DEBIAN-CVE-2026-43465

In the Linux kernel, the following vulnerability has been resolved:  net/mlx5e: RX, Fix XDP multi-buf frag counting for striding RQ  XDP multi-buf programs can modify the layout of the XDP buffer when the program calls bpfxdppulldata() or bpfxdpadjusttail(). The referenced commit in the fixes tag corrected the assumption in the mlx5 driver that the XDP buffer layout doesn't change during a program execution. However, this fix introduced another issue: the dropped fragments still need to be counted on the driver side to avoid page fragment reference counting issues.  The issue was discovered by the drivers/net/xdp.py selftest, more specifically the testxdpnativetxmb: - The mlx5 driver allocates a pagepool page and initializes it with   a frag counter of 64 (pprefcount=64) and the internal frag counter   to 0. - The test sends one packet with no payload. - On RX (mlx5eskbfromcqempwrqnonlinear()), mlx5 configures the XDP   buffer with the packet data starting in the first fragment which is the   page mentioned above. - The XDP program runs and calls bpfxdppulldata() which moves the   header into the linear part of the XDP buffer. As the packet doesn't   contain more data, the program drops the tail fragment since it no   longer contains any payload (pprefcount=63). - mlx5 device skips counting this fragment. Internal frag counter   remains 0. - mlx5 releases all 64 fragments of the page but page pprefcount is   63 => negative reference counting error.  Resulting splat during the test:    WARNING: CPU: 0 PID: 188225 at ./include/net/pagepool/helpers.h:297 mlx5epagereleasefragmented.isra.0+0xbd/0xe0 [mlx5core]   Modules linked in: [...]   CPU: 0 UID: 0 PID: 188225 Comm: ip Not tainted 6.18.0-rc7forupstreammindebug202512081144 #1 NONE   Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014   RIP: 0010:mlx5epagereleasefragmented.isra.0+0xbd/0xe0 [mlx5core]   [...]   Call Trace:    <TASK>    mlx5efreerxmpwqe+0x20a/0x250 [mlx5core]    mlx5edeallocrxmpwqe+0x37/0xb0 [mlx5core]    mlx5efreerxdescs+0x11a/0x170 [mlx5core]    mlx5ecloserq+0x78/0xa0 [mlx5core]    mlx5eclosequeues+0x46/0x2a0 [mlx5core]    mlx5eclosechannel+0x24/0x90 [mlx5core]    mlx5eclosechannels+0x5d/0xf0 [mlx5core]    mlx5esafeswitchparams+0x2ec/0x380 [mlx5core]    mlx5echangemtu+0x11d/0x490 [mlx5core]    mlx5echangenicmtu+0x19/0x30 [mlx5core]    netifsetmtuext+0xfc/0x240    dosetlink.isra.0+0x226/0x1100    rtnlnewlink+0x7a9/0xba0    rtnetlinkrcvmsg+0x220/0x3c0    netlinkrcvskb+0x4b/0xf0    netlinkunicast+0x255/0x380    netlinksendmsg+0x1f3/0x420    socksendmsg+0x38/0x60    syssendmsg+0x1e8/0x240    syssendmsg+0x7c/0xb0    [...]    syssendmsg+0x5f/0xb0    dosyscall64+0x55/0xc70  The problem applies for XDPPASS as well which is handled in a different code path in the driver.  This patch fixes the issue by doing page frag counting on all the original XDP buffer fragments for all relevant XDP actions (XDPTX , XDPREDIRECT and XDPPASS). This is basically reverting to the original counting before the commit in the fixes tag.  As fragpage is still pointing to the original tail, the nrfrags parameter to xdpupdateskbfragsinfo() needs to be calculated in a different way to reflect the new nr_frags.