Vulnerabilities

ONNX before 1.21.0 contains a symlink-following and time-of-check-to-time-of-use flaw in its save_external_data function, which builds the external-data output path from the model's external_data location field and opens it for writing without O_NOFOLLOW/O_EXCL after a non-atomic os.path.isfile() check. A local attacker with write access to the directory where a victim serializes a model's external data can pre-plant a symlink at the expected path, causing the victim's write to follow it. As a result, the attacker can append attacker-influenced content to any file the victim can write, such as ~/.ssh/authorized_keys, cron files, or application configs, potentially achieving code execution or persistence in the victim's user context. Any Python environment running a vulnerable onnx version and saving models with external data, particularly on shared or multi-user systems, is affected. No exploitation is known in the wild; one public advisory/PoC reference exists and EPSS is 0.1% (1st percentile).

Do: Upgrade the onnx pip package to 1.21.0 or later. As an interim mitigation, serialize model external data only into directories exclusively writable by the running user, and audit services, cron jobs, and pipelines on shared or multi-user hosts that call save_external_data for vulnerable usage.

PyTorch torchvision through 0.28.0, fixed in commit 4e05dc2, contains an out-of-bounds heap read vulnerability in the GIF decoder's read_from_tensor callback that passes unclamped length to memcpy. Attackers can supply malicious or truncated GIF files to cause denial of service via segmentation fault or disclose adjacent heap memory contents.

In the Linux kernel, the following vulnerability has been resolved: xfs: resample the data fork mapping after cycling ILOCK xfs_reflink_fill_{cow_hole,delalloc} are both presented with an inode, a data fork mapping, and a cow fork mapping. Unfortunately, these two helpers cycle the ILOCK to grab a transaction, which means that the mappings are stale as soon as we reacquire the ILOCK. Currently we refresh the cow fork mapping by re-calling xfs_find_trim_cow_extent, but we don't refresh the data fork mapping beforehand, which means that the xfs_bmap_trim_cow in that function queries the refcount btree about the wrong physical blocks and returns an inaccurate value in *shared. If *shared is now false, the directio write proceeds with a stale data fork mapping. Fix this by querying the data fork mapping if the sequence counter changes across the ILOCK cycle.

oras-go is a Go library for managing OCI artifacts. Prior to 2.6.1, registry/remote/repository.go in blobStore.completePushAfterInitialPost follows a registry-controlled Location header during monolithic blob upload and reuses the Authorization header from the initial POST request for the subsequent PUT request, allowing a malicious registry to return a cross-host Location and receive the caller's credentials at an attacker-controlled endpoint. This issue is fixed in version 2.6.1.

cert-manager adds certificates and certificate issuers as resource types in Kubernetes clusters, and simplifies the process of obtaining, renewing and using those certificates. From 1.18.0 until 1.19.6 and 1.20.3, Challenge resources under acme.cert-manager.io can be created directly by namespace users without admission validation tying the Challenge to an Order, owner reference, or Issuer-selected solver, allowing attacker-controlled Challenge.spec.solver values referencing a ClusterIssuer to bypass DNS01 solver selectors such as dnsZones, dnsNames, and matchLabels and cause cert-manager to use ClusterIssuer DNS credentials for attacker-selected provider settings and DNS names, including disclosure of X-Api-User and X-Api-Key headers for acme-dns. This issue is fixed in versions 1.19.6 and 1.20.3.

Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.9.0 before 1.22.0, onnx.version_converter.convert_version() can dereference a null pointer in Upsample_6_7::adapt_upsample_6_7() in onnx/version_converter/adapters/upsample_6_7.h when processing an untrusted model with an Upsample node that has zero inputs, causing an unrecoverable denial of service. This issue is fixed in version 1.22.0.

