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

4 files changed, 632 insertions, 0 deletions

diff --git a/arch/alpha/mm/Makefile b/arch/alpha/mm/Makefile
new file mode 100644
index 000000000..2d0566405
--- /dev/null
+++ b/arch/alpha/mm/Makefile
@@ -0,0 +1,6 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Makefile for the linux alpha-specific parts of the memory manager.
+#
+
+obj-y	:= init.o fault.o tlbflush.o
diff --git a/arch/alpha/mm/fault.c b/arch/alpha/mm/fault.c
new file mode 100644
index 000000000..a9816bbc9
--- /dev/null
+++ b/arch/alpha/mm/fault.c
@@ -0,0 +1,248 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  linux/arch/alpha/mm/fault.c
+ *
+ *  Copyright (C) 1995  Linus Torvalds
+ */
+
+#include <linux/sched/signal.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <asm/io.h>
+
+#define __EXTERN_INLINE inline
+#include <asm/mmu_context.h>
+#include <asm/tlbflush.h>
+#undef  __EXTERN_INLINE
+
+#include <linux/signal.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/smp.h>
+#include <linux/interrupt.h>
+#include <linux/extable.h>
+#include <linux/uaccess.h>
+#include <linux/perf_event.h>
+
+extern void die_if_kernel(char *,struct pt_regs *,long, unsigned long *);
+
+
+/*
+ * Force a new ASN for a task.
+ */
+
+#ifndef CONFIG_SMP
+unsigned long last_asn = ASN_FIRST_VERSION;
+#endif
+
+void
+__load_new_mm_context(struct mm_struct *next_mm)
+{
+	unsigned long mmc;
+	struct pcb_struct *pcb;
+
+	mmc = __get_new_mm_context(next_mm, smp_processor_id());
+	next_mm->context[smp_processor_id()] = mmc;
+
+	pcb = &current_thread_info()->pcb;
+	pcb->asn = mmc & HARDWARE_ASN_MASK;
+	pcb->ptbr = ((unsigned long) next_mm->pgd - IDENT_ADDR) >> PAGE_SHIFT;
+
+	__reload_thread(pcb);
+}
+
+
+/*
+ * This routine handles page faults.  It determines the address,
+ * and the problem, and then passes it off to handle_mm_fault().
+ *
+ * mmcsr:
+ *	0 = translation not valid
+ *	1 = access violation
+ *	2 = fault-on-read
+ *	3 = fault-on-execute
+ *	4 = fault-on-write
+ *
+ * cause:
+ *	-1 = instruction fetch
+ *	0 = load
+ *	1 = store
+ *
+ * Registers $9 through $15 are saved in a block just prior to `regs' and
+ * are saved and restored around the call to allow exception code to
+ * modify them.
+ */
+
+/* Macro for exception fixup code to access integer registers.  */
+#define dpf_reg(r)							\
+	(((unsigned long *)regs)[(r) <= 8 ? (r) : (r) <= 15 ? (r)-17 :	\
+				 (r) <= 18 ? (r)+11 : (r)-10])
+
+asmlinkage void
+do_page_fault(unsigned long address, unsigned long mmcsr,
+	      long cause, struct pt_regs *regs)
+{
+	struct vm_area_struct * vma;
+	struct mm_struct *mm = current->mm;
+	const struct exception_table_entry *fixup;
+	int si_code = SEGV_MAPERR;
+	vm_fault_t fault;
+	unsigned int flags = FAULT_FLAG_DEFAULT;
+
+	/* As of EV6, a load into $31/$f31 is a prefetch, and never faults
+	   (or is suppressed by the PALcode).  Support that for older CPUs
+	   by ignoring such an instruction.  */
+	if (cause == 0) {
+		unsigned int insn;
+		__get_user(insn, (unsigned int __user *)regs->pc);
+		if ((insn >> 21 & 0x1f) == 0x1f &&
+		    /* ldq ldl ldt lds ldg ldf ldwu ldbu */
+		    (1ul << (insn >> 26) & 0x30f00001400ul)) {
