Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 fixes from Ingo Molnar:
 "A handful of x86 fixes:

   - a syscall ABI fix, fixing an Android breakage
   - a Xen PV guest fix relating to the RTC device, causing a
     non-working console
   - a Xen guest syscall stack frame fix
   - an MCE hotplug CPU crash fix"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/numachip: Fix NumaConnect2 MMCFG PCI access
  x86/entry: Restore traditional SYSENTER calling convention
  x86/entry: Fix some comments
  x86/paravirt: Prevent rtc_cmos platform device init on PV guests
  x86/xen: Avoid fast syscall path for Xen PV guests
  x86/mce: Ensure offline CPUs don't participate in rendezvous process

arch/x86/entry/common.c

@@ -421,7 +421,7 @@ __visible long do_fast_syscall_32(struct pt_regs *regs)
 	regs->ip = landing_pad;
 
 	/*
-	 * Fetch ECX from where the vDSO stashed it.
+	 * Fetch EBP from where the vDSO stashed it.
 	 *
 	 * WARNING: We are in CONTEXT_USER and RCU isn't paying attention!
 	 */
@@ -432,10 +432,10 @@ __visible long do_fast_syscall_32(struct pt_regs *regs)
 		 * Micro-optimization: the pointer we're following is explicitly
 		 * 32 bits, so it can't be out of range.
 		 */
-		__get_user(*(u32 *)&regs->cx,
+		__get_user(*(u32 *)&regs->bp,
 			    (u32 __user __force *)(unsigned long)(u32)regs->sp)
 #else
-		get_user(*(u32 *)&regs->cx,
+		get_user(*(u32 *)&regs->bp,
 			 (u32 __user __force *)(unsigned long)(u32)regs->sp)
 #endif
 		) {

arch/x86/entry/entry_32.S

@@ -292,7 +292,7 @@ ENTRY(entry_SYSENTER_32)
 	movl	TSS_sysenter_sp0(%esp), %esp
 sysenter_past_esp:
 	pushl	$__USER_DS		/* pt_regs->ss */
-	pushl	%ecx			/* pt_regs->cx */
+	pushl	%ebp			/* pt_regs->sp (stashed in bp) */
 	pushfl				/* pt_regs->flags (except IF = 0) */
 	orl	$X86_EFLAGS_IF, (%esp)	/* Fix IF */
 	pushl	$__USER_CS		/* pt_regs->cs */
@@ -308,8 +308,9 @@ sysenter_past_esp:
 
 	movl	%esp, %eax
 	call	do_fast_syscall_32
-	testl	%eax, %eax
-	jz	.Lsyscall_32_done
+	/* XEN PV guests always use IRET path */
+	ALTERNATIVE "testl %eax, %eax; jz .Lsyscall_32_done", \
+		    "jmp .Lsyscall_32_done", X86_FEATURE_XENPV
 
 /* Opportunistic SYSEXIT */
 	TRACE_IRQS_ON			/* User mode traces as IRQs on. */

arch/x86/entry/entry_64_compat.S

@@ -63,7 +63,7 @@ ENTRY(entry_SYSENTER_compat)
 
 	/* Construct struct pt_regs on stack */
 	pushq	$__USER32_DS		/* pt_regs->ss */
-	pushq	%rcx			/* pt_regs->sp */
+	pushq	%rbp			/* pt_regs->sp (stashed in bp) */
 
 	/*
 	 * Push flags.  This is nasty.  First, interrupts are currently
@@ -82,14 +82,14 @@ ENTRY(entry_SYSENTER_compat)
 	pushq	%rdi			/* pt_regs->di */
 	pushq	%rsi			/* pt_regs->si */
 	pushq	%rdx			/* pt_regs->dx */
-	pushq	%rcx			/* pt_regs->cx (will be overwritten) */
+	pushq	%rcx			/* pt_regs->cx */
 	pushq	$-ENOSYS		/* pt_regs->ax */
 	pushq   %r8                     /* pt_regs->r8  = 0 */
 	pushq   %r8                     /* pt_regs->r9  = 0 */
 	pushq   %r8                     /* pt_regs->r10 = 0 */
 	pushq   %r8                     /* pt_regs->r11 = 0 */
 	pushq   %rbx                    /* pt_regs->rbx */
-	pushq   %rbp                    /* pt_regs->rbp */
+	pushq   %rbp                    /* pt_regs->rbp (will be overwritten) */
 	pushq   %r8                     /* pt_regs->r12 = 0 */
 	pushq   %r8                     /* pt_regs->r13 = 0 */
 	pushq   %r8                     /* pt_regs->r14 = 0 */
@@ -121,8 +121,9 @@ sysenter_flags_fixed:
 
