linux-stable.git, branch linux-6.12.y

Linux 6.12.107

2026-08-28T06:22:18+00:00

Signed-off-by: Greg Kroah-Hartman

inet: frags: strip GSO state from fragments before reassembly

2026-08-28T06:22:18+00:00

commit d5dc1e69fd7258ea605c9952e5d5947539159ae3 upstream.

A virtio_net_hdr (tun/tap, or AF_PACKET with PACKET_VNET_HDR) can mark
an IPv4 or IPv6 fragment as GSO; nothing relates gso_type to frag_off.
inet_frag_reasm_prepare()/inet_frag_reasm_finish() keep the first
fragment's skb as the head of the reassembled datagram, including its
shinfo->gso_size/gso_type/gso_segs, and chain the remaining fragments
on frag_list with whatever linear/paged layout they arrived with.

After ip_defrag() (ip_local_deliver(), nf_defrag_ipv4, ...) the
reassembled skb therefore still claims to be GSO (SKB_GSO_DODGY), and
the next software segmentation point - udp_rcv_segment() on local
delivery, validate_xmit_skb(), or the ip_finish_output_gso() slow
path - hands it to skb_segment(). skb_segment()'s frag_list walk
assumes GRO-shaped input and hits one of its BUG_ON()s. Two writes to
a tap by an unprivileged user in its own userns are enough:

  kernel BUG at net/core/skbuff.c:4899!
  Oops: invalid opcode: 0000 [#1] SMP KASAN NOPTI
  CPU: 0 UID: 1000 PID: 82 Comm: poc Not tainted 7.2.0-pentest+ #2
  RIP: 0010:skb_segment+0x20ca/0x48b0
  Call Trace:
   
   __udp_gso_segment+0x29a/0x27d0
   udp4_ufo_fragment+0x458/0x6c0
   inet_gso_segment+0x429/0x1340
   skb_mac_gso_segment+0x233/0x4f0
   __skb_gso_segment+0x308/0x660
   udp_queue_rcv_skb+0x440/0xad0
   udp_unicast_rcv_skb+0xc7/0x2c0
   udp_rcv+0x16ce/0x2260
   ip_protocol_deliver_rcu+0x197/0x2d0
   ip_local_deliver+0x430/0x690
   ip_rcv+0x16f/0x1f0
   __netif_receive_skb_one_core+0x15e/0x1c0
   __netif_receive_skb+0x1e/0x110
   netif_receive_skb+0xf6/0x5c0
   tun_rx_batched.isra.0+0x3ab/0x790
   tun_get_user+0x17c3/0x3550
   tun_chr_write_iter+0xba/0x1b0
   vfs_write+0x646/0x1130
   
  Kernel panic - not syncing: Fatal exception in interrupt

This runs with BH disabled, so it is a panic rather than an oops. The
same is reachable with CAP_NET_RAW in a netns where a defrag point
precedes a GSO point, and from a guest whose VMM forwards
virtio_net_hdr to a tap. The SKB_GSO_DODGY frag_list checks added by
commit 3dcbdb134f32 ("net: gso: Fix skb_segment splat when splitting
gso_size mangled skb having linear-headed frag_list") and by
commit 9e4b7a99a03a ("net: gso: fix panic on frag_list with mixed head
alloc types") do not cover it: page-backed heads skip them, and kmalloc
heads skip them when gso_size == skb_headlen(head), which the sender
controls.

An skb entering a frag queue is an IP fragment by definition and
cannot legitimately carry GSO state: GRO does not merge fragments and
the stack segments before it fragments, so only untrusted sources are
affected. This has been reachable since
commit f43798c27684 ("tun: Allow GSO using virtio_net_hdr"), the first
path that let userspace attach GSO metadata to an IP fragment. Reset
the GSO fields of every fragment as it is queued, in
inet_frag_queue_insert(), which IPv4, IPv6, nf_conntrack_reasm and
6lowpan reassembly share; then neither the head nor the frag_list
members of the reassembled skb carry them (the members matter too:
the ip_do_fragment()/ip6_fragment() fast paths send them out as they
are). The head may remain CHECKSUM_PARTIAL; that is already accepted
on receive and resolved by skb_checksum_help() in
ip_do_fragment()/ip6_fragment() on forward.

Tested on top of net.git (dc4b95b8fee9), x86_64: the tap reproducer
above, two further IPv4 frag_list geometries that reach
BUG_ON(i >= nfrags) and BUG_ON(!list_skb->head_frag), and an IPv6
fragment-header variant (udp6_ufo_fragment()) each panic the unpatched
kernel; with this patch all four datagrams are delivered intact and
nothing is logged.

