x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

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) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 548 insertions, 0 deletions

diff --git a/drivers/net/arcnet/rfc1201.c b/drivers/net/arcnet/rfc1201.c
new file mode 100644
index 000000000..f2ec58d06
--- /dev/null
+++ b/drivers/net/arcnet/rfc1201.c
@@ -0,0 +1,548 @@
+/*
+ * Linux ARCnet driver - RFC1201 (standard) packet encapsulation
+ *
+ * Written 1994-1999 by Avery Pennarun.
+ * Derived from skeleton.c by Donald Becker.
+ *
+ * Special thanks to Contemporary Controls, Inc. (www.ccontrols.com)
+ *  for sponsoring the further development of this driver.
+ *
+ * **********************
+ *
+ * The original copyright of skeleton.c was as follows:
+ *
+ * skeleton.c Written 1993 by Donald Becker.
+ * Copyright 1993 United States Government as represented by the
+ * Director, National Security Agency.  This software may only be used
+ * and distributed according to the terms of the GNU General Public License as
+ * modified by SRC, incorporated herein by reference.
+ *
+ * **********************
+ *
+ * For more details, see drivers/net/arcnet.c
+ *
+ * **********************
+ */
+
+#define pr_fmt(fmt) "arcnet:" KBUILD_MODNAME ": " fmt
+
+#include <linux/gfp.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/if_arp.h>
+#include <linux/netdevice.h>
+#include <linux/skbuff.h>
+
+#include "arcdevice.h"
+
+MODULE_DESCRIPTION("ARCNet packet format (RFC 1201) module");
+MODULE_LICENSE("GPL");
+
+static __be16 type_trans(struct sk_buff *skb, struct net_device *dev);
+static void rx(struct net_device *dev, int bufnum,
+	       struct archdr *pkthdr, int length);
+static int build_header(struct sk_buff *skb, struct net_device *dev,
+			unsigned short type, uint8_t daddr);
+static int prepare_tx(struct net_device *dev, struct archdr *pkt, int length,
+		      int bufnum);
+static int continue_tx(struct net_device *dev, int bufnum);
+
+static struct ArcProto rfc1201_proto = {
+	.suffix		= 'a',
+	.mtu		= 1500,	/* could be more, but some receivers can't handle it... */
+	.is_ip          = 1,    /* This is for sending IP and ARP packages */
+	.rx		= rx,
+	.build_header	= build_header,
+	.prepare_tx	= prepare_tx,
+	.continue_tx	= continue_tx,
+	.ack_tx         = NULL
+};
+
+static int __init arcnet_rfc1201_init(void)
+{
+	pr_info("%s\n", "RFC1201 \"standard\" (`a') encapsulation support loaded");
+
+	arc_proto_map[ARC_P_IP]
+	    = arc_proto_map[ARC_P_IPV6]
+	    = arc_proto_map[ARC_P_ARP]
+	    = arc_proto_map[ARC_P_RARP]
+	    = arc_proto_map[ARC_P_IPX]
+	    = arc_proto_map[ARC_P_NOVELL_EC]
+	    = &rfc1201_proto;
+
+	/* if someone else already owns the broadcast, we won't take it */
+	if (arc_bcast_proto == arc_proto_default)
+		arc_bcast_proto = &rfc1201_proto;
+
+	return 0;
+}
+
+static void __exit arcnet_rfc1201_exit(void)
+{
+	arcnet_unregister_proto(&rfc1201_proto);
+}
+
+module_init(arcnet_rfc1201_init);
+module_exit(arcnet_rfc1201_exit);
+
+/* Determine a packet's protocol ID.
+ *
+ * With ARCnet we have to convert everything to Ethernet-style stuff.