 	movq	%rsp, %rdi
 	call	do_fast_syscall_32
-	testl	%eax, %eax
-	jz	.Lsyscall_32_done
+	/* XEN PV guests always use IRET path */
+	ALTERNATIVE "testl %eax, %eax; jz .Lsyscall_32_done", \
+		    "jmp .Lsyscall_32_done", X86_FEATURE_XENPV
 	jmp	sysret32_from_system_call
 
 sysenter_fix_flags:
@@ -178,7 +179,7 @@ ENTRY(entry_SYSCALL_compat)
 	pushq	%rdi			/* pt_regs->di */
 	pushq	%rsi			/* pt_regs->si */
 	pushq	%rdx			/* pt_regs->dx */
-	pushq	%rcx			/* pt_regs->cx (will be overwritten) */
+	pushq	%rbp			/* pt_regs->cx (stashed in bp) */
 	pushq	$-ENOSYS		/* pt_regs->ax */
 	xorq    %r8,%r8
 	pushq   %r8                     /* pt_regs->r8  = 0 */
@@ -186,7 +187,7 @@ ENTRY(entry_SYSCALL_compat)
 	pushq   %r8                     /* pt_regs->r10 = 0 */
 	pushq   %r8                     /* pt_regs->r11 = 0 */
 	pushq   %rbx                    /* pt_regs->rbx */
-	pushq   %rbp                    /* pt_regs->rbp */
+	pushq   %rbp                    /* pt_regs->rbp (will be overwritten) */
 	pushq   %r8                     /* pt_regs->r12 = 0 */
 	pushq   %r8                     /* pt_regs->r13 = 0 */
 	pushq   %r8                     /* pt_regs->r14 = 0 */
@@ -200,8 +201,9 @@ ENTRY(entry_SYSCALL_compat)
 
 	movq	%rsp, %rdi
 	call	do_fast_syscall_32
-	testl	%eax, %eax
-	jz	.Lsyscall_32_done
+	/* XEN PV guests always use IRET path */
+	ALTERNATIVE "testl %eax, %eax; jz .Lsyscall_32_done", \
+		    "jmp .Lsyscall_32_done", X86_FEATURE_XENPV
 
 	/* Opportunistic SYSRET */
 sysret32_from_system_call:

arch/x86/entry/vdso/vdso32/system_call.S

 /*
- * Code for the vDSO.  This version uses the old int $0x80 method.
+ * AT_SYSINFO entry point
 */
 
 #include <asm/dwarf2.h>
@@ -21,35 +21,67 @@ __kernel_vsyscall:
 	/*
 	 * Reshuffle regs so that all of any of the entry instructions
 	 * will preserve enough state.
+	 *
+	 * A really nice entry sequence would be:
+	 *  pushl %edx
+	 *  pushl %ecx
+	 *  movl  %esp, %ecx
+	 *
+	 * Unfortunately, naughty Android versions between July and December
+	 * 2015 actually hardcode the traditional Linux SYSENTER entry
+	 * sequence.  That is severely broken for a number of reasons (ask
+	 * anyone with an AMD CPU, for example).  Nonetheless, we try to keep
+	 * it working approximately as well as it ever worked.
+	 *
+	 * This link may eludicate some of the history:
+	 *   https://android-review.googlesource.com/#/q/Iac3295376d61ef83e713ac9b528f3b50aa780cd7
+	 * personally, I find it hard to understand what's going on there.
+	 *
+	 * Note to future user developers: DO NOT USE SYSENTER IN YOUR CODE.
+	 * Execute an indirect call to the address in the AT_SYSINFO auxv
+	 * entry.  That is the ONLY correct way to make a fast 32-bit system
+	 * call on Linux.  (Open-coding int $0x80 is also fine, but it's
+	 * slow.)
 	 */
+	pushl	%ecx
+	CFI_ADJUST_CFA_OFFSET	4
+	CFI_REL_OFFSET		ecx, 0
 	pushl	%edx
 	CFI_ADJUST_CFA_OFFSET	4
 	CFI_REL_OFFSET		edx, 0
-	pushl	%ecx
+	pushl	%ebp
 	CFI_ADJUST_CFA_OFFSET	4
-	CFI_REL_OFFSET		ecx, 0
-	movl	%esp, %ecx
+	CFI_REL_OFFSET		ebp, 0
+
+	#define SYSENTER_SEQUENCE	"movl %esp, %ebp; sysenter"
+	#define SYSCALL_SEQUENCE	"movl %ecx, %ebp; syscall"
 
