Merge tag 'x86-entry-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 entry updates from Ingo Molnar:

 - Make IA32_EMULATION boot time configurable with
   the new ia32_emulation=<bool> boot option

 - Clean up fast syscall return validation code: convert
   it to C and refactor the code

 - As part of this, optimize the canonical RIP test code

* tag 'x86-entry-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/entry/32: Clean up syscall fast exit tests
  x86/entry/64: Use TASK_SIZE_MAX for canonical RIP test
  x86/entry/64: Convert SYSRET validation tests to C
  x86/entry/32: Remove SEP test for SYSEXIT
  x86/entry/32: Convert do_fast_syscall_32() to bool return type
  x86/entry/compat: Combine return value test from syscall handler
  x86/entry/64: Remove obsolete comment on tracing vs. SYSRET
  x86: Make IA32_EMULATION boot time configurable
  x86/entry: Make IA32 syscalls' availability depend on ia32_enabled()
  x86/elf: Make loading of 32bit processes depend on ia32_enabled()
  x86/entry: Compile entry_SYSCALL32_ignore() unconditionally
  x86/entry: Rename ignore_sysret()
  x86: Introduce ia32_enabled()

Documentation/admin-guide/kernel-parameters.txt

@@ -1893,6 +1893,12 @@
 			 0 -- machine default
 			 1 -- force brightness inversion
 
+	ia32_emulation=	[X86-64]
+			Format: <bool>
+			When true, allows loading 32-bit programs and executing 32-bit
+			syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at
+			boot time. When false, unconditionally disables IA32 emulation.
+
 	icn=		[HW,ISDN]
 			Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
 

arch/x86/Kconfig

@@ -2955,6 +2955,15 @@ config IA32_EMULATION
 	  64-bit kernel. You should likely turn this on, unless you're
 	  100% sure that you don't have any 32-bit programs left.
 
+config IA32_EMULATION_DEFAULT_DISABLED
+	bool "IA32 emulation disabled by default"
+	default n
+	depends on IA32_EMULATION
+	help
+	  Make IA32 emulation disabled by default. This prevents loading 32-bit
+	  processes and access to 32-bit syscalls. If unsure, leave it to its
+	  default value.
+
 config X86_X32_ABI
 	bool "x32 ABI for 64-bit mode"
 	depends on X86_64

arch/x86/entry/common.c

@@ -19,6 +19,7 @@
 #include <linux/nospec.h>
 #include <linux/syscalls.h>
 #include <linux/uaccess.h>
+#include <linux/init.h>
 
 #ifdef CONFIG_XEN_PV
 #include <xen/xen-ops.h>
@@ -70,7 +71,8 @@ static __always_inline bool do_syscall_x32(struct pt_regs *regs, int nr)
 	return false;
 }
 
-__visible noinstr void do_syscall_64(struct pt_regs *regs, int nr)
+/* Returns true to return using SYSRET, or false to use IRET */
+__visible noinstr bool do_syscall_64(struct pt_regs *regs, int nr)
 {
 	add_random_kstack_offset();
 	nr = syscall_enter_from_user_mode(regs, nr);
@@ -84,6 +86,46 @@ __visible noinstr void do_syscall_64(struct pt_regs *regs, int nr)
 
 	instrumentation_end();
 	syscall_exit_to_user_mode(regs);
+
+	/*
+	 * Check that the register state is valid for using SYSRET to exit
+	 * to userspace.  Otherwise use the slower but fully capable IRET
+	 * exit path.
+	 */
+
+	/* XEN PV guests always use the IRET path */
+	if (cpu_feature_enabled(X86_FEATURE_XENPV))
+		return false;
+
+	/* SYSRET requires RCX == RIP and R11 == EFLAGS */
+	if (unlikely(regs->cx != regs->ip || regs->r11 != regs->flags))
+		return false;
+
+	/* CS and SS must match the values set in MSR_STAR */
+	if (unlikely(regs->cs != __USER_CS || regs->ss != __USER_DS))
+		return false;
+
+	/*
+	 * On Intel CPUs, SYSRET with non-canonical RCX/RIP will #GP
+	 * in kernel space.  This essentially lets the user take over
+	 * the kernel, since userspace controls RSP.
+	 *
+	 * TASK_SIZE_MAX covers all user-accessible addresses other than
+	 * the deprecated vsyscall page.
+	 */
+	if (unlikely(regs->ip >= TASK_SIZE_MAX))
+		return false;
+
+	/*
+	 * SYSRET cannot restore RF.  It can restore TF, but unlike IRET,
+	 * restoring TF results in a trap from userspace immediately after
+	 * SYSRET.
+	 */
+	if (unlikely(regs->flags & (X86_EFLAGS_RF | X86_EFLAGS_TF)))
+		return false;
+
+	/* Use SYSRET to exit to userspace */
+	return true;
 }
 #endif
 
