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, 258 insertions, 0 deletions

diff --git a/drivers/clk/ti/gate.c b/drivers/clk/ti/gate.c
new file mode 100644
index 000000000..747412270
--- /dev/null
+++ b/drivers/clk/ti/gate.c
@@ -0,0 +1,258 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * OMAP gate clock support
+ *
+ * Copyright (C) 2013 Texas Instruments, Inc.
+ *
+ * Tero Kristo <t-kristo@ti.com>
+ */
+
+#include <linux/clk-provider.h>
+#include <linux/slab.h>
+#include <linux/io.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/clk/ti.h>
+
+#include "clock.h"
+
+#undef pr_fmt
+#define pr_fmt(fmt) "%s: " fmt, __func__
+
+static int omap36xx_gate_clk_enable_with_hsdiv_restore(struct clk_hw *clk);
+
+static const struct clk_ops omap_gate_clkdm_clk_ops = {
+	.init		= &omap2_init_clk_clkdm,
+	.enable		= &omap2_clkops_enable_clkdm,
+	.disable	= &omap2_clkops_disable_clkdm,
+	.restore_context = clk_gate_restore_context,
+};
+
+const struct clk_ops omap_gate_clk_ops = {
+	.init		= &omap2_init_clk_clkdm,
+	.enable		= &omap2_dflt_clk_enable,
+	.disable	= &omap2_dflt_clk_disable,
+	.is_enabled	= &omap2_dflt_clk_is_enabled,
+	.restore_context = clk_gate_restore_context,
+};
+
+static const struct clk_ops omap_gate_clk_hsdiv_restore_ops = {
+	.init		= &omap2_init_clk_clkdm,
+	.enable		= &omap36xx_gate_clk_enable_with_hsdiv_restore,
+	.disable	= &omap2_dflt_clk_disable,
+	.is_enabled	= &omap2_dflt_clk_is_enabled,
+	.restore_context = clk_gate_restore_context,
+};
+
+/**
+ * omap36xx_gate_clk_enable_with_hsdiv_restore - enable clocks suffering
+ *         from HSDivider PWRDN problem Implements Errata ID: i556.
+ * @hw: DPLL output struct clk_hw
+ *
+ * 3630 only: dpll3_m3_ck, dpll4_m2_ck, dpll4_m3_ck, dpll4_m4_ck,
+ * dpll4_m5_ck & dpll4_m6_ck dividers gets loaded with reset
+ * valueafter their respective PWRDN bits are set.  Any dummy write
+ * (Any other value different from the Read value) to the
+ * corresponding CM_CLKSEL register will refresh the dividers.
+ */
+static int omap36xx_gate_clk_enable_with_hsdiv_restore(struct clk_hw *hw)
+{
+	struct clk_omap_divider *parent;
+	struct clk_hw *parent_hw;
+	u32 dummy_v, orig_v;
+	int ret;
+
+	/* Clear PWRDN bit of HSDIVIDER */
+	ret = omap2_dflt_clk_enable(hw);
+
+	/* Parent is the x2 node, get parent of parent for the m2 div */
+	parent_hw = clk_hw_get_parent(clk_hw_get_parent(hw));
+	parent = to_clk_omap_divider(parent_hw);
+
+	/* Restore the dividers */
+	if (!ret) {
+		orig_v = ti_clk_ll_ops->clk_readl(&parent->reg);
+		dummy_v = orig_v;
+
+		/* Write any other value different from the Read value */
+		dummy_v ^= (1 << parent->shift);
+		ti_clk_ll_ops->clk_writel(dummy_v, &parent->reg);
+
+		/* Write the original divider */
+		ti_clk_ll_ops->clk_writel(orig_v, &parent->reg);
+	}
+
+	return ret;
+}
+
+static struct clk *_register_gate(struct device_node *node, const char *name,
+				  const char *parent_name, unsigned long flags,
+				  struct clk_omap_reg *reg, u8 bit_idx,
+				  u8 clk_gate_flags, const struct clk_ops *ops,
+				  const struct clk_hw_omap_ops *hw_ops)
+{
+	struct clk_init_data init = { NULL };
+	struct clk_hw_omap *clk_hw;
+	struct clk *clk;
+
+	clk_hw = kzalloc_obj(*clk_hw);
+	if (!clk_hw)
+		return ERR_PTR(-ENOMEM);
+
+	clk_hw->hw.init = &init;
+
+	init.name = name;
+	init.ops = ops;
+
+	memcpy(&clk_hw->enable_reg, reg, sizeof(*reg));
+	clk_hw->enable_bit = bit_idx;
+	clk_hw->ops = hw_ops;
+
+	clk_hw->flags = clk_gate_flags;
+
+	init.parent_names = &parent_name;
+	init.num_parents = 1;
+
+	init.flags = flags;
+
+	clk = of_ti_clk_register_omap_hw(node, &clk_hw->hw, name);
+
+	if (IS_ERR(clk))
+		kfree(clk_hw);
+
+	return clk;
+}
+
+static void __init _of_ti_gate_clk_setup(struct device_node *node,
+					 const struct clk_ops *ops,
