CVE-2024-46797: powerpc/qspinlock: Fix deadlock in MCS queue
In the Linux kernel, the following vulnerability has been resolved:
powerpc/qspinlock: Fix deadlock in MCS queue
If an interrupt occurs in queuedspinlockslowpath() after we increment qnodesp->count and before node->lock is initialized, another CPU might see stale lock values in gettailqnode(). If the stale lock value happens to match the lock on that CPU, then we write to the "next" pointer of the wrong qnode. This causes a deadlock as the former CPU, once it becomes the head of the MCS queue, will spin indefinitely until it's "next" pointer is set by its successor in the queue.
Running stress-ng on a 16 core (16EC/16VP) shared LPAR, results in occasional lockups similar to the following:
$ stress-ng --all 128 --vm-bytes 80% --aggressive \ --maximize --oomable --verify --syslog \ --metrics --times --timeout 5m
watchdog: CPU 15 Hard LOCKUP ...... NIP [c0000000000b78f4] queuedspinlockslowpath+0x1184/0x1490 LR [c000000001037c5c] rawspinlock+0x6c/0x90 Call Trace: 0xc000002cfffa3bf0 (unreliable) rawspinlock+0x6c/0x90 rawspinrqlocknested.part.135+0x4c/0xd0 schedttwupending+0x60/0x1f0 flushsmpcallfunctionqueue+0x1dc/0x670 smpipidemuxrelaxed+0xa4/0x100 xivemuxedipiaction+0x20/0x40 handleirqeventpercpu+0x80/0x240 handleirqeventpercpu+0x2c/0x80 handlepercpuirq+0x84/0xd0 generichandleirq+0x54/0x80 doirq+0xac/0x210 doIRQ+0x74/0xd0 0x0 doIRQ+0x8c/0x170 hardwareinterruptcommonvirt+0x29c/0x2a0 --- interrupt: 500 at queuedspinlockslowpath+0x4b8/0x1490 ...... NIP [c0000000000b6c28] queuedspinlockslowpath+0x4b8/0x1490 LR [c000000001037c5c] rawspinlock+0x6c/0x90 --- interrupt: 500 0xc0000029c1a41d00 (unreliable) rawspinlock+0x6c/0x90 futexwake+0x100/0x260 dofutex+0x21c/0x2a0 sysfutex+0x98/0x270 systemcallexception+0x14c/0x2f0 systemcallvectoredcommon+0x15c/0x2ec
The following code flow illustrates how the deadlock occurs. For the sake of brevity, assume that both locks (A and B) are contended and we call the queuedspinlockslowpath() function.
CPU0 CPU1 ---- ---- spinlockirqsave(A) | spinunlockirqrestore(A) | spinlock(B) | | | ▼ | id = qnodesp->count++; | (Note that nodes[0].lock == A) | | | ▼ | Interrupt | (happens before "nodes[0].lock = B") | | | ▼ | spinlockirqsave(A) | | | ▼ | id = qnodesp->count++ | nodes[1].lock = A | | | ▼ | Tail of MCS queue | | spinlockirqsave(A) ▼ | Head of MCS queue ▼ | CPU0 is previous tail ▼ | Spin indefinitely ▼ (until "nodes[1].next != NULL") prev = gettailqnode(A, CPU0) | ▼ prev == &qnodes[CPU0].nodes[0] (as qnodes ---truncated---
Other sources
This CVE was automatically created from a reference found in an email or other text. If you are reading this, then this CVE entry is probably erroneous, since this text should be replaced by the official CVE description automatically.
— Launchpad
Affected Software
Remediation
Event History
Frequently Asked Questions
What is the severity of CVE-2024-46797?
CVE-2024-46797 is regarded as a medium severity vulnerability in the Linux kernel.
How do I fix CVE-2024-46797?
To resolve CVE-2024-46797, update your Linux kernel to a patched version provided by your distribution.
Which versions of the Linux kernel are affected by CVE-2024-46797?
CVE-2024-46797 affects Linux kernel versions from 6.2 up to 6.6.51 and versions between 6.7 and 6.10.10.
Is CVE-2024-46797 present in the release candidates of the Linux kernel?
Yes, CVE-2024-46797 is also present in the release candidate versions of the Linux kernel 6.11 (rc1 to rc6).
What can CVE-2024-46797 lead to if exploited?
If exploited, CVE-2024-46797 could lead to a deadlock situation in the Linux kernel, potentially causing system instability.