From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions When using eBPF to probe CS CALL instructions within a function, a crash can be triggered. The eBPF tool probes offset 257 of the __hrtimer_run_queues() function: <__hrtimer_run_queues+249>: nopl 0x0(%rax,%rax,1) <__hrtimer_run_queues+254>: mov %r14,%rdi <__hrtimer_run_queues+257>: cs call <__x86_indirect_thunk_r12> <__hrtimer_run_queues+263>: mov %eax,%r12d <__hrtimer_run_queues+266>: xchg %ax,%ax <__hrtimer_run_queues+268>: mov %r13,%rdi Which triggers this crash: BUG: unable to handle page fault for address: 00000000000f41c9 #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD 0 P4D 0 Oops: 0002 [#1] SMP NOPTI CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P RIP: 0010:__hrtimer_run_queues+0x106/0x230 Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is at the 6th byte of the above CS CALL instruction. Since the CS CALL instruction occupies 6 bytes, the exception occurred in the middle of that call instruction. The root cause is that when using eBPF tools to probe in the middle of a function, a kprobe with INT3 is used as the underlying implementation. During single-step emulation of the original CALL instruction, int3_emulate_call() assumes that the probed CALL instruction is 5 bytes long. However, the actual CS-prefixed CALL instruction occupies 6 bytes, so it constructs an incorrect exception return address. When the CPU returns from the kprobe handler, the next instruction to be executed is at the address of the last byte of that CS CALL instruction. Coincidentally, starting from that address, the CPU fetches and decodes a completely different instruction, which ultimately triggers a kernel crash. Fix the issue by using the actual instruction length obtained from the instruction decoder when constructing the exception return address, rather than relying on the hardcoded CALL_INSN_SIZE macro. [ mingo: Refined the changelog ] Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step") Suggested-by: Masami Hiramatsu (Google) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- net/bridge/br_ioctl.c | 470 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 470 insertions(+) create mode 100644 net/bridge/br_ioctl.c (limited to 'net/bridge/br_ioctl.c') diff --git a/net/bridge/br_ioctl.c b/net/bridge/br_ioctl.c new file mode 100644 index 000000000..39f3ffcfa --- /dev/null +++ b/net/bridge/br_ioctl.c @@ -0,0 +1,470 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Ioctl handler + * Linux ethernet bridge + * + * Authors: + * Lennert Buytenhek + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "br_private.h" + +static int get_bridge_ifindices(struct net *net, int *indices, int num) +{ + struct net_device *dev; + int i = 0; + + rcu_read_lock(); + for_each_netdev_rcu(net, dev) { + if (i >= num) + break; + if (netif_is_bridge_master(dev)) + indices[i++] = dev->ifindex; + } + rcu_read_unlock(); + + return i; +} + +/* called with RTNL */ +static void get_port_ifindices(struct net_bridge *br, int *ifindices, int num) +{ + struct net_bridge_port *p; + + list_for_each_entry(p, &br->port_list, list) { + if (p->port_no < num) + ifindices[p->port_no] = p->dev->ifindex; + } +} + +/* + * Format up to a page worth of forwarding table entries + * userbuf -- where to copy result + * maxnum -- maximum number of entries desired + * (limited to a page for sanity) + * offset -- number of records to skip + */ +static int get_fdb_entries(struct net_bridge *br, void __user *userbuf, + unsigned long maxnum, unsigned long offset) +{ + int num; + void *buf; + size_t size; + + /* Clamp size to PAGE_SIZE, test maxnum to avoid overflow */ + if (maxnum > PAGE_SIZE/sizeof(struct __fdb_entry)) + maxnum = PAGE_SIZE/sizeof(struct __fdb_entry); + + size = maxnum * sizeof(struct __fdb_entry); + + buf = kmalloc(size, GFP_USER); + if (!buf) + return -ENOMEM; + + num = br_fdb_fillbuf(br, buf, maxnum, offset); + if (num > 0) { + if (copy_to_user(userbuf, buf, + array_size(num, sizeof(struct __fdb_entry)))) + num = -EFAULT; + } + kfree(buf); + + return num; +} + +/* called with RTNL */ +static int add_del_if(struct net_bridge *br, int ifindex, int isadd) +{ + struct net *net = dev_net(br->dev); + struct net_device *dev; + int ret; + + if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + dev = __dev_get_by_index(net, ifindex); + if (dev == NULL) + return -EINVAL; + + if (isadd) + ret = br_add_if(br, dev, NULL); + else + ret = br_del_if(br, dev); + + return ret; +} + +#define BR_UARGS_MAX 4 +static int br_dev_read_uargs(unsigned long *args, size_t nr_args, + void __user **argp, void __user *data) +{ + int ret; + + if (nr_args < 2 || nr_args > BR_UARGS_MAX) + return -EINVAL; + + if (in_compat_syscall()) { + unsigned int cargs[BR_UARGS_MAX]; + int i; + + ret = copy_from_user(cargs, data, nr_args * sizeof(*cargs)); + if (ret) + goto fault; + + for (i = 0; i < nr_args; ++i) + args[i] = cargs[i]; + + *argp = compat_ptr(args[1]); + } else { + ret = copy_from_user(args, data, nr_args * sizeof(*args)); + if (ret) + goto fault; + *argp = (void __user *)args[1]; + } + + return 0; +fault: + return -EFAULT; +} + +/* + * Legacy ioctl's through SIOCDEVPRIVATE + * This interface is deprecated because it was too difficult + * to do the translation for 32/64bit ioctl compatibility. + */ +int br_dev_siocdevprivate(struct net_device *dev, struct ifreq *rq, + void __user *data, int cmd) +{ + struct net_bridge *br = netdev_priv(dev); + struct net_bridge_port *p = NULL; + unsigned long args[4]; + void __user *argp; + int ret; + + ret = br_dev_read_uargs(args, ARRAY_SIZE(args), &argp, data); + if (ret) + return ret; + + switch (args[0]) { + case BRCTL_ADD_IF: + case BRCTL_DEL_IF: + return add_del_if(br, args[1], args[0] == BRCTL_ADD_IF); + + case BRCTL_GET_BRIDGE_INFO: + { + struct __bridge_info b; + + memset(&b, 0, sizeof(struct __bridge_info)); + rcu_read_lock(); + memcpy(&b.designated_root, &br->designated_root, 8); + memcpy(&b.bridge_id, &br->bridge_id, 8); + b.root_path_cost = br->root_path_cost; + b.max_age = jiffies_to_clock_t(br->max_age); + b.hello_time = jiffies_to_clock_t(br->hello_time); + b.forward_delay = br->forward_delay; + b.bridge_max_age = br->bridge_max_age; + b.bridge_hello_time = br->bridge_hello_time; + b.bridge_forward_delay = jiffies_to_clock_t(br->bridge_forward_delay); + b.topology_change = br->topology_change; + b.topology_change_detected = br->topology_change_detected; + b.root_port = br->root_port; + + b.stp_enabled = (br->stp_enabled != BR_NO_STP); + b.ageing_time = jiffies_to_clock_t(br->ageing_time); + b.hello_timer_value = br_timer_value(&br->hello_timer); + b.tcn_timer_value = br_timer_value(&br->tcn_timer); + b.topology_change_timer_value = br_timer_value(&br->topology_change_timer); + b.gc_timer_value = br_timer_value(&br->gc_work.timer); + rcu_read_unlock(); + + if (copy_to_user((void __user *)args[1], &b, sizeof(b))) + return -EFAULT; + + return 0; + } + + case BRCTL_GET_PORT_LIST: + { + int num, *indices; + + num = args[2]; + if (num < 0) + return -EINVAL; + if (num == 0) + num = 256; + if (num > BR_MAX_PORTS) + num = BR_MAX_PORTS; + + indices = kzalloc_objs(int, num); + if (indices == NULL) + return -ENOMEM; + + get_port_ifindices(br, indices, num); + if (copy_to_user(argp, indices, array_size(num, sizeof(int)))) + num = -EFAULT; + kfree(indices); + return num; + } + + case BRCTL_SET_BRIDGE_FORWARD_DELAY: + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + ret = br_set_forward_delay(br, args[1]); + break; + + case BRCTL_SET_BRIDGE_HELLO_TIME: + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + ret = br_set_hello_time(br, args[1]); + break; + + case BRCTL_SET_BRIDGE_MAX_AGE: + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + ret = br_set_max_age(br, args[1]); + break; + + case BRCTL_SET_AGEING_TIME: + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + ret = br_set_ageing_time(br, args[1]); + break; + + case BRCTL_GET_PORT_INFO: + { + struct __port_info p; + struct net_bridge_port *pt; + + rcu_read_lock(); + if ((pt = br_get_port(br, args[2])) == NULL) { + rcu_read_unlock(); + return -EINVAL; + } + + memset(&p, 0, sizeof(struct __port_info)); + memcpy(&p.designated_root, &pt->designated_root, 8); + memcpy(&p.designated_bridge, &pt->designated_bridge, 