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, 143 insertions, 0 deletions

diff --git a/arch/powerpc/boot/decompress.c b/arch/powerpc/boot/decompress.c
new file mode 100644
index 000000000..6835cb53f
--- /dev/null
+++ b/arch/powerpc/boot/decompress.c
@@ -0,0 +1,143 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Wrapper around the kernel's pre-boot decompression library.
+ *
+ * Copyright (C) IBM Corporation 2016.
+ */
+
+#include "elf.h"
+#include "page.h"
+#include "string.h"
+#include "stdio.h"
+#include "ops.h"
+#include "reg.h"
+#include "types.h"
+
+/*
+ * The decompressor_*.c files play #ifdef games so they can be used in both
+ * pre-boot and regular kernel code. We need these definitions to make the
+ * includes work.
+ */
+
+#define STATIC static
+#define INIT
+
+/*
+ * The build process will copy the required zlib source files and headers
+ * out of lib/ and "fix" the includes so they do not pull in other kernel
+ * headers.
+ */
+
+#ifdef CONFIG_KERNEL_GZIP
+#	include "decompress_inflate.c"
+#endif
+
+#ifdef CONFIG_KERNEL_XZ
+#	include "xz_config.h"
+#	include "../../../lib/decompress_unxz.c"
+#endif
+
+/* globals for tracking the state of the decompression */
+static unsigned long decompressed_bytes;
+static unsigned long limit;
+static unsigned long skip;
+static char *output_buffer;
+
+/*
+ * flush() is called by __decompress() when the decompressor's scratch buffer is
+ * full.
+ */
+static long flush(void *v, unsigned long buffer_size)
+{
+	unsigned long end = decompressed_bytes + buffer_size;
+	unsigned long size = buffer_size;
+	unsigned long offset = 0;
+	char *in = v;
+	char *out;
+
+	/*
+	 * if we hit our decompression limit, we need to fake an error to abort
+	 * the in-progress decompression.
+	 */
+	if (decompressed_bytes >= limit)
+		return -1;
+
+	/* skip this entire block */
+	if (end <= skip) {
+		decompressed_bytes += buffer_size;
+		return buffer_size;
+	}
+
+	/* skip some data at the start, but keep the rest of the block */
+	if (decompressed_bytes < skip && end > skip) {
+		offset = skip - decompressed_bytes;
+
+		in += offset;
+		size -= offset;
+		decompressed_bytes += offset;
+	}
+
+	out = &output_buffer[decompressed_bytes - skip];
+	size = min(decompressed_bytes + size, limit) - decompressed_bytes;
+
+	memcpy(out, in, size);
+	decompressed_bytes += size;
+
+	return buffer_size;
+}
+
+static void print_err(char *s)
+{
+	/* suppress the "error" when we terminate the decompressor */
+	if (decompressed_bytes >= limit)
+		return;
+
+	printf("Decompression error: '%s'\n\r", s);
+}
+
+/**
+ * partial_decompress - decompresses part or all of a compressed buffer
+ * @inbuf:       input buffer
+ * @input_size:  length of the input buffer
+ * @outbuf:      output buffer
+ * @output_size: length of the output buffer
+ * @_skip:       number of output bytes to ignore
+ *
+ * This function takes compressed data from inbuf, decompresses and write it to
+ * outbuf. Once output_size bytes are written to the output buffer, or the
+ * stream is exhausted the function will return the number of bytes that were
+ * decompressed. Otherwise it will return whatever error code the decompressor
+ * reported (NB: This is specific to each decompressor type).
+ *
+ * The skip functionality is mainly there so the program and discover
+ * the size of the compressed image so that it can ask firmware (if present)
+ * for an appropriately sized buffer.
+ */
+long partial_decompress(void *inbuf, unsigned long input_size,
+	void *outbuf, unsigned long output_size, unsigned long _skip)
+{
+	int ret;
+
+	/*
+	 * The skipped bytes needs to be included in the size of data we want
+	 * to decompress.
+	 */
+	output_size += _skip;
+
+	decompressed_bytes = 0;
+	output_buffer = outbuf;
+	limit = output_size;
+	skip = _skip;
+
+	ret = __decompress(inbuf, input_size, NULL, flush, outbuf,
+		output_size, NULL, print_err);
+
+	/*
+	 * If decompression was aborted due to an actual error rather than
+	 * a fake error that we used to abort, then we should report it.
+	 */
+	if (decompressed_bytes < limit)
+		return ret;
+
+	return decompressed_bytes - skip;
+}