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, 278 insertions, 0 deletions

diff --git a/net/sunrpc/socklib.c b/net/sunrpc/socklib.c
new file mode 100644
index 000000000..d8d8842c7
--- /dev/null
+++ b/net/sunrpc/socklib.c
@@ -0,0 +1,278 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * linux/net/sunrpc/socklib.c
+ *
+ * Common socket helper routines for RPC client and server
+ *
+ * Copyright (C) 1995, 1996 Olaf Kirch <okir@monad.swb.de>
+ */
+
+#include <linux/compiler.h>
+#include <linux/netdevice.h>
+#include <linux/gfp.h>
+#include <linux/skbuff.h>
+#include <linux/types.h>
+#include <linux/pagemap.h>
+#include <linux/udp.h>
+#include <linux/sunrpc/msg_prot.h>
+#include <linux/sunrpc/sched.h>
+#include <linux/sunrpc/xdr.h>
+#include <linux/export.h>
+
+#include "socklib.h"
+
+/*
+ * Helper structure for copying from an sk_buff.
+ */
+struct xdr_skb_reader {
+	struct sk_buff	*skb;
+	unsigned int	offset;
+	bool		need_checksum;
+	size_t		count;
+	__wsum		csum;
+};
+
+/**
+ * xdr_skb_read_bits - copy some data bits from skb to internal buffer
+ * @desc: sk_buff copy helper
+ * @to: copy destination
+ * @len: number of bytes to copy
+ *
+ * Possibly called several times to iterate over an sk_buff and copy data out of
+ * it.
+ */
+static size_t
+xdr_skb_read_bits(struct xdr_skb_reader *desc, void *to, size_t len)
+{
+	len = min(len, desc->count);
+
+	if (desc->need_checksum) {
+		__wsum csum;
+
+		csum = skb_copy_and_csum_bits(desc->skb, desc->offset, to, len);
+		desc->csum = csum_block_add(desc->csum, csum, desc->offset);
+	} else {
+		if (unlikely(skb_copy_bits(desc->skb, desc->offset, to, len)))
+			return 0;
+	}
+
+	desc->count -= len;
+	desc->offset += len;
+	return len;
+}
+
+static ssize_t
+xdr_partial_copy_from_skb(struct xdr_buf *xdr, struct xdr_skb_reader *desc)
+{
+	struct page **ppage = xdr->pages + (xdr->page_base >> PAGE_SHIFT);
+	unsigned int poff = xdr->page_base & ~PAGE_MASK;
+	unsigned int pglen = xdr->page_len;
+	ssize_t copied = 0;
+	size_t ret;
+
+	if (xdr->head[0].iov_len == 0)
+		return 0;
+
+	ret = xdr_skb_read_bits(desc, xdr->head[0].iov_base,
+				xdr->head[0].iov_len);
+	if (ret != xdr->head[0].iov_len || !desc->count)
+		return ret;
+	copied += ret;
+
+	while (pglen) {
+		unsigned int len = min(PAGE_SIZE - poff, pglen);
+		char *kaddr;
+
+		/* ACL likes to be lazy in allocating pages - ACLs
+		 * are small by default but can get huge. */
+		if ((xdr->flags & XDRBUF_SPARSE_PAGES) && *ppage == NULL) {
+			*ppage = alloc_page(GFP_NOWAIT);
+			if (unlikely(*ppage == NULL)) {
+				if (copied == 0)
+					return -ENOMEM;
+				return copied;
+			}
+		}
+
+		kaddr = kmap_atomic(*ppage);
+		ret = xdr_skb_read_bits(desc, kaddr + poff, len);
+		flush_dcache_page(*ppage);
+		kunmap_atomic(kaddr);
+
+		copied += ret;
+		if (ret != len || !desc->count)
+			return copied;
+		ppage++;
+		pglen -= len;
+		poff = 0;
+	}
+
+	if (xdr->tail[0].iov_len) {
+		copied += xdr_skb_read_bits(desc, xdr->tail[0].iov_base,
+					xdr->tail[0].iov_len);
+	}
+
+	return copied;
+}
+
+/**
+ * csum_partial_copy_to_xdr - checksum and copy data
+ * @xdr: target XDR buffer
+ * @skb: source skb
+ *
+ * We have set things up such that we perform the checksum of the UDP
+ * packet in parallel with the copies into the RPC client iovec.  -DaveM
+ */
+int csum_partial_copy_to_xdr(struct xdr_buf *xdr, struct sk_buff *skb)
+{
+	struct xdr_skb_reader desc = {
+		.skb		= skb,
+		.count		= skb->len - desc.offset,
+	};
+
+	if (skb_csum_unnecessary(skb)) {
+		if (xdr_partial_copy_from_skb(xdr, &desc) < 0)
+			return -1;
+		if (desc.count)
+			return -1;
+		return 0;
+	}
+
+	desc.need_checksum = true;
