CVE-2024-39474: mm/vmalloc: fix vmalloc which may return null if called with __GFP_NOFAIL
In the Linux kernel, the following vulnerability has been resolved:
mm/vmalloc: fix vmalloc which may return null if called with GFPNOFAIL
commit a421ef303008 ("mm: allow !GFPKERNEL allocations for kvmalloc") includes support for GFPNOFAIL, but it presents a conflict with commit dd544141b9eb ("vmalloc: back off when the current task is OOM-killed"). A possible scenario is as follows:
process-a vmallocnoderange(GFPKERNEL | GFPNOFAIL) vmallocareanode() vmareaallocpages() --> oom-killer send SIGKILL to process-a if (fatalsignalpending(current)) break; --> return NULL;
To fix this, do not check fatalsignalpending() in vmareaallocpages() if GFPNOFAIL set.
This issue occurred during OPLUS KASAN TEST. Below is part of the log -> oom-killer sends signal to process [65731.222840] [ T1308] oom-kill:constraint=CONSTRAINTNONE,nodemask=(null),cpuset=/,memsallowed=0,globaloom,taskmemcg=/apps/uid10198,task=gs.intelligence,pid=32454,uid=10198
[65731.259685] [T32454] Call trace: [65731.259698] [T32454] dumpbacktrace+0xf4/0x118 [65731.259734] [T32454] showstack+0x18/0x24 [65731.259756] [T32454] dumpstacklvl+0x60/0x7c [65731.259781] [T32454] dumpstack+0x18/0x38 [65731.259800] [T32454] mrdumpcommondie+0x250/0x39c [mrdump] [65731.259936] [T32454] ipanicdie+0x20/0x34 [mrdump] [65731.260019] [T32454] atomicnotifiercallchain+0xb4/0xfc [65731.260047] [T32454] notifydie+0x114/0x198 [65731.260073] [T32454] die+0xf4/0x5b4 [65731.260098] [T32454] diekernelfault+0x80/0x98 [65731.260124] [T32454] dokernelfault+0x160/0x2a8 [65731.260146] [T32454] dobadarea+0x68/0x148 [65731.260174] [T32454] domemabort+0x151c/0x1b34 [65731.260204] [T32454] el1abort+0x3c/0x5c [65731.260227] [T32454] el1h64synchandler+0x54/0x90 [65731.260248] [T32454] el1h64sync+0x68/0x6c
[65731.260269] [T32454] zerofsdecompressqueue+0x7f0/0x2258 --> be->decompressedpages = kvcalloc(be->nrpages, sizeof(struct page ), GFPKERNEL | GFPNOFAIL); kernel panic by NULL pointer dereference. erofs assume kvmalloc with GFPNOFAIL never return NULL. [65731.260293] [T32454] zerofsrunqueue+0xf30/0x104c [65731.260314] [T32454] zerofsreadahead+0x4f0/0x968 [65731.260339] [T32454] readpages+0x170/0xadc [65731.260364] [T32454] pagecacheraunbounded+0x874/0xf30 [65731.260388] [T32454] pagecacheraorder+0x24c/0x714 [65731.260411] [T32454] filemapfault+0xbf0/0x1a74 [65731.260437] [T32454] dofault+0xd0/0x33c [65731.260462] [T32454] handlemmfault+0xf74/0x3fe0 [65731.260486] [T32454] domemabort+0x54c/0x1b34 [65731.260509] [T32454] el0da+0x44/0x94 [65731.260531] [T32454] el0t64synchandler+0x98/0xb4 [65731.260553] [T32454] el0t64sync+0x198/0x19c
Other sources
In the Linux kernel, the following vulnerability has been resolved:
mm/vmalloc: fix vmalloc which may return null if called with GFPNOFAIL
The Linux kernel CVE team has assigned CVE-2024-39474 to this issue.
Upstream advisory: https://lore.kernel.org/linux-cve-announce/2024070518-CVE-2024-39474-97cf@gregkh/T
— Red Hat
Affected Software
Remediation
Event History
Frequently Asked Questions
What is the severity of CVE-2024-39474?
CVE-2024-39474 has been classified with a moderate severity level due to its impact on memory allocation in the Linux kernel.
How do I fix CVE-2024-39474?
To fix CVE-2024-39474, update to a patched version of the Linux kernel, such as 6.1.95 or later.
What systems are affected by CVE-2024-39474?
CVE-2024-39474 affects various versions of the Linux kernel, particularly those prior to 6.1.95.
Who is responsible for addressing CVE-2024-39474?
Linux kernel maintainers are responsible for addressing CVE-2024-39474 through patches and updates.
What are the potential risks of CVE-2024-39474?
The potential risks of CVE-2024-39474 include system instability and security vulnerabilities related to improper memory management.