@@ -96,6 +138,16 @@ static __always_inline int syscall_32_enter(struct pt_regs *regs)
 	return (int)regs->orig_ax;
 }
 
+#ifdef CONFIG_IA32_EMULATION
+bool __ia32_enabled __ro_after_init = !IS_ENABLED(CONFIG_IA32_EMULATION_DEFAULT_DISABLED);
+
+static int ia32_emulation_override_cmdline(char *arg)
+{
+	return kstrtobool(arg, &__ia32_enabled);
+}
+early_param("ia32_emulation", ia32_emulation_override_cmdline);
+#endif
+
 /*
  * Invoke a 32-bit syscall.  Called with IRQs on in CONTEXT_KERNEL.
  */
@@ -182,8 +234,8 @@ static noinstr bool __do_fast_syscall_32(struct pt_regs *regs)
 	return true;
 }
 
-/* Returns 0 to return using IRET or 1 to return using SYSEXIT/SYSRETL. */
-__visible noinstr long do_fast_syscall_32(struct pt_regs *regs)
+/* Returns true to return using SYSEXIT/SYSRETL, or false to use IRET */
+__visible noinstr bool do_fast_syscall_32(struct pt_regs *regs)
 {
 	/*
 	 * Called using the internal vDSO SYSENTER/SYSCALL32 calling
@@ -201,41 +253,36 @@ __visible noinstr long do_fast_syscall_32(struct pt_regs *regs)
 
 	/* Invoke the syscall. If it failed, keep it simple: use IRET. */
 	if (!__do_fast_syscall_32(regs))
-		return 0;
+		return false;
 
-#ifdef CONFIG_X86_64
 	/*
-	 * Opportunistic SYSRETL: if possible, try to return using SYSRETL.
-	 * SYSRETL is available on all 64-bit CPUs, so we don't need to
-	 * bother with SYSEXIT.
-	 *
-	 * Unlike 64-bit opportunistic SYSRET, we can't check that CX == IP,
-	 * because the ECX fixup above will ensure that this is essentially
-	 * never the case.
-	 */
-	return regs->cs == __USER32_CS && regs->ss == __USER_DS &&
-		regs->ip == landing_pad &&
-		(regs->flags & (X86_EFLAGS_RF | X86_EFLAGS_TF)) == 0;
-#else
-	/*
-	 * Opportunistic SYSEXIT: if possible, try to return using SYSEXIT.
-	 *
-	 * Unlike 64-bit opportunistic SYSRET, we can't check that CX == IP,
-	 * because the ECX fixup above will ensure that this is essentially
-	 * never the case.
-	 *
-	 * We don't allow syscalls at all from VM86 mode, but we still
-	 * need to check VM, because we might be returning from sys_vm86.
+	 * Check that the register state is valid for using SYSRETL/SYSEXIT
+	 * to exit to userspace.  Otherwise use the slower but fully capable
+	 * IRET exit path.
 	 */
-	return static_cpu_has(X86_FEATURE_SEP) &&
-		regs->cs == __USER_CS && regs->ss == __USER_DS &&
-		regs->ip == landing_pad &&
-		(regs->flags & (X86_EFLAGS_RF | X86_EFLAGS_TF | X86_EFLAGS_VM)) == 0;
-#endif
+
+	/* XEN PV guests always use the IRET path */
+	if (cpu_feature_enabled(X86_FEATURE_XENPV))
+		return false;
+
+	/* EIP must point to the VDSO landing pad */
+	if (unlikely(regs->ip != landing_pad))
+		return false;
+
+	/* CS and SS must match the values set in MSR_STAR */
+	if (unlikely(regs->cs != __USER32_CS || regs->ss != __USER_DS))
+		return false;
+
+	/* If the TF, RF, or VM flags are set, use IRET */
+	if (unlikely(regs->flags & (X86_EFLAGS_RF | X86_EFLAGS_TF | X86_EFLAGS_VM)))
+		return false;
+
+	/* Use SYSRETL/SYSEXIT to exit to userspace */
+	return true;
 }
 
