LKML Archive on lore.kernel.org
 help / color / Atom feed
From: Andy Lutomirski <luto@kernel.org>
To: X86 ML <x86@kernel.org>
Cc: Borislav Petkov <bpetkov@suse.de>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Brian Gerst <brgerst@gmail.com>,
	Dave Hansen <dave.hansen@intel.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Josh Poimboeuf <jpoimboe@redhat.com>,
	Andy Lutomirski <luto@kernel.org>
Subject: [PATCH v3 12/19] x86/asm/64: Use a percpu trampoline stack for IDT entries
Date: Thu, 23 Nov 2017 20:32:58 -0800
Message-ID: <fa3958723a1a85baeaf309c735b775841205800e.1511497875.git.luto@kernel.org> (raw)
In-Reply-To: <cover.1511497874.git.luto@kernel.org>
In-Reply-To: <cover.1511497874.git.luto@kernel.org>

Historically, IDT entries from usermode have always gone directly
to the running task's kernel stack.  Rearrange it so that we enter on
a percpu 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 <luto@kernel.org>
---
 arch/x86/entry/entry_64.S        | 67 ++++++++++++++++++++++++++++++----------
 arch/x86/entry/entry_64_compat.S |  5 ++-
 arch/x86/include/asm/switch_to.h |  2 +-
 arch/x86/include/asm/traps.h     |  1 -
 arch/x86/kernel/cpu/common.c     |  6 ++--
 arch/x86/kernel/traps.c          | 18 +++++------
 6 files changed, 68 insertions(+), 31 deletions(-)

diff --git a/arch/x86/entry/entry_64.S b/arch/x86/entry/entry_64.S
index f81d50d7ceac..7d47199f405f 100644
--- a/arch/x86/entry/entry_64.S
+++ b/arch/x86/entry/entry_64.S
@@ -563,6 +563,13 @@ 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
@@ -572,12 +579,8 @@ 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,
@@ -831,6 +834,32 @@ 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
@@ -848,11 +877,12 @@ 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
@@ -894,20 +924,15 @@ 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
@@ -1170,6 +1195,14 @@ 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
diff --git a/arch/x86/entry/entry_64_compat.S b/arch/x86/entry/entry_64_compat.S
index dcc6987f9bae..95ad40eb7eff 100644
--- a/arch/x86/entry/entry_64_compat.S
+++ b/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 */
diff --git a/arch/x86/include/asm/switch_to.h b/arch/x86/include/asm/switch_to.h
index 8c6bd6863db9..a6796ac8d311 100644
--- a/arch/x86/include/asm/switch_to.h
+++ b/arch/x86/include/asm/switch_to.h
@@ -93,7 +93,7 @@ static inline void update_sp0(struct task_struct *task)
 #ifdef CONFIG_X86_32
 	load_sp0(task->thread.sp0);
 #else
-	load_sp0(task_top_of_stack(task));
+	/* On x86_64, sp0 always points to the entry trampoline stack. */
 #endif
 }
 
diff --git a/arch/x86/include/asm/traps.h b/arch/x86/include/asm/traps.h
index 1fadd310ff68..31051f35cbb7 100644
--- a/arch/x86/include/asm/traps.h
+++ b/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);
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index c67742df569a..7c82a8a8bfda 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -1645,11 +1645,13 @@ 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);
 
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 1bd43f044c62..cbc4272bb9dd 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -359,8 +359,7 @@ dotraplinkage void do_double_fault(struct pt_regs *regs, long error_code)
 		regs->cs == __KERNEL_CS &&
 		regs->ip == (unsigned long)native_irq_return_iret)
 	{
-		struct pt_regs *normal_regs =
-			(struct pt_regs *)this_cpu_read(cpu_tss.x86_tss.sp0) - 1;
+		struct pt_regs *normal_regs = (struct pt_regs *)this_cpu_read(cpu_tss.x86_tss.sp0) - 1;
 
