Merge tag 'for-linus-6.3-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip

Pull xen updates from Juergen Gross:

 - help deprecate the /proc/xen files by making the related information
   available via sysfs

 - mark the Xen variants of play_dead "noreturn"

 - support a shared Xen platform interrupt

 - several small cleanups and fixes

* tag 'for-linus-6.3-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
  xen: sysfs: make kobj_type structure constant
  x86/Xen: drop leftover VM-assist uses
  xen: Replace one-element array with flexible-array member
  xen/grant-dma-iommu: Implement a dummy probe_device() callback
  xen/pvcalls-back: fix permanently masked event channel
  xen: Allow platform PCI interrupt to be shared
  x86/xen/time: prefer tsc as clocksource when it is invariant
  x86/xen: mark xen_pv_play_dead() as __noreturn
  x86/xen: don't let xen_pv_play_dead() return
  drivers/xen/hypervisor: Expose Xen SIF flags to userspace

Documentation/ABI/stable/sysfs-hypervisor-xen

@@ -120,3 +120,16 @@ Contact:	xen-devel@lists.xenproject.org
 Description:	If running under Xen:
 		The Xen version is in the format <major>.<minor><extra>
 		This is the <minor> part of it.
+
+What:		/sys/hypervisor/start_flags/*
+Date:		March 2023
+KernelVersion:	6.3.0
+Contact:	xen-devel@lists.xenproject.org
+Description:	If running under Xen:
+		All bits in Xen's start-flags are represented as
+		boolean files, returning '1' if set, '0' otherwise.
+		This takes the place of the defunct /proc/xen/capabilities,
+		which would contain "control_d" on dom0, and be empty
+		otherwise.  This flag is now exposed as "initdomain" in
+		addition to the "privileged" flag; all other possible flags
+		are accessible as "unknownXX".

arch/x86/xen/setup.c

@@ -934,12 +934,8 @@ void xen_enable_syscall(void)
 
 static void __init xen_pvmmu_arch_setup(void)
 {
-	HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_4gb_segments);
 	HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_writable_pagetables);
 
-	HYPERVISOR_vm_assist(VMASST_CMD_enable,
-			     VMASST_TYPE_pae_extended_cr3);
-
 	if (register_callback(CALLBACKTYPE_event,
 			      xen_asm_exc_xen_hypervisor_callback) ||
 	    register_callback(CALLBACKTYPE_failsafe, xen_failsafe_callback))

arch/x86/xen/smp.h

@@ -21,6 +21,8 @@ void xen_smp_send_reschedule(int cpu);
 void xen_smp_send_call_function_ipi(const struct cpumask *mask);
 void xen_smp_send_call_function_single_ipi(int cpu);
 
+void __noreturn xen_cpu_bringup_again(unsigned long stack);
+
 struct xen_common_irq {
 	int irq;
 	char *name;

arch/x86/xen/smp_pv.c

@@ -381,21 +381,12 @@ static void xen_pv_cpu_die(unsigned int cpu)
 	}
 }
 
-static void xen_pv_play_dead(void) /* used only with HOTPLUG_CPU */
+static void __noreturn xen_pv_play_dead(void) /* used only with HOTPLUG_CPU */
 {
 	play_dead_common();
 	HYPERVISOR_vcpu_op(VCPUOP_down, xen_vcpu_nr(smp_processor_id()), NULL);
-	cpu_bringup();
-	/*
-	 * commit 4b0c0f294 (tick: Cleanup NOHZ per cpu data on cpu down)
-	 * clears certain data that the cpu_idle loop (which called us
-	 * and that we return from) expects. The only way to get that
-	 * data back is to call:
-	 */
-	tick_nohz_idle_enter();
-	tick_nohz_idle_stop_tick_protected();
-
-	cpuhp_online_idle(CPUHP_AP_ONLINE_IDLE);
+	xen_cpu_bringup_again((unsigned long)task_pt_regs(current));
+	BUG();
 }
 
 #else /* !CONFIG_HOTPLUG_CPU */
@@ -409,7 +400,7 @@ static void xen_pv_cpu_die(unsigned int cpu)
 	BUG();
 }
 