 #ifdef CONFIG_X86_64
 	/* If SYSENTER (Intel) or SYSCALL32 (AMD) is available, use it. */
-	ALTERNATIVE_2 "", "sysenter", X86_FEATURE_SYSENTER32, \
-	                  "syscall",  X86_FEATURE_SYSCALL32
+	ALTERNATIVE_2 "", SYSENTER_SEQUENCE, X86_FEATURE_SYSENTER32, \
+	                  SYSCALL_SEQUENCE,  X86_FEATURE_SYSCALL32
 #else
-	ALTERNATIVE "", "sysenter", X86_FEATURE_SEP
+	ALTERNATIVE "", SYSENTER_SEQUENCE, X86_FEATURE_SEP
 #endif
 
 	/* Enter using int $0x80 */
-	movl	(%esp), %ecx
 	int	$0x80
 GLOBAL(int80_landing_pad)
 
-	/* Restore ECX and EDX in case they were clobbered. */
-	popl	%ecx
-	CFI_RESTORE		ecx
+	/*
+	 * Restore EDX and ECX in case they were clobbered.  EBP is not
+	 * clobbered (the kernel restores it), but it's cleaner and
+	 * probably faster to pop it than to adjust ESP using addl.
+	 */
+	popl	%ebp
+	CFI_RESTORE		ebp
 	CFI_ADJUST_CFA_OFFSET	-4
 	popl	%edx
 	CFI_RESTORE		edx
 	CFI_ADJUST_CFA_OFFSET	-4
+	popl	%ecx
+	CFI_RESTORE		ecx
+	CFI_ADJUST_CFA_OFFSET	-4
 	ret
 	CFI_ENDPROC
 

arch/x86/include/asm/cpufeature.h

@@ -216,6 +216,7 @@
 #define X86_FEATURE_PAUSEFILTER ( 8*32+13) /* AMD filtered pause intercept */
 #define X86_FEATURE_PFTHRESHOLD ( 8*32+14) /* AMD pause filter threshold */
 #define X86_FEATURE_VMMCALL     ( 8*32+15) /* Prefer vmmcall to vmcall */
+#define X86_FEATURE_XENPV       ( 8*32+16) /* "" Xen paravirtual guest */
 
 
 /* Intel-defined CPU features, CPUID level 0x00000007:0 (ebx), word 9 */

arch/x86/include/asm/paravirt.h

@@ -19,6 +19,12 @@ static inline int paravirt_enabled(void)
 	return pv_info.paravirt_enabled;
 }
 
+static inline int paravirt_has_feature(unsigned int feature)
+{
+	WARN_ON_ONCE(!pv_info.paravirt_enabled);
+	return (pv_info.features & feature);
+}
+
 static inline void load_sp0(struct tss_struct *tss,
 			     struct thread_struct *thread)
 {

arch/x86/include/asm/paravirt_types.h

@@ -70,9 +70,14 @@ struct pv_info {
 #endif
 
 	int paravirt_enabled;
+	unsigned int features;	  /* valid only if paravirt_enabled is set */
 	const char *name;
 };
 