-/* Returns 0 to return using IRET or 1 to return using SYSEXIT/SYSRETL. */
-__visible noinstr long do_SYSENTER_32(struct pt_regs *regs)
+/* Returns true to return using SYSEXIT/SYSRETL, or false to use IRET */
+__visible noinstr bool do_SYSENTER_32(struct pt_regs *regs)
 {
 	/* SYSENTER loses RSP, but the vDSO saved it in RBP. */
 	regs->sp = regs->bp;

arch/x86/entry/entry_32.S

@@ -837,7 +837,7 @@ SYM_FUNC_START(entry_SYSENTER_32)
 
 	movl	%esp, %eax
 	call	do_SYSENTER_32
-	testl	%eax, %eax
+	testb	%al, %al
 	jz	.Lsyscall_32_done
 
 	STACKLEAK_ERASE

arch/x86/entry/entry_64.S

@@ -126,70 +126,8 @@ SYM_INNER_LABEL(entry_SYSCALL_64_after_hwframe, SYM_L_GLOBAL)
 	 * In the Xen PV case we must use iret anyway.
 	 */
 
-	ALTERNATIVE "", "jmp	swapgs_restore_regs_and_return_to_usermode", \
-		X86_FEATURE_XENPV
-
-	movq	RCX(%rsp), %rcx
-	movq	RIP(%rsp), %r11
-
-	cmpq	%rcx, %r11	/* SYSRET requires RCX == RIP */
-	jne	swapgs_restore_regs_and_return_to_usermode
-
-	/*
-	 * On Intel CPUs, SYSRET with non-canonical RCX/RIP will #GP
-	 * in kernel space.  This essentially lets the user take over
-	 * the kernel, since userspace controls RSP.
-	 *
-	 * If width of "canonical tail" ever becomes variable, this will need
-	 * to be updated to remain correct on both old and new CPUs.
-	 *
-	 * Change top bits to match most significant bit (47th or 56th bit
-	 * depending on paging mode) in the address.
-	 */
-#ifdef CONFIG_X86_5LEVEL
-	ALTERNATIVE "shl $(64 - 48), %rcx; sar $(64 - 48), %rcx", \
-		"shl $(64 - 57), %rcx; sar $(64 - 57), %rcx", X86_FEATURE_LA57
-#else
-	shl	$(64 - (__VIRTUAL_MASK_SHIFT+1)), %rcx
-	sar	$(64 - (__VIRTUAL_MASK_SHIFT+1)), %rcx
-#endif
-
-	/* If this changed %rcx, it was not canonical */
-	cmpq	%rcx, %r11
-	jne	swapgs_restore_regs_and_return_to_usermode
-
-	cmpq	$__USER_CS, CS(%rsp)		/* CS must match SYSRET */
-	jne	swapgs_restore_regs_and_return_to_usermode
-
-	movq	R11(%rsp), %r11
-	cmpq	%r11, EFLAGS(%rsp)		/* R11 == RFLAGS */
-	jne	swapgs_restore_regs_and_return_to_usermode
-
-	/*
-	 * SYSCALL clears RF when it saves RFLAGS in R11 and SYSRET cannot
-	 * restore RF properly. If the slowpath sets it for whatever reason, we
-	 * need to restore it correctly.
-	 *
-	 * SYSRET can restore TF, but unlike IRET, restoring TF results in a
-	 * trap from userspace immediately after SYSRET.  This would cause an
-	 * infinite loop whenever #DB happens with register state that satisfies
-	 * the opportunistic SYSRET conditions.  For example, single-stepping
-	 * this user code:
-	 *
-	 *           movq	$stuck_here, %rcx
-	 *           pushfq
-	 *           popq %r11
-	 *   stuck_here:
-	 *
-	 * would never get past 'stuck_here'.
-	 */
-	testq	$(X86_EFLAGS_RF|X86_EFLAGS_TF), %r11
-	jnz	swapgs_restore_regs_and_return_to_usermode
-
-	/* nothing to check for RSP */
-
-	cmpq	$__USER_DS, SS(%rsp)		/* SS must match SYSRET */
-	jne	swapgs_restore_regs_and_return_to_usermode
+	ALTERNATIVE "testb %al, %al; jz swapgs_restore_regs_and_return_to_usermode", \
+		"jmp swapgs_restore_regs_and_return_to_usermode", X86_FEATURE_XENPV
 
