linux/include
Richard Gobert af276a5ac8 net: gro: fix udp bad offset in socket lookup by adding {inner_}network_offset to napi_gro_cb
[ Upstream commit 5ef31ea5d0 ]

Commits a602456 ("udp: Add GRO functions to UDP socket") and 57c67ff ("udp:
additional GRO support") introduce incorrect usage of {ip,ipv6}_hdr in the
complete phase of gro. The functions always return skb->network_header,
which in the case of encapsulated packets at the gro complete phase, is
always set to the innermost L3 of the packet. That means that calling
{ip,ipv6}_hdr for skbs which completed the GRO receive phase (both in
gro_list and *_gro_complete) when parsing an encapsulated packet's _outer_
L3/L4 may return an unexpected value.

This incorrect usage leads to a bug in GRO's UDP socket lookup.
udp{4,6}_lib_lookup_skb functions use ip_hdr/ipv6_hdr respectively. These
*_hdr functions return network_header which will point to the innermost L3,
resulting in the wrong offset being used in __udp{4,6}_lib_lookup with
encapsulated packets.

This patch adds network_offset and inner_network_offset to napi_gro_cb, and
makes sure both are set correctly.

To fix the issue, network_offsets union is used inside napi_gro_cb, in
which both the outer and the inner network offsets are saved.

Reproduction example:

Endpoint configuration example (fou + local address bind)

    # ip fou add port 6666 ipproto 4
    # ip link add name tun1 type ipip remote 2.2.2.1 local 2.2.2.2 encap fou encap-dport 5555 encap-sport 6666 mode ipip
    # ip link set tun1 up
    # ip a add 1.1.1.2/24 dev tun1

Netperf TCP_STREAM result on net-next before patch is applied:

net-next main, GRO enabled:
    $ netperf -H 1.1.1.2 -t TCP_STREAM -l 5
    Recv   Send    Send
    Socket Socket  Message  Elapsed
    Size   Size    Size     Time     Throughput
    bytes  bytes   bytes    secs.    10^6bits/sec

    131072  16384  16384    5.28        2.37

net-next main, GRO disabled:
    $ netperf -H 1.1.1.2 -t TCP_STREAM -l 5
    Recv   Send    Send
    Socket Socket  Message  Elapsed
    Size   Size    Size     Time     Throughput
    bytes  bytes   bytes    secs.    10^6bits/sec

    131072  16384  16384    5.01     2745.06

patch applied, GRO enabled:
    $ netperf -H 1.1.1.2 -t TCP_STREAM -l 5
    Recv   Send    Send
    Socket Socket  Message  Elapsed
    Size   Size    Size     Time     Throughput
    bytes  bytes   bytes    secs.    10^6bits/sec

    131072  16384  16384    5.01     2877.38

Fixes: a6024562ff ("udp: Add GRO functions to UDP socket")
Signed-off-by: Richard Gobert <richardbgobert@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
History 2024-05-17 12:02:07 +02:00
..
acpi ACPI: x86: Move acpi_quirk_skip_serdev_enumeration() out of CONFIG_X86_ANDROID_TABLETS 2024-04-13 13:07:31 +02:00
asm-generic sched: Add missing memory barrier in switch_mm_cid 2024-04-27 17:11:41 +02:00
clocksource clocksource: hyper-v: Adjust hv_read_tsc_page_tsc() to avoid special casing U64_MAX 2023-06-05 21:11:07 +02:00
crypto crypto: af_alg - Disallow multiple in-flight AIO requests 2024-01-25 15:35:16 -08:00
drm mm, treewide: introduce NR_PAGE_ORDERS 2024-05-02 16:32:41 +02:00
dt-bindings clk: renesas: r8a779g0: Correct PFC/GPIO parent clocks 2024-03-26 18:19:47 -04:00
keys integrity: PowerVM support for loading third party code signing keys 2023-08-17 20:12:35 +00:00
kunit - An extensive rework of kexec and crash Kconfig from Eric DeVolder 2023-08-29 14:53:51 -07:00
kvm KVM: arm64: Fix host-programmed guest events in nVHE 2024-04-10 16:35:48 +02:00
linux regmap: Add regmap_read_bypassed() 2024-05-17 12:01:59 +02:00
math-emu powerpc/math-emu: Remove -w build flag and fix warnings 2022-09-05 17:28:25 +10:00
media media: mc: Add num_links flag to media_pad 2024-04-03 15:28:17 +02:00
memory memory: renesas-rpc-if: Remove Runtime PM wrappers 2023-01-23 11:48:00 +01:00
misc cxl: fix typo in comment 2022-11-24 23:12:19 +11:00
net net: gro: fix udp bad offset in socket lookup by adding {inner_}network_offset to napi_gro_cb 2024-05-17 12:02:07 +02:00
pcmcia ARM: pxa/sa1100: move I/O space to PCI_IOBASE 2022-05-07 22:56:17 +02:00
ras mm, hwpoison: enable memory error handling on 1GB hugepage 2022-08-08 18:06:44 -07:00
rdma RDMA/core: Fix umem iterator when PAGE_SIZE is greater then HCA pgsz 2023-12-13 18:45:16 +01:00
rv rv: Set variable 'da_mon_##name' to static 2023-09-01 21:00:00 -04:00
scsi scsi: sd: Fix TCG OPAL unlock on system resume 2024-04-03 15:28:59 +02:00
soc soc: qcom: socinfo: rename PM2250 to PM4125 2024-03-26 18:19:23 -04:00
sound ALSA: emu10k1: move the whole GPIO event handling to the workqueue 2024-05-17 12:02:03 +02:00
target scsi: target: iscsi: Remove the unused netif_timeout attribute 2023-07-11 12:33:32 -04:00
trace mm: turn folio_test_hugetlb into a PageType 2024-05-02 16:32:47 +02:00
uapi PCI/DPC: Use FIELD_GET() 2024-04-27 17:11:36 +02:00
ufs Merge branch 'fixes' into misc 2023-09-02 08:25:19 +01:00
vdso vdso/bits.h: Add BIT_ULL() for the sake of consistency 2023-01-31 14:42:10 +01:00
video fbdev: stifb: Make the STI next font pointer a 32-bit signed offset 2023-11-28 17:19:58 +00:00
xen xen/events: reduce externally visible helper functions 2024-03-01 13:34:57 +01:00