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, 278 insertions, 0 deletions

diff --git a/drivers/base/regmap/regmap-sdw-mbq.c b/drivers/base/regmap/regmap-sdw-mbq.c
new file mode 100644
index 000000000..b0312f57c
--- /dev/null
+++ b/drivers/base/regmap/regmap-sdw-mbq.c
@@ -0,0 +1,278 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright(c) 2020 Intel Corporation.
+
+#include <linux/bits.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/errno.h>
+#include <linux/iopoll.h>
+#include <linux/module.h>
+#include <linux/regmap.h>
+#include <linux/soundwire/sdw.h>
+#include <linux/soundwire/sdw_registers.h>
+#include <sound/sdca_function.h>
+#include "internal.h"
+
+struct regmap_mbq_context {
+	struct device *dev;
+	struct sdw_slave *sdw;
+
+	bool (*readable_reg)(struct device *dev, unsigned int reg);
+
+	struct regmap_sdw_mbq_cfg cfg;
+
+	int val_size;
+};
+
+static int regmap_sdw_mbq_size(struct regmap_mbq_context *ctx, unsigned int reg)
+{
+	int size = ctx->val_size;
+
+	if (ctx->cfg.mbq_size) {
+		size = ctx->cfg.mbq_size(ctx->dev, reg);
+		if (!size || size > ctx->val_size)
+			return -EINVAL;
+	}
+
+	return size;
+}
+
+static bool regmap_sdw_mbq_deferrable(struct regmap_mbq_context *ctx, unsigned int reg)
+{
+	if (ctx->cfg.deferrable)
+		return ctx->cfg.deferrable(ctx->dev, reg);
+
+	return false;
+}
+
+static int regmap_sdw_mbq_poll_busy(struct sdw_slave *slave, unsigned int reg,
+				    struct regmap_mbq_context *ctx)
+{
+	struct device *dev = ctx->dev;
+	int val, ret = 0;
+
+	dev_dbg(dev, "Deferring transaction for 0x%x\n", reg);
+
+	reg = SDW_SDCA_CTL(SDW_SDCA_CTL_FUNC(reg), 0,
+			   SDCA_CTL_ENTITY_0_FUNCTION_STATUS, 0);
+
+	if (!ctx->readable_reg || ctx->readable_reg(dev, reg)) {
+		ret = read_poll_timeout(sdw_read_no_pm, val,
+					val < 0 || !(val & SDCA_CTL_ENTITY_0_FUNCTION_BUSY),
+					ctx->cfg.retry_us, ctx->cfg.timeout_us,
+					false, slave, reg);
+		if (val < 0)
+			return val;
+		if (ret)
+			dev_err(dev, "Function busy timed out 0x%x: %d\n", reg, val);
+	} else {
+		fsleep(ctx->cfg.timeout_us);
+	}
+
+	return ret;
+}
+
+static int regmap_sdw_mbq_write_impl(struct sdw_slave *slave,
+				     unsigned int reg, unsigned int val,
+				     int mbq_size)
+{
+	int shift = mbq_size * BITS_PER_BYTE;
+	int ret;
+
+	while (--mbq_size > 0) {
+		shift -= BITS_PER_BYTE;
+
+		ret = sdw_write_no_pm(slave, SDW_SDCA_MBQ_CTL(reg),
+				      (val >> shift) & 0xff);
+		if (ret < 0)
+			return ret;
+	}
+
+	return sdw_write_no_pm(slave, reg, val & 0xff);
+}
+
+static int regmap_sdw_mbq_write(void *context, unsigned int reg, unsigned int val)
+{
+	struct regmap_mbq_context *ctx = context;
+	struct sdw_slave *slave = ctx->sdw;
+	struct device *dev = ctx->dev;
+	bool deferrable = regmap_sdw_mbq_deferrable(ctx, reg);
+	int mbq_size = regmap_sdw_mbq_size(ctx, reg);
+	int ret;
+
+	if (mbq_size < 0)
+		return mbq_size;
+
+	/*
+	 * Technically the spec does allow a device to set itself to busy for
+	 * internal reasons, but since it doesn't provide any information on
+	 * how to handle timeouts in that case, for now the code will only
+	 * process a single wait/timeout on function busy and a single retry
+	 * of the transaction.
