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, 256 insertions, 0 deletions

diff --git a/arch/xtensa/mm/fault.c b/arch/xtensa/mm/fault.c
new file mode 100644
index 000000000..16e11b6f6
--- /dev/null
+++ b/arch/xtensa/mm/fault.c
@@ -0,0 +1,256 @@
+// TODO VM_EXEC flag work-around, cache aliasing
+/*
+ * arch/xtensa/mm/fault.c
+ *
+ * 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) 2001 - 2010 Tensilica Inc.
+ *
+ * Chris Zankel <chris@zankel.net>
+ * Joe Taylor	<joe@tensilica.com, joetylr@yahoo.com>
+ */
+
+#include <linux/mm.h>
+#include <linux/extable.h>
+#include <linux/hardirq.h>
+#include <linux/perf_event.h>
+#include <linux/uaccess.h>
+#include <asm/mmu_context.h>
+#include <asm/cacheflush.h>
+#include <asm/hardirq.h>
+#include <asm/traps.h>
+
+void bad_page_fault(struct pt_regs*, unsigned long, int);
+
+static void vmalloc_fault(struct pt_regs *regs, unsigned int address)
+{
+#ifdef CONFIG_MMU
+	/* Synchronize this task's top level page-table
+	 * with the 'reference' page table.
+	 */
+	struct mm_struct *act_mm = current->active_mm;
+	int index = pgd_index(address);
+	pgd_t *pgd, *pgd_k;
+	p4d_t *p4d, *p4d_k;
+	pud_t *pud, *pud_k;
+	pmd_t *pmd, *pmd_k;
+	pte_t *pte_k;
+
+	if (act_mm == NULL)
+		goto bad_page_fault;
+
+	pgd = act_mm->pgd + index;
+	pgd_k = init_mm.pgd + index;
+
+	if (!pgd_present(*pgd_k))
+		goto bad_page_fault;
+
+	pgd_val(*pgd) = pgd_val(*pgd_k);
+
+	p4d = p4d_offset(pgd, address);
+	p4d_k = p4d_offset(pgd_k, address);
+	if (!p4d_present(*p4d) || !p4d_present(*p4d_k))
+		goto bad_page_fault;
+
+	pud = pud_offset(p4d, address);
+	pud_k = pud_offset(p4d_k, address);
+	if (!pud_present(*pud) || !pud_present(*pud_k))
+		goto bad_page_fault;
+
+	pmd = pmd_offset(pud, address);
+	pmd_k = pmd_offset(pud_k, address);
+	if (!pmd_present(*pmd) || !pmd_present(*pmd_k))
+		goto bad_page_fault;
+
+	pmd_val(*pmd) = pmd_val(*pmd_k);
+	pte_k = pte_offset_kernel(pmd_k, address);
+
+	if (!pte_present(*pte_k))
+		goto bad_page_fault;
+	return;
+
+bad_page_fault:
+	bad_page_fault(regs, address, SIGKILL);
+#else
+	WARN_ONCE(1, "%s in noMMU configuration\n", __func__);
+#endif
+}
+/*
+ * This routine handles page faults.  It determines the address,
+ * and the problem, and then passes it off to one of the appropriate
+ * routines.
+ *
+ * Note: does not handle Miss and MultiHit.
+ */
+
+void do_page_fault(struct pt_regs *regs)
+{
+	struct vm_area_struct * vma;
+	struct mm_struct *mm = current->mm;
+	unsigned int exccause = regs->exccause;
+	unsigned int address = regs->excvaddr;
+	int code;
+
+	int is_write, is_exec;
+	vm_fault_t fault;
+	unsigned int flags = FAULT_FLAG_DEFAULT;
+
+	code = SEGV_MAPERR;
+
+	/* We fault-in kernel-space virtual memory on-demand. The
+	 * 'reference' page table is init_mm.pgd.
+	 */
+	if (address >= TASK_SIZE && !user_mode(regs)) {
+		vmalloc_fault(regs, address);
+		return;
+	}
+
+	/* If we're in an interrupt or have no user
+	 * context, we must not take the fault..
