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, 167 insertions, 0 deletions

diff --git a/sound/soc/sof/ops.c b/sound/soc/sof/ops.c
new file mode 100644
index 000000000..74c04dcf4
--- /dev/null
+++ b/sound/soc/sof/ops.c
@@ -0,0 +1,167 @@
+// SPDX-License-Identifier: (GPL-2.0-only OR BSD-3-Clause)
+//
+// This file is provided under a dual BSD/GPLv2 license.  When using or
+// redistributing this file, you may do so under either license.
+//
+// Copyright(c) 2018 Intel Corporation
+//
+// Author: Liam Girdwood <liam.r.girdwood@linux.intel.com>
+//
+
+#include <linux/pci.h>
+#include "ops.h"
+
+static
+bool snd_sof_pci_update_bits_unlocked(struct snd_sof_dev *sdev, u32 offset,
+				      u32 mask, u32 value)
+{
+	struct pci_dev *pci = to_pci_dev(sdev->dev);
+	unsigned int old, new;
+	u32 ret = 0;
+
+	pci_read_config_dword(pci, offset, &ret);
+	old = ret;
+	dev_dbg(sdev->dev, "Debug PCIR: %8.8x at  %8.8x\n", old & mask, offset);
+
+	new = (old & ~mask) | (value & mask);
+
+	if (old == new)
+		return false;
+
+	pci_write_config_dword(pci, offset, new);
+	dev_dbg(sdev->dev, "Debug PCIW: %8.8x at  %8.8x\n", value,
+		offset);
+
+	return true;
+}
+
+bool snd_sof_pci_update_bits(struct snd_sof_dev *sdev, u32 offset,
+			     u32 mask, u32 value)
+{
+	guard(spinlock_irqsave)(&sdev->hw_lock);
+	return snd_sof_pci_update_bits_unlocked(sdev, offset, mask, value);
+}
+EXPORT_SYMBOL(snd_sof_pci_update_bits);
+
+bool snd_sof_dsp_update_bits_unlocked(struct snd_sof_dev *sdev, u32 bar,
+				      u32 offset, u32 mask, u32 value)
+{
+	unsigned int old, new;
+	u32 ret;
+
+	ret = snd_sof_dsp_read(sdev, bar, offset);
+
+	old = ret;
+	new = (old & ~mask) | (value & mask);
+
+	if (old == new)
+		return false;
+
+	snd_sof_dsp_write(sdev, bar, offset, new);
+
+	return true;
+}
+EXPORT_SYMBOL(snd_sof_dsp_update_bits_unlocked);
+
+bool snd_sof_dsp_update_bits64_unlocked(struct snd_sof_dev *sdev, u32 bar,
+					u32 offset, u64 mask, u64 value)
+{
+	u64 old, new;
+
+	old = snd_sof_dsp_read64(sdev, bar, offset);
+
+	new = (old & ~mask) | (value & mask);
+
+	if (old == new)
+		return false;
+
+	snd_sof_dsp_write64(sdev, bar, offset, new);
+
+	return true;
+}
+EXPORT_SYMBOL(snd_sof_dsp_update_bits64_unlocked);
+
+/* This is for registers bits with attribute RWC */
+bool snd_sof_dsp_update_bits(struct snd_sof_dev *sdev, u32 bar, u32 offset,
+			     u32 mask, u32 value)
+{
+	guard(spinlock_irqsave)(&sdev->hw_lock);
+	return snd_sof_dsp_update_bits_unlocked(sdev, bar, offset, mask, value);
+}
+EXPORT_SYMBOL(snd_sof_dsp_update_bits);
+
+bool snd_sof_dsp_update_bits64(struct snd_sof_dev *sdev, u32 bar, u32 offset,
+			       u64 mask, u64 value)
+{
+	guard(spinlock_irqsave)(&sdev->hw_lock);
+	return snd_sof_dsp_update_bits64_unlocked(sdev, bar, offset, mask, value);
+}
+EXPORT_SYMBOL(snd_sof_dsp_update_bits64);
+
+static
+void snd_sof_dsp_update_bits_forced_unlocked(struct snd_sof_dev *sdev, u32 bar,
+					     u32 offset, u32 mask, u32 value)
+{
+	unsigned int old, new;
+	u32 ret;
+
+	ret = snd_sof_dsp_read(sdev, bar, offset);
+
+	old = ret;
+	new = (old & ~mask) | (value & mask);
+
+	snd_sof_dsp_write(sdev, bar, offset, new);
+}
+
+/* This is for registers bits with attribute RWC */
+void snd_sof_dsp_update_bits_forced(struct snd_sof_dev *sdev, u32 bar,
+				    u32 offset, u32 mask, u32 value)
+{
+	guard(spinlock_irqsave)(&sdev->hw_lock);
+	snd_sof_dsp_update_bits_forced_unlocked(sdev, bar, offset, mask, value);
+}
+EXPORT_SYMBOL(snd_sof_dsp_update_bits_forced);
+
+/**
+ * snd_sof_dsp_panic - handle a received DSP panic message
+ * @sdev: Pointer to the device's sdev
+ * @offset: offset of panic information
+ * @non_recoverable: the panic is fatal, no recovery will be done by the caller
+ */
+void snd_sof_dsp_panic(struct snd_sof_dev *sdev, u32 offset, bool non_recoverable)
+{
+	/*
+	 * if DSP is not ready and the dsp_oops_offset is not yet set, use the
+	 * offset from the panic message.
+	 */
+	if (!sdev->dsp_oops_offset)
+		sdev->dsp_oops_offset = offset;
+
+	/*
+	 * Print warning if the offset from the panic message differs from
+	 * dsp_oops_offset
+	 */
+	if (sdev->dsp_oops_offset != offset)
+		dev_warn(sdev->dev,
+			 "%s: dsp_oops_offset %zu differs from panic offset %u\n",
+			 __func__, sdev->dsp_oops_offset, offset);
+
+	/*
+	 * Set the fw_state to crashed only in case of non recoverable DSP panic
+	 * event.
+	 * Use different message within the snd_sof_dsp_dbg_dump() depending on
+	 * the non_recoverable flag.
+	 */
+	sdev->dbg_dump_printed = false;
+	if (non_recoverable) {
+		snd_sof_dsp_dbg_dump(sdev, "DSP panic!",
+				     SOF_DBG_DUMP_REGS | SOF_DBG_DUMP_MBOX);
+		sof_set_fw_state(sdev, SOF_FW_CRASHED);
+		sof_fw_trace_fw_crashed(sdev);
+	} else {
+		snd_sof_dsp_dbg_dump(sdev,
+				     "DSP panic (recovery will be attempted)",
+				     SOF_DBG_DUMP_REGS | SOF_DBG_DUMP_MBOX);
+	}
+}
+EXPORT_SYMBOL(snd_sof_dsp_panic);