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, 276 insertions, 0 deletions

diff --git a/sound/firewire/fireface/ff-stream.c b/sound/firewire/fireface/ff-stream.c
new file mode 100644
index 000000000..ba42490f2
--- /dev/null
+++ b/sound/firewire/fireface/ff-stream.c
@@ -0,0 +1,276 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * ff-stream.c - a part of driver for RME Fireface series
+ *
+ * Copyright (c) 2015-2017 Takashi Sakamoto
+ */
+
+#include "ff.h"
+
+#define READY_TIMEOUT_MS	200
+
+int snd_ff_stream_get_multiplier_mode(enum cip_sfc sfc,
+				      enum snd_ff_stream_mode *mode)
+{
+	static const enum snd_ff_stream_mode modes[] = {
+		[CIP_SFC_32000] = SND_FF_STREAM_MODE_LOW,
+		[CIP_SFC_44100] = SND_FF_STREAM_MODE_LOW,
+		[CIP_SFC_48000] = SND_FF_STREAM_MODE_LOW,
+		[CIP_SFC_88200] = SND_FF_STREAM_MODE_MID,
+		[CIP_SFC_96000] = SND_FF_STREAM_MODE_MID,
+		[CIP_SFC_176400] = SND_FF_STREAM_MODE_HIGH,
+		[CIP_SFC_192000] = SND_FF_STREAM_MODE_HIGH,
+	};
+
+	if (sfc >= CIP_SFC_COUNT)
+		return -EINVAL;
+
+	*mode = modes[sfc];
+
+	return 0;
+}
+
+static inline void finish_session(struct snd_ff *ff)
+{
+	ff->spec->protocol->finish_session(ff);
+	ff->spec->protocol->switch_fetching_mode(ff, false);
+}
+
+static int init_stream(struct snd_ff *ff, struct amdtp_stream *s)
+{
+	struct fw_iso_resources *resources;
+	enum amdtp_stream_direction dir;
+	int err;
+
+	if (s == &ff->tx_stream) {
+		resources = &ff->tx_resources;
+		dir = AMDTP_IN_STREAM;
+	} else {
+		resources = &ff->rx_resources;
+		dir = AMDTP_OUT_STREAM;
+	}
+
+	err = fw_iso_resources_init(resources, ff->unit);
+	if (err < 0)
+		return err;
+
+	err = amdtp_ff_init(s, ff->unit, dir);
+	if (err < 0)
+		fw_iso_resources_destroy(resources);
+
+	return err;
+}
+
+static void destroy_stream(struct snd_ff *ff, struct amdtp_stream *s)
+{
+	amdtp_stream_destroy(s);
+
+	if (s == &ff->tx_stream)
+		fw_iso_resources_destroy(&ff->tx_resources);
+	else
+		fw_iso_resources_destroy(&ff->rx_resources);
+}
+
+int snd_ff_stream_init_duplex(struct snd_ff *ff)
+{
+	int err;
+
+	err = init_stream(ff, &ff->rx_stream);
+	if (err < 0)
+		return err;
+
+	err = init_stream(ff, &ff->tx_stream);
+	if (err < 0) {
+		destroy_stream(ff, &ff->rx_stream);
+		return err;
+	}
+
+	err = amdtp_domain_init(&ff->domain);
+	if (err < 0) {
+		destroy_stream(ff, &ff->rx_stream);
+		destroy_stream(ff, &ff->tx_stream);
+	}
+
+	return err;
+}
+
+/*
+ * This function should be called before starting streams or after stopping
+ * streams.