+			regs->pc += 4;
+			return;
+		}
+	}
+
+	/* If we're in an interrupt context, or have no user context,
+	   we must not take the fault.  */
+	if (!mm || faulthandler_disabled())
+		goto no_context;
+
+#ifdef CONFIG_ALPHA_LARGE_VMALLOC
+	if (address >= TASK_SIZE)
+		goto vmalloc_fault;
+#endif
+	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.  */
+	si_code = SEGV_ACCERR;
+	if (cause < 0) {
+		if (!(vma->vm_flags & VM_EXEC))
+			goto bad_area;
+	} else if (!cause) {
+		/* Allow reads even for write-only mappings */
+		if (!(vma->vm_flags & (VM_READ | VM_WRITE)))
+			goto bad_area;
+	} else {
+		if (!(vma->vm_flags & VM_WRITE))
+			goto bad_area;
+		flags |= FAULT_FLAG_WRITE;
+	}
+
+	/* 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))
+			goto no_context;
+		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))
+		goto do_sigsegv;
+
+ no_context:
+	/* Are we prepared to handle this fault as an exception?  */
+	if ((fixup = search_exception_tables(regs->pc)) != 0) {
+		unsigned long newpc;
+		newpc = fixup_exception(dpf_reg, fixup, regs->pc);
+		regs->pc = newpc;
+		return;
+	}
+
+	/* Oops. The kernel tried to access some bad page. We'll have to
+	   terminate things with extreme prejudice.  */
+	printk(KERN_ALERT "Unable to handle kernel paging request at "
+	       "virtual address %016lx\n", address);
+	die_if_kernel("Oops", regs, cause, (unsigned long*)regs - 16);
+	make_task_dead(SIGKILL);
+
+	/* 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))
+		goto no_context;
+	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 __user *) address);
+	if (!user_mode(regs))
+		goto no_context;
+	return;
+
+ do_sigsegv:
+	force_sig_fault(SIGSEGV, si_code, (void __user *) address);
+	return;
+
+#ifdef CONFIG_ALPHA_LARGE_VMALLOC
+ vmalloc_fault:
+	if (user_mode(regs))
+		goto do_sigsegv;
+	else {
+		/* Synchronize this task's top level page-table
+		   with the "reference" page table from init.  */
+		long index = pgd_index(address);
+		pgd_t *pgd, *pgd_k;
+
+		pgd = current->active_mm->pgd + index;
+		pgd_k = swapper_pg_dir + index;
+		if (!pgd_present(*pgd) && pgd_present(*pgd_k)) {
+			pgd_val(*pgd) = pgd_val(*pgd_k);
+			return;
+		}
+		goto no_context;
+	}
+#endif
+}
diff --git a/arch/alpha/mm/init.c b/arch/alpha/mm/init.c
new file mode 100644
index 000000000..f4d65a60c
--- /dev/null
+++ b/arch/alpha/mm/init.c
@@ -0,0 +1,266 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  linux/arch/alpha/mm/init.c
+ *
+ *  Copyright (C) 1995  Linus Torvalds
+ */
+
+/* 2.3.x zone allocator, 1999 Andrea Arcangeli <andrea@suse.de> */
+
+#include <linux/pagemap.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/mm.h>
+#include <linux/swap.h>
+#include <linux/init.h>
+#include <linux/memblock.h> /* max_low_pfn */
+#include <linux/vmalloc.h>
+#include <linux/gfp.h>
+
+#include <linux/uaccess.h>
+#include <asm/pgalloc.h>
+#include <asm/hwrpb.h>
+#include <asm/dma.h>
+#include <asm/mmu_context.h>
+#include <asm/console.h>
+#include <asm/tlb.h>
+#include <asm/setup.h>
+#include <asm/sections.h>
+
+#include "../kernel/proto.h"
+
+static struct pcb_struct original_pcb;
+
+pgd_t *
+pgd_alloc(struct mm_struct *mm)
+{
+	pgd_t *ret, *init;
+