In the Linux kernel, the following vulnerability has been resolved: sctp: purge outqueue on stale COOKIE-ECHO handling sctp_stream_update() is only invoked when the association is moved into COOKIE_WAIT during association setup/reconfiguration. In this path, the outbound stream scheduler state (stream->out_curr) is expected to be clean, since no user data should have been transmitted yet unless the state machine has already partially progressed. However, a corner case exists in sctp_sf_do_5_2_6_stale(): when a Stale Cookie ERROR is received, the association is rolled back from COOKIE_ECHOED to COOKIE_WAIT. In this scenario, user data may already have been queued and even bundled with the COOKIE-ECHO chunk. During the rollback, sctp_stream_update() frees the old stream table and installs a new one, but it does not invalidate stream->out_curr. As a result, out_curr may still point to a freed sctp_stream_out entry from the previous stream state. Later, SCTP scheduler dequeue paths (FCFS, RR, PRIO, etc.) rely on stream->out_curr->ext, which can lead to use-after-free once the old stream state has been released via sctp_stream_free(). This results in crashes such as (reported by Yuqi): BUG: KASAN: slab-use-after-free in sctp_sched_fcfs_dequeue+0x13a/0x140 Read of size 8 at addr ff1100004d4d3208 by task mini_poc/9312 CPU: 1 UID: 1001 PID: 9312 Comm: mini_poc Not tainted 7.1.0-rc1-00305-gbd3a4795d574 #5 PREEMPT(full) sctp_sched_fcfs_dequeue+0x13a/0x140 sctp_outq_flush+0x1603/0x33e0 sctp_do_sm+0x31c9/0x5d30 sctp_assoc_bh_rcv+0x392/0x6f0 sctp_inq_push+0x1db/0x270 sctp_rcv+0x138d/0x3c10 Fix this by fully purging the association outqueue when handling the Stale Cookie case. This ensures all pending transmit and retransmit state is dropped, and any scheduler cached pointers are invalidated, making it safe to rebuild stream state during COOKIE_WAIT restart. Updating only stream->out_curr would be insufficient, since queued and retransmittable data would still reference the old stream state and trigger later use-after-free in dequeue paths.

