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, 189 insertions, 0 deletions

diff --git a/drivers/firmware/imx/imx-dsp.c b/drivers/firmware/imx/imx-dsp.c
new file mode 100644
index 000000000..ed79e8231
--- /dev/null
+++ b/drivers/firmware/imx/imx-dsp.c
@@ -0,0 +1,189 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2019 NXP
+ *  Author: Daniel Baluta <daniel.baluta@nxp.com>
+ *
+ * Implementation of the DSP IPC interface (host side)
+ */
+
+#include <linux/firmware/imx/dsp.h>
+#include <linux/kernel.h>
+#include <linux/mailbox_client.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+
+/*
+ * imx_dsp_ring_doorbell - triggers an interrupt on the other side (DSP)
+ *
+ * @dsp: DSP IPC handle
+ * @chan_idx: index of the channel where to trigger the interrupt
+ *
+ * Returns non-negative value for success, negative value for error
+ */
+int imx_dsp_ring_doorbell(struct imx_dsp_ipc *ipc, unsigned int idx)
+{
+	int ret;
+	struct imx_dsp_chan *dsp_chan;
+
+	if (idx >= DSP_MU_CHAN_NUM)
+		return -EINVAL;
+
+	dsp_chan = &ipc->chans[idx];
+	ret = mbox_send_message(dsp_chan->ch, NULL);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+EXPORT_SYMBOL(imx_dsp_ring_doorbell);
+
+/*
+ * imx_dsp_handle_rx - rx callback used by imx mailbox
+ *
+ * @c: mbox client
+ * @msg: message received
+ *
+ * Users of DSP IPC will need to privde handle_reply and handle_request
+ * callbacks.
+ */
+static void imx_dsp_handle_rx(struct mbox_client *c, void *msg)
+{
+	struct imx_dsp_chan *chan = container_of(c, struct imx_dsp_chan, cl);
+
+	if (chan->idx == 0) {
+		chan->ipc->ops->handle_reply(chan->ipc);
+	} else {
+		chan->ipc->ops->handle_request(chan->ipc);
+		imx_dsp_ring_doorbell(chan->ipc, 1);
+	}
+}
+
+struct mbox_chan *imx_dsp_request_channel(struct imx_dsp_ipc *dsp_ipc, int idx)
+{
+	struct imx_dsp_chan *dsp_chan;
+
+	if (idx >= DSP_MU_CHAN_NUM)
+		return ERR_PTR(-EINVAL);
+
+	dsp_chan = &dsp_ipc->chans[idx];
+	dsp_chan->ch = mbox_request_channel_byname(&dsp_chan->cl, dsp_chan->name);
+	return dsp_chan->ch;
+}
+EXPORT_SYMBOL(imx_dsp_request_channel);
+
+void imx_dsp_free_channel(struct imx_dsp_ipc *dsp_ipc, int idx)
+{
+	struct imx_dsp_chan *dsp_chan;
+
+	if (idx >= DSP_MU_CHAN_NUM)
+		return;
+
+	dsp_chan = &dsp_ipc->chans[idx];
+	mbox_free_channel(dsp_chan->ch);
+}
+EXPORT_SYMBOL(imx_dsp_free_channel);
+
+static int imx_dsp_setup_channels(struct imx_dsp_ipc *dsp_ipc)
+{
+	struct device *dev = dsp_ipc->dev;
+	struct imx_dsp_chan *dsp_chan;
+	struct mbox_client *cl;
+	char *chan_name;
+	int ret;
+	int i, j;
+
+	for (i = 0; i < DSP_MU_CHAN_NUM; i++) {
+		if (i < 2)
+			chan_name = kasprintf(GFP_KERNEL, "txdb%d", i);
+		else
+			chan_name = kasprintf(GFP_KERNEL, "rxdb%d", i - 2);
+
+		if (!chan_name)
+			return -ENOMEM;
+
+		dsp_chan = &dsp_ipc->chans[i];
+		dsp_chan->name = chan_name;
+		cl = &dsp_chan->cl;
+		cl->dev = dev;
+		cl->tx_block = false;
+		cl->knows_txdone = true;
+		cl->rx_callback = imx_dsp_handle_rx;
+
+		dsp_chan->ipc = dsp_ipc;
+		dsp_chan->idx = i % 2;
+		dsp_chan->ch = mbox_request_channel_byname(cl, chan_name);
+		if (IS_ERR(dsp_chan->ch)) {
+			ret = PTR_ERR(dsp_chan->ch);
+			if (ret != -EPROBE_DEFER)
+				dev_err(dev, "Failed to request mbox chan %s ret %d\n",
+					chan_name, ret);
+			kfree(dsp_chan->name);
+			goto out;
+		}
+
+		dev_dbg(dev, "request mbox chan %s\n", chan_name);
+	}
+
+	return 0;
+out:
+	for (j = 0; j < i; j++) {
+		dsp_chan = &dsp_ipc->chans[j];
+		mbox_free_channel(dsp_chan->ch);
+		kfree(dsp_chan->name);
+	}
+
+	return ret;
+}
+
+static int imx_dsp_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct imx_dsp_ipc *dsp_ipc;
+	int ret;
+
+	device_set_of_node_from_dev(&pdev->dev, pdev->dev.parent);
+
+	dsp_ipc = devm_kzalloc(dev, sizeof(*dsp_ipc), GFP_KERNEL);
+	if (!dsp_ipc)
+		return -ENOMEM;
+
+	dsp_ipc->dev = dev;
+	dev_set_drvdata(dev, dsp_ipc);
+
+	ret = imx_dsp_setup_channels(dsp_ipc);
+	if (ret < 0)
+		return ret;
+
+	dev_info(dev, "NXP i.MX DSP IPC initialized\n");
+
+	return 0;
+}
+
+static void imx_dsp_remove(struct platform_device *pdev)
+{
+	struct imx_dsp_chan *dsp_chan;
+	struct imx_dsp_ipc *dsp_ipc;
+	int i;
+
+	dsp_ipc = dev_get_drvdata(&pdev->dev);
+
+	for (i = 0; i < DSP_MU_CHAN_NUM; i++) {
+		dsp_chan = &dsp_ipc->chans[i];
+		mbox_free_channel(dsp_chan->ch);
+		kfree(dsp_chan->name);
+	}
+}
+
+static struct platform_driver imx_dsp_driver = {
+	.driver = {
+		.name = "imx-dsp",
+	},
+	.probe = imx_dsp_probe,
+	.remove = imx_dsp_remove,
+};
+builtin_platform_driver(imx_dsp_driver);
+
+MODULE_AUTHOR("Daniel Baluta <daniel.baluta@nxp.com>");
+MODULE_DESCRIPTION("IMX DSP IPC protocol driver");
+MODULE_LICENSE("GPL v2");