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, 151 insertions, 0 deletions

diff --git a/arch/powerpc/boot/mktree.c b/arch/powerpc/boot/mktree.c
new file mode 100644
index 000000000..dc603f3c1
--- /dev/null
+++ b/arch/powerpc/boot/mktree.c
@@ -0,0 +1,151 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Makes a tree bootable image for IBM Evaluation boards.
+ * Basically, just take a zImage, skip the ELF header, and stuff
+ * a 32 byte header on the front.
+ *
+ * We use htonl, which is a network macro, to make sure we're doing
+ * The Right Thing on an LE machine.  It's non-obvious, but it should
+ * work on anything BSD'ish.
+ */
+
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <netinet/in.h>
+#ifdef __sun__
+#include <inttypes.h>
+#else
+#include <stdint.h>
+#endif
+
+/* This gets tacked on the front of the image.  There are also a few
+ * bytes allocated after the _start label used by the boot rom (see
+ * head.S for details).
+ */
+typedef struct boot_block {
+	uint32_t bb_magic;		/* 0x0052504F */
+	uint32_t bb_dest;		/* Target address of the image */
+	uint32_t bb_num_512blocks;	/* Size, rounded-up, in 512 byte blks */
+	uint32_t bb_debug_flag;	/* Run debugger or image after load */
+	uint32_t bb_entry_point;	/* The image address to start */
+	uint32_t bb_checksum;	/* 32 bit checksum including header */
+	uint32_t reserved[2];
+} boot_block_t;
+
+#define IMGBLK	512
+unsigned int	tmpbuf[IMGBLK / sizeof(unsigned int)];
+
+int main(int argc, char *argv[])
+{
+	int	in_fd, out_fd;
+	int	nblks, i;
+	unsigned int	cksum, *cp;
+	struct	stat	st;
+	boot_block_t	bt;
+
+	if (argc < 5) {
+		fprintf(stderr, "usage: %s <zImage-file> <boot-image> <load address> <entry point>\n",argv[0]);
+		exit(1);
+	}
+
+	if (stat(argv[1], &st) < 0) {
+		perror("stat");
+		exit(2);
+	}
+
+	nblks = (st.st_size + IMGBLK) / IMGBLK;
+
+	bt.bb_magic = htonl(0x0052504F);
+
+	/* If we have the optional entry point parameter, use it */
+	bt.bb_dest = htonl(strtoul(argv[3], NULL, 0));
+	bt.bb_entry_point = htonl(strtoul(argv[4], NULL, 0));
+
+	/* We know these from the linker command.
+	 * ...and then move it up into memory a little more so the
+	 * relocation can happen.
+	 */
+	bt.bb_num_512blocks = htonl(nblks);
+	bt.bb_debug_flag = 0;
+
+	bt.bb_checksum = 0;
+
+	/* To be neat and tidy :-).
+	*/
+	bt.reserved[0] = 0;
+	bt.reserved[1] = 0;
+
+	if ((in_fd = open(argv[1], O_RDONLY)) < 0) {
+		perror("zImage open");
+		exit(3);
+	}
+
+	if ((out_fd = open(argv[2], (O_RDWR | O_CREAT | O_TRUNC), 0666)) < 0) {
+		perror("bootfile open");
+		exit(3);
+	}
+
+	cksum = 0;
+	cp = (void *)&bt;
+	for (i = 0; i < sizeof(bt) / sizeof(unsigned int); i++)
+		cksum += *cp++;
+
+	/* Assume zImage is an ELF file, and skip the 64K header.
+	*/
+	if (read(in_fd, tmpbuf, sizeof(tmpbuf)) != sizeof(tmpbuf)) {
+		fprintf(stderr, "%s is too small to be an ELF image\n",
+				argv[1]);
+		exit(4);
+	}
+
+	if (tmpbuf[0] != htonl(0x7f454c46)) {
+		fprintf(stderr, "%s is not an ELF image\n", argv[1]);
+		exit(4);
+	}
+
+	if (lseek(in_fd, (64 * 1024), SEEK_SET) < 0) {
+		fprintf(stderr, "%s failed to seek in ELF image\n", argv[1]);
+		exit(4);
+	}
+
+	nblks -= (64 * 1024) / IMGBLK;
+
+	/* And away we go......
+	*/
+	if (write(out_fd, &bt, sizeof(bt)) != sizeof(bt)) {
+		perror("boot-image write");
+		exit(5);
+	}
+
+	while (nblks-- > 0) {
+		if (read(in_fd, tmpbuf, sizeof(tmpbuf)) < 0) {
+			perror("zImage read");
+			exit(5);
+		}
+		cp = tmpbuf;
+		for (i = 0; i < sizeof(tmpbuf) / sizeof(unsigned int); i++)
+			cksum += *cp++;
+		if (write(out_fd, tmpbuf, sizeof(tmpbuf)) != sizeof(tmpbuf)) {
+			perror("boot-image write");
+			exit(5);
+		}
+	}
+
+	/* rewrite the header with the computed checksum.
+	*/
+	bt.bb_checksum = htonl(cksum);
+	if (lseek(out_fd, 0, SEEK_SET) < 0) {
+		perror("rewrite seek");
+		exit(1);
+	}
+	if (write(out_fd, &bt, sizeof(bt)) != sizeof(bt)) {
+		perror("boot-image rewrite");
+		exit(1);
+	}
+
+	exit(0);
+}