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, 296 insertions, 0 deletions

diff --git a/fs/proc/task_nommu.c b/fs/proc/task_nommu.c
new file mode 100644
index 000000000..d362919f4
--- /dev/null
+++ b/fs/proc/task_nommu.c
@@ -0,0 +1,296 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/mm.h>
+#include <linux/file.h>
+#include <linux/fdtable.h>
+#include <linux/fs_struct.h>
+#include <linux/mount.h>
+#include <linux/ptrace.h>
+#include <linux/slab.h>
+#include <linux/seq_file.h>
+#include <linux/sched/mm.h>
+
+#include "internal.h"
+
+/*
+ * Logic: we've got two memory sums for each process, "shared", and
+ * "non-shared". Shared memory may get counted more than once, for
+ * each process that owns it. Non-shared memory is counted
+ * accurately.
+ */
+void task_mem(struct seq_file *m, struct mm_struct *mm)
+{
+	VMA_ITERATOR(vmi, mm, 0);
+	struct vm_area_struct *vma;
+	struct vm_region *region;
+	unsigned long bytes = 0, sbytes = 0, slack = 0, size;
+
+	mmap_read_lock(mm);
+	for_each_vma(vmi, vma) {
+		bytes += kobjsize(vma);
+
+		region = vma->vm_region;
+		if (region) {
+			size = kobjsize(region);
+			size += region->vm_end - region->vm_start;
+		} else {
+			size = vma->vm_end - vma->vm_start;
+		}
+
+		if (atomic_read(&mm->mm_count) > 1 ||
+		    is_nommu_shared_mapping(vma->vm_flags)) {
+			sbytes += size;
+		} else {
+			bytes += size;
+			if (region)
+				slack = region->vm_end - vma->vm_end;
+		}
+	}
+
+	if (atomic_read(&mm->mm_count) > 1)
+		sbytes += kobjsize(mm);
+	else
+		bytes += kobjsize(mm);
+
+	if (current->fs && current->fs->users > 1)
+		sbytes += kobjsize(current->fs);
+	else
+		bytes += kobjsize(current->fs);
+
+	if (current->files && atomic_read(&current->files->count) > 1)
+		sbytes += kobjsize(current->files);
+	else
+		bytes += kobjsize(current->files);
+
+	if (current->sighand && refcount_read(&current->sighand->count) > 1)
+		sbytes += kobjsize(current->sighand);
+	else
+		bytes += kobjsize(current->sighand);
+
+	bytes += kobjsize(current); /* includes kernel stack */
+
+	mmap_read_unlock(mm);
+
+	seq_printf(m,
+		"Mem:\t%8lu bytes\n"
+		"Slack:\t%8lu bytes\n"
+		"Shared:\t%8lu bytes\n",
+		bytes, slack, sbytes);
+}
+
+unsigned long task_vsize(struct mm_struct *mm)
+{
+	VMA_ITERATOR(vmi, mm, 0);
+	struct vm_area_struct *vma;
+	unsigned long vsize = 0;
+
+	mmap_read_lock(mm);
+	for_each_vma(vmi, vma)
+		vsize += vma->vm_end - vma->vm_start;
+	mmap_read_unlock(mm);
+	return vsize;
+}
+
+unsigned long task_statm(struct mm_struct *mm,
+			 unsigned long *shared, unsigned long *text,
+			 unsigned long *data, unsigned long *resident)
+{
+	VMA_ITERATOR(vmi, mm, 0);
+	struct vm_area_struct *vma;
+	struct vm_region *region;
+	unsigned long size = kobjsize(mm);
+
+	mmap_read_lock(mm);
+	for_each_vma(vmi, vma) {
+		size += kobjsize(vma);
+		region = vma->vm_region;
+		if (region) {
+			size += kobjsize(region);
+			size += region->vm_end - region->vm_start;
+		}
+	}
+
+	*text = (PAGE_ALIGN(mm->end_code) - (mm->start_code & PAGE_MASK))
+		>> PAGE_SHIFT;
+	*data = (PAGE_ALIGN(mm->start_stack) - (mm->start_data & PAGE_MASK))
+		>> PAGE_SHIFT;
+	mmap_read_unlock(mm);
+	size >>= PAGE_SHIFT;
+	size += *text + *data;
+	*resident = size;
+	return size;
+}
+
+/*
+ * display a single VMA to a sequenced file
+ */
+static int nommu_vma_show(struct seq_file *m, struct vm_area_struct *vma)
+{
+	struct mm_struct *mm = vma->vm_mm;
+	unsigned long ino = 0;
+	struct file *file;
+	dev_t dev = 0;
+	int flags;
+	unsigned long long pgoff = 0;
+
+	flags = vma->vm_flags;