Fixes: f43798c27684 ("tun: Allow GSO using virtio_net_hdr")
Cc: stable@kernel.org
Suggested-by: Eric Dumazet 
Signed-off-by: Xinyang Ge 
Signed-off-by: Paolo Abeni 
Reviewed-by: Eric Dumazet 
Link: https://patch.msgid.link/937926e509f2acd8e0e66520dc2b30fd6b4d1687.1787839506.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski 
Signed-off-by: Greg Kroah-Hartman

Linux 6.12.106

2026-08-27T12:32:09+00:00

Link: https://lore.kernel.org/r/20260825132541.580275153@linuxfoundation.org
Tested-by: Florian Fainelli 
Tested-by: Shuah Khan 
Tested-by: Ron Economos 
Tested-by: Pavel Machek (CIP) 
Tested-by: Dominique Martinet 
Tested-by: Salvatore Bonaccorso 
Tested-by: Brett A C Sheffield 
Tested-by: Miguel Ojeda 
Tested-by: Peter Schneider 
Signed-off-by: Greg Kroah-Hartman

net: gro: properly validate BIG TCP aggregation criteria

2026-08-27T12:32:08+00:00

When GRO attempts to aggregate packets beyond GRO_LEGACY_MAX_SIZE (64KB),
BIG TCP should only be permitted for plain IPv4 TCP and plain IPv6 TCP
(with sufficient MAC header room to insert the temporary HBH jumbo header).

However, commit b1a78b9b9886 ("net: add support for ipv4 big tcp")
loosened the check in skb_gro_receive(), leading to several issues:

1. skb_gro_receive() checked skb_headroom(p) instead of the actual space
   before the MAC header (p->mac_header). Because skb_headroom(p) includes
   mac_len, crafted frames (e.g. injected via AF_PACKET) can pass the check
   with p->mac_header < 8 bytes. When ipv6_gro_complete() inserts the
   temporary HBH jumbo header, the memmove() starts before skb->head,
   causing an out-of-bounds write and wrapping skb->mac_header.
2. It allowed non-IP protocols such as software VLAN (ETH_P_8021Q /
   ETH_P_8021AD) to aggregate beyond 64KB because
   p->protocol != ETH_P_IPV6 was true.
3. It checked p->encapsulation instead of NAPI_GRO_CB(skb)->encap_mark,
   allowing encapsulated flows (e.g. SIT / IPv6-in-IPv4) to aggregate
   beyond 64KB.

Fix skb_gro_receive() to strictly enforce:
- NAPI_GRO_CB(skb)->proto == IPPROTO_TCP
- Not encapsulated (!NAPI_GRO_CB(skb)->encap_mark && !p->encapsulation)
- Protocol must be either ETH_P_IP or ETH_P_IPV6
- If ETH_P_IPV6, p->mac_header must be at least
  sizeof(struct hop_jumbo_hdr)

Returning -E2BIG from skb_gro_receive() ensures that packets which cannot
become BIG TCP are cleanly flushed at <= 64KB and delivered intact without
dropping.

This issue does not exist in mainline (7.0+) because the subsystem was
rewritten in commit 81be30c1f5f2 ("net/ipv6: Drop HBH for BIG TCP on RX
side"), making this fix relevant only for older stable branches like
6.18.y.

Fixes: 0fe79f28bfaf ("net: allow gro_max_size to exceed 65536")
Fixes: b1a78b9b9886 ("net: add support for ipv4 big tcp")
Reported-by: Sam Dlinn 
Cc: stable@vger.kernel.org
Signed-off-by: Eric Dumazet 
Signed-off-by: Greg Kroah-Hartman

Bluetooth: hci_aml: validate firmware segment lengths

2026-08-27T12:32:08+00:00

commit 2bf6b9baca9372ea51b6d0f2820dc9bf29a83ef4 upstream.

aml_download_firmware() reads two lengths from the firmware header and
uses them to build pointers before checking that the header and segment
data are present. A truncated or inconsistent firmware image can make
the driver read past firmware->data while constructing TCI commands.

Reject images shorter than the header and ensure that the ICCM and DCCM
ranges fit within the loaded firmware before downloading either segment.