+	ret = __pgd_alloc(mm, 0);
+	init = pgd_offset(&init_mm, 0UL);
+	if (ret) {
+#ifdef CONFIG_ALPHA_LARGE_VMALLOC
+		memcpy (ret + USER_PTRS_PER_PGD, init + USER_PTRS_PER_PGD,
+			(PTRS_PER_PGD - USER_PTRS_PER_PGD - 1)*sizeof(pgd_t));
+#else
+		pgd_val(ret[PTRS_PER_PGD-2]) = pgd_val(init[PTRS_PER_PGD-2]);
+#endif
+
+		/* The last PGD entry is the VPTB self-map.  */
+		pgd_val(ret[PTRS_PER_PGD-1])
+		  = pte_val(mk_pte(virt_to_page(ret), PAGE_KERNEL));
+	}
+	return ret;
+}
+
+
+static inline unsigned long
+load_PCB(struct pcb_struct *pcb)
+{
+	pcb->ksp = (unsigned long)current_stack_pointer;
+	return __reload_thread(pcb);
+}
+
+/* Set up initial PCB, VPTB, and other such nicities.  */
+
+static inline void
+switch_to_system_map(void)
+{
+	unsigned long newptbr;
+	unsigned long original_pcb_ptr;
+
+	/* Initialize the kernel's page tables.  Linux puts the vptb in
+	   the last slot of the L1 page table.  */
+	memset(swapper_pg_dir, 0, PAGE_SIZE);
+	newptbr = ((unsigned long) swapper_pg_dir - PAGE_OFFSET) >> PAGE_SHIFT;
+	pgd_val(swapper_pg_dir[1023]) =
+		(newptbr << 32) | pgprot_val(PAGE_KERNEL);
+
+	/* Set the vptb.  This is often done by the bootloader, but 
+	   shouldn't be required.  */
+	if (hwrpb->vptb != 0xfffffffe00000000UL) {
+		wrvptptr(0xfffffffe00000000UL);
+		hwrpb->vptb = 0xfffffffe00000000UL;
+		hwrpb_update_checksum(hwrpb);
+	}
+
+	/* Also set up the real kernel PCB while we're at it.  */
+	init_thread_info.pcb.ptbr = newptbr;
+	init_thread_info.pcb.flags = 1;	/* set FEN, clear everything else */
+	original_pcb_ptr = load_PCB(&init_thread_info.pcb);
+	tbia();
+
+	/* Save off the contents of the original PCB so that we can
+	   restore the original console's page tables for a clean reboot.
+
+	   Note that the PCB is supposed to be a physical address, but
+	   since KSEG values also happen to work, folks get confused.
+	   Check this here.  */
+
+	if (original_pcb_ptr < PAGE_OFFSET) {
+		original_pcb_ptr = (unsigned long)
+			phys_to_virt(original_pcb_ptr);
+	}
+	original_pcb = *(struct pcb_struct *) original_pcb_ptr;
+}
+
+int callback_init_done;
+
+void * __init
+callback_init(void * kernel_end)
+{
+	struct crb_struct * crb;
+	pgd_t *pgd;
+	p4d_t *p4d;
+	pud_t *pud;
+	pmd_t *pmd;
+	void *two_pages;
+
+	/* Starting at the HWRPB, locate the CRB. */
+	crb = (struct crb_struct *)((char *)hwrpb + hwrpb->crb_offset);
+
+	if (alpha_using_srm) {
+		/* Tell the console whither it is to be remapped. */
+		if (srm_fixup(VMALLOC_START, (unsigned long)hwrpb))
+			__halt();		/* "We're boned."  --Bender */
+
+		/* Edit the procedure descriptors for DISPATCH and FIXUP. */
+		crb->dispatch_va = (struct procdesc_struct *)
+			(VMALLOC_START + (unsigned long)crb->dispatch_va
+			 - crb->map[0].va);
+		crb->fixup_va = (struct procdesc_struct *)
+			(VMALLOC_START + (unsigned long)crb->fixup_va
+			 - crb->map[0].va);
+	}
+
+	switch_to_system_map();
+
+	/* Allocate one PGD and one PMD.  In the case of SRM, we'll need
+	   these to actually remap the console.  There is an assumption
+	   here that only one of each is needed, and this allows for 8MB.
+	   On systems with larger consoles, additional pages will be
+	   allocated as needed during the mapping process.