+	desc.csum = csum_partial(skb->data, desc.offset, skb->csum);
+	if (xdr_partial_copy_from_skb(xdr, &desc) < 0)
+		return -1;
+	if (desc.offset != skb->len) {
+		__wsum csum2;
+		csum2 = skb_checksum(skb, desc.offset, skb->len - desc.offset, 0);
+		desc.csum = csum_block_add(desc.csum, csum2, desc.offset);
+	}
+	if (desc.count)
+		return -1;
+	if (csum_fold(desc.csum))
+		return -1;
+	if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE) &&
+	    !skb->csum_complete_sw)
+		netdev_rx_csum_fault(skb->dev, skb);
+	return 0;
+}
+
+static inline int xprt_sendmsg(struct socket *sock, struct msghdr *msg,
+			       size_t seek)
+{
+	if (seek)
+		iov_iter_advance(&msg->msg_iter, seek);
+	return sock_sendmsg(sock, msg);
+}
+
+static int xprt_send_kvec(struct socket *sock, struct msghdr *msg,
+			  struct kvec *vec, size_t seek)
+{
+	iov_iter_kvec(&msg->msg_iter, ITER_SOURCE, vec, 1, vec->iov_len);
+	return xprt_sendmsg(sock, msg, seek);
+}
+
+static int xprt_send_pagedata(struct socket *sock, struct msghdr *msg,
+			      struct xdr_buf *xdr, size_t base)
+{
+	iov_iter_bvec(&msg->msg_iter, ITER_SOURCE, xdr->bvec, xdr_buf_pagecount(xdr),
+		      xdr->page_len + xdr->page_base);
+	return xprt_sendmsg(sock, msg, base + xdr->page_base);
+}
+
+/* Common case:
+ *  - stream transport
+ *  - sending from byte 0 of the message
+ *  - the message is wholly contained in @xdr's head iovec
+ */
+static int xprt_send_rm_and_kvec(struct socket *sock, struct msghdr *msg,
+				 rpc_fraghdr marker, struct kvec *vec,
+				 size_t base)
+{
+	struct kvec iov[2] = {
+		[0] = {
+			.iov_base	= &marker,
+			.iov_len	= sizeof(marker)
+		},
+		[1] = *vec,
+	};
+	size_t len = iov[0].iov_len + iov[1].iov_len;
+
+	iov_iter_kvec(&msg->msg_iter, ITER_SOURCE, iov, 2, len);
+	return xprt_sendmsg(sock, msg, base);
+}
+
+/**
+ * xprt_sock_sendmsg - write an xdr_buf directly to a socket
+ * @sock: open socket to send on
+ * @msg: socket message metadata
+ * @xdr: xdr_buf containing this request
+ * @base: starting position in the buffer
+ * @marker: stream record marker field
+ * @sent_p: return the total number of bytes successfully queued for sending
+ *
+ * Return values:
+ *   On success, returns zero and fills in @sent_p.
+ *   %-ENOTSOCK if  @sock is not a struct socket.
+ */
+int xprt_sock_sendmsg(struct socket *sock, struct msghdr *msg,
+		      struct xdr_buf *xdr, unsigned int base,
+		      rpc_fraghdr marker, unsigned int *sent_p)
+{
+	unsigned int rmsize = marker ? sizeof(marker) : 0;
+	unsigned int remainder = rmsize + xdr->len - base;
+	unsigned int want;
+	int err = 0;
+
+	*sent_p = 0;
+
+	if (unlikely(!sock))
+		return -ENOTSOCK;
+
+	msg->msg_flags |= MSG_MORE;
+	want = xdr->head[0].iov_len + rmsize;
+	if (base < want) {
+		unsigned int len = want - base;
+
+		remainder -= len;
+		if (remainder == 0)
+			msg->msg_flags &= ~MSG_MORE;
+		if (rmsize)
+			err = xprt_send_rm_and_kvec(sock, msg, marker,
+						    &xdr->head[0], base);
+		else
+			err = xprt_send_kvec(sock, msg, &xdr->head[0], base);
+		if (remainder == 0 || err != len)
+			goto out;
+		*sent_p += err;
+		base = 0;
+	} else {
+		base -= want;
+	}
+
+	if (base < xdr->page_len) {
+		unsigned int len = xdr->page_len - base;
+
+		remainder -= len;
+		if (remainder == 0)
+			msg->msg_flags &= ~MSG_MORE;
+		err = xprt_send_pagedata(sock, msg, xdr, base);
+		if (remainder == 0 || err != len)
+			goto out;
+		*sent_p += err;
+		base = 0;
+	} else {
+		base -= xdr->page_len;
+	}
+
+	if (base >= xdr->tail[0].iov_len)
+		return 0;
+	msg->msg_flags &= ~MSG_MORE;
+	err = xprt_send_kvec(sock, msg, &xdr->tail[0], base);
+out:
+	if (err > 0) {
+		*sent_p += err;
+		err = 0;
+	}
+	return err;
+}