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, 195 insertions, 0 deletions

diff --git a/arch/microblaze/lib/memmove.c b/arch/microblaze/lib/memmove.c
new file mode 100644
index 000000000..c1f08c484
--- /dev/null
+++ b/arch/microblaze/lib/memmove.c
@@ -0,0 +1,195 @@
+/*
+ * Copyright (C) 2008-2009 Michal Simek <monstr@monstr.eu>
+ * Copyright (C) 2008-2009 PetaLogix
+ * Copyright (C) 2007 John Williams
+ *
+ * Reasonably optimised generic C-code for memcpy on Microblaze
+ * This is generic C code to do efficient, alignment-aware memmove.
+ *
+ * It is based on demo code originally Copyright 2001 by Intel Corp, taken from
+ * http://www.embedded.com/showArticle.jhtml?articleID=19205567
+ *
+ * Attempts were made, unsuccessfully, to contact the original
+ * author of this code (Michael Morrow, Intel).  Below is the original
+ * copyright notice.
+ *
+ * This software has been developed by Intel Corporation.
+ * Intel specifically disclaims all warranties, express or
+ * implied, and all liability, including consequential and
+ * other indirect damages, for the use of this program, including
+ * liability for infringement of any proprietary rights,
+ * and including the warranties of merchantability and fitness
+ * for a particular purpose. Intel does not assume any
+ * responsibility for and errors which may appear in this program
+ * not any responsibility to update it.
+ */
+
+#include <linux/export.h>
+#include <linux/types.h>
+#include <linux/stddef.h>
+#include <linux/compiler.h>
+#include <linux/string.h>
+
+#ifdef CONFIG_OPT_LIB_FUNCTION
+void *memmove(void *v_dst, const void *v_src, __kernel_size_t c)
+{
+	const char *src = v_src;
+	char *dst = v_dst;
+	const uint32_t *i_src;
+	uint32_t *i_dst;
+
+	if (!c)
+		return v_dst;
+
+	/* Use memcpy when source is higher than dest */
+	if (v_dst <= v_src)
+		return memcpy(v_dst, v_src, c);
+
+	/* The following code tries to optimize the copy by using unsigned
+	 * alignment. This will work fine if both source and destination are
+	 * aligned on the same boundary. However, if they are aligned on
+	 * different boundaries shifts will be necessary. This might result in
+	 * bad performance on MicroBlaze systems without a barrel shifter.
+	 */
+	/* FIXME this part needs more test */
+	/* Do a descending copy - this is a bit trickier! */
+	dst += c;
+	src += c;
+
+	if (c >= 4) {
+		unsigned  value, buf_hold;
+
+		/* Align the destination to a word boundary. */
+		/* This is done in an endian independent manner. */
+
+		switch ((unsigned long)dst & 3) {
+		case 3:
+			*--dst = *--src;
+			--c;
+			fallthrough;
+		case 2:
+			*--dst = *--src;
+			--c;
+			fallthrough;
+		case 1:
+			*--dst = *--src;
+			--c;
+		}
+
+		i_dst = (void *)dst;
+		/* Choose a copy scheme based on the source */
+		/* alignment relative to destination. */
+		switch ((unsigned long)src & 3) {
+		case 0x0:	/* Both byte offsets are aligned */
+
+			i_src  = (const void *)src;
+
+			for (; c >= 4; c -= 4)
+				*--i_dst = *--i_src;
+
+			src  = (const void *)i_src;
+			break;
+		case 0x1:	/* Unaligned - Off by 1 */
+			/* Word align the source */
+			i_src = (const void *) (((unsigned)src + 4) & ~3);
+#ifndef __MICROBLAZEEL__
+			/* Load the holding buffer */
+			buf_hold = *--i_src >> 24;
+
+			for (; c >= 4; c -= 4) {
+				value = *--i_src;
+				*--i_dst = buf_hold << 8 | value;
+				buf_hold = value >> 24;
+			}
+#else
+			/* Load the holding buffer */
+			buf_hold = (*--i_src & 0xFF) << 24;
+
+			for (; c >= 4; c -= 4) {
+				value = *--i_src;
+				*--i_dst = buf_hold |
+						((value & 0xFFFFFF00) >> 8);
+				buf_hold = (value  & 0xFF) << 24;
+			}
+#endif
+			/* Realign the source */
+			src = (const void *)i_src;
+			src += 1;
+			break;
+		case 0x2:	/* Unaligned - Off by 2 */
+			/* Word align the source */
+			i_src = (const void *) (((unsigned)src + 4) & ~3);
+#ifndef __MICROBLAZEEL__
+			/* Load the holding buffer */
+			buf_hold = *--i_src >> 16;
+
+			for (; c >= 4; c -= 4) {
+				value = *--i_src;
+				*--i_dst = buf_hold << 16 | value;
+				buf_hold = value >> 16;
+			}
+#else
+			/* Load the holding buffer */
+			buf_hold = (*--i_src & 0xFFFF) << 16;
+
+			for (; c >= 4; c -= 4) {
+				value = *--i_src;
+				*--i_dst = buf_hold |
+						((value & 0xFFFF0000) >> 16);
+				buf_hold = (value & 0xFFFF) << 16;
+			}
+#endif
+			/* Realign the source */
+			src = (const void *)i_src;
+			src += 2;
+			break;
+		case 0x3:	/* Unaligned - Off by 3 */
+			/* Word align the source */
+			i_src = (const void *) (((unsigned)src + 4) & ~3);
+#ifndef __MICROBLAZEEL__
+			/* Load the holding buffer */
+			buf_hold = *--i_src >> 8;
+
+			for (; c >= 4; c -= 4) {
+				value = *--i_src;
+				*--i_dst = buf_hold << 24 | value;
+				buf_hold = value >> 8;
+			}
+#else
+			/* Load the holding buffer */
+			buf_hold = (*--i_src & 0xFFFFFF) << 8;
+
+			for (; c >= 4; c -= 4) {
+				value = *--i_src;
+				*--i_dst = buf_hold |
+						((value & 0xFF000000) >> 24);
+				buf_hold = (value & 0xFFFFFF) << 8;
+			}
+#endif
+			/* Realign the source */
+			src = (const void *)i_src;
+			src += 3;
+			break;
+		}
+		dst = (void *)i_dst;
+	}
+
+	/* simple fast copy, ... unless a cache boundary is crossed */
+	/* Finish off any remaining bytes */
+	switch (c) {
+	case 4:
+		*--dst = *--src;
+		fallthrough;
+	case 3:
+		*--dst = *--src;
+		fallthrough;
+	case 2:
+		*--dst = *--src;
+		fallthrough;
+	case 1:
+		*--dst = *--src;
+	}
+	return v_dst;
+}
+EXPORT_SYMBOL(memmove);
+#endif /* CONFIG_OPT_LIB_FUNCTION */