-static void xen_pv_play_dead(void)
+static void __noreturn xen_pv_play_dead(void)
 {
 	BUG();
 }

arch/x86/xen/time.c

@@ -482,15 +482,51 @@ static void xen_setup_vsyscall_time_info(void)
 	xen_clocksource.vdso_clock_mode = VDSO_CLOCKMODE_PVCLOCK;
 }
 
+/*
+ * Check if it is possible to safely use the tsc as a clocksource.  This is
+ * only true if the hypervisor notifies the guest that its tsc is invariant,
+ * the tsc is stable, and the tsc instruction will never be emulated.
+ */
+static int __init xen_tsc_safe_clocksource(void)
+{
+	u32 eax, ebx, ecx, edx;
+
+	if (!(boot_cpu_has(X86_FEATURE_CONSTANT_TSC)))
+		return 0;
+
+	if (!(boot_cpu_has(X86_FEATURE_NONSTOP_TSC)))
+		return 0;
+
+	if (check_tsc_unstable())
+		return 0;
+
+	/* Leaf 4, sub-leaf 0 (0x40000x03) */
+	cpuid_count(xen_cpuid_base() + 3, 0, &eax, &ebx, &ecx, &edx);
+
+	/* tsc_mode = no_emulate (2) */
+	if (ebx != 2)
+		return 0;
+
+	return 1;
+}
+
 static void __init xen_time_init(void)
 {
 	struct pvclock_vcpu_time_info *pvti;
 	int cpu = smp_processor_id();
 	struct timespec64 tp;
 
-	/* As Dom0 is never moved, no penalty on using TSC there */
+	/*
+	 * As Dom0 is never moved, no penalty on using TSC there.
+	 *
+	 * If it is possible for the guest to determine that the tsc is a safe
+	 * clocksource, then set xen_clocksource rating below that of the tsc
+	 * so that the system prefers tsc instead.
+	 */
 	if (xen_initial_domain())
 		xen_clocksource.rating = 275;
+	else if (xen_tsc_safe_clocksource())
+		xen_clocksource.rating = 299;
 
 	clocksource_register_hz(&xen_clocksource, NSEC_PER_SEC);
 

arch/x86/xen/xen-head.S

@@ -76,6 +76,13 @@ SYM_CODE_START(asm_cpu_bringup_and_idle)
 
 	call cpu_bringup_and_idle
 SYM_CODE_END(asm_cpu_bringup_and_idle)
+
+SYM_CODE_START(xen_cpu_bringup_again)
+	UNWIND_HINT_FUNC
+	mov	%rdi, %rsp
+	UNWIND_HINT_REGS
+	call	cpu_bringup_and_idle
+SYM_CODE_END(xen_cpu_bringup_again)
 .popsection
 #endif
 #endif

drivers/xen/events/events_base.c

@@ -1710,9 +1710,10 @@ void handle_irq_for_port(evtchn_port_t port, struct evtchn_loop_ctrl *ctrl)
 	generic_handle_irq(irq);
 }
 
-static void __xen_evtchn_do_upcall(void)
+static int __xen_evtchn_do_upcall(void)
 {
 	struct vcpu_info *vcpu_info = __this_cpu_read(xen_vcpu);
+	int ret = vcpu_info->evtchn_upcall_pending ? IRQ_HANDLED : IRQ_NONE;
 	int cpu = smp_processor_id();
 	struct evtchn_loop_ctrl ctrl = { 0 };
 
@@ -1737,6 +1738,8 @@ static void __xen_evtchn_do_upcall(void)
 	 * above.
 	 */
 	__this_cpu_inc(irq_epoch);
+
+	return ret;
 }
 
 void xen_evtchn_do_upcall(struct pt_regs *regs)
@@ -1751,9 +1754,9 @@ void xen_evtchn_do_upcall(struct pt_regs *regs)
 	set_irq_regs(old_regs);
 }
 
-void xen_hvm_evtchn_do_upcall(void)
+int xen_hvm_evtchn_do_upcall(void)
 {
-	__xen_evtchn_do_upcall();
+	return __xen_evtchn_do_upcall();
 }
 EXPORT_SYMBOL_GPL(xen_hvm_evtchn_do_upcall);
 

drivers/xen/grant-dma-iommu.c

@@ -16,8 +16,15 @@ struct grant_dma_iommu_device {
 	struct iommu_device iommu;
 };
 