+
+	   In the case of not SRM, but not CONFIG_ALPHA_LARGE_VMALLOC,
+	   we need to allocate the PGD we use for vmalloc before we start
+	   forking other tasks.  */
+
+	two_pages = (void *)
+	  (((unsigned long)kernel_end + ~PAGE_MASK) & PAGE_MASK);
+	kernel_end = two_pages + 2*PAGE_SIZE;
+	memset(two_pages, 0, 2*PAGE_SIZE);
+
+	pgd = pgd_offset_k(VMALLOC_START);
+	p4d = p4d_offset(pgd, VMALLOC_START);
+	pud = pud_offset(p4d, VMALLOC_START);
+	pud_set(pud, (pmd_t *)two_pages);
+	pmd = pmd_offset(pud, VMALLOC_START);
+	pmd_set(pmd, (pte_t *)(two_pages + PAGE_SIZE));
+
+	if (alpha_using_srm) {
+		static struct vm_struct console_remap_vm;
+		unsigned long nr_pages = 0;
+		unsigned long vaddr;
+		unsigned long i, j;
+
+		/* calculate needed size */
+		for (i = 0; i < crb->map_entries; ++i)
+			nr_pages += crb->map[i].count;
+
+		/* register the vm area */
+		console_remap_vm.flags = VM_ALLOC;
+		console_remap_vm.size = nr_pages << PAGE_SHIFT;
+		vm_area_register_early(&console_remap_vm, PAGE_SIZE);
+
+		vaddr = (unsigned long)console_remap_vm.addr;
+
+		/* Set up the third level PTEs and update the virtual
+		   addresses of the CRB entries.  */
+		for (i = 0; i < crb->map_entries; ++i) {
+			unsigned long pfn = crb->map[i].pa >> PAGE_SHIFT;
+			crb->map[i].va = vaddr;
+			for (j = 0; j < crb->map[i].count; ++j) {
+				/* Newer consoles (especially on larger
+				   systems) may require more pages of
+				   PTEs. Grab additional pages as needed. */
+				if (pmd != pmd_offset(pud, vaddr)) {
+					memset(kernel_end, 0, PAGE_SIZE);
+					pmd = pmd_offset(pud, vaddr);
+					pmd_set(pmd, (pte_t *)kernel_end);
+					kernel_end += PAGE_SIZE;
+				}
+				set_pte(pte_offset_kernel(pmd, vaddr),
+					pfn_pte(pfn, PAGE_KERNEL));
+				pfn++;
+				vaddr += PAGE_SIZE;
+			}
+		}
+	}
+
+	callback_init_done = 1;
+	return kernel_end;
+}
+
+void __init arch_zone_limits_init(unsigned long *max_zone_pfn)
+{
+	unsigned long dma_pfn;
+
+	dma_pfn = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT;
+	max_pfn = max_low_pfn;
+
+	max_zone_pfn[ZONE_DMA] = dma_pfn;
+	max_zone_pfn[ZONE_NORMAL] = max_pfn;
+}
+
+/*
+ * paging_init() initializes the kernel's ZERO_PGE.
+ */
+void __init paging_init(void)
+{
+	memset(absolute_pointer(ZERO_PGE), 0, PAGE_SIZE);
+}
+
+#if defined(CONFIG_ALPHA_GENERIC) || defined(CONFIG_ALPHA_SRM)
+void
+srm_paging_stop (void)
+{
+	/* Move the vptb back to where the SRM console expects it.  */
+	swapper_pg_dir[1] = swapper_pg_dir[1023];
+	tbia();
+	wrvptptr(0x200000000UL);
+	hwrpb->vptb = 0x200000000UL;
+	hwrpb_update_checksum(hwrpb);
+
+	/* Reload the page tables that the console had in use.  */
+	load_PCB(&original_pcb);
+	tbia();
+}
+#endif
+
+static const pgprot_t protection_map[16] = {
+	[VM_NONE]					= _PAGE_P(_PAGE_FOE | _PAGE_FOW |
+								  _PAGE_FOR),
+	[VM_READ]					= _PAGE_P(_PAGE_FOE | _PAGE_FOW),
+	[VM_WRITE]					= _PAGE_P(_PAGE_FOE),
+	[VM_WRITE | VM_READ]				= _PAGE_P(_PAGE_FOE),
+	[VM_EXEC]					= _PAGE_P(_PAGE_FOW | _PAGE_FOR),
+	[VM_EXEC | VM_READ]				= _PAGE_P(_PAGE_FOW),
+	[VM_EXEC | VM_WRITE]				= _PAGE_P(0),
+	[VM_EXEC | VM_WRITE | VM_READ]			= _PAGE_P(0),
+	[VM_SHARED]					= _PAGE_S(_PAGE_FOE | _PAGE_FOW |
+								  _PAGE_FOR),
+	[VM_SHARED | VM_READ]				= _PAGE_S(_PAGE_FOE | _PAGE_FOW),
+	[VM_SHARED | VM_WRITE]				= _PAGE_S(_PAGE_FOE),
+	[VM_SHARED | VM_WRITE | VM_READ]		= _PAGE_S(_PAGE_FOE),
+	[VM_SHARED | VM_EXEC]				= _PAGE_S(_PAGE_FOW | _PAGE_FOR),
+	[VM_SHARED | VM_EXEC | VM_READ]			= _PAGE_S(_PAGE_FOW),
+	[VM_SHARED | VM_EXEC | VM_WRITE]		= _PAGE_S(0),
+	[VM_SHARED | VM_EXEC | VM_WRITE | VM_READ]	= _PAGE_S(0)
+};
+DECLARE_VM_GET_PAGE_PROT
diff --git a/arch/alpha/mm/tlbflush.c b/arch/alpha/mm/tlbflush.c
new file mode 100644
index 000000000..ccbc317b9
--- /dev/null
+++ b/arch/alpha/mm/tlbflush.c
@@ -0,0 +1,112 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Alpha TLB shootdown helpers
+ *
+ * Copyright (C) 2025 Magnus Lindholm <linmag7@gmail.com>
+ *
+ * Alpha-specific TLB flush helpers that cannot be expressed purely
+ * as inline functions.
