x86/entry/64: Use a per-CPU trampoline stack for IDT entries

Historically, IDT entries from usermode have always gone directly
to the running task's kernel stack.  Rearrange it so that we enter on
a per-CPU trampoline stack and then manually switch to the task's stack.
This touches a couple of extra cachelines, but it gives us a chance
to run some code before we touch the kernel stack.

The asm isn't exactly beautiful, but I think that fully refactoring
it can wait.

Signed-off-by: Andy Lutomirski <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Reviewed-by: Borislav Petkov <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Cc: Boris Ostrovsky <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: Brian Gerst <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: David Laight <[email protected]>
Cc: Denys Vlasenko <[email protected]>
Cc: Eduardo Valentin <[email protected]>
Cc: Greg KH <[email protected]>
Cc: H. Peter Anvin <[email protected]>
Cc: Josh Poimboeuf <[email protected]>
Cc: Juergen Gross <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Ingo Molnar <[email protected]>

Author: amluto

arch/x86/entry/entry_64.S

@@ -560,6 +560,13 @@ END(irq_entries_start)
 /* 0(%rsp): ~(interrupt number) */
 	.macro interrupt func
 	cld
+
+	testb	$3, CS-ORIG_RAX(%rsp)
+	jz	1f
+	SWAPGS
+	call	switch_to_thread_stack
+1:
+
 	ALLOC_PT_GPREGS_ON_STACK
 	SAVE_C_REGS
 	SAVE_EXTRA_REGS
@@ -569,12 +576,8 @@ END(irq_entries_start)
 	jz	1f
 
 	/*
-	 * IRQ from user mode.  Switch to kernel gsbase and inform context
-	 * tracking that we're in kernel mode.
-	 */
-	SWAPGS
-
-	/*
+	 * IRQ from user mode.
+	 *
 	 * We need to tell lockdep that IRQs are off.  We can't do this until
 	 * we fix gsbase, and we should do it before enter_from_user_mode
 	 * (which can take locks).  Since TRACE_IRQS_OFF idempotent,
@@ -828,6 +831,32 @@ apicinterrupt IRQ_WORK_VECTOR			irq_work_interrupt		smp_irq_work_interrupt
  */
 #define CPU_TSS_IST(x) PER_CPU_VAR(cpu_tss) + (TSS_ist + ((x) - 1) * 8)
 
+/*
+ * Switch to the thread stack.  This is called with the IRET frame and
+ * orig_ax on the stack.  (That is, RDI..R12 are not on the stack and
+ * space has not been allocated for them.)
+ */
+ENTRY(switch_to_thread_stack)
+	UNWIND_HINT_FUNC
+
+	pushq	%rdi
+	movq	%rsp, %rdi
+	movq	PER_CPU_VAR(cpu_current_top_of_stack), %rsp
+	UNWIND_HINT sp_offset=16 sp_reg=ORC_REG_DI
+
+	pushq	7*8(%rdi)		/* regs->ss */
+	pushq	6*8(%rdi)		/* regs->rsp */
+	pushq	5*8(%rdi)		/* regs->eflags */
+	pushq	4*8(%rdi)		/* regs->cs */
+	pushq	3*8(%rdi)		/* regs->ip */
+	pushq	2*8(%rdi)		/* regs->orig_ax */
+	pushq	8(%rdi)			/* return address */
+	UNWIND_HINT_FUNC
+
+	movq	(%rdi), %rdi
+	ret
+END(switch_to_thread_stack)
+
 .macro idtentry sym do_sym has_error_code:req paranoid=0 shift_ist=-1
 ENTRY(\sym)
 	UNWIND_HINT_IRET_REGS offset=\has_error_code*8
@@ -845,11 +874,12 @@ ENTRY(\sym)
 
 	ALLOC_PT_GPREGS_ON_STACK
 
-	.if \paranoid
-	.if \paranoid == 1
+	.if \paranoid < 2
 	testb	$3, CS(%rsp)			/* If coming from userspace, switch stacks */
-	jnz	1f
+	jnz	.Lfrom_usermode_switch_stack_\@
 	.endif
+
+	.if \paranoid
 	call	paranoid_entry
 	.else
 	call	error_entry
@@ -891,20 +921,15 @@ ENTRY(\sym)
 	jmp	error_exit
 	.endif
 
-	.if \paranoid == 1
+	.if \paranoid < 2
 	/*
-	 * Paranoid entry from userspace.  Switch stacks and treat it
+	 * Entry from userspace.  Switch stacks and treat it
 	 * as a normal entry.  This means that paranoid handlers
 	 * run in real process context if user_mode(regs).
 	 */
-1:
+.Lfrom_usermode_switch_stack_\@:
 	call	error_entry
 
-
-	movq	%rsp, %rdi			/* pt_regs pointer */
-	call	sync_regs
-	movq	%rax, %rsp			/* switch stack */
-
 	movq	%rsp, %rdi			/* pt_regs pointer */
 