+ */
+static __be16 type_trans(struct sk_buff *skb, struct net_device *dev)
+{
+	struct archdr *pkt = (struct archdr *)skb->data;
+	struct arc_rfc1201 *soft = &pkt->soft.rfc1201;
+	int hdr_size = ARC_HDR_SIZE + RFC1201_HDR_SIZE;
+
+	/* Pull off the arcnet header. */
+	skb_reset_mac_header(skb);
+	skb_pull(skb, hdr_size);
+
+	if (pkt->hard.dest == 0) {
+		skb->pkt_type = PACKET_BROADCAST;
+	} else if (dev->flags & IFF_PROMISC) {
+		/* if we're not sending to ourselves :) */
+		if (pkt->hard.dest != dev->dev_addr[0])
+			skb->pkt_type = PACKET_OTHERHOST;
+	}
+	/* now return the protocol number */
+	switch (soft->proto) {
+	case ARC_P_IP:
+		return htons(ETH_P_IP);
+	case ARC_P_IPV6:
+		return htons(ETH_P_IPV6);
+	case ARC_P_ARP:
+		return htons(ETH_P_ARP);
+	case ARC_P_RARP:
+		return htons(ETH_P_RARP);
+
+	case ARC_P_IPX:
+	case ARC_P_NOVELL_EC:
+		return htons(ETH_P_802_3);
+	default:
+		dev->stats.rx_errors++;
+		dev->stats.rx_crc_errors++;
+		return 0;
+	}
+
+	return htons(ETH_P_IP);
+}
+
+/* packet receiver */
+static void rx(struct net_device *dev, int bufnum,
+	       struct archdr *pkthdr, int length)
+{
+	struct arcnet_local *lp = netdev_priv(dev);
+	struct sk_buff *skb;
+	struct archdr *pkt = pkthdr;
+	struct arc_rfc1201 *soft = &pkthdr->soft.rfc1201;
+	int saddr = pkt->hard.source, ofs;
+	struct Incoming *in = &lp->rfc1201.incoming[saddr];
+
+	arc_printk(D_DURING, dev, "it's an RFC1201 packet (length=%d)\n",
+		   length);
+
+	if (length >= MinTU)
+		ofs = 512 - length;
+	else
+		ofs = 256 - length;
+
+	if (soft->split_flag == 0xFF) {		/* Exception Packet */
+		if (length >= 4 + RFC1201_HDR_SIZE) {
+			arc_printk(D_DURING, dev, "compensating for exception packet\n");
+		} else {
+			arc_printk(D_EXTRA, dev, "short RFC1201 exception packet from %02Xh",
+				   saddr);
+			return;
+		}
+
+		/* skip over 4-byte junkola */
+		length -= 4;
+		ofs += 4;
+		lp->hw.copy_from_card(dev, bufnum, 512 - length,
+				      soft, sizeof(pkt->soft));
+	}
+	if (!soft->split_flag) {	/* not split */
+		arc_printk(D_RX, dev, "incoming is not split (splitflag=%d)\n",
+			   soft->split_flag);
+
+		if (in->skb) {	/* already assembling one! */
+			arc_printk(D_EXTRA, dev, "aborting assembly (seq=%d) for unsplit packet (splitflag=%d, seq=%d)\n",
+				   in->sequence, soft->split_flag,
+				   soft->sequence);
+			lp->rfc1201.aborted_seq = soft->sequence;
+			dev_kfree_skb_irq(in->skb);
+			dev->stats.rx_errors++;
+			dev->stats.rx_missed_errors++;
+			in->skb = NULL;
+		}
+		in->sequence = soft->sequence;
+
+		skb = alloc_skb(length + ARC_HDR_SIZE, GFP_ATOMIC);
+		if (!skb) {
+			dev->stats.rx_dropped++;
+			return;
+		}
+		skb_put(skb, length + ARC_HDR_SIZE);
+		skb->dev = dev;
+
+		pkt = (struct archdr *)skb->data;
+		soft = &pkt->soft.rfc1201;
+
+		/* up to sizeof(pkt->soft) has already
+		 * been copied from the card
+		 */
+		memcpy(pkt, pkthdr, sizeof(struct archdr));
+		if (length > sizeof(pkt->soft))
+			lp->hw.copy_from_card(dev, bufnum,
+					      ofs + sizeof(pkt->soft),
+					      pkt->soft.raw + sizeof(pkt->soft),
+					      length - sizeof(pkt->soft));
+
+		/* ARP packets have problems when sent from some DOS systems:
+		 * the source address is always 0!
+		 * So we take the hardware source addr (which is impossible
+		 * to fumble) and insert it ourselves.
+		 */
+		if (soft->proto == ARC_P_ARP) {
+			struct arphdr *arp = (struct arphdr *)soft->payload;
+
+			/* make sure addresses are the right length */
+			if (arp->ar_hln == 1 && arp->ar_pln == 4) {
+				uint8_t *cptr = (uint8_t *)arp + sizeof(struct arphdr);
+
+				if (!*cptr) {	/* is saddr = 00? */
+					arc_printk(D_EXTRA, dev,
+						   "ARP source address was 00h, set to %02Xh\n",
+						   saddr);
+					dev->stats.rx_crc_errors++;
+					*cptr = saddr;
+				} else {
+					arc_printk(D_DURING, dev, "ARP source address (%Xh) is fine.\n",
+						   *cptr);
+				}
+			} else {
+				arc_printk(D_NORMAL, dev, "funny-shaped ARP packet. (%Xh, %Xh)\n",
+					   arp->ar_hln, arp->ar_pln);
+				dev->stats.rx_errors++;
+				dev->stats.rx_crc_errors++;
+			}
+		}
+		if (BUGLVL(D_SKB))
+			arcnet_dump_skb(dev, skb, "rx");
+
+		skb->protocol = type_trans(skb, dev);
+		netif_rx(skb);
+	} else {		/* split packet */
+		/* NOTE: MS-DOS ARP packet correction should only need to
+		 * apply to unsplit packets, since ARP packets are so short.