+#define paravirt_has(x) paravirt_has_feature(PV_SUPPORTED_##x)
+/* Supported features */
+#define PV_SUPPORTED_RTC        (1<<0)
+
 struct pv_init_ops {
 	/*
 	 * Patch may replace one of the defined code sequences with

arch/x86/include/asm/processor.h

@@ -472,6 +472,7 @@ static inline unsigned long current_top_of_stack(void)
 #else
 #define __cpuid			native_cpuid
 #define paravirt_enabled()	0
+#define paravirt_has(x) 	0
 
 static inline void load_sp0(struct tss_struct *tss,
 			    struct thread_struct *thread)

arch/x86/kernel/apic/apic_numachip.c

@@ -193,20 +193,17 @@ static int __init numachip_system_init(void)
 	case 1:
 		init_extra_mapping_uc(NUMACHIP_LCSR_BASE, NUMACHIP_LCSR_SIZE);
 		numachip_apic_icr_write = numachip1_apic_icr_write;
-		x86_init.pci.arch_init = pci_numachip_init;
 		break;
 	case 2:
 		init_extra_mapping_uc(NUMACHIP2_LCSR_BASE, NUMACHIP2_LCSR_SIZE);
 		numachip_apic_icr_write = numachip2_apic_icr_write;
-
-		/* Use MCFG config cycles rather than locked CF8 cycles */
-		raw_pci_ops = &pci_mmcfg;
 		break;
 	default:
 		return 0;
 	}
 
 	x86_cpuinit.fixup_cpu_id = fixup_cpu_id;
+	x86_init.pci.arch_init = pci_numachip_init;
 
 	return 0;
 }

arch/x86/kernel/cpu/mcheck/mce.c

@@ -999,6 +999,17 @@ void do_machine_check(struct pt_regs *regs, long error_code)
 	int flags = MF_ACTION_REQUIRED;
 	int lmce = 0;
 
+	/* If this CPU is offline, just bail out. */
+	if (cpu_is_offline(smp_processor_id())) {
+		u64 mcgstatus;
+
+		mcgstatus = mce_rdmsrl(MSR_IA32_MCG_STATUS);
+		if (mcgstatus & MCG_STATUS_RIPV) {
+			mce_wrmsrl(MSR_IA32_MCG_STATUS, 0);
+			return;
+		}
+	}
+
 	ist_enter(regs);
 
 	this_cpu_inc(mce_exception_count);

arch/x86/kernel/rtc.c

@@ -200,6 +200,9 @@ static __init int add_rtc_cmos(void)
 	}
 #endif
 
+	if (paravirt_enabled() && !paravirt_has(RTC))
+		return -ENODEV;
+
 	platform_device_register(&rtc_device);
 	dev_info(&rtc_device.dev,
 		 "registered platform RTC device (no PNP device found)\n");

arch/x86/lguest/boot.c

@@ -1414,6 +1414,7 @@ __init void lguest_init(void)
 	pv_info.kernel_rpl = 1;
 	/* Everyone except Xen runs with this set. */
 	pv_info.shared_kernel_pmd = 1;
+	pv_info.features = 0;
 
 	/*
 	 * We set up all the lguest overrides for sensitive operations.  These

arch/x86/xen/enlighten.c

@@ -1192,7 +1192,7 @@ static const struct pv_info xen_info __initconst = {
 #ifdef CONFIG_X86_64
 	.extra_user_64bit_cs = FLAT_USER_CS64,
 #endif
-
+	.features = 0,
 	.name = "Xen",
 };
 
@@ -1535,6 +1535,8 @@ asmlinkage __visible void __init xen_start_kernel(void)
 
 	/* Install Xen paravirt ops */
 	pv_info = xen_info;
+	if (xen_initial_domain())
+		pv_info.features |= PV_SUPPORTED_RTC;
 	pv_init_ops = xen_init_ops;
 	pv_apic_ops = xen_apic_ops;
 	if (!xen_pvh_domain()) {
@@ -1886,8 +1888,10 @@ EXPORT_SYMBOL_GPL(xen_hvm_need_lapic);
 
 static void xen_set_cpu_features(struct cpuinfo_x86 *c)
 {
-	if (xen_pv_domain())
+	if (xen_pv_domain()) {
 		clear_cpu_bug(c, X86_BUG_SYSRET_SS_ATTRS);
+		set_cpu_cap(c, X86_FEATURE_XENPV);
+	}
 }
 
 const struct hypervisor_x86 x86_hyper_xen = {