CVE-2024-58070

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> bpf: bpf_local_storage: Always use bpf_mem_alloc in PREEMPT_RT<br /> <br /> In PREEMPT_RT, kmalloc(GFP_ATOMIC) is still not safe in non preemptible<br /> context. bpf_mem_alloc must be used in PREEMPT_RT. This patch is<br /> to enforce bpf_mem_alloc in the bpf_local_storage when CONFIG_PREEMPT_RT<br /> is enabled.<br /> <br /> [ 35.118559] BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:48<br /> [ 35.118566] in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 1832, name: test_progs<br /> [ 35.118569] preempt_count: 1, expected: 0<br /> [ 35.118571] RCU nest depth: 1, expected: 1<br /> [ 35.118577] INFO: lockdep is turned off.<br /> ...<br /> [ 35.118647] __might_resched+0x433/0x5b0<br /> [ 35.118677] rt_spin_lock+0xc3/0x290<br /> [ 35.118700] ___slab_alloc+0x72/0xc40<br /> [ 35.118723] __kmalloc_noprof+0x13f/0x4e0<br /> [ 35.118732] bpf_map_kzalloc+0xe5/0x220<br /> [ 35.118740] bpf_selem_alloc+0x1d2/0x7b0<br /> [ 35.118755] bpf_local_storage_update+0x2fa/0x8b0<br /> [ 35.118784] bpf_sk_storage_get_tracing+0x15a/0x1d0<br /> [ 35.118791] bpf_prog_9a118d86fca78ebb_trace_inet_sock_set_state+0x44/0x66<br /> [ 35.118795] bpf_trace_run3+0x222/0x400<br /> [ 35.118820] __bpf_trace_inet_sock_set_state+0x11/0x20<br /> [ 35.118824] trace_inet_sock_set_state+0x112/0x130<br /> [ 35.118830] inet_sk_state_store+0x41/0x90<br /> [ 35.118836] tcp_set_state+0x3b3/0x640<br /> <br /> There is no need to adjust the gfp_flags passing to the<br /> bpf_mem_cache_alloc_flags() which only honors the GFP_KERNEL.<br /> The verifier has ensured GFP_KERNEL is passed only in sleepable context.<br /> <br /> It has been an old issue since the first introduction of the<br /> bpf_local_storage ~5 years ago, so this patch targets the bpf-next.<br /> <br /> bpf_mem_alloc is needed to solve it, so the Fixes tag is set<br /> to the commit when bpf_mem_alloc was first used in the bpf_local_storage.