+ *
+ * These helpers provide combined MM context handling (ASN rollover)
+ * and immediate TLB invalidation for page migration and memory
+ * compaction paths, where lazy shootdowns are insufficient.
+ */
+
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/sched.h>
+#include <asm/tlbflush.h>
+#include <asm/pal.h>
+#include <asm/mmu_context.h>
+
+#define asn_locked() (cpu_data[smp_processor_id()].asn_lock)
+
+/*
+ * Migration/compaction helper: combine mm context (ASN) handling with an
+ * immediate per-page TLB invalidate and (for exec) an instruction barrier.
+ *
+ * This mirrors the SMP combined IPI handler semantics, but runs locally on UP.
+ */
+#ifndef CONFIG_SMP
+void migrate_flush_tlb_page(struct vm_area_struct *vma,
+					   unsigned long addr)
+{
+	struct mm_struct *mm = vma->vm_mm;
+	int tbi_type = (vma->vm_flags & VM_EXEC) ? 3 : 2;
+
+	/*
+	 * First do the mm-context side:
+	 * If we're currently running this mm, reload a fresh context ASN.
+	 * Otherwise, mark context invalid.
+	 *
+	 * On UP, this is mostly about matching the SMP semantics and ensuring
+	 * exec/i-cache tagging assumptions hold when compaction migrates pages.
+	 */
+	if (mm == current->active_mm)
+		flush_tlb_current(mm);
+	else
+		flush_tlb_other(mm);
+
+	/*
+	 * Then do the immediate translation kill for this VA.
+	 * For exec mappings, order instruction fetch after invalidation.
+	 */
+	tbi(tbi_type, addr);
+}
+
+#else
+struct tlb_mm_and_addr {
+	struct mm_struct *mm;
+	unsigned long addr;
+	int tbi_type;	/* 2 = DTB, 3 = ITB+DTB */
+};
+
+static void ipi_flush_mm_and_page(void *x)
+{
+	struct tlb_mm_and_addr *d = x;
+
+	/* Part 1: mm context side (Alpha uses ASN/context as a key mechanism). */
+	if (d->mm == current->active_mm && !asn_locked())
+		__load_new_mm_context(d->mm);
+	else
+		flush_tlb_other(d->mm);
+
+	/* Part 2: immediate per-VA invalidation on this CPU. */
+	tbi(d->tbi_type, d->addr);
+}
+
+void migrate_flush_tlb_page(struct vm_area_struct *vma, unsigned long addr)
+{
+	struct mm_struct *mm = vma->vm_mm;
+	struct tlb_mm_and_addr d = {
+		.mm = mm,
+		.addr = addr,
+		.tbi_type = (vma->vm_flags & VM_EXEC) ? 3 : 2,
+	};
+
+	/*
+	 * One synchronous rendezvous: every CPU runs ipi_flush_mm_and_page().
+	 * This is the "combined" version of flush_tlb_mm + per-page invalidate.
+	 */
+	preempt_disable();
+	on_each_cpu(ipi_flush_mm_and_page, &d, 1);
+
+	/*
+	 * mimic flush_tlb_mm()'s mm_users<=1 optimization.
+	 */
+	if (atomic_read(&mm->mm_users) <= 1) {
+
+		int cpu, this_cpu;
+		this_cpu = smp_processor_id();
+
+		for (cpu = 0; cpu < NR_CPUS; cpu++) {
+			if (!cpu_online(cpu) || cpu == this_cpu)
+				continue;
+			if (READ_ONCE(mm->context[cpu]))
+				WRITE_ONCE(mm->context[cpu], 0);
+		}
+	}
+	preempt_enable();
+}
+
+#endif