4 files changed, 105 insertions, 11 deletions

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index c8e1c7b95c3b..ddbd54a9b845 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -31,6 +31,7 @@
 #include <linux/pci.h>
 #include <linux/gfp.h>
 #include <linux/memblock.h>
+#include <linux/edd.h>
 
 #include <xen/xen.h>
 #include <xen/events.h>
@@ -1306,6 +1307,55 @@ static const struct machine_ops xen_machine_ops __initconst = {
         .emergency_restart = xen_emergency_restart,
 };
 
+static void __init xen_boot_params_init_edd(void)
+{
+#if IS_ENABLED(CONFIG_EDD)
+        struct xen_platform_op op;
+        struct edd_info *edd_info;
+        u32 *mbr_signature;
+        unsigned nr;
+        int ret;
+
+        edd_info = boot_params.eddbuf;
+        mbr_signature = boot_params.edd_mbr_sig_buffer;
+
+        op.cmd = XENPF_firmware_info;
+
+        op.u.firmware_info.type = XEN_FW_DISK_INFO;
+        for (nr = 0; nr < EDDMAXNR; nr++) {
+                struct edd_info *info = edd_info + nr;
+
+                op.u.firmware_info.index = nr;
+                info->params.length = sizeof(info->params);
+                set_xen_guest_handle(op.u.firmware_info.u.disk_info.edd_params,
+                                     &info->params);
+                ret = HYPERVISOR_dom0_op(&op);
+                if (ret)
+                        break;
+
+#define C(x) info->x = op.u.firmware_info.u.disk_info.x
+                C(device);
+                C(version);
+                C(interface_support);
+                C(legacy_max_cylinder);
+                C(legacy_max_head);
+                C(legacy_sectors_per_track);
+#undef C
+        }
+        boot_params.eddbuf_entries = nr;
+
+        op.u.firmware_info.type = XEN_FW_DISK_MBR_SIGNATURE;
+        for (nr = 0; nr < EDD_MBR_SIG_MAX; nr++) {
+                op.u.firmware_info.index = nr;
+                ret = HYPERVISOR_dom0_op(&op);
+                if (ret)
+                        break;
+                mbr_signature[nr] = op.u.firmware_info.u.disk_mbr_signature.mbr_signature;
+        }
+        boot_params.edd_mbr_sig_buf_entries = nr;
+#endif
+}
+
 /*
  * Set up the GDT and segment registers for -fstack-protector.  Until
  * we do this, we have to be careful not to call any stack-protected
@@ -1508,6 +1558,8 @@ asmlinkage void __init xen_start_kernel(void)
                 /* Avoid searching for BIOS MP tables */
                 x86_init.mpparse.find_smp_config = x86_init_noop;
                 x86_init.mpparse.get_smp_config = x86_init_uint_noop;
+
+                xen_boot_params_init_edd();
         }
 #ifdef CONFIG_PCI
         /* PCI BIOS service won't work from a PV guest. */
@@ -1589,8 +1641,11 @@ static int __cpuinit xen_hvm_cpu_notify(struct notifier_block *self,
         switch (action) {
         case CPU_UP_PREPARE:
                 xen_vcpu_setup(cpu);
-                if (xen_have_vector_callback)
+                if (xen_have_vector_callback) {
                         xen_init_lock_cpu(cpu);
+                        if (xen_feature(XENFEAT_hvm_safe_pvclock))
+                                xen_setup_timer(cpu);
+                }
                 break;
         default:
                 break;
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 09ea61d2e02f..0d466d7c7175 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -144,6 +144,13 @@ static int xen_smp_intr_init(unsigned int cpu)
                 goto fail;
         per_cpu(xen_callfuncsingle_irq, cpu) = rc;
 
+        /*
+         * The IRQ worker on PVHVM goes through the native path and uses the
+         * IPI mechanism.
+         */
+        if (xen_hvm_domain())
+                return 0;
+
         callfunc_name = kasprintf(GFP_KERNEL, "irqwork%d", cpu);
         rc = bind_ipi_to_irqhandler(XEN_IRQ_WORK_VECTOR,
                                     cpu,
@@ -167,6 +174,9 @@ static int xen_smp_intr_init(unsigned int cpu)
         if (per_cpu(xen_callfuncsingle_irq, cpu) >= 0)
                 unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu),
                                        NULL);
+        if (xen_hvm_domain())
+                return rc;
+
         if (per_cpu(xen_irq_work, cpu) >= 0)
                 unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
 
