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, 196 insertions, 0 deletions

diff --git a/drivers/spi/spi-sh-sci.c b/drivers/spi/spi-sh-sci.c
new file mode 100644
index 000000000..f66efaabc
--- /dev/null
+++ b/drivers/spi/spi-sh-sci.c
@@ -0,0 +1,196 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * SH SCI SPI interface
+ *
+ * Copyright (c) 2008 Magnus Damm
+ *
+ * Based on S3C24XX GPIO based SPI driver, which is:
+ *   Copyright (c) 2006 Ben Dooks
+ *   Copyright (c) 2006 Simtec Electronics
+ */
+
+#include <linux/kernel.h>
+#include <linux/delay.h>
+#include <linux/spinlock.h>
+#include <linux/platform_device.h>
+
+#include <linux/spi/spi.h>
+#include <linux/spi/spi_bitbang.h>
+#include <linux/module.h>
+
+#include <asm/spi.h>
+#include <asm/io.h>
+
+struct sh_sci_spi {
+	struct spi_bitbang bitbang;
+
+	void __iomem *membase;
+	unsigned char val;
+	struct sh_spi_info *info;
+	struct platform_device *dev;
+};
+
+#define SCSPTR(sp)	(sp->membase + 0x1c)
+#define PIN_SCK		(1 << 2)
+#define PIN_TXD		(1 << 0)
+#define PIN_RXD		PIN_TXD
+#define PIN_INIT	((1 << 1) | (1 << 3) | PIN_SCK | PIN_TXD)
+
+static inline void setbits(struct sh_sci_spi *sp, int bits, int on)
+{
+	/*
+	 * We are the only user of SCSPTR so no locking is required.
+	 * Reading bit 2 and 0 in SCSPTR gives pin state as input.
+	 * Writing the same bits sets the output value.
+	 * This makes regular read-modify-write difficult so we
+	 * use sp->val to keep track of the latest register value.
+	 */
+
+	if (on)
+		sp->val |= bits;
+	else
+		sp->val &= ~bits;
+
+	iowrite8(sp->val, SCSPTR(sp));
+}
+
+static inline void setsck(struct spi_device *dev, int on)
+{
+	setbits(spi_controller_get_devdata(dev->controller), PIN_SCK, on);
+}
+
+static inline void setmosi(struct spi_device *dev, int on)
+{
+	setbits(spi_controller_get_devdata(dev->controller), PIN_TXD, on);
+}
+
+static inline u32 getmiso(struct spi_device *dev)
+{
+	struct sh_sci_spi *sp = spi_controller_get_devdata(dev->controller);
+
+	return (ioread8(SCSPTR(sp)) & PIN_RXD) ? 1 : 0;
+}
+
+#define spidelay(x) ndelay(x)
+
+#include "spi-bitbang-txrx.h"
+
+static u32 sh_sci_spi_txrx_mode0(struct spi_device *spi,
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
+{
+	return bitbang_txrx_be_cpha0(spi, nsecs, 0, flags, word, bits);
+}
+
+static u32 sh_sci_spi_txrx_mode1(struct spi_device *spi,
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
+{
+	return bitbang_txrx_be_cpha1(spi, nsecs, 0, flags, word, bits);
+}
+
+static u32 sh_sci_spi_txrx_mode2(struct spi_device *spi,
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
+{
+	return bitbang_txrx_be_cpha0(spi, nsecs, 1, flags, word, bits);
+}
+
+static u32 sh_sci_spi_txrx_mode3(struct spi_device *spi,
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
+{
+	return bitbang_txrx_be_cpha1(spi, nsecs, 1, flags, word, bits);
+}
+
+static void sh_sci_spi_chipselect(struct spi_device *dev, int value)
+{
+	struct sh_sci_spi *sp = spi_controller_get_devdata(dev->controller);
+
+	if (sp->info->chip_select)
+		(sp->info->chip_select)(sp->info, spi_get_chipselect(dev, 0), value);
+}
+
+static int sh_sci_spi_probe(struct platform_device *dev)
+{
+	struct resource	*r;
+	struct spi_controller *host;
+	struct sh_sci_spi *sp;
+	int ret;
+
+	host = spi_alloc_host(&dev->dev, sizeof(struct sh_sci_spi));
+	if (host == NULL) {
+		dev_err(&dev->dev, "failed to allocate spi host\n");
+		ret = -ENOMEM;
+		goto err0;
+	}
+
+	sp = spi_controller_get_devdata(host);
+
+	platform_set_drvdata(dev, sp);
+	sp->info = dev_get_platdata(&dev->dev);
+	if (!sp->info) {
+		dev_err(&dev->dev, "platform data is missing\n");
+		ret = -ENOENT;
+		goto err1;
+	}
+
+	/* setup spi bitbang adaptor */
+	sp->bitbang.ctlr = host;
+	sp->bitbang.ctlr->bus_num = sp->info->bus_num;
+	sp->bitbang.ctlr->num_chipselect = sp->info->num_chipselect;
+	sp->bitbang.chipselect = sh_sci_spi_chipselect;
+
+	sp->bitbang.txrx_word[SPI_MODE_0] = sh_sci_spi_txrx_mode0;
+	sp->bitbang.txrx_word[SPI_MODE_1] = sh_sci_spi_txrx_mode1;
+	sp->bitbang.txrx_word[SPI_MODE_2] = sh_sci_spi_txrx_mode2;
+	sp->bitbang.txrx_word[SPI_MODE_3] = sh_sci_spi_txrx_mode3;
+
+	r = platform_get_resource(dev, IORESOURCE_MEM, 0);
+	if (r == NULL) {
+		ret = -ENOENT;
+		goto err1;
+	}
+	sp->membase = ioremap(r->start, resource_size(r));
+	if (!sp->membase) {
+		ret = -ENXIO;
+		goto err1;
+	}
+	sp->val = ioread8(SCSPTR(sp));
+	setbits(sp, PIN_INIT, 1);
+
+	ret = spi_bitbang_start(&sp->bitbang);
+	if (!ret)
+		return 0;
+
+	setbits(sp, PIN_INIT, 0);
+	iounmap(sp->membase);
+ err1:
+	spi_controller_put(sp->bitbang.ctlr);
+ err0:
+	return ret;
+}
+
+static void sh_sci_spi_remove(struct platform_device *dev)
+{
+	struct sh_sci_spi *sp = platform_get_drvdata(dev);
+
+	spi_bitbang_stop(&sp->bitbang);
+	setbits(sp, PIN_INIT, 0);
+	iounmap(sp->membase);
+	spi_controller_put(sp->bitbang.ctlr);
+}
+
+static struct platform_driver sh_sci_spi_drv = {
+	.probe		= sh_sci_spi_probe,
+	.remove		= sh_sci_spi_remove,
+	.driver		= {
+		.name	= "spi_sh_sci",
+	},
+};
+module_platform_driver(sh_sci_spi_drv);
+
+MODULE_DESCRIPTION("SH SCI SPI Driver");
+MODULE_AUTHOR("Magnus Damm <damm@opensource.se>");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("platform:spi_sh_sci");