+		 *
+		 * My interpretation of the RFC1201 document is that if a
+		 * packet is received out of order, the entire assembly
+		 * process should be aborted.
+		 *
+		 * The RFC also mentions "it is possible for successfully
+		 * received packets to be retransmitted." As of 0.40 all
+		 * previously received packets are allowed, not just the
+		 * most recent one.
+		 *
+		 * We allow multiple assembly processes, one for each
+		 * ARCnet card possible on the network.
+		 * Seems rather like a waste of memory, but there's no
+		 * other way to be reliable.
+		 */
+
+		arc_printk(D_RX, dev, "packet is split (splitflag=%d, seq=%d)\n",
+			   soft->split_flag, in->sequence);
+
+		if (in->skb && in->sequence != soft->sequence) {
+			arc_printk(D_EXTRA, dev, "wrong seq number (saddr=%d, expected=%d, seq=%d, splitflag=%d)\n",
+				   saddr, in->sequence, soft->sequence,
+				   soft->split_flag);
+			dev_kfree_skb_irq(in->skb);
+			in->skb = NULL;
+			dev->stats.rx_errors++;
+			dev->stats.rx_missed_errors++;
+			in->lastpacket = in->numpackets = 0;
+		}
+		if (soft->split_flag & 1) {	/* first packet in split */
+			arc_printk(D_RX, dev, "brand new splitpacket (splitflag=%d)\n",
+				   soft->split_flag);
+			if (in->skb) {	/* already assembling one! */
+				arc_printk(D_EXTRA, dev, "aborting previous (seq=%d) assembly (splitflag=%d, seq=%d)\n",
+					   in->sequence, soft->split_flag,
+					   soft->sequence);
+				dev->stats.rx_errors++;
+				dev->stats.rx_missed_errors++;
+				dev_kfree_skb_irq(in->skb);
+			}
+			in->sequence = soft->sequence;
+			in->numpackets = ((unsigned)soft->split_flag >> 1) + 2;
+			in->lastpacket = 1;
+
+			if (in->numpackets > 16) {
+				arc_printk(D_EXTRA, dev, "incoming packet more than 16 segments; dropping. (splitflag=%d)\n",
+					   soft->split_flag);
+				lp->rfc1201.aborted_seq = soft->sequence;
+				dev->stats.rx_errors++;
+				dev->stats.rx_length_errors++;
+				return;
+			}
+			in->skb = skb = alloc_skb(508 * in->numpackets + ARC_HDR_SIZE,
+						  GFP_ATOMIC);
+			if (!skb) {
+				arc_printk(D_NORMAL, dev, "(split) memory squeeze, dropping packet.\n");
+				lp->rfc1201.aborted_seq = soft->sequence;
+				dev->stats.rx_dropped++;
+				return;
+			}
+			skb->dev = dev;
+			pkt = (struct archdr *)skb->data;
+			soft = &pkt->soft.rfc1201;
+
+			memcpy(pkt, pkthdr, ARC_HDR_SIZE + RFC1201_HDR_SIZE);
+			skb_put(skb, ARC_HDR_SIZE + RFC1201_HDR_SIZE);
+
+			soft->split_flag = 0;	/* end result won't be split */
+		} else {	/* not first packet */
+			int packetnum = ((unsigned)soft->split_flag >> 1) + 1;
+
+			/* if we're not assembling, there's no point trying to
+			 * continue.