 	/*
 	 * We win! This label is here just for ease of understanding
@@ -1509,18 +1447,16 @@ nmi_restore:
 	iretq
 SYM_CODE_END(asm_exc_nmi)
 
-#ifndef CONFIG_IA32_EMULATION
 /*
  * This handles SYSCALL from 32-bit code.  There is no way to program
  * MSRs to fully disable 32-bit SYSCALL.
  */
-SYM_CODE_START(ignore_sysret)
+SYM_CODE_START(entry_SYSCALL32_ignore)
 	UNWIND_HINT_END_OF_STACK
 	ENDBR
 	mov	$-ENOSYS, %eax
 	sysretl
-SYM_CODE_END(ignore_sysret)
-#endif
+SYM_CODE_END(entry_SYSCALL32_ignore)
 
 .pushsection .text, "ax"
 	__FUNC_ALIGN

arch/x86/entry/entry_64_compat.S

@@ -118,9 +118,6 @@ SYM_INNER_LABEL(entry_SYSENTER_compat_after_hwframe, SYM_L_GLOBAL)
 
 	movq	%rsp, %rdi
 	call	do_SYSENTER_32
-	/* XEN PV guests always use IRET path */
-	ALTERNATIVE "testl %eax, %eax; jz swapgs_restore_regs_and_return_to_usermode", \
-		    "jmp swapgs_restore_regs_and_return_to_usermode", X86_FEATURE_XENPV
 	jmp	sysret32_from_system_call
 
 .Lsysenter_fix_flags:
@@ -212,13 +209,15 @@ SYM_INNER_LABEL(entry_SYSCALL_compat_after_hwframe, SYM_L_GLOBAL)
 
 	movq	%rsp, %rdi
 	call	do_fast_syscall_32
+
+sysret32_from_system_call:
 	/* XEN PV guests always use IRET path */
-	ALTERNATIVE "testl %eax, %eax; jz swapgs_restore_regs_and_return_to_usermode", \
+	ALTERNATIVE "testb %al, %al; jz swapgs_restore_regs_and_return_to_usermode", \
 		    "jmp swapgs_restore_regs_and_return_to_usermode", X86_FEATURE_XENPV
 
-	/* Opportunistic SYSRET */
-sysret32_from_system_call:
 	/*
+	 * Opportunistic SYSRET
+	 *
 	 * We are not going to return to userspace from the trampoline
 	 * stack. So let's erase the thread stack right now.
 	 */

arch/x86/include/asm/elf.h

@@ -7,6 +7,7 @@
  */
 #include <linux/thread_info.h>
 
+#include <asm/ia32.h>
 #include <asm/ptrace.h>
 #include <asm/user.h>
 #include <asm/auxvec.h>
@@ -149,7 +150,7 @@ do {						\
 	((x)->e_machine == EM_X86_64)
 
 #define compat_elf_check_arch(x)					\
-	(elf_check_arch_ia32(x) ||					\
+	((elf_check_arch_ia32(x) && ia32_enabled()) ||			\
 	 (IS_ENABLED(CONFIG_X86_X32_ABI) && (x)->e_machine == EM_X86_64))
 
 static inline void elf_common_init(struct thread_struct *t,

arch/x86/include/asm/ia32.h

@@ -68,6 +68,20 @@ extern void ia32_pick_mmap_layout(struct mm_struct *mm);
 
 #endif
 
-#endif /* CONFIG_IA32_EMULATION */
+extern bool __ia32_enabled;
+
+static inline bool ia32_enabled(void)
+{
+	return __ia32_enabled;
+}
+
+#else /* !CONFIG_IA32_EMULATION */
+
+static inline bool ia32_enabled(void)
+{
+	return IS_ENABLED(CONFIG_X86_32);
+}
+
+#endif
 
 #endif /* _ASM_X86_IA32_H */

arch/x86/include/asm/processor.h

@@ -399,7 +399,7 @@ static inline unsigned long cpu_kernelmode_gs_base(int cpu)
 	return (unsigned long)per_cpu(fixed_percpu_data.gs_base, cpu);
 }
 
-extern asmlinkage void ignore_sysret(void);
+extern asmlinkage void entry_SYSCALL32_ignore(void);
 
