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, 197 insertions, 0 deletions

diff --git a/drivers/video/fbdev/sunxvr1000.c b/drivers/video/fbdev/sunxvr1000.c
new file mode 100644
index 000000000..a93b06614
--- /dev/null
+++ b/drivers/video/fbdev/sunxvr1000.c
@@ -0,0 +1,197 @@
+/* sunxvr1000.c: Sun XVR-1000 fb driver for sparc64 systems
+ *
+ * License: GPL
+ *
+ * Copyright (C) 2010 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/kernel.h>
+#include <linux/fb.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+
+struct gfb_info {
+	struct fb_info		*info;
+
+	char __iomem		*fb_base;
+	unsigned long		fb_base_phys;
+
+	struct device_node	*of_node;
+
+	unsigned int		width;
+	unsigned int		height;
+	unsigned int		depth;
+	unsigned int		fb_size;
+
+	u32			pseudo_palette[16];
+};
+
+static int gfb_get_props(struct gfb_info *gp)
+{
+	gp->width = of_getintprop_default(gp->of_node, "width", 0);
+	gp->height = of_getintprop_default(gp->of_node, "height", 0);
+	gp->depth = of_getintprop_default(gp->of_node, "depth", 32);
+
+	if (!gp->width || !gp->height) {
+		printk(KERN_ERR "gfb: Critical properties missing for %pOF\n",
+		       gp->of_node);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int gfb_setcolreg(unsigned regno,
+			 unsigned red, unsigned green, unsigned blue,
+			 unsigned transp, struct fb_info *info)
+{
+	u32 value;
+
+	if (regno < 16) {
+		red >>= 8;
+		green >>= 8;
+		blue >>= 8;
+
+		value = (blue << 16) | (green << 8) | red;
+		((u32 *)info->pseudo_palette)[regno] = value;
+	}
+
+	return 0;
+}
+
+static const struct fb_ops gfb_ops = {
+	.owner			= THIS_MODULE,
+	FB_DEFAULT_IOMEM_OPS,
+	.fb_setcolreg		= gfb_setcolreg,
+};
+
+static int gfb_set_fbinfo(struct gfb_info *gp)
+{
+	struct fb_info *info = gp->info;
+	struct fb_var_screeninfo *var = &info->var;
+
+	info->fbops = &gfb_ops;
+	info->screen_base = gp->fb_base;
+	info->screen_size = gp->fb_size;
+
+	info->pseudo_palette = gp->pseudo_palette;
+
+	/* Fill fix common fields */
+	strscpy(info->fix.id, "gfb", sizeof(info->fix.id));
+        info->fix.smem_start = gp->fb_base_phys;
+        info->fix.smem_len = gp->fb_size;
+        info->fix.type = FB_TYPE_PACKED_PIXELS;
+	if (gp->depth == 32 || gp->depth == 24)
+		info->fix.visual = FB_VISUAL_TRUECOLOR;
+	else
+		info->fix.visual = FB_VISUAL_PSEUDOCOLOR;
+
+	var->xres = gp->width;
+	var->yres = gp->height;
+	var->xres_virtual = var->xres;
+	var->yres_virtual = var->yres;
+	var->bits_per_pixel = gp->depth;
+
+	var->red.offset = 0;
+	var->red.length = 8;
+	var->green.offset = 8;
+	var->green.length = 8;
+	var->blue.offset = 16;
+	var->blue.length = 8;
+	var->transp.offset = 0;
+	var->transp.length = 0;
+
+	if (fb_alloc_cmap(&info->cmap, 256, 0)) {
+		printk(KERN_ERR "gfb: Cannot allocate color map.\n");
+		return -ENOMEM;
+	}
+
+        return 0;
+}
+
+static int gfb_probe(struct platform_device *op)
+{
+	struct device_node *dp = op->dev.of_node;
+	struct fb_info *info;
+	struct gfb_info *gp;
+	int err;
+
+	info = framebuffer_alloc(sizeof(struct gfb_info), &op->dev);
+	if (!info) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+
+	gp = info->par;
+	gp->info = info;
+	gp->of_node = dp;
+
+	gp->fb_base_phys = op->resource[6].start;
+
+	err = gfb_get_props(gp);
+	if (err)
+		goto err_release_fb;
+
+	/* Framebuffer length is the same regardless of resolution. */
+	info->fix.line_length = 16384;
+	gp->fb_size = info->fix.line_length * gp->height;
+
+	gp->fb_base = of_ioremap(&op->resource[6], 0,
+				 gp->fb_size, "gfb fb");
+	if (!gp->fb_base) {
+		err = -ENOMEM;
+		goto err_release_fb;
+	}
+
+	err = gfb_set_fbinfo(gp);
+	if (err)
+		goto err_unmap_fb;
+
+	printk("gfb: Found device at %pOF\n", dp);
+
+	err = register_framebuffer(info);
+	if (err < 0) {
+		printk(KERN_ERR "gfb: Could not register framebuffer %pOF\n",
+		       dp);
+		goto err_unmap_fb;
+	}
+
+	dev_set_drvdata(&op->dev, info);
+
+	return 0;
+
+err_unmap_fb:
+	of_iounmap(&op->resource[6], gp->fb_base, gp->fb_size);
+
+err_release_fb:
+        framebuffer_release(info);
+
+err_out:
+	return err;
+}
+
+static const struct of_device_id gfb_match[] = {
+	{
+		.name = "SUNW,gfb",
+	},
+	{},
+};
+
+static struct platform_driver gfb_driver = {
+	.probe		= gfb_probe,
+	.driver = {
+		.name			= "gfb",
+		.of_match_table		= gfb_match,
+		.suppress_bind_attrs	= true,
+	},
+};
+
+static int __init gfb_init(void)
+{
+	if (fb_get_options("gfb", NULL))
+		return -ENODEV;
+
+	return platform_driver_register(&gfb_driver);
+}
+device_initcall(gfb_init);