-/* Nothing is really needed here */
-static const struct iommu_ops grant_dma_iommu_ops;
+static struct iommu_device *grant_dma_iommu_probe_device(struct device *dev)
+{
+	return ERR_PTR(-ENODEV);
+}
+
+/* Nothing is really needed here except a dummy probe_device callback */
+static const struct iommu_ops grant_dma_iommu_ops = {
+	.probe_device = grant_dma_iommu_probe_device,
+};
 
 static const struct of_device_id grant_dma_iommu_of_match[] = {
 	{ .compatible = "xen,grant-dma" },

drivers/xen/platform-pci.c

@@ -64,14 +64,13 @@ static uint64_t get_callback_via(struct pci_dev *pdev)
 
 static irqreturn_t do_hvm_evtchn_intr(int irq, void *dev_id)
 {
-	xen_hvm_evtchn_do_upcall();
-	return IRQ_HANDLED;
+	return xen_hvm_evtchn_do_upcall();
 }
 
 static int xen_allocate_irq(struct pci_dev *pdev)
 {
 	return request_irq(pdev->irq, do_hvm_evtchn_intr,
-			IRQF_NOBALANCING | IRQF_TRIGGER_RISING,
+			IRQF_NOBALANCING | IRQF_SHARED,
 			"xen-platform-pci", pdev);
 }
 

drivers/xen/pvcalls-back.c

@@ -173,6 +173,8 @@ static bool pvcalls_conn_back_write(struct sock_mapping *map)
 	RING_IDX cons, prod, size, array_size;
 	int ret;
 
+	atomic_set(&map->write, 0);
+
 	cons = intf->out_cons;
 	prod = intf->out_prod;
 	/* read the indexes before dealing with the data */
@@ -197,7 +199,6 @@ static bool pvcalls_conn_back_write(struct sock_mapping *map)
 		iov_iter_kvec(&msg.msg_iter, ITER_SOURCE, vec, 2, size);
 	}
 
-	atomic_set(&map->write, 0);
 	ret = inet_sendmsg(map->sock, &msg, size);
 	if (ret == -EAGAIN) {
 		atomic_inc(&map->write);

drivers/xen/sys-hypervisor.c

@@ -31,7 +31,10 @@ struct hyp_sysfs_attr {
 	struct attribute attr;
 	ssize_t (*show)(struct hyp_sysfs_attr *, char *);
 	ssize_t (*store)(struct hyp_sysfs_attr *, const char *, size_t);
-	void *hyp_attr_data;
+	union {
+		void *hyp_attr_data;
+		unsigned long hyp_attr_value;
+	};
 };
 
 static ssize_t type_show(struct hyp_sysfs_attr *attr, char *buffer)
@@ -399,6 +402,60 @@ static int __init xen_sysfs_properties_init(void)
 	return sysfs_create_group(hypervisor_kobj, &xen_properties_group);
 }
 