+			 */
+			if (!in->skb) {
+				if (lp->rfc1201.aborted_seq != soft->sequence) {
+					arc_printk(D_EXTRA, dev, "can't continue split without starting first! (splitflag=%d, seq=%d, aborted=%d)\n",
+						   soft->split_flag,
+						   soft->sequence,
+						   lp->rfc1201.aborted_seq);
+					dev->stats.rx_errors++;
+					dev->stats.rx_missed_errors++;
+				}
+				return;
+			}
+			in->lastpacket++;
+			/* if not the right flag */
+			if (packetnum != in->lastpacket) {
+				/* harmless duplicate? ignore. */
+				if (packetnum <= in->lastpacket - 1) {
+					arc_printk(D_EXTRA, dev, "duplicate splitpacket ignored! (splitflag=%d)\n",
+						   soft->split_flag);
+					dev->stats.rx_errors++;
+					dev->stats.rx_frame_errors++;
+					return;
+				}
+				/* "bad" duplicate, kill reassembly */
+				arc_printk(D_EXTRA, dev, "out-of-order splitpacket, reassembly (seq=%d) aborted (splitflag=%d, seq=%d)\n",
+					   in->sequence, soft->split_flag,
+					   soft->sequence);
+				lp->rfc1201.aborted_seq = soft->sequence;
+				dev_kfree_skb_irq(in->skb);
+				in->skb = NULL;
+				dev->stats.rx_errors++;
+				dev->stats.rx_missed_errors++;
+				in->lastpacket = in->numpackets = 0;
+				return;
+			}
+			pkt = (struct archdr *)in->skb->data;
+			soft = &pkt->soft.rfc1201;
+		}
+
+		skb = in->skb;
+
+		lp->hw.copy_from_card(dev, bufnum, ofs + RFC1201_HDR_SIZE,
+				      skb->data + skb->len,
+				      length - RFC1201_HDR_SIZE);
+		skb_put(skb, length - RFC1201_HDR_SIZE);
+
+		/* are we done? */
+		if (in->lastpacket == in->numpackets) {
+			in->skb = NULL;
+			in->lastpacket = in->numpackets = 0;
+
+			arc_printk(D_SKB_SIZE, dev, "skb: received %d bytes from %02X (unsplit)\n",
+				   skb->len, pkt->hard.source);
+			arc_printk(D_SKB_SIZE, dev, "skb: received %d bytes from %02X (split)\n",
+				   skb->len, pkt->hard.source);
+			if (BUGLVL(D_SKB))
+				arcnet_dump_skb(dev, skb, "rx");
+
+			skb->protocol = type_trans(skb, dev);
+			netif_rx(skb);
+		}
+	}
+}
+
+/* Create the ARCnet hard/soft headers for RFC1201. */
+static int build_header(struct sk_buff *skb, struct net_device *dev,
+			unsigned short type, uint8_t daddr)
+{
+	struct arcnet_local *lp = netdev_priv(dev);
+	int hdr_size = ARC_HDR_SIZE + RFC1201_HDR_SIZE;
+	struct archdr *pkt = skb_push(skb, hdr_size);
+	struct arc_rfc1201 *soft = &pkt->soft.rfc1201;
+
+	/* set the protocol ID according to RFC1201 */
+	switch (type) {
+	case ETH_P_IP:
+		soft->proto = ARC_P_IP;
+		break;
+	case ETH_P_IPV6:
+		soft->proto = ARC_P_IPV6;
+		break;
+	case ETH_P_ARP:
+		soft->proto = ARC_P_ARP;
+		break;
+	case ETH_P_RARP:
+		soft->proto = ARC_P_RARP;
+		break;
+	case ETH_P_IPX:
+	case ETH_P_802_3:
+	case ETH_P_802_2:
+		soft->proto = ARC_P_IPX;
+		break;
+	case ETH_P_ATALK:
+		soft->proto = ARC_P_ATALK;
+		break;
+	default:
+		arc_printk(D_NORMAL, dev, "RFC1201: I don't understand protocol %d (%Xh)\n",
+			   type, type);
+		dev->stats.tx_errors++;
+		dev->stats.tx_aborted_errors++;
+		return 0;
+	}
+
+	/* Set the source hardware address.
+	 *
+	 * This is pretty pointless for most purposes, but it can help in
+	 * debugging.  ARCnet does not allow us to change the source address
+	 * in the actual packet sent.
+	 */
+	pkt->hard.source = *dev->dev_addr;
+
+	soft->sequence = htons(lp->rfc1201.sequence++);
+	soft->split_flag = 0;	/* split packets are done elsewhere */
+
+	/* see linux/net/ethernet/eth.c to see where I got the following */
+
+	if (dev->flags & (IFF_LOOPBACK | IFF_NOARP)) {
+		/* FIXME: fill in the last byte of the dest ipaddr here
+		 * to better comply with RFC1051 in "noarp" mode.