@@ -418,7 +428,7 @@ static int xen_cpu_disable(void)
 
 static void xen_cpu_die(unsigned int cpu)
 {
-        while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
+        while (xen_pv_domain() && HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
                 current->state = TASK_UNINTERRUPTIBLE;
                 schedule_timeout(HZ/10);
         }
@@ -426,7 +436,8 @@ static void xen_cpu_die(unsigned int cpu)
         unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
         unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
         unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
-        unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
+        if (!xen_hvm_domain())
+                unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
         xen_uninit_lock_cpu(cpu);
         xen_teardown_timer(cpu);
 }
@@ -657,11 +668,7 @@ static int __cpuinit xen_hvm_cpu_up(unsigned int cpu, struct task_struct *tidle)
 
 static void xen_hvm_cpu_die(unsigned int cpu)
 {
-        unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
-        unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
-        unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
-        unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
-        unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
+        xen_cpu_die(cpu);
         native_cpu_die(cpu);
 }
 
diff --git a/arch/x86/xen/spinlock.c b/arch/x86/xen/spinlock.c
index f7a080ef0354..8b54603ce816 100644
--- a/arch/x86/xen/spinlock.c
+++ b/arch/x86/xen/spinlock.c
@@ -364,6 +364,16 @@ void __cpuinit xen_init_lock_cpu(int cpu)
         int irq;
         const char *name;
 
+        WARN(per_cpu(lock_kicker_irq, cpu) > 0, "spinlock on CPU%d exists on IRQ%d!\n",
+             cpu, per_cpu(lock_kicker_irq, cpu));
+
+        /*
+         * See git commit f10cd522c5fbfec9ae3cc01967868c9c2401ed23
+         * (xen: disable PV spinlocks on HVM)
+         */
+        if (xen_hvm_domain())
+                return;
+
         name = kasprintf(GFP_KERNEL, "spinlock%d", cpu);
         irq = bind_ipi_to_irqhandler(XEN_SPIN_UNLOCK_VECTOR,
                                      cpu,
@@ -382,11 +392,26 @@ void __cpuinit xen_init_lock_cpu(int cpu)
 
 void xen_uninit_lock_cpu(int cpu)
 {
+        /*
+         * See git commit f10cd522c5fbfec9ae3cc01967868c9c2401ed23
+         * (xen: disable PV spinlocks on HVM)
+         */
+        if (xen_hvm_domain())
+                return;
+
         unbind_from_irqhandler(per_cpu(lock_kicker_irq, cpu), NULL);
+        per_cpu(lock_kicker_irq, cpu) = -1;
 }
 
 void __init xen_init_spinlocks(void)
 {
+        /*
+         * See git commit f10cd522c5fbfec9ae3cc01967868c9c2401ed23
+         * (xen: disable PV spinlocks on HVM)
+         */
+        if (xen_hvm_domain())
+                return;
+
         BUILD_BUG_ON(sizeof(struct xen_spinlock) > sizeof(arch_spinlock_t));
 
         pv_lock_ops.spin_is_locked = xen_spin_is_locked;
diff --git a/arch/x86/xen/time.c b/arch/x86/xen/time.c
index 0296a9522501..3d88bfdf9e1c 100644
--- a/arch/x86/xen/time.c
+++ b/arch/x86/xen/time.c
@@ -377,7 +377,7 @@ static const struct clock_event_device xen_vcpuop_clockevent = {
 
 static const struct clock_event_device *xen_clockevent =
         &xen_timerop_clockevent;
-static DEFINE_PER_CPU(struct clock_event_device, xen_clock_events);
+static DEFINE_PER_CPU(struct clock_event_device, xen_clock_events) = { .irq = -1 };
 
 static irqreturn_t xen_timer_interrupt(int irq, void *dev_id)
 {
@@ -401,6 +401,9 @@ void xen_setup_timer(int cpu)
         struct clock_event_device *evt;
         int irq;
 
+        evt = &per_cpu(xen_clock_events, cpu);
+        WARN(evt->irq >= 0, "IRQ%d for CPU%d is already allocated\n", evt->irq, cpu);
+
         printk(KERN_INFO "installing Xen timer for CPU %d\n", cpu);
 
         name = kasprintf(GFP_KERNEL, "timer%d", cpu);
@@ -413,7 +416,6 @@ void xen_setup_timer(int cpu)
                                       IRQF_FORCE_RESUME,
                                       name, NULL);
 
-        evt = &per_cpu(xen_clock_events, cpu);
         memcpy(evt, xen_clockevent, sizeof(*evt));
 
         evt->cpumask = cpumask_of(cpu);
@@ -426,6 +428,7 @@ void xen_teardown_timer(int cpu)
         BUG_ON(cpu == 0);
         evt = &per_cpu(xen_clock_events, cpu);
         unbind_from_irqhandler(evt->irq, NULL);
+        evt->irq = -1;
 }
 
 void xen_setup_cpu_clockevents(void)
@@ -497,7 +500,11 @@ static void xen_hvm_setup_cpu_clockevents(void)
 {
         int cpu = smp_processor_id();
         xen_setup_runstate_info(cpu);
-        xen_setup_timer(cpu);
+        /*
+         * xen_setup_timer(cpu) - snprintf is bad in atomic context. Hence
+         * doing it xen_hvm_cpu_notify (which gets called by smp_init during
+         * early bootup and also during CPU hotplug events).
+         */
         xen_setup_cpu_clockevents();
 }