Fixes: 37bac77e4649 ("Bluetooth: hci_uart: Add support for Amlogic HCI UART")
Cc: stable@vger.kernel.org
Signed-off-by: Laxman Acharya Padhya 
Signed-off-by: Luiz Augusto von Dentz 
Signed-off-by: Greg Kroah-Hartman

Bluetooth: ISO: do not force BT_LISTEN after a failed BIG sync

2026-08-27T12:32:08+00:00

commit 9838a80096ba472d5e03057136a112631aabae6e upstream.

iso_sock_recvmsg() handles the deferred setup of a broadcast sink by
dropping the socket lock, calling iso_conn_big_sync() and taking the
lock again:

	release_sock(sk);
	iso_conn_big_sync(sk);
	lock_sock(sk);

	sk->sk_state = BT_LISTEN;

The state is written unconditionally, but iso_conn_big_sync() returns
void and has paths that do nothing at all: hci_get_route() may fail, and
after re-acquiring the socket lock the connection may already be gone,
in which case it bails out without ever issuing an LE BIG Create Sync.

While the lock is dropped the connection can be torn down, for example
when the controller reports HCI_EV_LE_PA_SYNC_LOST:

	hci_le_pa_sync_lost_evt()
	  hci_disconn_cfm() -> iso_disconn_cfm() -> iso_conn_del()
	    iso_chan_del()
	      iso_pi(sk)->conn = NULL
	      sk->sk_state = BT_CLOSED
	      sock_set_flag(sk, SOCK_ZAPPED)

iso_conn_big_sync() then finds conn == NULL and returns, but the caller
still overwrites the BT_CLOSED that iso_chan_del() has just set. The
socket ends up marked BT_LISTEN with no connection, so recvmsg() reports
success for a setup that never happened and a later accept() waits for
BIS connections that can never arrive instead of failing.

A concurrent shutdown() reaches the same write by another route:
__iso_sock_close() takes the BT_CONNECT2 PA sync path to
iso_sock_disconn(), which sets BT_DISCONN but leaves conn and
conn->hcon in place, so iso_conn_big_sync() succeeds and BT_LISTEN is
written over BT_DISCONN. Both the BT_CONNECT2 and the BT_CONNECTED case
write the state the same way.

Let iso_conn_big_sync() report whether the BIG sync was started, and
only move the socket to BT_LISTEN when it was and when the state has not
changed while the lock was dropped, mirroring what the BT_CONNECT case
of the same switch already does with iso_connect_cis(). Both conditions
are needed, the error alone does not cover the shutdown() race.

This corrupts the socket state machine only, it is not a memory safety
issue. KASAN and lockdep stayed quiet in all of the runs below.

Reproduced with an emulated controller over /dev/vhci on a KASAN +
PROVE_LOCKING kernel. A PA sync broadcast sink socket is driven to
BT_CONNECT2 and recvmsg() on it is raced against teardown, with a debug
delay inside the lock-dropped section to widen the window:

 - HCI_EV_LE_PA_SYNC_LOST injected: 64 of 64 rounds left the socket in
   BT_LISTEN with the connection gone, recvmsg() returned 0 and accept()
   on that fd returned EAGAIN, which iso_sock_accept() can only do while
   the socket is BT_LISTEN. With this patch, 0 of 64, recvmsg() returns
   an error and accept() returns EBADFD.

 - shutdown() instead of a controller event: 24 of 32 rounds wedged in
   BT_LISTEN, 0 of 32 with this patch. With only the error check in
   place and a short window, one round still wedged while recvmsg()
   returned 0, which is the case the state re-check covers.

An unraced control round behaves the same before and after: recvmsg()
returns 0, the socket reaches BT_LISTEN and an LE BIG Create Sync is
issued.

Fixes: 7a17308c1788 ("Bluetooth: iso: Fix circular lock in iso_conn_big_sync")
Cc: stable@vger.kernel.org
Signed-off-by: Ali Ahmet Memis 
Signed-off-by: Luiz Augusto von Dentz 
Signed-off-by: Greg Kroah-Hartman

Bluetooth: hci_event: validate LE Set CIG Parameters response

2026-08-27T12:32:08+00:00

commit 0acd4eeb4b225b9bebbf9ef96cc10cdd79b94899 upstream.

The Command Complete dispatch validates only the fixed part of the LE Set
CIG Parameters response. After that part is pulled from the skb,
hci_cc_le_set_cig_params() trusts num_handles and reads each entry in the
trailing handle array.

Matching num_handles against the command's num_cis does not guarantee
that the response contains the advertised handles. A truncated response
from a malfunctioning controller can therefore make the handler read
beyond the skb data.

Validate that the remaining skb data contains all advertised handles.
Include this in the existing response validation so malformed responses
also follow the established CIG failure handling.