+#define FLAG_UNAME "unknown"
+#define FLAG_UNAME_FMT FLAG_UNAME "%02u"
+#define FLAG_UNAME_MAX sizeof(FLAG_UNAME "XX")
+#define FLAG_COUNT (sizeof(xen_start_flags) * BITS_PER_BYTE)
+static_assert(sizeof(xen_start_flags) <=
+	      sizeof_field(struct hyp_sysfs_attr, hyp_attr_value));
+
+static ssize_t flag_show(struct hyp_sysfs_attr *attr, char *buffer)
+{
+	char *p = buffer;
+
+	*p++ = '0' + ((xen_start_flags & attr->hyp_attr_value) != 0);
+	*p++ = '\n';
+	return p - buffer;
+}
+
+#define FLAG_NODE(flag, node)				\
+	[ilog2(flag)] = {				\
+		.attr = { .name = #node, .mode = 0444 },\
+		.show = flag_show,			\
+		.hyp_attr_value = flag			\
+	}
+
+/*
+ * Add new, known flags here.  No other changes are required, but
+ * note that each known flag wastes one entry in flag_unames[].
+ * The code/complexity machinations to avoid this isn't worth it
+ * for a few entries, but keep it in mind.
+ */
+static struct hyp_sysfs_attr flag_attrs[FLAG_COUNT] = {
+	FLAG_NODE(SIF_PRIVILEGED, privileged),
+	FLAG_NODE(SIF_INITDOMAIN, initdomain)
+};
+static struct attribute_group xen_flags_group = {
+	.name = "start_flags",
+	.attrs = (struct attribute *[FLAG_COUNT + 1]){}
+};
+static char flag_unames[FLAG_COUNT][FLAG_UNAME_MAX];
+
+static int __init xen_sysfs_flags_init(void)
+{
+	for (unsigned fnum = 0; fnum != FLAG_COUNT; fnum++) {
+		if (likely(flag_attrs[fnum].attr.name == NULL)) {
+			sprintf(flag_unames[fnum], FLAG_UNAME_FMT, fnum);
+			flag_attrs[fnum].attr.name = flag_unames[fnum];
+			flag_attrs[fnum].attr.mode = 0444;
+			flag_attrs[fnum].show = flag_show;
+			flag_attrs[fnum].hyp_attr_value = 1 << fnum;
+		}
+		xen_flags_group.attrs[fnum] = &flag_attrs[fnum].attr;
+	}
+	return sysfs_create_group(hypervisor_kobj, &xen_flags_group);
+}
+
 #ifdef CONFIG_XEN_HAVE_VPMU
 struct pmu_mode {
 	const char *name;
@@ -539,18 +596,22 @@ static int __init hyper_sysfs_init(void)
 	ret = xen_sysfs_properties_init();
 	if (ret)
 		goto prop_out;
+	ret = xen_sysfs_flags_init();
+	if (ret)
+		goto flags_out;
 #ifdef CONFIG_XEN_HAVE_VPMU
 	if (xen_initial_domain()) {
 		ret = xen_sysfs_pmu_init();
 		if (ret) {
-			sysfs_remove_group(hypervisor_kobj,
-					   &xen_properties_group);
-			goto prop_out;
+			sysfs_remove_group(hypervisor_kobj, &xen_flags_group);
+			goto flags_out;
 		}
 	}
 #endif
 	goto out;
 
+flags_out:
+	sysfs_remove_group(hypervisor_kobj, &xen_properties_group);
 prop_out:
 	sysfs_remove_file(hypervisor_kobj, &uuid_attr.attr);
 uuid_out:
@@ -594,7 +655,7 @@ static const struct sysfs_ops hyp_sysfs_ops = {
 	.store = hyp_sysfs_store,
 };
 
-static struct kobj_type hyp_sysfs_kobj_type = {
+static const struct kobj_type hyp_sysfs_kobj_type = {
 	.sysfs_ops = &hyp_sysfs_ops,
 };
 

drivers/xen/xen-front-pgdir-shbuf.c

@@ -30,7 +30,7 @@
 struct xen_page_directory {
 	grant_ref_t gref_dir_next_page;
 #define XEN_GREF_LIST_END	0
-	grant_ref_t gref[1]; /* Variable length */
+	grant_ref_t gref[]; /* Variable length */
 };
 
 /**

include/xen/events.h

@@ -107,7 +107,7 @@ evtchn_port_t evtchn_from_irq(unsigned irq);
 
 int xen_set_callback_via(uint64_t via);
 void xen_evtchn_do_upcall(struct pt_regs *regs);
-void xen_hvm_evtchn_do_upcall(void);
+int xen_hvm_evtchn_do_upcall(void);
 
 /* Bind a pirq for a physical interrupt to an irq. */
 int xen_bind_pirq_gsi_to_irq(unsigned gsi,

tools/objtool/check.c

@@ -186,6 +186,7 @@ static bool __dead_end_function(struct objtool_file *file, struct symbol *func,
 		"snp_abort",
 		"stop_this_cpu",
 		"usercopy_abort",
+		"xen_cpu_bringup_again",
 		"xen_start_kernel",
 	};