+		 * For now, always broadcasting will probably at least get
+		 * packets sent out :)
+		 */
+		pkt->hard.dest = 0;
+		return hdr_size;
+	}
+	/* otherwise, drop in the dest address */
+	pkt->hard.dest = daddr;
+	return hdr_size;
+}
+
+static void load_pkt(struct net_device *dev, struct arc_hardware *hard,
+		     struct arc_rfc1201 *soft, int softlen, int bufnum)
+{
+	struct arcnet_local *lp = netdev_priv(dev);
+	int ofs;
+
+	/* assume length <= XMTU: someone should have handled that by now. */
+
+	if (softlen > MinTU) {
+		hard->offset[0] = 0;
+		hard->offset[1] = ofs = 512 - softlen;
+	} else if (softlen > MTU) {	/* exception packet - add an extra header */
+		struct arc_rfc1201 excsoft;
+
+		excsoft.proto = soft->proto;
+		excsoft.split_flag = 0xff;
+		excsoft.sequence = htons(0xffff);
+
+		hard->offset[0] = 0;
+		ofs = 512 - softlen;
+		hard->offset[1] = ofs - RFC1201_HDR_SIZE;
+		lp->hw.copy_to_card(dev, bufnum, ofs - RFC1201_HDR_SIZE,
+				    &excsoft, RFC1201_HDR_SIZE);
+	} else {
+		hard->offset[0] = ofs = 256 - softlen;
+	}
+
+	lp->hw.copy_to_card(dev, bufnum, 0, hard, ARC_HDR_SIZE);
+	lp->hw.copy_to_card(dev, bufnum, ofs, soft, softlen);
+
+	lp->lastload_dest = hard->dest;
+}
+
+static int prepare_tx(struct net_device *dev, struct archdr *pkt, int length,
+		      int bufnum)
+{
+	struct arcnet_local *lp = netdev_priv(dev);
+	const int maxsegsize = XMTU - RFC1201_HDR_SIZE;
+	struct Outgoing *out;
+
+	arc_printk(D_DURING, dev, "prepare_tx: txbufs=%d/%d/%d\n",
+		   lp->next_tx, lp->cur_tx, bufnum);
+
+	/* hard header is not included in packet length */
+	length -= ARC_HDR_SIZE;
+	pkt->soft.rfc1201.split_flag = 0;
+
+	/* need to do a split packet? */
+	if (length > XMTU) {
+		out = &lp->outgoing;
+
+		out->length = length - RFC1201_HDR_SIZE;
+		out->dataleft = lp->outgoing.length;
+		out->numsegs = (out->dataleft + maxsegsize - 1) / maxsegsize;
+		out->segnum = 0;
+
+		arc_printk(D_DURING, dev, "rfc1201 prep_tx: ready for %d-segment split (%d bytes, seq=%d)\n",
+			   out->numsegs, out->length,
+			   pkt->soft.rfc1201.sequence);
+
+		return 0;	/* not done */
+	}
+	/* just load the packet into the buffers and send it off */
+	load_pkt(dev, &pkt->hard, &pkt->soft.rfc1201, length, bufnum);
+
+	return 1;		/* done */
+}
+
+static int continue_tx(struct net_device *dev, int bufnum)
+{
+	struct arcnet_local *lp = netdev_priv(dev);
+	struct Outgoing *out = &lp->outgoing;
+	struct arc_hardware *hard = &out->pkt->hard;
+	struct arc_rfc1201 *soft = &out->pkt->soft.rfc1201, *newsoft;
+	int maxsegsize = XMTU - RFC1201_HDR_SIZE;
+	int seglen;
+
+	arc_printk(D_DURING, dev,
+		   "rfc1201 continue_tx: loading segment %d(+1) of %d (seq=%d)\n",
+		   out->segnum, out->numsegs, soft->sequence);
+
+	/* the "new" soft header comes right before the data chunk */
+	newsoft = (struct arc_rfc1201 *)
+	    (out->pkt->soft.raw + out->length - out->dataleft);
+
+	if (!out->segnum)	/* first packet; newsoft == soft */
+		newsoft->split_flag = ((out->numsegs - 2) << 1) | 1;
+	else {
+		newsoft->split_flag = out->segnum << 1;
+		newsoft->proto = soft->proto;
+		newsoft->sequence = soft->sequence;
+	}
+
+	seglen = maxsegsize;
+	if (seglen > out->dataleft)
+		seglen = out->dataleft;
+	out->dataleft -= seglen;
+
+	load_pkt(dev, hard, newsoft, seglen + RFC1201_HDR_SIZE, bufnum);
+
+	out->segnum++;
+	if (out->segnum >= out->numsegs)
+		return 1;
+	else
+		return 0;
+}