+ */
+void snd_ff_stream_destroy_duplex(struct snd_ff *ff)
+{
+	amdtp_domain_destroy(&ff->domain);
+
+	destroy_stream(ff, &ff->rx_stream);
+	destroy_stream(ff, &ff->tx_stream);
+}
+
+int snd_ff_stream_reserve_duplex(struct snd_ff *ff, unsigned int rate,
+				 unsigned int frames_per_period,
+				 unsigned int frames_per_buffer)
+{
+	unsigned int curr_rate;
+	enum snd_ff_clock_src src;
+	int err;
+
+	err = ff->spec->protocol->get_clock(ff, &curr_rate, &src);
+	if (err < 0)
+		return err;
+
+	if (ff->substreams_counter == 0 || curr_rate != rate) {
+		enum snd_ff_stream_mode mode;
+		int i;
+
+		amdtp_domain_stop(&ff->domain);
+		finish_session(ff);
+
+		fw_iso_resources_free(&ff->tx_resources);
+		fw_iso_resources_free(&ff->rx_resources);
+
+		for (i = 0; i < CIP_SFC_COUNT; ++i) {
+			if (amdtp_rate_table[i] == rate)
+				break;
+		}
+		if (i >= CIP_SFC_COUNT)
+			return -EINVAL;
+
+		err = snd_ff_stream_get_multiplier_mode(i, &mode);
+		if (err < 0)
+			return err;
+
+		err = amdtp_ff_set_parameters(&ff->tx_stream, rate,
+					ff->spec->pcm_capture_channels[mode]);
+		if (err < 0)
+			return err;
+
+		err = amdtp_ff_set_parameters(&ff->rx_stream, rate,
+					ff->spec->pcm_playback_channels[mode]);
+		if (err < 0)
+			return err;
+
+		err = ff->spec->protocol->allocate_resources(ff, rate);
+		if (err < 0)
+			return err;
+
+		err = amdtp_domain_set_events_per_period(&ff->domain,
+					frames_per_period, frames_per_buffer);
+		if (err < 0) {
+			fw_iso_resources_free(&ff->tx_resources);
+			fw_iso_resources_free(&ff->rx_resources);
+			return err;
+		}
+	}
+
+	return 0;
+}
+
+int snd_ff_stream_start_duplex(struct snd_ff *ff, unsigned int rate)
+{
+	int err;
+
+	if (ff->substreams_counter == 0)
+		return 0;
+
+	if (amdtp_streaming_error(&ff->tx_stream) ||
+	    amdtp_streaming_error(&ff->rx_stream)) {
+		amdtp_domain_stop(&ff->domain);
+		finish_session(ff);
+	}
+
+	/*
+	 * Regardless of current source of clock signal, drivers transfer some
+	 * packets. Then, the device transfers packets.
+	 */
+	if (!amdtp_stream_running(&ff->rx_stream)) {
+		int spd = fw_parent_device(ff->unit)->max_speed;
+
+		err = ff->spec->protocol->begin_session(ff, rate);
+		if (err < 0)
+			goto error;
+
+		err = amdtp_domain_add_stream(&ff->domain, &ff->rx_stream,
+					      ff->rx_resources.channel, spd);
+		if (err < 0)
+			goto error;
+
+		err = amdtp_domain_add_stream(&ff->domain, &ff->tx_stream,
+					      ff->tx_resources.channel, spd);
+		if (err < 0)
+			goto error;
+
+		// NOTE: The device doesn't transfer packets unless receiving any packet. The
+		// sequence of tx packets includes cycle skip corresponding to empty packet or
+		// NODATA packet in IEC 61883-1/6. The sequence of the number of data blocks per
+		// packet is important for media clock recovery.
+		err = amdtp_domain_start(&ff->domain, 0, true, true);
+		if (err < 0)
+			goto error;
+
+		if (!amdtp_domain_wait_ready(&ff->domain, READY_TIMEOUT_MS)) {
+			err = -ETIMEDOUT;
+			goto error;
+		}
+
+		err = ff->spec->protocol->switch_fetching_mode(ff, true);
+		if (err < 0)
+			goto error;
+	}
+
+	return 0;
+error:
+	amdtp_domain_stop(&ff->domain);
+	finish_session(ff);
+
+	return err;
+}
+
+void snd_ff_stream_stop_duplex(struct snd_ff *ff)
+{
+	if (ff->substreams_counter == 0) {
+		amdtp_domain_stop(&ff->domain);
+		finish_session(ff);
+
+		fw_iso_resources_free(&ff->tx_resources);
+		fw_iso_resources_free(&ff->rx_resources);
+	}
+}
+
+void snd_ff_stream_update_duplex(struct snd_ff *ff)
+{
+	amdtp_domain_stop(&ff->domain);
+
+	// The device discontinue to transfer packets.
+	amdtp_stream_pcm_abort(&ff->tx_stream);
+	amdtp_stream_pcm_abort(&ff->rx_stream);
+}
+
+void snd_ff_stream_lock_changed(struct snd_ff *ff)
+{
+	ff->dev_lock_changed = true;
+	wake_up(&ff->hwdep_wait);
+}
+
+int snd_ff_stream_lock_try(struct snd_ff *ff)
+{
+	guard(spinlock_irq)(&ff->lock);
+
+	/* user land lock this */
+	if (ff->dev_lock_count < 0)
+		return -EBUSY;
+
+	/* this is the first time */
+	if (ff->dev_lock_count++ == 0)
+		snd_ff_stream_lock_changed(ff);
+	return 0;
+}
+
+void snd_ff_stream_lock_release(struct snd_ff *ff)
+{
+	guard(spinlock_irq)(&ff->lock);
+
+	if (WARN_ON(ff->dev_lock_count <= 0))
+		return;
+	if (--ff->dev_lock_count == 0)
+		snd_ff_stream_lock_changed(ff);
+}