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, 295 insertions, 0 deletions

diff --git a/arch/mips/sni/pcit.c b/arch/mips/sni/pcit.c
new file mode 100644
index 000000000..b331fe22c
--- /dev/null
+++ b/arch/mips/sni/pcit.c
@@ -0,0 +1,295 @@
+/*
+ * PCI Tower specific code
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2006 Thomas Bogendoerfer (tsbogend@alpha.franken.de)
+ */
+
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+#include <linux/pci.h>
+#include <linux/serial_8250.h>
+
+#include <asm/sni.h>
+#include <asm/time.h>
+#include <asm/irq_cpu.h>
+
+
+#define PORT(_base,_irq)				\
+	{						\
+		.iobase		= _base,		\
+		.irq		= _irq,			\
+		.uartclk	= 1843200,		\
+		.iotype		= UPIO_PORT,		\
+		.flags		= UPF_BOOT_AUTOCONF,	\
+	}
+
+static struct plat_serial8250_port pcit_data[] = {
+	PORT(0x3f8, 0),
+	PORT(0x2f8, 3),
+	{ },
+};
+
+static struct platform_device pcit_serial8250_device = {
+	.name			= "serial8250",
+	.id			= PLAT8250_DEV_PLATFORM,
+	.dev			= {
+		.platform_data	= pcit_data,
+	},
+};
+
+static struct plat_serial8250_port pcit_cplus_data[] = {
+	PORT(0x3f8, 0),
+	PORT(0x2f8, 3),
+	PORT(0x3e8, 4),
+	PORT(0x2e8, 3),
+	{ },
+};
+
+static struct platform_device pcit_cplus_serial8250_device = {
+	.name			= "serial8250",
+	.id			= PLAT8250_DEV_PLATFORM,
+	.dev			= {
+		.platform_data	= pcit_cplus_data,
+	},
+};
+
+static struct resource pcit_cmos_rsrc[] = {
+	{
+		.start = 0x70,
+		.end   = 0x71,
+		.flags = IORESOURCE_IO
+	},
+	{
+		.start = 8,
+		.end   = 8,
+		.flags = IORESOURCE_IRQ
+	}
+};
+
+static struct platform_device pcit_cmos_device = {
+	.name		= "rtc_cmos",
+	.num_resources	= ARRAY_SIZE(pcit_cmos_rsrc),
+	.resource	= pcit_cmos_rsrc
+};
+
+static struct platform_device pcit_pcspeaker_pdev = {
+	.name		= "pcspkr",
+	.id		= -1,
+};
+
+static struct resource sni_io_resource = {
+	.start	= 0x00000000UL,
+	.end	= 0x03bfffffUL,
+	.name	= "PCIT IO",
+	.flags	= IORESOURCE_IO,
+};
+
+static struct resource pcit_io_resources[] = {
+	{
+		.start	= 0x00,
+		.end	= 0x1f,
+		.name	= "dma1",
+		.flags	= IORESOURCE_BUSY
+	}, {
+		.start	=  0x40,
+		.end	= 0x5f,
+		.name	= "timer",
+		.flags	= IORESOURCE_BUSY
+	}, {
+		.start	=  0x60,
+		.end	= 0x6f,
+		.name	= "keyboard",
+		.flags	= IORESOURCE_BUSY
+	}, {
+		.start	=  0x80,
+		.end	= 0x8f,
+		.name	= "dma page reg",
+		.flags	= IORESOURCE_BUSY
+	}, {
+		.start	=  0xc0,
+		.end	= 0xdf,
+		.name	= "dma2",
+		.flags	= IORESOURCE_BUSY
+	}, {
+		.start	=  0xcf8,
+		.end	= 0xcfb,
+		.name	= "PCI config addr",
+		.flags	= IORESOURCE_BUSY
+	}, {
+		.start	=  0xcfc,
+		.end	= 0xcff,
+		.name	= "PCI config data",
+		.flags	= IORESOURCE_BUSY
+	}
+};
+
+static void __init sni_pcit_resource_init(void)
+{
+	int i;
+
+	/* request I/O space for devices used on all i[345]86 PCs */
+	for (i = 0; i < ARRAY_SIZE(pcit_io_resources); i++)
+		request_resource(&sni_io_resource, pcit_io_resources + i);
+}
+
+
+extern struct pci_ops sni_pcit_ops;
+
+#ifdef CONFIG_PCI
+static struct resource sni_mem_resource = {
+	.start	= 0x18000000UL,
+	.end	= 0x1fbfffffUL,
+	.name	= "PCIT PCI MEM",
+	.flags	= IORESOURCE_MEM
+};
+
+static struct pci_controller sni_pcit_controller = {
+	.pci_ops	= &sni_pcit_ops,
+	.mem_resource	= &sni_mem_resource,
+	.mem_offset	= 0x00000000UL,
+	.io_resource	= &sni_io_resource,
+	.io_offset	= 0x00000000UL,
+	.io_map_base	= SNI_PORT_BASE
+};