+	 */
+	ret = regmap_sdw_mbq_write_impl(slave, reg, val, mbq_size);
+	if (ret == -ENODATA) {
+		if (!deferrable)
+			dev_warn(dev, "Defer on undeferrable control: %x\n", reg);
+
+		ret = regmap_sdw_mbq_poll_busy(slave, reg, ctx);
+		if (ret)
+			return ret;
+
+		ret = regmap_sdw_mbq_write_impl(slave, reg, val, mbq_size);
+	}
+
+	return ret;
+}
+
+static int regmap_sdw_mbq_read_impl(struct sdw_slave *slave,
+				    unsigned int reg, unsigned int *val,
+				    int mbq_size)
+{
+	int shift = BITS_PER_BYTE;
+	int read;
+
+	read = sdw_read_no_pm(slave, reg);
+	if (read < 0)
+		return read;
+
+	*val = read;
+
+	while (--mbq_size > 0) {
+		read = sdw_read_no_pm(slave, SDW_SDCA_MBQ_CTL(reg));
+		if (read < 0)
+			return read;
+
+		*val |= read << shift;
+		shift += BITS_PER_BYTE;
+	}
+
+	return 0;
+}
+
+static int regmap_sdw_mbq_read(void *context, unsigned int reg, unsigned int *val)
+{
+	struct regmap_mbq_context *ctx = context;
+	struct sdw_slave *slave = ctx->sdw;
+	struct device *dev = ctx->dev;
+	bool deferrable = regmap_sdw_mbq_deferrable(ctx, reg);
+	int mbq_size = regmap_sdw_mbq_size(ctx, reg);
+	int ret;
+
+	if (mbq_size < 0)
+		return mbq_size;
+
+	/*
+	 * Technically the spec does allow a device to set itself to busy for
+	 * internal reasons, but since it doesn't provide any information on
+	 * how to handle timeouts in that case, for now the code will only
+	 * process a single wait/timeout on function busy and a single retry
+	 * of the transaction.
+	 */
+	ret = regmap_sdw_mbq_read_impl(slave, reg, val, mbq_size);
+	if (ret == -ENODATA) {
+		if (!deferrable)
+			dev_warn(dev, "Defer on undeferrable control: %x\n", reg);
+
+		ret = regmap_sdw_mbq_poll_busy(slave, reg, ctx);
+		if (ret)
+			return ret;
+
+		ret = regmap_sdw_mbq_read_impl(slave, reg, val, mbq_size);
+	}
+
+	return ret;
+}
+
+static const struct regmap_bus regmap_sdw_mbq = {
+	.reg_read = regmap_sdw_mbq_read,
+	.reg_write = regmap_sdw_mbq_write,
+	.reg_format_endian_default = REGMAP_ENDIAN_LITTLE,
+	.val_format_endian_default = REGMAP_ENDIAN_LITTLE,
+};
+
+static int regmap_sdw_mbq_config_check(const struct regmap_config *config)
+{
+	if (config->val_bits > (sizeof(unsigned int) * BITS_PER_BYTE))
+		return -ENOTSUPP;
+
+	/* Registers are 32 bits wide */
+	if (config->reg_bits != 32)
+		return -ENOTSUPP;
+
+	if (config->pad_bits != 0)
+		return -ENOTSUPP;
+
+	return 0;
+}
+
+static struct regmap_mbq_context *
+regmap_sdw_mbq_gen_context(struct device *dev,
+			   struct sdw_slave *sdw,
+			   const struct regmap_config *config,
+			   const struct regmap_sdw_mbq_cfg *mbq_config)
+{
+	struct regmap_mbq_context *ctx;
+
+	ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return ERR_PTR(-ENOMEM);
+
+	ctx->dev = dev;
+	ctx->sdw = sdw;
+
+	if (mbq_config)
+		ctx->cfg = *mbq_config;
+
+	ctx->val_size = config->val_bits / BITS_PER_BYTE;
+	ctx->readable_reg = config->readable_reg;
+
+	return ctx;
+}
+
+struct regmap *__regmap_init_sdw_mbq(struct device *dev, struct sdw_slave *sdw,
+				     const struct regmap_config *config,
+				     const struct regmap_sdw_mbq_cfg *mbq_config,
+				     struct lock_class_key *lock_key,
+				     const char *lock_name)
+{
+	struct regmap_mbq_context *ctx;
+	int ret;
+
+	ret = regmap_sdw_mbq_config_check(config);
+	if (ret)
+		return ERR_PTR(ret);
+
+	ctx = regmap_sdw_mbq_gen_context(dev, sdw, config, mbq_config);
+	if (IS_ERR(ctx))
+		return ERR_CAST(ctx);
+
+	return __regmap_init(dev, &regmap_sdw_mbq, ctx,
+			     config, lock_key, lock_name);
+}
+EXPORT_SYMBOL_GPL(__regmap_init_sdw_mbq);
+
+struct regmap *__devm_regmap_init_sdw_mbq(struct device *dev, struct sdw_slave *sdw,
+					  const struct regmap_config *config,
+					  const struct regmap_sdw_mbq_cfg *mbq_config,
+					  struct lock_class_key *lock_key,
+					  const char *lock_name)
+{
+	struct regmap_mbq_context *ctx;
+	int ret;
+
+	ret = regmap_sdw_mbq_config_check(config);
+	if (ret)
+		return ERR_PTR(ret);
+
+	ctx = regmap_sdw_mbq_gen_context(dev, sdw, config, mbq_config);
+	if (IS_ERR(ctx))
+		return ERR_CAST(ctx);
+
+	return __devm_regmap_init(dev, &regmap_sdw_mbq, ctx,
+				  config, lock_key, lock_name);
+}
+EXPORT_SYMBOL_GPL(__devm_regmap_init_sdw_mbq);
+
+MODULE_DESCRIPTION("regmap SoundWire MBQ Module");
+MODULE_LICENSE("GPL");