8); + p.port_id = READ_ONCE(pt->port_id); + p.designated_port = READ_ONCE(pt->designated_port); + p.path_cost = READ_ONCE(pt->path_cost); + p.designated_cost = READ_ONCE(pt->designated_cost); + p.state = pt->state; + p.top_change_ack = pt->topology_change_ack; + p.config_pending = READ_ONCE(pt->config_pending); + p.message_age_timer_value = br_timer_value(&pt->message_age_timer); + p.forward_delay_timer_value = br_timer_value(&pt->forward_delay_timer); + p.hold_timer_value = br_timer_value(&pt->hold_timer); + + rcu_read_unlock(); + + if (copy_to_user(argp, &p, sizeof(p))) + return -EFAULT; + + return 0; + } + + case BRCTL_SET_BRIDGE_STP_STATE: + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + ret = br_stp_set_enabled(br, args[1], NULL); + break; + + case BRCTL_SET_BRIDGE_PRIORITY: + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + br_stp_set_bridge_priority(br, args[1]); + ret = 0; + break; + + case BRCTL_SET_PORT_PRIORITY: + { + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + spin_lock_bh(&br->lock); + if ((p = br_get_port(br, args[1])) == NULL) + ret = -EINVAL; + else + ret = br_stp_set_port_priority(p, args[2]); + spin_unlock_bh(&br->lock); + break; + } + + case BRCTL_SET_PATH_COST: + { + if (!ns_capable(dev_net(dev)->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + spin_lock_bh(&br->lock); + if ((p = br_get_port(br, args[1])) == NULL) + ret = -EINVAL; + else + ret = br_stp_set_path_cost(p, args[2]); + spin_unlock_bh(&br->lock); + break; + } + + case BRCTL_GET_FDB_ENTRIES: + return get_fdb_entries(br, argp, args[2], args[3]); + + default: + ret = -EOPNOTSUPP; + } + + if (!ret) { + if (p) + br_ifinfo_notify(RTM_NEWLINK, NULL, p); + else + netdev_state_change(br->dev); + } + + return ret; +} + +static int old_deviceless(struct net *net, void __user *data) +{ + unsigned long args[3]; + void __user *argp; + int ret; + + ret = br_dev_read_uargs(args, ARRAY_SIZE(args), &argp, data); + if (ret) + return ret; + + switch (args[0]) { + case BRCTL_GET_VERSION: + return BRCTL_VERSION; + + case BRCTL_GET_BRIDGES: + { + int *indices; + int ret = 0; + + if (args[2] >= 2048) + return -ENOMEM; + indices = kzalloc_objs(int, args[2]); + if (indices == NULL) + return -ENOMEM; + + args[2] = get_bridge_ifindices(net, indices, args[2]); + + ret = copy_to_user(argp, indices, + array_size(args[2], sizeof(int))) + ? -EFAULT : args[2]; + + kfree(indices); + return ret; + } + + case BRCTL_ADD_BRIDGE: + case BRCTL_DEL_BRIDGE: + { + char buf[IFNAMSIZ]; + + if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + if (copy_from_user(buf, argp, IFNAMSIZ)) + return -EFAULT; + + buf[IFNAMSIZ-1] = 0; + + if (args[0] == BRCTL_ADD_BRIDGE) + return br_add_bridge(net, buf); + + return br_del_bridge(net, buf); + } + } + + return -EOPNOTSUPP; +} + +int br_ioctl_stub(struct net *net, unsigned int cmd, void __user *uarg) +{ + int ret = -EOPNOTSUPP; + struct ifreq ifr; + + if (cmd == SIOCBRADDIF || cmd == SIOCBRDELIF) { + void __user *data; + char *colon; + + if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) + return -EPERM; + + if (get_user_ifreq(&ifr, &data, uarg)) + return -EFAULT; + + ifr.ifr_name[IFNAMSIZ - 1] = 0; + colon = strchr(ifr.ifr_name, ':'); + if (colon) + *colon = 0; + } + + rtnl_lock(); + + switch (cmd) { + case SIOCGIFBR: + case SIOCSIFBR: + ret = old_deviceless(net, uarg); + break; + case SIOCBRADDBR: + case SIOCBRDELBR: + { + char buf[IFNAMSIZ]; + + if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) { + ret = -EPERM; + break; + } + + if (copy_from_user(buf, uarg, IFNAMSIZ)) { + ret = -EFAULT; + break; + } + + buf[IFNAMSIZ-1] = 0; + if (cmd == SIOCBRADDBR) + ret = br_add_bridge(net, buf); + else + ret = br_del_bridge(net, buf); + } + break; + case SIOCBRADDIF: + case SIOCBRDELIF: + { + struct net_device *dev; + + dev = __dev_get_by_name(net, ifr.ifr_name); + if (!dev || !netif_device_present(dev)) { + ret = -ENODEV; + break; + } + if (!netif_is_bridge_master(dev)) { + ret = -EOPNOTSUPP; + break; + } + + ret = add_del_if(netdev_priv(dev), ifr.ifr_ifindex, cmd == SIOCBRADDIF); + } + break; + } + + rtnl_unlock(); + + return ret; +} -- cgit v1.3.1