 /* Save actual FS/GS selectors and bases to current->thread */
 void current_save_fsgs(void);

arch/x86/include/asm/proto.h

@@ -36,6 +36,9 @@ void entry_INT80_compat(void);
 #ifdef CONFIG_XEN_PV
 void xen_entry_INT80_compat(void);
 #endif
+#else /* !CONFIG_IA32_EMULATION */
+#define entry_SYSCALL_compat NULL
+#define entry_SYSENTER_compat NULL
 #endif
 
 void x86_configure_nx(void);

arch/x86/include/asm/syscall.h

@@ -126,12 +126,12 @@ static inline int syscall_get_arch(struct task_struct *task)
 		? AUDIT_ARCH_I386 : AUDIT_ARCH_X86_64;
 }
 
-void do_syscall_64(struct pt_regs *regs, int nr);
+bool do_syscall_64(struct pt_regs *regs, int nr);
 
 #endif	/* CONFIG_X86_32 */
 
 void do_int80_syscall_32(struct pt_regs *regs);
-long do_fast_syscall_32(struct pt_regs *regs);
-long do_SYSENTER_32(struct pt_regs *regs);
+bool do_fast_syscall_32(struct pt_regs *regs);
+bool do_SYSENTER_32(struct pt_regs *regs);
 
 #endif	/* _ASM_X86_SYSCALL_H */

arch/x86/kernel/cpu/common.c

@@ -62,6 +62,7 @@
 #include <asm/intel-family.h>
 #include <asm/cpu_device_id.h>
 #include <asm/uv/uv.h>
+#include <asm/ia32.h>
 #include <asm/set_memory.h>
 #include <asm/traps.h>
 #include <asm/sev.h>
@@ -2074,7 +2075,7 @@ void syscall_init(void)
 	wrmsr(MSR_STAR, 0, (__USER32_CS << 16) | __KERNEL_CS);
 	wrmsrl(MSR_LSTAR, (unsigned long)entry_SYSCALL_64);
 
-#ifdef CONFIG_IA32_EMULATION
+	if (ia32_enabled()) {
 		wrmsrl_cstar((unsigned long)entry_SYSCALL_compat);
 		/*
 		 * This only works on Intel CPUs.
@@ -2086,12 +2087,12 @@ void syscall_init(void)
 		wrmsrl_safe(MSR_IA32_SYSENTER_ESP,
 			    (unsigned long)(cpu_entry_stack(smp_processor_id()) + 1));
 		wrmsrl_safe(MSR_IA32_SYSENTER_EIP, (u64)entry_SYSENTER_compat);
-#else
-	wrmsrl_cstar((unsigned long)ignore_sysret);
+	} else {
+		wrmsrl_cstar((unsigned long)entry_SYSCALL32_ignore);
 		wrmsrl_safe(MSR_IA32_SYSENTER_CS, (u64)GDT_ENTRY_INVALID_SEG);
 		wrmsrl_safe(MSR_IA32_SYSENTER_ESP, 0ULL);
 		wrmsrl_safe(MSR_IA32_SYSENTER_EIP, 0ULL);
-#endif
+	}
 
 	/*
 	 * Flags to clear on syscall; clear as much as possible

arch/x86/kernel/idt.c

@@ -10,6 +10,7 @@
 #include <asm/proto.h>
 #include <asm/desc.h>
 #include <asm/hw_irq.h>
+#include <asm/ia32.h>
 #include <asm/idtentry.h>
 
 #define DPL0		0x0
@@ -116,6 +117,9 @@ static const __initconst struct idt_data def_idts[] = {
 #endif
 
 	SYSG(X86_TRAP_OF,		asm_exc_overflow),
+};
+
+static const struct idt_data ia32_idt[] __initconst = {
 #if defined(CONFIG_IA32_EMULATION)
 	SYSG(IA32_SYSCALL_VECTOR,	entry_INT80_compat),
 #elif defined(CONFIG_X86_32)
@@ -225,6 +229,9 @@ void __init idt_setup_early_traps(void)
 void __init idt_setup_traps(void)
 {
 	idt_setup_from_table(idt_table, def_idts, ARRAY_SIZE(def_idts), true);
+
+	if (ia32_enabled())
+		idt_setup_from_table(idt_table, ia32_idt, ARRAY_SIZE(ia32_idt), true);
 }
 
 #ifdef CONFIG_X86_64