In the Linux kernel, the following vulnerability has been resolved:

ip_gre: test csum_start instead of transport header

GRE with TUNNEL_CSUM will apply local checksum offload on CHECKSUM_PARTIAL packets.

ipgre_xmit must validate csum_start after an optional skb_pull, else lco_csum may trigger an overflow. The original check was

if (csum && skb_checksum_start(skb) < skb->data) return -EINVAL;

This had false positives when skb_checksum_start is undefined: when ip_summed is not CHECKSUM_PARTIAL. A discussed refinement was straightforward

if (csum && skb->ip_summed == CHECKSUM_PARTIAL && skb_checksum_start(skb) < skb->data) return -EINVAL;

But was eventually revised more thoroughly: - restrict the check to the only branch where needed, in an uncommon GRE path that uses header_ops and calls skb_pull. - test skb_transport_header, which is set along with csum_start in skb_partial_csum_set in the normal header_ops datapath.

Turns out skbs can arrive in this branch without the transport header set, e.g., through BPF redirection.

Revise the check back to check csum_start directly, and only if CHECKSUM_PARTIAL. Do leave the check in the updated location. Check field regardless of whether TUNNEL_CSUM is configured.

In the Linux kernel, the following vulnerability has been resolved:

ip_gre: test csum_start instead of transport header

GRE with TUNNEL_CSUM will apply local checksum offload on CHECKSUM_PARTIAL packets.

ipgre_xmit must validate csum_start after an optional skb_pull, else lco_csum may trigger an overflow. The original check was

if (csum && skb_checksum_start(skb) < skb->data) return -EINVAL;

This had false positives when skb_checksum_start is undefined: when ip_summed is not CHECKSUM_PARTIAL. A discussed refinement was straightforward

if (csum && skb->ip_summed == CHECKSUM_PARTIAL && skb_checksum_start(skb) < skb->data) return -EINVAL;

But was eventually revised more thoroughly: - restrict the check to the only branch where needed, in an uncommon GRE path that uses header_ops and calls skb_pull. - test skb_transport_header, which is set along with csum_start in skb_partial_csum_set in the normal header_ops datapath.

Turns out skbs can arrive in this branch without the transport header set, e.g., through BPF redirection.

Revise the check back to check csum_start directly, and only if CHECKSUM_PARTIAL. Do leave the check in the updated location. Check field regardless of whether TUNNEL_CSUM is configured.

kernel: off-path attacker may inject data or terminate victim's TCP session kernel: use-after-free vulnerability in function sco_sock_sendmsg() kernel: fix 'struct pid' leaks in 'dbgfs_target_ids_write()' kernel: veth: ensure skb entering GRO are not cloned. kernel: inet: fully convert sk->sk_rx_dst to RCU rules kernel: irqchip/gic-v3-its: Fix potential VPE leak on error kernel: nvme-rdma: destroy cm id before destroy qp to avoid use after free kernel: ptp: Fix possible memory leak in ptp_clock_register() kernel: ethtool: do not perform operations on net devices being unregistered kernel: ethtool: ioctl: fix potential NULL deref in ethtool_set_coalesce() kernel: net: nexthop: fix null pointer dereference when IPv6 is not enabled kernel: scsi: scsi_debug: Fix type in min_t to avoid stack OOB kernel: KVM: x86/mmu: Zap _all_ roots when unmapping gfn range in TDP MMU kernel: Revert "Revert "block, bfq: honor already-setup queue merges"" kernel: udmabuf: validate ubuf->pagecount kernel: drm/virtio: Ensure that objs is not NULL in virtio_gpu_array_put_free() kernel: smb2_ioctl_query_info NULL pointer dereference kernel: NULL pointer dereference in udf_expand_file_adinicbdue() during writeback kernel: swiotlb information leak with DMA_FROM_DEVICE kernel: uninitialized registers on stack in nft_do_chain can cause kernel pointer leakage to UM kernel: race condition in snd_pcm_hw_free leading to use-after-free kernel: KVM: cmpxchg_gpte can write to pfns outside the userspace region ke…

Affected: Ubuntu:Pro:14.04:LTS: linux-azure, Ubuntu:Pro:16.04:LTS: linux-aws-hwe, Ubuntu:Pro:16.04:LTS: linux-azure, Ubuntu:Pro:16.04:LTS: linux-gcp, Ubuntu:Pro:16.04:LTS: linux-hwe, Ubuntu:Pro:16.04:LTS: linux-oracle, Ubuntu:16.04:LTS: linux-hwe-edge, Ubuntu:18.04:LTS: linux, Ubuntu:18.04:LTS: linux-aws, Ubuntu:18.04:LTS: linux-aws-5.4 and 129 more

In the Linux kernel, the following vulnerability has been resolved: ip_gre: test csum_start instead of transport header GRE with TUNNEL_CSUM will apply local checksum offload on CHECKSUM_PARTIAL packets. ipgre_xmit must validate csum_start after an optional skb_pull, else lco_csum may trigger an overflow. The original check was if (csum && skb_checksum_start(skb) < skb->data) return -EINVAL; This had false positives when skb_checksum_start is undefined: when ip_summed is not CHECKSUM_PARTIAL. A discussed refinement was straightforward if (csum && skb->ip_summed == CHECKSUM_PARTIAL && skb_checksum_start(skb) < skb->data) return -EINVAL; But was eventually revised more thoroughly: - restrict the check to the only branch where needed, in an uncommon GRE path that uses header_ops and calls skb_pull. - test skb_transport_header, which is set along with csum_start in skb_partial_csum_set in the normal header_ops datapath. Turns out skbs can arrive in this branch without the transport header set, e.g., through BPF redirection. Revise the check back to check csum_start directly, and only if CHECKSUM_PARTIAL. Do leave the check in the updated location. Check field regardless of whether TUNNEL_CSUM is configured.