In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
bpf: Fix UAF due to concurrent consumption of ttrace lists in alloc_bulk<\/p>\n
Syzkaller repeatedly triggered UAF splats related to nodes in
\nwaiting_for_gp_ttrace within the bpf memalloc:<\/p>\n
BUG: KASAN: slab-use-after-free in llist_del_first+0x85\/0x110 lib\/llist.c:61
\nRead of size 8 at addr ffff8881572cd080 by task syz.4.470\/5112
\n …
\n llist_del_first+0x85\/0x110 lib\/llist.c:61
\n alloc_bulk+0x193\/0x460 kernel\/bpf\/memalloc.c:229
\n bpf_mem_refill+0x386\/0x560 kernel\/bpf\/memalloc.c:436<\/p>\n
Freed by task 14:
\n …
\n __free_rcu kernel\/bpf\/memalloc.c:281 [inline]
\n __free_rcu_tasks_trace+0x48\/0xd0 kernel\/bpf\/memalloc.c:291
\n rcu_tasks_invoke_cbs+0x1ec\/0x3e0 kernel\/rcu\/tasks.h:571
\n rcu_tasks_one_gp+0x13d\/0x220 kernel\/rcu\/tasks.h:621
\n rcu_tasks_kthread+0xf3\/0x120 kernel\/rcu\/tasks.h:651<\/p>\n
The reason is that the UAF occurs after the RCU Tasks Trace GP expires:
\nwhen the __free_rcu() callback runs, there is no synchronization
\nprotecting llist_del_all() against concurrent alloc_bulk() operating on
\nwaiting_for_gp_ttrace, leading to the race condition below:<\/p>\n
CPU0 CPU1
\n __free_rcu (RCU Tasks Trace callback)
\nalloc_bulk
\n llist_del_first(&c->waiting_for_gp_ttrace)
\n entry = smp_load_acquire(&head->first);
\n do {
\n if (entry == NULL)
\n return NULL;
\n free_all(llist_del_all(&c->waiting_for_gp_ttrace))
\n llist_for_each_safe(pos, t, llnode)
\n free_one(pos);
\n next = READ_ONCE(entry->next); first, &entry, next));<\/p>\n
In addition, there is also a theoretical race condition on the
\nfree_by_rcu_ttrace list. This race requires two preconditions: an
\nin-flight Tasks Trace GP keeping c->call_rcu_ttrace_in_progress == 1,
\nand concurrent cross-CPU frees repopulating c->free_by_rcu_ttrace with
\nnew nodes. Under these conditions, the following scenario triggers UAF:<\/p>\n
\/\/ CPU0
\n\/\/ irq work is still busy (on PREEMPT_RT)
\nalloc_bulk()
\n llist_del_first(&c->free_by_rcu_ttrace)
\n entry = smp_load_acquire(&head->first);
\n do {
\n if (entry == NULL)
\n return NULL;<\/p>\n
\/\/ CPU1
\n bpf_mem_alloc_destroy()
\n WRITE_ONCE(c->draining, true)
\n \/\/ wait for CPU0
\n irq_work_sync()<\/p>\n
\/\/ CPU2
\n do_call_rcu_ttrace(tgt(CPU0))
\n if (c->draining) {
\n llist_del_all(&c->free_by_rcu_ttrace)
\n free_all()
\n }<\/p>\n
\/\/ CPU0 continue
\n next = READ_ONCE(entry->next); first, &entry, next));<\/p>\n
Fix this by introducing a raw spinlock to synchronize the concurrent
\nconsumption on waiting_for_gp_ttrace and free_by_rcu_ttrace.<\/p>\n","protected":false},"excerpt":{"rendered":"
In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
bpf: Fix UAF due to concurrent consumption of ttrace lists in alloc_bulk<\/p>\n
Syzkaller repeatedly triggered UAF splats related to node<\/p>\n","protected":false},"template":"","meta":[],"categories":[],"tags":[],"class_list":["post-41114","vuln","type-vuln","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln\/41114","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln"}],"about":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/types\/vuln"}],"wp:attachment":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=41114"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=41114"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=41114"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}