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, 265 insertions, 0 deletions

diff --git a/drivers/soundwire/sysfs_slave.c b/drivers/soundwire/sysfs_slave.c
new file mode 100644
index 000000000..a2ee06414
--- /dev/null
+++ b/drivers/soundwire/sysfs_slave.c
@@ -0,0 +1,265 @@
+// SPDX-License-Identifier: GPL-2.0-only
+// Copyright(c) 2015-2020 Intel Corporation.
+
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/sysfs.h>
+#include <linux/soundwire/sdw.h>
+#include <linux/soundwire/sdw_type.h>
+#include "bus.h"
+#include "sysfs_local.h"
+
+/*
+ * Slave sysfs
+ */
+
+/*
+ * The sysfs for Slave reflects the MIPI description as given
+ * in the MIPI DisCo spec.
+ * status and device_number come directly from the MIPI SoundWire
+ * 1.x specification.
+ *
+ * Base file is device
+ *	|---- status
+ *	|---- device_number
+ *	|---- modalias
+ *	|---- dev-properties
+ *		|---- mipi_revision
+ *		|---- wake_capable
+ *		|---- test_mode_capable
+ *		|---- clk_stop_mode1
+ *		|---- simple_clk_stop_capable
+ *		|---- clk_stop_timeout
+ *		|---- ch_prep_timeout
+ *		|---- reset_behave
+ *		|---- high_PHY_capable
+ *		|---- paging_support
+ *		|---- bank_delay_support
+ *		|---- p15_behave
+ *		|---- master_count
+ *		|---- source_ports
+ *		|---- sink_ports
+ *	|---- dp0
+ *		|---- max_word
+ *		|---- min_word
+ *		|---- words
+ *		|---- BRA_flow_controlled
+ *		|---- simple_ch_prep_sm
+ *		|---- imp_def_interrupts
+ *	|---- dpN_<sink/src>
+ *		|---- max_word
+ *		|---- min_word
+ *		|---- words
+ *		|---- type
+ *		|---- max_grouping
+ *		|---- simple_ch_prep_sm
+ *		|---- ch_prep_timeout
+ *		|---- imp_def_interrupts
+ *		|---- min_ch
+ *		|---- max_ch
+ *		|---- channels
+ *		|---- ch_combinations
+ *		|---- max_async_buffer
+ *		|---- block_pack_mode
+ *		|---- port_encoding
+ *
+ */
+
+#define sdw_slave_attr(field, format_string)			\
+static ssize_t field##_show(struct device *dev,			\
+			    struct device_attribute *attr,	\
+			    char *buf)				\
+{								\
+	struct sdw_slave *slave = dev_to_sdw_dev(dev);		\
+	return sprintf(buf, format_string, slave->prop.field);	\
+}								\
+static DEVICE_ATTR_RO(field)
+
+sdw_slave_attr(mipi_revision, "0x%x\n");
+sdw_slave_attr(wake_capable, "%d\n");
+sdw_slave_attr(test_mode_capable, "%d\n");
+sdw_slave_attr(clk_stop_mode1, "%d\n");
+sdw_slave_attr(simple_clk_stop_capable, "%d\n");
+sdw_slave_attr(clk_stop_timeout, "%d\n");
+sdw_slave_attr(ch_prep_timeout, "%d\n");
+sdw_slave_attr(reset_behave, "%d\n");
+sdw_slave_attr(high_PHY_capable, "%d\n");
+sdw_slave_attr(paging_support, "%d\n");
+sdw_slave_attr(bank_delay_support, "%d\n");
+sdw_slave_attr(p15_behave, "%d\n");
+sdw_slave_attr(master_count, "%d\n");
+sdw_slave_attr(source_ports, "0x%x\n");
+sdw_slave_attr(sink_ports, "0x%x\n");
+
+static ssize_t modalias_show(struct device *dev,
+			     struct device_attribute *attr, char *buf)
+{
+	struct sdw_slave *slave = dev_to_sdw_dev(dev);
+
+	return sdw_slave_modalias(slave, buf, 256);
+}
+static DEVICE_ATTR_RO(modalias);
+
+static struct attribute *slave_attrs[] = {
+	&dev_attr_modalias.attr,
+	NULL,
+};
+
+static const struct attribute_group slave_attr_group = {
+	.attrs = slave_attrs,
+};
+
+static struct attribute *slave_dev_attrs[] = {
+	&dev_attr_mipi_revision.attr,
+	&dev_attr_wake_capable.attr,
+	&dev_attr_test_mode_capable.attr,
+	&dev_attr_clk_stop_mode1.attr,
+	&dev_attr_simple_clk_stop_capable.attr,
+	&dev_attr_clk_stop_timeout.attr,
+	&dev_attr_ch_prep_timeout.attr,
+	&dev_attr_reset_behave.attr,
+	&dev_attr_high_PHY_capable.attr,
+	&dev_attr_paging_support.attr,
+	&dev_attr_bank_delay_support.attr,
