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, 102 insertions, 0 deletions

diff --git a/drivers/clk/versatile/icst.c b/drivers/clk/versatile/icst.c
new file mode 100644
index 000000000..307cb3774
--- /dev/null
+++ b/drivers/clk/versatile/icst.c
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ *  linux/arch/arm/common/icst307.c
+ *
+ *  Copyright (C) 2003 Deep Blue Solutions, Ltd, All Rights Reserved.
+ *
+ *  Support functions for calculating clocks/divisors for the ICST307
+ *  clock generators.  See https://www.idt.com/ for more information
+ *  on these devices.
+ *
+ *  This is an almost identical implementation to the ICST525 clock generator.
+ *  The s2div and idx2s files are different
+ */
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <asm/div64.h>
+#include "icst.h"
+
+/*
+ * Divisors for each OD setting.
+ */
+const unsigned char icst307_s2div[8] = { 10, 2, 8, 4, 5, 7, 3, 6 };
+const unsigned char icst525_s2div[8] = { 10, 2, 8, 4, 5, 7, 9, 6 };
+EXPORT_SYMBOL(icst307_s2div);
+EXPORT_SYMBOL(icst525_s2div);
+
+unsigned long icst_hz(const struct icst_params *p, struct icst_vco vco)
+{
+	u64 dividend = p->ref * 2 * (u64)(vco.v + 8);
+	u32 divisor = (vco.r + 2) * p->s2div[vco.s];
+
+	do_div(dividend, divisor);
+	return (unsigned long)dividend;
+}
+
+EXPORT_SYMBOL(icst_hz);
+
+/*
+ * Ascending divisor S values.
+ */
+const unsigned char icst307_idx2s[8] = { 1, 6, 3, 4, 7, 5, 2, 0 };
+const unsigned char icst525_idx2s[8] = { 1, 3, 4, 7, 5, 2, 6, 0 };
+EXPORT_SYMBOL(icst307_idx2s);
+EXPORT_SYMBOL(icst525_idx2s);
+
+struct icst_vco
+icst_hz_to_vco(const struct icst_params *p, unsigned long freq)
+{
+	struct icst_vco vco = { .s = 1, .v = p->vd_max, .r = p->rd_max };
+	unsigned long f;
+	unsigned int i = 0, rd, best = (unsigned int)-1;
+
+	/*
+	 * First, find the PLL output divisor such
+	 * that the PLL output is within spec.
+	 */
+	do {
+		f = freq * p->s2div[p->idx2s[i]];
+
+		if (f > p->vco_min && f <= p->vco_max)
+			break;
+		i++;
+	} while (i < 8);
+
+	if (i >= 8)
+		return vco;
+
+	vco.s = p->idx2s[i];
+
+	/*
+	 * Now find the closest divisor combination
+	 * which gives a PLL output of 'f'.
+	 */
+	for (rd = p->rd_min; rd <= p->rd_max; rd++) {
+		unsigned long fref_div, f_pll;
+		unsigned int vd;
+		int f_diff;
+
+		fref_div = (2 * p->ref) / rd;
+
+		vd = (f + fref_div / 2) / fref_div;
+		if (vd < p->vd_min || vd > p->vd_max)
+			continue;
+
+		f_pll = fref_div * vd;
+		f_diff = f_pll - f;
+		if (f_diff < 0)
+			f_diff = -f_diff;
+
+		if ((unsigned)f_diff < best) {
+			vco.v = vd - 8;
+			vco.r = rd - 2;
+			if (f_diff == 0)
+				break;
+			best = f_diff;
+		}
+	}
+
+	return vco;
+}
+
+EXPORT_SYMBOL(icst_hz_to_vco);