mirror of
https://github.com/fkie-cad/nvd-json-data-feeds.git
synced 2025-05-31 02:31:22 +00:00
21 lines
3.5 KiB
JSON
21 lines
3.5 KiB
JSON
|
{
|
||
|
|
"id": "CVE-2024-41010",
|
||
|
|
"sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
|
||
|
|
"published": "2024-07-17T07:15:02.183",
|
||
|
|
"lastModified": "2024-07-17T07:15:02.183",
|
||
|
|
"vulnStatus": "Received",
|
||
|
|
"cveTags": [],
|
||
|
|
"descriptions": [
|
||
|
|
{
|
||
|
|
"lang": "en",
|
||
|
|
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: Fix too early release of tcx_entry\n\nPedro Pinto and later independently also Hyunwoo Kim and Wongi Lee reported\nan issue that the tcx_entry can be released too early leading to a use\nafter free (UAF) when an active old-style ingress or clsact qdisc with a\nshared tc block is later replaced by another ingress or clsact instance.\n\nEssentially, the sequence to trigger the UAF (one example) can be as follows:\n\n 1. A network namespace is created\n 2. An ingress qdisc is created. This allocates a tcx_entry, and\n &tcx_entry->miniq is stored in the qdisc's miniqp->p_miniq. At the\n same time, a tcf block with index 1 is created.\n 3. chain0 is attached to the tcf block. chain0 must be connected to\n the block linked to the ingress qdisc to later reach the function\n tcf_chain0_head_change_cb_del() which triggers the UAF.\n 4. Create and graft a clsact qdisc. This causes the ingress qdisc\n created in step 1 to be removed, thus freeing the previously linked\n tcx_entry:\n\n rtnetlink_rcv_msg()\n => tc_modify_qdisc()\n => qdisc_create()\n => clsact_init() [a]\n => qdisc_graft()\n => qdisc_destroy()\n => __qdisc_destroy()\n => ingress_destroy() [b]\n => tcx_entry_free()\n => kfree_rcu() // tcx_entry freed\n\n 5. Finally, the network namespace is closed. This registers the\n cleanup_net worker, and during the process of releasing the\n remaining clsact qdisc, it accesses the tcx_entry that was\n already freed in step 4, causing the UAF to occur:\n\n cleanup_net()\n => ops_exit_list()\n => default_device_exit_batch()\n => unregister_netdevice_many()\n => unregister_netdevice_many_notify()\n => dev_shutdown()\n => qdisc_put()\n => clsact_destroy() [c]\n => tcf_block_put_ext()\n => tcf_chain0_head_change_cb_del()\n => tcf_chain_head_change_item()\n => clsact_chain_head_change()\n => mini_qdisc_pair_swap() // UAF\n\nThere are also other variants, the gist is to add an ingress (or clsact)\nqdisc with a specific shared block, then to replace that qdisc, waiting\nfor the tcx_entry kfree_rcu() to be executed and subsequently accessing\nthe current active qdisc's miniq one way or another.\n\nThe correct fix is to turn the miniq_active boolean into a counter. What\ncan be observed, at step 2 above, the counter transitions from 0->1, at\nstep [a] from 1->2 (in order for the miniq object to remain active during\nthe replacement), then in [b] from 2->1 and finally [c] 1->0 with the\neventual release. The reference counter in general ranges from [0,2] and\nit does not need to be atomic since all access to the counter is protected\nby the rtnl mutex. With this in place, there is no longer a UAF happening\nand the tcx_entry is freed at the correct time."
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"metrics": {},
|
||
|
|
"references": [
|
||
|
|
{
|
||
|
|
"url": "https://git.kernel.org/stable/c/1cb6f0bae50441f4b4b32a28315853b279c7404e",
|
||
|
|
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
|
||
|
|
}
|
||
|
|
]
|
||
|
|
}
|