)]}'
{
  "commit": "21646bdd971e420c24d223810be1eb29308ea687",
  "tree": "40ba1fddbaec1c38cf30cc26d644c2dadb799f1b",
  "parents": [
    "5da04622a099fa0dfdfba3abf2378b826e2a7841"
  ],
  "author": {
    "name": "Octavian Purdila",
    "email": "tavip@google.com",
    "time": "Tue Dec 03 19:05:19 2024 -0800"
  },
  "committer": {
    "name": "Arnav Kansal",
    "email": "rnv@google.com",
    "time": "Tue Mar 18 14:16:00 2025 -0700"
  },
  "message": "net_sched: sch_sfq: don\u0027t allow 1 packet limit\n\n[ Upstream commit 10685681bafce6febb39770f3387621bf5d67d0b ]\n\nThe current implementation does not work correctly with a limit of\n1. iproute2 actually checks for this and this patch adds the check in\nkernel as well.\n\nThis fixes the following syzkaller reported crash:\n\nUBSAN: array-index-out-of-bounds in net/sched/sch_sfq.c:210:6\nindex 65535 is out of range for type \u0027struct sfq_head[128]\u0027\nCPU: 0 PID: 2569 Comm: syz-executor101 Not tainted 5.10.0-smp-DEV #1\nHardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024\nCall Trace:\n  __dump_stack lib/dump_stack.c:79 [inline]\n  dump_stack+0x125/0x19f lib/dump_stack.c:120\n  ubsan_epilogue lib/ubsan.c:148 [inline]\n  __ubsan_handle_out_of_bounds+0xed/0x120 lib/ubsan.c:347\n  sfq_link net/sched/sch_sfq.c:210 [inline]\n  sfq_dec+0x528/0x600 net/sched/sch_sfq.c:238\n  sfq_dequeue+0x39b/0x9d0 net/sched/sch_sfq.c:500\n  sfq_reset+0x13/0x50 net/sched/sch_sfq.c:525\n  qdisc_reset+0xfe/0x510 net/sched/sch_generic.c:1026\n  tbf_reset+0x3d/0x100 net/sched/sch_tbf.c:319\n  qdisc_reset+0xfe/0x510 net/sched/sch_generic.c:1026\n  dev_reset_queue+0x8c/0x140 net/sched/sch_generic.c:1296\n  netdev_for_each_tx_queue include/linux/netdevice.h:2350 [inline]\n  dev_deactivate_many+0x6dc/0xc20 net/sched/sch_generic.c:1362\n  __dev_close_many+0x214/0x350 net/core/dev.c:1468\n  dev_close_many+0x207/0x510 net/core/dev.c:1506\n  unregister_netdevice_many+0x40f/0x16b0 net/core/dev.c:10738\n  unregister_netdevice_queue+0x2be/0x310 net/core/dev.c:10695\n  unregister_netdevice include/linux/netdevice.h:2893 [inline]\n  __tun_detach+0x6b6/0x1600 drivers/net/tun.c:689\n  tun_detach drivers/net/tun.c:705 [inline]\n  tun_chr_close+0x104/0x1b0 drivers/net/tun.c:3640\n  __fput+0x203/0x840 fs/file_table.c:280\n  task_work_run+0x129/0x1b0 kernel/task_work.c:185\n  exit_task_work include/linux/task_work.h:33 [inline]\n  do_exit+0x5ce/0x2200 kernel/exit.c:931\n  do_group_exit+0x144/0x310 kernel/exit.c:1046\n  __do_sys_exit_group kernel/exit.c:1057 [inline]\n  __se_sys_exit_group kernel/exit.c:1055 [inline]\n  __x64_sys_exit_group+0x3b/0x40 kernel/exit.c:1055\n do_syscall_64+0x6c/0xd0\n entry_SYSCALL_64_after_hwframe+0x61/0xcb\nRIP: 0033:0x7fe5e7b52479\nCode: Unable to access opcode bytes at RIP 0x7fe5e7b5244f.\nRSP: 002b:00007ffd3c800398 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7\nRAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fe5e7b52479\nRDX: 000000000000003c RSI: 00000000000000e7 RDI: 0000000000000000\nRBP: 00007fe5e7bcd2d0 R08: ffffffffffffffb8 R09: 0000000000000014\nR10: 0000000000000000 R11: 0000000000000246 R12: 00007fe5e7bcd2d0\nR13: 0000000000000000 R14: 00007fe5e7bcdd20 R15: 00007fe5e7b24270\n\nThe crash can be also be reproduced with the following (with a tc\nrecompiled to allow for sfq limits of 1):\n\ntc qdisc add dev dummy0 handle 1: root tbf rate 1Kbit burst 100b lat 1s\n../iproute2-6.9.0/tc/tc qdisc add dev dummy0 handle 2: parent 1:10 sfq limit 1\nifconfig dummy0 up\nping -I dummy0 -f -c2 -W0.1 8.8.8.8\nsleep 1\n\nScenario that triggers the crash:\n\n* the first packet is sent and queued in TBF and SFQ; qdisc qlen is 1\n\n* TBF dequeues: it peeks from SFQ which moves the packet to the\n  gso_skb list and keeps qdisc qlen set to 1. TBF is out of tokens so\n  it schedules itself for later.\n\n* the second packet is sent and TBF tries to queues it to SFQ. qdisc\n  qlen is now 2 and because the SFQ limit is 1 the packet is dropped\n  by SFQ. At this point qlen is 1, and all of the SFQ slots are empty,\n  however q-\u003etail is not NULL.\n\nAt this point, assuming no more packets are queued, when sch_dequeue\nruns again it will decrement the qlen for the current empty slot\ncausing an underflow and the subsequent out of bounds access.\n\nBUG\u003db/401712081\nTEST\u003dpresubmit\nRELEASE_NOTE\u003dFixed CVE-2024-57996 in the Linux kernel.\n\ncos-patch: security-moderate\nReported-by: syzbot \u003csyzkaller@googlegroups.com\u003e\nFixes: 1da177e4c3f4 (\"Linux-2.6.12-rc2\")\nChange-Id: I97a5dec499810f5c2c49df010d9abda4356e4c3d\nSigned-off-by: Octavian Purdila \u003ctavip@google.com\u003e\nReviewed-by: Eric Dumazet \u003cedumazet@google.com\u003e\nLink: https://patch.msgid.link/20241204030520.2084663-2-tavip@google.com\nSigned-off-by: Jakub Kicinski \u003ckuba@kernel.org\u003e\nSigned-off-by: Sasha Levin \u003csashal@kernel.org\u003e\nSigned-off-by: Kernel CVE Triage Automation \u003ccloud-image-kernel-cve-triage-automation@prod.google.com\u003e\nReviewed-on: https://cos-review.googlesource.com/c/third_party/kernel/+/95483\nReviewed-by: Arnav Kansal \u003crnv@google.com\u003e\nReviewed-by: Kevin Berry \u003ckpberry@google.com\u003e\nTested-by: Cusky Presubmit Bot \u003cpresubmit@cos-infra-prod.iam.gserviceaccount.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "7d4feae2fae369e5d9d9a7fa02a3390785ace366",
      "old_mode": 33188,
      "old_path": "net/sched/sch_sfq.c",
      "new_id": "60754f366ab7bc27fa44fdc19a581f80cb178328",
      "new_mode": 33188,
      "new_path": "net/sched/sch_sfq.c"
    }
  ]
}