-trampoline>\n\nUnoptimize path (int3_update_unoptimize):\n\n 1) Initial optimized state:\n 48 8d 64 24 80 e8 d0 d1 d2 d3\n Same as 3) above.\n\n 2) Trap new entries before restoring the NOP bytes:\n [cc] 8d 64 24 80 e8 d0 d1 d2 d3\n\n From offset 0 this traps. A thread that had already executed the\n LEA can still reach the intact CALL at offset 5.\n\n 3) Restore bytes 1..4 of the original NOP while keeping byte 0 trapped\n and byte 5 as CALL.\n cc [2e 0f 1f 84] e8 d0 d1 d2 d3\n\n From offset 0 this still traps. Offset 5 is still the CALL for any\n thread that was already past the first LEA byte.\n\n 4) Publish the first byte of the original NOP:\n [66] 2e 0f 1f 84 e8 d0 d1 d2 d3\n\n From offset 0 this is the restored 10-byte NOP; the CALL opcode and\n displacement are now only NOP operands. Offset 5 still decodes as\n CALL for a thread that was already there.\n\n Tthere is only a single target uprobe-trampoline for the given nop10\n instruction address, so the CALL instruction will not be changed across\n unoptimization/optimization cycles.\n Therefore, any task that is preempted at the CALL instruction is guaranteed\n to observe that CALL and not anything else.\n\nNote as explained in [2] we need to use following nop10:\n PF1 PF2 ESC NOPL MOD SIB DISP32\nNOP10: 0x66, 0x2e, 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 -- cs nopw 0x00000000(%rax,%rax,1)\n\nwhich means we need to allow 0x2e prefix which maps to INAT_PFX_CS\nattribute in is_prefix_bad function.\n\nAlso changing the uprobe syscall error when called out of uprobe\ntrampoline to -EPROTO, so we are able to detect the fixed kernel.\n\nThe optimized uprobe performance stays the same:\n\n uprobe-nop : 3.129 ± 0.013M/s\n uprobe-push : 3.045 ± 0.006M/s\n uprobe-ret : 1.095 ± 0.004M/s\n --> uprobe-nop10 : 7.170 ± 0.020M/s\n uretprobe-nop : 2.143 ± 0.021M/s\n uretprobe-push : 2.090 ± 0.000M/s\n uretprobe-ret : 0.942 ± 0.000M/s\n --> uretprobe-nop10: 3.381 ± 0.003M/s\n usdt-nop : 3.245 ± 0.004M/s\n --> usdt-nop10 : 7.256 ± 0.023M/s\n\n[1] https://lore.kernel.org/bpf/20260509003146.976844-1-andrii@kernel.org/\n[2] https://lore.kernel.org/bpf/20260518104306.GU3102624@noisy.programming.kicks-ass.net/#t"
}
],
"lastModified": "2026-09-17T17:18:11.890",
"sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
}