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, 144 insertions, 0 deletions

diff --git a/drivers/acpi/acpi_pcc.c b/drivers/acpi/acpi_pcc.c
new file mode 100644
index 000000000..438c67189
--- /dev/null
+++ b/drivers/acpi/acpi_pcc.c
@@ -0,0 +1,144 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Author: Sudeep Holla <sudeep.holla@arm.com>
+ * Copyright 2021 Arm Limited
+ *
+ * The PCC Address Space also referred as PCC Operation Region pertains to the
+ * region of PCC subspace that succeeds the PCC signature. The PCC Operation
+ * Region works in conjunction with the PCC Table(Platform Communications
+ * Channel Table). PCC subspaces that are marked for use as PCC Operation
+ * Regions must not be used as PCC subspaces for the standard ACPI features
+ * such as CPPC, RASF, PDTT and MPST. These standard features must always use
+ * the PCC Table instead.
+ *
+ * This driver sets up the PCC Address Space and installs an handler to enable
+ * handling of PCC OpRegion in the firmware.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/acpi.h>
+#include <linux/completion.h>
+#include <linux/idr.h>
+#include <linux/io.h>
+
+#include <acpi/pcc.h>
+
+/*
+ * Arbitrary retries in case the remote processor is slow to respond
+ * to PCC commands
+ */
+#define PCC_CMD_WAIT_RETRIES_NUM	500ULL
+
+struct pcc_data {
+	struct pcc_mbox_chan *pcc_chan;
+	struct completion done;
+	struct mbox_client cl;
+	struct acpi_pcc_info ctx;
+};
+
+static struct acpi_pcc_info pcc_ctx;
+
+static void pcc_rx_callback(struct mbox_client *cl, void *m)
+{
+	struct pcc_data *data = container_of(cl, struct pcc_data, cl);
+
+	complete(&data->done);
+}
+
+static acpi_status
+acpi_pcc_address_space_setup(acpi_handle region_handle, u32 function,
+			     void *handler_context,  void **region_context)
+{
+	struct pcc_data *data;
+	struct acpi_pcc_info *ctx = handler_context;
+	struct pcc_mbox_chan *pcc_chan;
+	acpi_status ret;
+
+	data = kzalloc_obj(*data);
+	if (!data)
+		return AE_NO_MEMORY;
+
+	data->cl.rx_callback = pcc_rx_callback;
+	data->cl.knows_txdone = true;
+	data->ctx.length = ctx->length;
+	data->ctx.subspace_id = ctx->subspace_id;
+	data->ctx.internal_buffer = ctx->internal_buffer;
+
+	init_completion(&data->done);
+	data->pcc_chan = pcc_mbox_request_channel(&data->cl, ctx->subspace_id);
+	if (IS_ERR(data->pcc_chan)) {
+		pr_err("Failed to find PCC channel for subspace %d\n",
+		       ctx->subspace_id);
+		ret = AE_NOT_FOUND;
+		goto err_free_data;
+	}
+
+	pcc_chan = data->pcc_chan;
+	if (!pcc_chan->mchan->mbox->txdone_irq) {
+		pr_err("This channel-%d does not support interrupt.\n",
+		       ctx->subspace_id);
+		ret = AE_SUPPORT;
+		goto err_free_channel;
+	}
+
+	*region_context = data;
+	return AE_OK;
+
+err_free_channel:
+	pcc_mbox_free_channel(data->pcc_chan);
+err_free_data:
+	kfree(data);
+
+	return ret;
+}
+
+static acpi_status
+acpi_pcc_address_space_handler(u32 function, acpi_physical_address addr,
+			       u32 bits, acpi_integer *value,
+			       void *handler_context, void *region_context)
+{
+	int ret;
+	struct pcc_data *data = region_context;
+	u64 usecs_lat;
+
+	reinit_completion(&data->done);
+
+	/* Write to Shared Memory */
+	memcpy_toio(data->pcc_chan->shmem, (void *)value, data->ctx.length);
+
+	ret = mbox_send_message(data->pcc_chan->mchan, NULL);
+	if (ret < 0)
+		return AE_ERROR;
+
+	/*
+	 * pcc_chan->latency is just a Nominal value. In reality the remote
+	 * processor could be much slower to reply. So add an arbitrary
+	 * amount of wait on top of Nominal.
+	 */
+	usecs_lat = PCC_CMD_WAIT_RETRIES_NUM * data->pcc_chan->latency;
+	ret = wait_for_completion_timeout(&data->done,
+						usecs_to_jiffies(usecs_lat));
+	if (ret == 0) {
+		pr_err("PCC command executed timeout!\n");
+		return AE_TIME;
+	}
+
+	mbox_chan_txdone(data->pcc_chan->mchan, ret);
+
+	memcpy_fromio(value, data->pcc_chan->shmem, data->ctx.length);
+
+	return AE_OK;
+}
+
+void __init acpi_init_pcc(void)
+{
+	acpi_status status;
+
+	status = acpi_install_address_space_handler(ACPI_ROOT_OBJECT,
+						    ACPI_ADR_SPACE_PLATFORM_COMM,
+						    &acpi_pcc_address_space_handler,
+						    &acpi_pcc_address_space_setup,
+						    &pcc_ctx);
+	if (ACPI_FAILURE(status))
+		pr_alert("OperationRegion handler could not be installed\n");
+}