 	.if \has_error_code
@@ -1165,6 +1190,14 @@ ENTRY(error_entry)
 	SWAPGS
 
 .Lerror_entry_from_usermode_after_swapgs:
+	/* Put us onto the real thread stack. */
+	popq	%r12				/* save return addr in %12 */
+	movq	%rsp, %rdi			/* arg0 = pt_regs pointer */
+	call	sync_regs
+	movq	%rax, %rsp			/* switch stack */
+	ENCODE_FRAME_POINTER
+	pushq	%r12
+
 	/*
 	 * We need to tell lockdep that IRQs are off.  We can't do this until
 	 * we fix gsbase, and we should do it before enter_from_user_mode

arch/x86/entry/entry_64_compat.S

@@ -306,8 +306,11 @@ ENTRY(entry_INT80_compat)
 	 */
 	movl	%eax, %eax
 
-	/* Construct struct pt_regs on stack (iret frame is already on stack) */
 	pushq	%rax			/* pt_regs->orig_ax */
+
+	/* switch to thread stack expects orig_ax to be pushed */
+	call	switch_to_thread_stack
+
 	pushq	%rdi			/* pt_regs->di */
 	pushq	%rsi			/* pt_regs->si */
 	pushq	%rdx			/* pt_regs->dx */

arch/x86/include/asm/switch_to.h

@@ -90,10 +90,12 @@ static inline void refresh_sysenter_cs(struct thread_struct *thread)
 /* This is used when switching tasks or entering/exiting vm86 mode. */
 static inline void update_sp0(struct task_struct *task)
 {
+	/* On x86_64, sp0 always points to the entry trampoline stack, which is constant: */
 #ifdef CONFIG_X86_32
 	load_sp0(task->thread.sp0);
 #else
-	load_sp0(task_top_of_stack(task));
+	if (static_cpu_has(X86_FEATURE_XENPV))
+		load_sp0(task_top_of_stack(task));
 #endif
 }
 

arch/x86/include/asm/traps.h

@@ -75,7 +75,6 @@ dotraplinkage void do_segment_not_present(struct pt_regs *, long);
 dotraplinkage void do_stack_segment(struct pt_regs *, long);
 #ifdef CONFIG_X86_64
 dotraplinkage void do_double_fault(struct pt_regs *, long);
-asmlinkage struct pt_regs *sync_regs(struct pt_regs *);
 #endif
 dotraplinkage void do_general_protection(struct pt_regs *, long);
 dotraplinkage void do_page_fault(struct pt_regs *, unsigned long);

arch/x86/kernel/cpu/common.c

@@ -1623,11 +1623,13 @@ void cpu_init(void)
 	setup_cpu_entry_area(cpu);
 
 	/*
-	 * Initialize the TSS.  Don't bother initializing sp0, as the initial
-	 * task never enters user mode.
+	 * Initialize the TSS.  sp0 points to the entry trampoline stack
+	 * regardless of what task is running.
 	 */
 	set_tss_desc(cpu, &get_cpu_entry_area(cpu)->tss.x86_tss);
 	load_TR_desc();
+	load_sp0((unsigned long)&get_cpu_entry_area(cpu)->tss +
+		 offsetofend(struct tss_struct, SYSENTER_stack));
 
 	load_mm_ldt(&init_mm);
 

arch/x86/kernel/traps.c

@@ -619,14 +619,15 @@ NOKPROBE_SYMBOL(do_int3);
 
 #ifdef CONFIG_X86_64
 /*
- * Help handler running on IST stack to switch off the IST stack if the
- * interrupted code was in user mode. The actual stack switch is done in
- * entry_64.S
+ * Help handler running on a per-cpu (IST or entry trampoline) stack
+ * to switch to the normal thread stack if the interrupted code was in
+ * user mode. The actual stack switch is done in entry_64.S
  */
 asmlinkage __visible notrace struct pt_regs *sync_regs(struct pt_regs *eregs)
 {
-	struct pt_regs *regs = task_pt_regs(current);
-	*regs = *eregs;
+	struct pt_regs *regs = (struct pt_regs *)this_cpu_read(cpu_current_top_of_stack) - 1;
+	if (regs != eregs)
+		*regs = *eregs;
 	return regs;
 }
 NOKPROBE_SYMBOL(sync_regs);
@@ -642,13 +643,13 @@ struct bad_iret_stack *fixup_bad_iret(struct bad_iret_stack *s)
 	/*
 	 * This is called from entry_64.S early in handling a fault
 	 * caused by a bad iret to user mode.  To handle the fault
-	 * correctly, we want move our stack frame to task_pt_regs
-	 * and we want to pretend that the exception came from the
-	 * iret target.
+	 * correctly, we want to move our stack frame to where it would
+	 * be had we entered directly on the entry stack (rather than
+	 * just below the IRET frame) and we want to pretend that the
+	 * exception came from the IRET target.
 	 */
 	struct bad_iret_stack *new_stack =
-		container_of(task_pt_regs(current),
-			     struct bad_iret_stack, regs);
+		(struct bad_iret_stack *)this_cpu_read(cpu_tss.x86_tss.sp0) - 1;
 
 	/* Copy the IRET target to the new stack. */
 	memmove(&new_stack->regs.ip, (void *)s->regs.sp, 5*8);