+	&dev_attr_p15_behave.attr,
+	&dev_attr_master_count.attr,
+	&dev_attr_source_ports.attr,
+	&dev_attr_sink_ports.attr,
+	NULL,
+};
+
+static const struct attribute_group sdw_slave_dev_attr_group = {
+	.attrs	= slave_dev_attrs,
+	.name = "dev-properties",
+};
+
+/*
+ * DP0 sysfs
+ */
+
+#define sdw_dp0_attr(field, format_string)				\
+static ssize_t field##_show(struct device *dev,				\
+			    struct device_attribute *attr,		\
+			    char *buf)					\
+{									\
+	struct sdw_slave *slave = dev_to_sdw_dev(dev);			\
+	return sprintf(buf, format_string, slave->prop.dp0_prop->field);\
+}									\
+static DEVICE_ATTR_RO(field)
+
+sdw_dp0_attr(max_word, "%d\n");
+sdw_dp0_attr(min_word, "%d\n");
+sdw_dp0_attr(BRA_flow_controlled, "%d\n");
+sdw_dp0_attr(simple_ch_prep_sm, "%d\n");
+sdw_dp0_attr(imp_def_interrupts, "0x%x\n");
+
+static ssize_t words_show(struct device *dev,
+			  struct device_attribute *attr, char *buf)
+{
+	struct sdw_slave *slave = dev_to_sdw_dev(dev);
+	ssize_t size = 0;
+	int i;
+
+	for (i = 0; i < slave->prop.dp0_prop->num_words; i++)
+		size += sprintf(buf + size, "%d ",
+				slave->prop.dp0_prop->words[i]);
+	size += sprintf(buf + size, "\n");
+
+	return size;
+}
+static DEVICE_ATTR_RO(words);
+
+static struct attribute *dp0_attrs[] = {
+	&dev_attr_max_word.attr,
+	&dev_attr_min_word.attr,
+	&dev_attr_words.attr,
+	&dev_attr_BRA_flow_controlled.attr,
+	&dev_attr_simple_ch_prep_sm.attr,
+	&dev_attr_imp_def_interrupts.attr,
+	NULL,
+};
+
+static umode_t dp0_attr_visible(struct kobject *kobj, struct attribute *attr,
+			      int n)
+{
+	struct sdw_slave *slave = dev_to_sdw_dev(kobj_to_dev(kobj));
+
+	if (slave->prop.dp0_prop)
+		return attr->mode;
+	return 0;
+}
+
+static bool dp0_group_visible(struct kobject *kobj)
+{
+	struct sdw_slave *slave = dev_to_sdw_dev(kobj_to_dev(kobj));
+
+	if (slave->prop.dp0_prop)
+		return true;
+	return false;
+}
+DEFINE_SYSFS_GROUP_VISIBLE(dp0);
+
+static const struct attribute_group dp0_group = {
+	.attrs = dp0_attrs,
+	.is_visible = SYSFS_GROUP_VISIBLE(dp0),
+	.name = "dp0",
+};
+
+const struct attribute_group *sdw_attr_groups[] = {
+	&slave_attr_group,
+	&sdw_slave_dev_attr_group,
+	&dp0_group,
+	NULL,
+};
+
+/*
+ * the status is shown in capital letters for UNATTACHED and RESERVED
+ * on purpose, to highlight users to the fact that these status values
+ * are not expected.
+ */
+static const char *const slave_status[] = {
+	[SDW_SLAVE_UNATTACHED] =  "UNATTACHED",
+	[SDW_SLAVE_ATTACHED] = "Attached",
+	[SDW_SLAVE_ALERT] = "Alert",
+	[SDW_SLAVE_RESERVED] = "RESERVED",
+};
+
+static ssize_t status_show(struct device *dev,
+			   struct device_attribute *attr, char *buf)
+{
+	struct sdw_slave *slave = dev_to_sdw_dev(dev);
+
+	return sprintf(buf, "%s\n", slave_status[slave->status]);
+}
+static DEVICE_ATTR_RO(status);
+
+static ssize_t device_number_show(struct device *dev,
+				  struct device_attribute *attr, char *buf)
+{
+	struct sdw_slave *slave = dev_to_sdw_dev(dev);
+
+	if (slave->status == SDW_SLAVE_UNATTACHED)
+		return sprintf(buf, "%s", "N/A");
+	else
+		return sprintf(buf, "%d", slave->dev_num);
+}
+static DEVICE_ATTR_RO(device_number);
+
+static struct attribute *slave_status_attrs[] = {
+	&dev_attr_status.attr,
+	&dev_attr_device_number.attr,
+	NULL,
+};
+
+/*
+ * we don't use ATTRIBUTES_GROUP here since the group is used in a
+ * separate file and can't be handled as a static.
+ */
+static const struct attribute_group sdw_slave_status_attr_group = {
+	.attrs	= slave_status_attrs,
+};
+
+const struct attribute_group *sdw_slave_status_attr_groups[] = {
+	&sdw_slave_status_attr_group,
+	NULL
+};