+	 */
+	if (faulthandler_disabled() || !mm) {
+		bad_page_fault(regs, address, SIGSEGV);
+		return;
+	}
+
+	is_write = (exccause == EXCCAUSE_STORE_CACHE_ATTRIBUTE) ? 1 : 0;
+	is_exec =  (exccause == EXCCAUSE_ITLB_PRIVILEGE ||
+		    exccause == EXCCAUSE_ITLB_MISS ||
+		    exccause == EXCCAUSE_FETCH_CACHE_ATTRIBUTE) ? 1 : 0;
+
+	pr_debug("[%s:%d:%08x:%d:%08lx:%s%s]\n",
+		 current->comm, current->pid,
+		 address, exccause, regs->pc,
+		 is_write ? "w" : "", is_exec ? "x" : "");
+
+	if (user_mode(regs))
+		flags |= FAULT_FLAG_USER;
+
+	perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
+
+retry:
+	vma = lock_mm_and_find_vma(mm, address, regs);
+	if (!vma)
+		goto bad_area_nosemaphore;
+
+	/* Ok, we have a good vm_area for this memory access, so
+	 * we can handle it..
+	 */
+
+	code = SEGV_ACCERR;
+
+	if (is_write) {
+		if (!(vma->vm_flags & VM_WRITE))
+			goto bad_area;
+		flags |= FAULT_FLAG_WRITE;
+	} else if (is_exec) {
+		if (!(vma->vm_flags & VM_EXEC))
+			goto bad_area;
+	} else	/* Allow read even from write-only pages. */
+		if (!(vma->vm_flags & (VM_READ | VM_WRITE)))
+			goto bad_area;
+
+	/* If for any reason at all we couldn't handle the fault,
+	 * make sure we exit gracefully rather than endlessly redo
+	 * the fault.
+	 */
+	fault = handle_mm_fault(vma, address, flags, regs);
+
+	if (fault_signal_pending(fault, regs)) {
+		if (!user_mode(regs))
+			bad_page_fault(regs, address, SIGKILL);
+		return;
+	}
+
+	/* The fault is fully completed (including releasing mmap lock) */
+	if (fault & VM_FAULT_COMPLETED)
+		return;
+
+	if (unlikely(fault & VM_FAULT_ERROR)) {
+		if (fault & VM_FAULT_OOM)
+			goto out_of_memory;
+		else if (fault & VM_FAULT_SIGSEGV)
+			goto bad_area;
+		else if (fault & VM_FAULT_SIGBUS)
+			goto do_sigbus;
+		BUG();
+	}
+
+	if (fault & VM_FAULT_RETRY) {
+		flags |= FAULT_FLAG_TRIED;
+
+		/* No need to mmap_read_unlock(mm) as we would
+		 * have already released it in __lock_page_or_retry
+		 * in mm/filemap.c.
+		 */
+
+		goto retry;
+	}
+
+	mmap_read_unlock(mm);
+	return;
+
+	/* Something tried to access memory that isn't in our memory map..
+	 * Fix it, but check if it's kernel or user first..
+	 */
+bad_area:
+	mmap_read_unlock(mm);
+bad_area_nosemaphore:
+	if (user_mode(regs)) {
+		force_sig_fault(SIGSEGV, code, (void *) address);
+		return;
+	}
+	bad_page_fault(regs, address, SIGSEGV);
+	return;
+
+
+	/* We ran out of memory, or some other thing happened to us that made
+	 * us unable to handle the page fault gracefully.
+	 */
+out_of_memory:
+	mmap_read_unlock(mm);
+	if (!user_mode(regs))
+		bad_page_fault(regs, address, SIGKILL);
+	else
+		pagefault_out_of_memory();
+	return;
+
+do_sigbus:
+	mmap_read_unlock(mm);
+
+	/* Send a sigbus, regardless of whether we were in kernel
+	 * or user mode.
+	 */
+	force_sig_fault(SIGBUS, BUS_ADRERR, (void *) address);
+
+	/* Kernel mode? Handle exceptions or die */
+	if (!user_mode(regs))
+		bad_page_fault(regs, address, SIGBUS);
+	return;
+}
+
+
+void
+bad_page_fault(struct pt_regs *regs, unsigned long address, int sig)
+{
+	extern void __noreturn die(const char*, struct pt_regs*, long);
+	const struct exception_table_entry *entry;
+
+	/* Are we prepared to handle this kernel fault?  */
+	if ((entry = search_exception_tables(regs->pc)) != NULL) {
+		pr_debug("%s: Exception at pc=%#010lx (%lx)\n",
+			 current->comm, regs->pc, entry->fixup);
+		regs->pc = entry->fixup;
+		return;
+	}
+
+	/* Oops. The kernel tried to access some bad page. We'll have to
+	 * terminate things with extreme prejudice.
+	 */
+	pr_alert("Unable to handle kernel paging request at virtual "
+		 "address %08lx\n pc = %08lx, ra = %08lx\n",
+		 address, regs->pc, regs->areg[0]);
+	die("Oops", regs, sig);
+}