+	file = vma->vm_file;
+
+	if (file) {
+		struct inode *inode = file_inode(vma->vm_file);
+		dev = inode->i_sb->s_dev;
+		ino = inode->i_ino;
+		pgoff = (loff_t)vma->vm_pgoff << PAGE_SHIFT;
+	}
+
+	seq_setwidth(m, 25 + sizeof(void *) * 6 - 1);
+	seq_printf(m,
+		   "%08lx-%08lx %c%c%c%c %08llx %02x:%02x %lu ",
+		   vma->vm_start,
+		   vma->vm_end,
+		   flags & VM_READ ? 'r' : '-',
+		   flags & VM_WRITE ? 'w' : '-',
+		   flags & VM_EXEC ? 'x' : '-',
+		   flags & VM_MAYSHARE ? flags & VM_SHARED ? 'S' : 's' : 'p',
+		   pgoff,
+		   MAJOR(dev), MINOR(dev), ino);
+
+	if (file) {
+		seq_pad(m, ' ');
+		seq_path(m, file_user_path(file), "");
+	} else if (mm && vma_is_initial_stack(vma)) {
+		seq_pad(m, ' ');
+		seq_puts(m, "[stack]");
+	}
+
+	seq_putc(m, '\n');
+	return 0;
+}
+
+/*
+ * display mapping lines for a particular process's /proc/pid/maps
+ */
+static int show_map(struct seq_file *m, void *_p)
+{
+	return nommu_vma_show(m, _p);
+}
+
+static struct vm_area_struct *proc_get_vma(struct proc_maps_private *priv,
+						loff_t *ppos)
+{
+	struct vm_area_struct *vma = vma_next(&priv->iter);
+
+	if (vma) {
+		*ppos = vma->vm_start;
+	} else {
+		*ppos = -1UL;
+	}
+
+	return vma;
+}
+
+static void *m_start(struct seq_file *m, loff_t *ppos)
+{
+	struct proc_maps_private *priv = m->private;
+	unsigned long last_addr = *ppos;
+	struct mm_struct *mm;
+
+	/* See proc_get_vma(). Zero at the start or after lseek. */
+	if (last_addr == -1UL)
+		return NULL;
+
+	/* pin the task and mm whilst we play with them */
+	priv->task = get_proc_task(priv->inode);
+	if (!priv->task)
+		return ERR_PTR(-ESRCH);
+
+	mm = priv->lock_ctx.mm;
+	if (!mm || !mmget_not_zero(mm)) {
+		put_task_struct(priv->task);
+		priv->task = NULL;
+		return NULL;
+	}
+
+	if (mmap_read_lock_killable(mm)) {
+		mmput(mm);
+		put_task_struct(priv->task);
+		priv->task = NULL;
+		return ERR_PTR(-EINTR);
+	}
+
+	vma_iter_init(&priv->iter, mm, last_addr);
+
+	return proc_get_vma(priv, ppos);
+}
+
+static void m_stop(struct seq_file *m, void *v)
+{
+	struct proc_maps_private *priv = m->private;
+	struct mm_struct *mm = priv->lock_ctx.mm;
+
+	if (!priv->task)
+		return;
+
+	mmap_read_unlock(mm);
+	mmput(mm);
+	put_task_struct(priv->task);
+	priv->task = NULL;
+}
+
+static void *m_next(struct seq_file *m, void *_p, loff_t *ppos)
+{
+	return proc_get_vma(m->private, ppos);
+}
+
+static const struct seq_operations proc_pid_maps_ops = {
+	.start	= m_start,
+	.next	= m_next,
+	.stop	= m_stop,
+	.show	= show_map
+};
+
+static int maps_open(struct inode *inode, struct file *file,
+		     const struct seq_operations *ops)
+{
+	struct proc_maps_private *priv;
+
+	priv = __seq_open_private(file, ops, sizeof(*priv));
+	if (!priv)
+		return -ENOMEM;
+
+	priv->inode = inode;
+	priv->lock_ctx.mm = proc_mem_open(inode, PTRACE_MODE_READ);
+	if (IS_ERR_OR_NULL(priv->lock_ctx.mm)) {
+		int err = priv->lock_ctx.mm ? PTR_ERR(priv->lock_ctx.mm) : -ESRCH;
+
+		seq_release_private(inode, file);
+		return err;
+	}
+
+	return 0;
+}
+
+
+static int map_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq = file->private_data;
+	struct proc_maps_private *priv = seq->private;
+
+	if (priv->lock_ctx.mm)
+		mmdrop(priv->lock_ctx.mm);
+
+	return seq_release_private(inode, file);
+}
+
+static int pid_maps_open(struct inode *inode, struct file *file)
+{
+	return maps_open(inode, file, &proc_pid_maps_ops);
+}
+
+const struct file_operations proc_pid_maps_operations = {
+	.open		= pid_maps_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= map_release,
+};
+