Fixes: 26afbd826ee3 ("Bluetooth: Add initial implementation of CIS connections")
Cc: stable@vger.kernel.org
Signed-off-by: Laxman Acharya Padhya 
Signed-off-by: Luiz Augusto von Dentz 
Signed-off-by: Greg Kroah-Hartman

Bluetooth: hci_event: fix LE list UAF on reset

2026-08-27T12:32:08+00:00

commit 33af47e847fe4a28b109673affb5874015d54f5a upstream.

hci_cc_reset() clears the LE accept and resolving lists without taking
hdev->lock. Other command-complete handlers serialize updates to these
lists with that lock, and the debugfs readers hold it while walking them.

This permits the reset completion and a debugfs read to interleave as
follows:

  hci_rx_work                 debugfs reader
  -----------                 --------------
                              lock hdev->lock
                              fetch current entry
  list_del(entry)
  kfree(entry)
                              read entry fields

The reader then dereferences a freed list entry and may follow its stale
next pointer.

KASAN reported:

  BUG: KASAN: slab-use-after-free in white_list_show+0x15f/0x180
  Read of size 1 at addr ffff8881015dab16 by task poc/95

  Call Trace:
   white_list_show+0x15f/0x180
   seq_read_iter+0x3ff/0x1190
   seq_read+0x267/0x3d0
   vfs_read+0x177/0xa20
   ksys_read+0xf7/0x1c0

  Allocated by task 91:
   hci_bdaddr_list_add+0x1a6/0x3a0
   hci_cc_le_add_to_accept_list+0xab/0x140
   hci_cmd_complete_evt+0x26c/0x9a0
   hci_event_packet+0x454/0xb20
   hci_rx_work+0x293/0x730

  Freed by task 90:
   kfree+0x131/0x3c0
   hci_bdaddr_list_clear+0xd8/0x160
   hci_cc_reset+0x28a/0x370
   hci_cmd_complete_evt+0x26c/0x9a0
   hci_event_packet+0x454/0xb20
   hci_rx_work+0x293/0x730

Take hdev->lock around both list clears. This matches the existing
mutation and traversal locking convention.

Fixes: a4d5504d5c39 ("Bluetooth: Clear LE white list when resetting controller")
Fixes: cfdb0c2d095a ("Bluetooth: Store Resolv list size")
Cc: stable@vger.kernel.org
Signed-off-by: Chengfeng Ye 
Signed-off-by: Luiz Augusto von Dentz 
Signed-off-by: Greg Kroah-Hartman

HID: hyperv: validate initial device info bounds

2026-08-27T12:32:08+00:00

commit 934b7778aa7b7c8f6bb073d2a73ba3674885bae0 upstream.

The Hyper-V synthetic HID host supplies SYNTH_HID_INITIAL_DEVICE_INFO
messages that contain a HID descriptor followed by the report descriptor
bytes. mousevsc_on_receive_device_info() trusts bLength and
wDescriptorLength without checking that the received packet contains both
byte ranges.

A malformed host or backend message can therefore make the guest read
past the received VMBus packet while copying the report descriptor. Pass
the received initial-device-info size into the parser and reject
descriptor lengths that exceed the packet.

Impact: A malicious Hyper-V host or backend can crash a guest by sending
a short initial device-info message with an oversized HID report
descriptor length.

Fixes: b95f5bcb811e ("HID: Move the hid-hyperv driver out of staging")
Cc: stable@vger.kernel.org
Assisted-by: Codex:gpt-5-5-xhigh
Signed-off-by: Michael Bommarito 
Signed-off-by: Jiri Kosina 
Signed-off-by: Greg Kroah-Hartman

HID: sensor: custom: Fix use-after-free in enable_sensor

2026-08-27T12:32:08+00:00

commit ad8fb82b04422f49530d2aa2753cc81d1c60102c upstream.

enable_sensor_store() can call set_power_report_state(), which
dereferences sensor_inst->power_state and sensor_inst->report_state.
These pointers refer to entries in sensor_inst->fields.

Create the field attributes before exposing the enable_sensor sysfs
attribute, so enable_sensor cannot be accessed before the state it
depends on has been initialized.

On remove, delete enable_sensor before freeing the field attributes,
so a concurrent sysfs write cannot dereference freed memory through
power_state or report_state.

Reported-by: Sashiko AI Review 
Link: https://sashiko.dev/#/patchset/20260623021950.1736413-1-haoxiang_li2024@163.com?part=1
Fixes: 4a7de0519df5 ("HID: sensor: Custom and Generic sensor support")
Cc: stable@vger.kernel.org
Signed-off-by: Haoxiang Li 
Acked-by: Srinivas Pandruvada 
Signed-off-by: Jiri Kosina 
Signed-off-by: Greg Kroah-Hartman