+					 const struct clk_hw_omap_ops *hw_ops)
+{
+	struct clk *clk;
+	const char *parent_name;
+	struct clk_omap_reg reg;
+	const char *name;
+	u8 enable_bit = 0;
+	u32 flags = 0;
+	u8 clk_gate_flags = 0;
+
+	if (ops != &omap_gate_clkdm_clk_ops) {
+		if (ti_clk_get_reg_addr(node, 0, &reg))
+			return;
+
+		enable_bit = reg.bit;
+	}
+
+	if (of_clk_get_parent_count(node) != 1) {
+		pr_err("%pOFn must have 1 parent\n", node);
+		return;
+	}
+
+	parent_name = of_clk_get_parent_name(node, 0);
+
+	if (of_property_read_bool(node, "ti,set-rate-parent"))
+		flags |= CLK_SET_RATE_PARENT;
+
+	if (of_property_read_bool(node, "ti,set-bit-to-disable"))
+		clk_gate_flags |= INVERT_ENABLE;
+
+	name = ti_dt_clk_name(node);
+	clk = _register_gate(node, name, parent_name, flags, &reg,
+			     enable_bit, clk_gate_flags, ops, hw_ops);
+
+	if (!IS_ERR(clk))
+		of_clk_add_provider(node, of_clk_src_simple_get, clk);
+}
+
+static void __init
+_of_ti_composite_gate_clk_setup(struct device_node *node,
+				const struct clk_hw_omap_ops *hw_ops)
+{
+	struct clk_hw_omap *gate;
+
+	gate = kzalloc_obj(*gate);
+	if (!gate)
+		return;
+
+	if (ti_clk_get_reg_addr(node, 0, &gate->enable_reg))
+		goto cleanup;
+
+	gate->enable_bit = gate->enable_reg.bit;
+	gate->ops = hw_ops;
+
+	if (!ti_clk_add_component(node, &gate->hw, CLK_COMPONENT_TYPE_GATE))
+		return;
+
+cleanup:
+	kfree(gate);
+}
+
+static void __init
+of_ti_composite_no_wait_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_composite_gate_clk_setup(node, NULL);
+}
+CLK_OF_DECLARE(ti_composite_no_wait_gate_clk, "ti,composite-no-wait-gate-clock",
+	       of_ti_composite_no_wait_gate_clk_setup);
+
+#if defined(CONFIG_ARCH_OMAP2) || defined(CONFIG_ARCH_OMAP3)
+static void __init of_ti_composite_interface_clk_setup(struct device_node *node)
+{
+	_of_ti_composite_gate_clk_setup(node, &clkhwops_iclk_wait);
+}
+CLK_OF_DECLARE(ti_composite_interface_clk, "ti,composite-interface-clock",
+	       of_ti_composite_interface_clk_setup);
+#endif
+
+static void __init of_ti_composite_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_composite_gate_clk_setup(node, &clkhwops_wait);
+}
+CLK_OF_DECLARE(ti_composite_gate_clk, "ti,composite-gate-clock",
+	       of_ti_composite_gate_clk_setup);
+
+
+static void __init of_ti_clkdm_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_gate_clk_setup(node, &omap_gate_clkdm_clk_ops, NULL);
+}
+CLK_OF_DECLARE(ti_clkdm_gate_clk, "ti,clkdm-gate-clock",
+	       of_ti_clkdm_gate_clk_setup);
+
+static void __init of_ti_hsdiv_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_gate_clk_setup(node, &omap_gate_clk_hsdiv_restore_ops,
+			      &clkhwops_wait);
+}
+CLK_OF_DECLARE(ti_hsdiv_gate_clk, "ti,hsdiv-gate-clock",
+	       of_ti_hsdiv_gate_clk_setup);
+
+static void __init of_ti_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_gate_clk_setup(node, &omap_gate_clk_ops, NULL);
+}
+CLK_OF_DECLARE(ti_gate_clk, "ti,gate-clock", of_ti_gate_clk_setup);
+
+static void __init of_ti_wait_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_gate_clk_setup(node, &omap_gate_clk_ops, &clkhwops_wait);
+}
+CLK_OF_DECLARE(ti_wait_gate_clk, "ti,wait-gate-clock",
+	       of_ti_wait_gate_clk_setup);
+
+#ifdef CONFIG_ARCH_OMAP3
+static void __init of_ti_am35xx_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_gate_clk_setup(node, &omap_gate_clk_ops,
+			      &clkhwops_am35xx_ipss_module_wait);
+}
+CLK_OF_DECLARE(ti_am35xx_gate_clk, "ti,am35xx-gate-clock",
+	       of_ti_am35xx_gate_clk_setup);
+
+static void __init of_ti_dss_gate_clk_setup(struct device_node *node)
+{
+	_of_ti_gate_clk_setup(node, &omap_gate_clk_ops,
+			      &clkhwops_omap3430es2_dss_usbhost_wait);
+}
+CLK_OF_DECLARE(ti_dss_gate_clk, "ti,dss-gate-clock",
+	       of_ti_dss_gate_clk_setup);
+#endif