ghsa-47fv-c4cg-c767
Vulnerability from github
In the Linux kernel, the following vulnerability has been resolved:
ipc/mqueue, msg, sem: avoid relying on a stack reference past its expiry
do_mq_timedreceive calls wq_sleep with a stack local address. The sender (do_mq_timedsend) uses this address to later call pipelined_send.
This leads to a very hard to trigger race where a do_mq_timedreceive call might return and leave do_mq_timedsend to rely on an invalid address, causing the following crash:
RIP: 0010:wake_q_add_safe+0x13/0x60 Call Trace: __x64_sys_mq_timedsend+0x2a9/0x490 do_syscall_64+0x80/0x680 entry_SYSCALL_64_after_hwframe+0x44/0xa9 RIP: 0033:0x7f5928e40343
The race occurs as:
-
do_mq_timedreceive calls wq_sleep with the address of
struct ext_wait_queueon function stack (aliased asewq_addrhere) - it holds a validstruct ext_wait_queue *as long as the stack has not been overwritten. -
ewq_addrgets added to info->e_wait_q[RECV].list in wq_add, and do_mq_timedsend receives it via wq_get_first_waiter(info, RECV) to call __pipelined_op. -
Sender calls __pipelined_op::smp_store_release(&this->state, STATE_READY). Here is where the race window begins. (
thisisewq_addr.) -
If the receiver wakes up now in do_mq_timedreceive::wq_sleep, it will see
state == STATE_READYand break. -
do_mq_timedreceive returns, and
ewq_addris no longer guaranteed to be astruct ext_wait_queue *since it was on do_mq_timedreceive's stack. (Although the address may not get overwritten until another function happens to touch it, which means it can persist around for an indefinite time.) -
do_mq_timedsend::__pipelined_op() still believes
ewq_addris astruct ext_wait_queue *, and uses it to find a task_struct to pass to the wake_q_add_safe call. In the lucky case where nothing has overwrittenewq_addryet,ewq_addr->taskis the right task_struct. In the unlucky case, __pipelined_op::wake_q_add_safe gets handed a bogus address as the receiver's task_struct causing the crash.
do_mq_timedsend::__pipelined_op() should not dereference this after
setting STATE_READY, as the receiver counterpart is now free to return.
Change __pipelined_op to call wake_q_add_safe on the receiver's
task_struct returned by get_task_struct, instead of dereferencing this
which sits on the receiver's stack.
As Manfred pointed out, the race potentially also exists in ipc/msg.c::expunge_all and ipc/sem.c::wake_up_sem_queue_prepare. Fix those in the same way.
{
"affected": [],
"aliases": [
"CVE-2021-47069"
],
"database_specific": {
"cwe_ids": [],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2024-03-01T22:15:46Z",
"severity": null
},
"details": "In the Linux kernel, the following vulnerability has been resolved:\n\nipc/mqueue, msg, sem: avoid relying on a stack reference past its expiry\n\ndo_mq_timedreceive calls wq_sleep with a stack local address. The\nsender (do_mq_timedsend) uses this address to later call pipelined_send.\n\nThis leads to a very hard to trigger race where a do_mq_timedreceive\ncall might return and leave do_mq_timedsend to rely on an invalid\naddress, causing the following crash:\n\n RIP: 0010:wake_q_add_safe+0x13/0x60\n Call Trace:\n __x64_sys_mq_timedsend+0x2a9/0x490\n do_syscall_64+0x80/0x680\n entry_SYSCALL_64_after_hwframe+0x44/0xa9\n RIP: 0033:0x7f5928e40343\n\nThe race occurs as:\n\n1. do_mq_timedreceive calls wq_sleep with the address of `struct\n ext_wait_queue` on function stack (aliased as `ewq_addr` here) - it\n holds a valid `struct ext_wait_queue *` as long as the stack has not\n been overwritten.\n\n2. `ewq_addr` gets added to info-\u003ee_wait_q[RECV].list in wq_add, and\n do_mq_timedsend receives it via wq_get_first_waiter(info, RECV) to call\n __pipelined_op.\n\n3. Sender calls __pipelined_op::smp_store_release(\u0026this-\u003estate,\n STATE_READY). Here is where the race window begins. (`this` is\n `ewq_addr`.)\n\n4. If the receiver wakes up now in do_mq_timedreceive::wq_sleep, it\n will see `state == STATE_READY` and break.\n\n5. do_mq_timedreceive returns, and `ewq_addr` is no longer guaranteed\n to be a `struct ext_wait_queue *` since it was on do_mq_timedreceive\u0027s\n stack. (Although the address may not get overwritten until another\n function happens to touch it, which means it can persist around for an\n indefinite time.)\n\n6. do_mq_timedsend::__pipelined_op() still believes `ewq_addr` is a\n `struct ext_wait_queue *`, and uses it to find a task_struct to pass to\n the wake_q_add_safe call. In the lucky case where nothing has\n overwritten `ewq_addr` yet, `ewq_addr-\u003etask` is the right task_struct.\n In the unlucky case, __pipelined_op::wake_q_add_safe gets handed a\n bogus address as the receiver\u0027s task_struct causing the crash.\n\ndo_mq_timedsend::__pipelined_op() should not dereference `this` after\nsetting STATE_READY, as the receiver counterpart is now free to return.\nChange __pipelined_op to call wake_q_add_safe on the receiver\u0027s\ntask_struct returned by get_task_struct, instead of dereferencing `this`\nwhich sits on the receiver\u0027s stack.\n\nAs Manfred pointed out, the race potentially also exists in\nipc/msg.c::expunge_all and ipc/sem.c::wake_up_sem_queue_prepare. Fix\nthose in the same way.",
"id": "GHSA-47fv-c4cg-c767",
"modified": "2024-03-02T00:31:30Z",
"published": "2024-03-02T00:31:30Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2021-47069"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/4528c0c323085e645b8765913b4a7fd42cf49b65"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/807fa14536b26803b858da878b643be72952a097"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/a11ddb37bf367e6b5239b95ca759e5389bb46048"
}
],
"schema_version": "1.4.0",
"severity": []
}