 		/* Fake a #GP(0) from userspace. */
 		memmove(&normal_regs->ip, (void *)regs->sp, 5*8);
@@ -611,9 +610,10 @@ NOKPROBE_SYMBOL(do_int3);
  * 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)
+asmlinkage __visible notrace
+struct pt_regs *sync_regs(struct pt_regs *eregs)
 {
-	struct pt_regs *regs = task_pt_regs(current);
+	struct pt_regs *regs = (struct pt_regs *)this_cpu_read(cpu_current_top_of_stack) - 1;
 	*regs = *eregs;
 	return regs;
 }
@@ -630,13 +630,13 @@ 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 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);
-- 
2.13.6

  parent reply index

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-11-24  4:32 [PATCH v3 00/19] Entry stack switching Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 01/19] x86/asm/64: Allocate and enable the SYSENTER stack Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 02/19] x86/dumpstack: Add get_stack_info() support for " Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 03/19] x86/gdt: Put per-cpu GDT remaps in ascending order Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 04/19] x86/fixmap: Generalize the GDT fixmap mechanism Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 05/19] x86/kasan/64: Teach KASAN about the cpu_entry_area Andy Lutomirski
2017-11-24 13:16   ` Andrey Ryabinin
2017-11-24 16:27     ` Andy Lutomirski
2017-11-24 17:21       ` [PATCH v4] x86/mm/kasan: " Andrey Ryabinin
2017-11-24 19:26         ` Andy Lutomirski
2017-11-24 17:26       ` [PATCH v3 05/19] x86/kasan/64: " Andrey Ryabinin
2017-11-24  4:32 ` [PATCH v3 06/19] x86/asm: Fix assumptions that the HW TSS is at the beginning of cpu_tss Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 07/19] x86/dumpstack: Handle stack overflow on all stacks Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 08/19] x86/asm: Move SYSENTER_stack to the beginning of struct tss_struct Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 09/19] x86/asm: Remap the TSS into the cpu entry area Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 10/19] x86/asm/64: Separate cpu_current_top_of_stack from TSS.sp0 Andy Lutomirski
2017-11-24  4:32 ` [PATCH v3 11/19] x86/espfix/64: Stop assuming that pt_regs is on the entry stack Andy Lutomirski
2017-11-24  4:32 ` Andy Lutomirski [this message]
2017-11-24  4:32 ` [PATCH v3 13/19] x86/asm/64: Return to userspace from the trampoline stack Andy Lutomirski
2017-11-24  4:33 ` [PATCH v3 14/19] x86/entry/64: Create a percpu SYSCALL entry trampoline Andy Lutomirski
2017-11-24 11:39   ` David Laight
2017-12-01  5:56     ` Andy Lutomirski
2017-11-24  4:33 ` [PATCH v3 15/19] x86/irq: Remove an old outdated comment about context tracking races Andy Lutomirski
2017-11-24  4:33 ` [PATCH v3 16/19] x86/irq/64: In the stack overflow warning, print the offending IP Andy Lutomirski
2017-11-24  4:33 ` [PATCH v3 17/19] x86/entry/64: Move the IST stacks into cpu_entry_area Andy Lutomirski
2017-11-24  4:33 ` [PATCH v3 18/19] x86/entry/64: Remove the SYSENTER stack canary Andy Lutomirski
2017-11-24  4:33 ` [PATCH v3 19/19] x86/entry: Clean up SYSENTER_stack code Andy Lutomirski

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=fa3958723a1a85baeaf309c735b775841205800e.1511497875.git.luto@kernel.org \
    --to=luto@kernel.org \
    --cc=bpetkov@suse.de \
    --cc=brgerst@gmail.com \
    --cc=dave.hansen@intel.com \
    --cc=jpoimboe@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

LKML Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/lkml/0 lkml/git/0.git
	git clone --mirror https://lore.kernel.org/lkml/1 lkml/git/1.git
	git clone --mirror https://lore.kernel.org/lkml/2 lkml/git/2.git
	git clone --mirror https://lore.kernel.org/lkml/3 lkml/git/3.git
	git clone --mirror https://lore.kernel.org/lkml/4 lkml/git/4.git
	git clone --mirror https://lore.kernel.org/lkml/5 lkml/git/5.git
	git clone --mirror https://lore.kernel.org/lkml/6 lkml/git/6.git
	git clone --mirror https://lore.kernel.org/lkml/7 lkml/git/7.git
	git clone --mirror https://lore.kernel.org/lkml/8 lkml/git/8.git
	git clone --mirror https://lore.kernel.org/lkml/9 lkml/git/9.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 lkml lkml/ https://lore.kernel.org/lkml \
		linux-kernel@vger.kernel.org
	public-inbox-index lkml

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-kernel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git