+#endif /* CONFIG_PCI */
+
+static void enable_pcit_irq(struct irq_data *d)
+{
+	u32 mask = 1 << (d->irq - SNI_PCIT_INT_START + 24);
+
+	*(volatile u32 *)SNI_PCIT_INT_REG |= mask;
+}
+
+void disable_pcit_irq(struct irq_data *d)
+{
+	u32 mask = 1 << (d->irq - SNI_PCIT_INT_START + 24);
+
+	*(volatile u32 *)SNI_PCIT_INT_REG &= ~mask;
+}
+
+static struct irq_chip pcit_irq_type = {
+	.name = "PCIT",
+	.irq_mask = disable_pcit_irq,
+	.irq_unmask = enable_pcit_irq,
+};
+
+static void pcit_hwint1(void)
+{
+	u32 pending = *(volatile u32 *)SNI_PCIT_INT_REG;
+	int irq;
+
+	clear_c0_status(IE_IRQ1);
+	irq = ffs((pending >> 16) & 0x7f);
+
+	if (likely(irq > 0))
+		do_IRQ(irq + SNI_PCIT_INT_START - 1);
+	set_c0_status(IE_IRQ1);
+}
+
+static void pcit_hwint0(void)
+{
+	u32 pending = *(volatile u32 *)SNI_PCIT_INT_REG;
+	int irq;
+
+	clear_c0_status(IE_IRQ0);
+	irq = ffs((pending >> 16) & 0x3f);
+
+	if (likely(irq > 0))
+		do_IRQ(irq + SNI_PCIT_INT_START - 1);
+	set_c0_status(IE_IRQ0);
+}
+
+static void sni_pcit_hwint(void)
+{
+	u32 pending = read_c0_cause() & read_c0_status();
+
+	if (pending & C_IRQ1)
+		pcit_hwint1();
+	else if (pending & C_IRQ2)
+		do_IRQ(MIPS_CPU_IRQ_BASE + 4);
+	else if (pending & C_IRQ3)
+		do_IRQ(MIPS_CPU_IRQ_BASE + 5);
+	else if (pending & C_IRQ5)
+		do_IRQ(MIPS_CPU_IRQ_BASE + 7);
+}
+
+static void sni_pcit_hwint_cplus(void)
+{
+	u32 pending = read_c0_cause() & read_c0_status();
+
+	if (pending & C_IRQ0)
+		pcit_hwint0();
+	else if (pending & C_IRQ1)
+		do_IRQ(MIPS_CPU_IRQ_BASE + 3);
+	else if (pending & C_IRQ2)
+		do_IRQ(MIPS_CPU_IRQ_BASE + 4);
+	else if (pending & C_IRQ3)
+		do_IRQ(MIPS_CPU_IRQ_BASE + 5);
+	else if (pending & C_IRQ5)
+		do_IRQ(MIPS_CPU_IRQ_BASE + 7);
+}
+
+void __init sni_pcit_irq_init(void)
+{
+	int i;
+
+	mips_cpu_irq_init();
+	for (i = SNI_PCIT_INT_START; i <= SNI_PCIT_INT_END; i++)
+		irq_set_chip_and_handler(i, &pcit_irq_type, handle_level_irq);
+	*(volatile u32 *)SNI_PCIT_INT_REG = 0;
+	sni_hwint = sni_pcit_hwint;
+	change_c0_status(ST0_IM, IE_IRQ1);
+	if (request_irq(SNI_PCIT_INT_START + 6, sni_isa_irq_handler, 0, "ISA",
+			NULL))
+		pr_err("Failed to register ISA interrupt\n");
+}
+
+void __init sni_pcit_cplus_irq_init(void)
+{
+	int i;
+
+	mips_cpu_irq_init();
+	for (i = SNI_PCIT_INT_START; i <= SNI_PCIT_INT_END; i++)
+		irq_set_chip_and_handler(i, &pcit_irq_type, handle_level_irq);
+	*(volatile u32 *)SNI_PCIT_INT_REG = 0x40000000;
+	sni_hwint = sni_pcit_hwint_cplus;
+	change_c0_status(ST0_IM, IE_IRQ0);
+	if (request_irq(MIPS_CPU_IRQ_BASE + 3, sni_isa_irq_handler, 0, "ISA",
+			NULL))
+		pr_err("Failed to register ISA interrupt\n");
+}
+
+void __init sni_pcit_init(void)
+{
+	ioport_resource.end = sni_io_resource.end;
+#ifdef CONFIG_PCI
+	PCIBIOS_MIN_IO = 0x9000;
+	register_pci_controller(&sni_pcit_controller);
+#endif
+	sni_pcit_resource_init();
+}
+
+static int __init snirm_pcit_setup_devinit(void)
+{
+	switch (sni_brd_type) {
+	case SNI_BRD_PCI_TOWER:
+		platform_device_register(&pcit_serial8250_device);
+		platform_device_register(&pcit_cmos_device);
+		platform_device_register(&pcit_pcspeaker_pdev);
+		break;
+
+	case SNI_BRD_PCI_TOWER_CPLUS:
+		platform_device_register(&pcit_cplus_serial8250_device);
+		platform_device_register(&pcit_cmos_device);
+		platform_device_register(&pcit_pcspeaker_pdev);
+		break;
+	}
+	return 0;
+}
+
+device_initcall(snirm_pcit_setup_devinit);