In the Linux kernel, the following vulnerability has been resolved: bpf: Free reuseport cBPF prog after RCU grace period. Eulgyu Kim reported the splat below with a repro. [0] The repro sets up a UDP reuseport group with a cBPF prog and replaces it with a new one while another thread is sending a UDP packet to the group. The reuseport prog is freed by sk_reuseport_prog_free(). bpf_prog_put() is called for "e"BPF prog to destruct through multiple stages while cBPF prog is freed immediately by bpf_release_orig_filter() and bpf_prog_free(). If a reuseport prog is detached from the setsockopt() path (reuseport_attach_prog() or reuseport_detach_prog()), sk_reuseport_prog_free() is called without waiting for RCU readers to complete, resulting in various bugs. Let's defer freeing the reuseport cBPF prog after one RCU grace period. Note "e"BPF prog is safe as is unless the fast path starts to touch fields destroyed in bpf_prog_put_deferred() and __bpf_prog_put_noref(). [0]: BUG: KASAN: vmalloc-out-of-bounds in reuseport_select_sock+0xedc/0x1220 net/core/sock_reuseport.c:596 Read of size 4 at addr ffffc9000051e004 by task slowme/10208 CPU: 6 UID: 1000 PID: 10208 Comm: slowme Not tainted 7.0.0-geb7ac95ff75e #32 PREEMPT(full) Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch_caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 Call Trace: dump_stack_lvl+0xe8/0x150 lib/dump_stack.c:120 print_address_description mm/kasan/report.c:378 [inline] print_report+0xca/0x240 mm/kasan/report.c:482 kasan_report+0x118/0x150 mm/kasan/report.c:595 reuseport_select_sock+0xedc/0x1220 net/core/sock_reuseport.c:596 udp4_lib_lookup2+0x3bc/0x950 net/ipv4/udp.c:495 __udp4_lib_lookup+0x768/0xe20 net/ipv4/udp.c:723 __udp4_lib_lookup_skb+0x297/0x390 net/ipv4/udp.c:752 __udp4_lib_rcv+0x1312/0x2620 net/ipv4/udp.c:2752 ip_protocol_deliver_rcu+0x282/0x440 net/ipv4/ip_input.c:207 ip_local_deliver_finish+0x3bb/0x6f0 net/ipv4/ip_input.c:241 NF_HOOK+0x30c/0x3a0 include/linux/netfilter.h:318 NF_HOOK+0x30c/0x3a0 include/linux/netfilter.h:318 __netif_receive_skb_one_core net/core/dev.c:6181 [inline] __netif_receive_skb net/core/dev.c:6294 [inline] process_backlog+0xaa4/0x1960 net/core/dev.c:6645 __napi_poll+0xae/0x340 net/core/dev.c:7709 napi_poll net/core/dev.c:7772 [inline] net_rx_action+0x5d7/0xf50 net/core/dev.c:7929 handle_softirqs+0x22b/0x870 kernel/softirq.c:622 do_softirq+0x76/0xd0 kernel/softirq.c:523 __local_bh_enable_ip+0xf8/0x130 kernel/softirq.c:450 local_bh_enable include/linux/bottom_half.h:33 [inline] rcu_read_unlock_bh include/linux/rcupdate.h:924 [inline] __dev_queue_xmit+0x1dd7/0x3710 net/core/dev.c:4890 neigh_output include/net/neighbour.h:556 [inline] ip_finish_output2+0xca9/0x1070 net/ipv4/ip_output.c:237 NF_HOOK_COND include/linux/netfilter.h:307 [inline] ip_output+0x29f/0x450 net/ipv4/ip_output.c:438 ip_send_skb+0x45/0xc0 net/ipv4/ip_output.c:1508 udp_send_skb+0xb04/0x1510 net/ipv4/udp.c:1195 udp_sendmsg+0x1a71/0x2350 net/ipv4/udp.c:1485 sock_sendmsg_nosec net/socket.c:727 [inline] __sock_sendmsg net/socket.c:742 [inline] __sys_sendto+0x554/0x680 net/socket.c:2206 __do_sys_sendto net/socket.c:2213 [inline] __se_sys_sendto net/socket.c:2209 [inline] __x64_sys_sendto+0xde/0x100 net/socket.c:2209 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x160/0xf80 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x415a2d Code: b3 66 2e 0f 1f 84 00 00 00 00 00 66 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007f6bc31e41e8 EFLAGS: 00000212 ORIG_RAX: 000000000000002c RAX: ffffffffffffffda RBX: 00007f6bc31e4cdc RCX: 0000000000415a2d RDX: 0000000000000001 RSI: 00007f6bc31e421f RDI: 0000000000000003 RBP: 00007f6bc31e4240 R08: 00007f6bc31e4220 R09: 0000000000000010 R10: 0000000000000000 R11: ---truncated---

CVE-2022-0995 is an out-of-bounds (OOB) memory write in the Linux kernel's watch_queue event notification subsystem (CWE-787) that can overwrite parts of kernel state. A local user can trigger it through the watch_queue interface, for example by supplying a crafted event filter definition, causing the kernel to write beyond allocated memory when event notifications are processed. Successful exploitation may allow the local user to gain privileged (kernel/root) access or crash the system, yielding high impact on confidentiality, integrity, and availability (CVSS 3.1: 7.8, local attack vector). Affected systems include mainstream Linux kernels (Fedora is explicitly listed) and NetApp HCI appliance firmware products that ship the affected kernel. The flaw is confirmed exploited in the wild: CISA added it to the Known Exploited Vulnerabilities catalog on 2026-08-26, two public proof-of-concepts are available, and EPSS puts 30-day exploitation probability at 9.5% (95th percentile), though ransomware use is unconfirmed.

Do: Apply updated kernels from your distribution or vendor (Fedora kernel updates; NetApp fixes are delivered via the HCI firmware updates in its security advisory) and prioritize internet-exposed, multi-user hosts per CISA BOD 26-04, including the required forensics triage checks. Because exploitation requires local access, restrict untrusted local code execution and shell access on unpatched systems, and hunt for signs of local privilege escalation on high-exposure hosts. After remediation, verify the patched kernel package is installed and that the running kernel has actually been booted into the fixed version.