#include <linux/init.h>
#include <linux/smp.h>
#include <linux/preempt.h>
+#include <linux/hardirq.h>
#include <linux/percpu.h>
#include <linux/delay.h>
#include <linux/start_kernel.h>
#include <linux/mm.h>
#include <linux/page-flags.h>
#include <linux/highmem.h>
+#include <linux/smp.h>
#include <xen/interface/xen.h>
#include <xen/interface/physdev.h>
#include <xen/interface/vcpu.h>
+#include <xen/interface/sched.h>
#include <xen/features.h>
#include <xen/page.h>
#include <asm/setup.h>
#include <asm/desc.h>
#include <asm/pgtable.h>
+#include <asm/tlbflush.h>
+#include <asm/reboot.h>
#include "xen-ops.h"
#include "mmu.h"
struct start_info *xen_start_info;
EXPORT_SYMBOL_GPL(xen_start_info);
-static void xen_vcpu_setup(int cpu)
+static /* __initdata */ struct shared_info dummy_shared_info;
+
+/*
+ * Point at some empty memory to start with. We map the real shared_info
+ * page as soon as fixmap is up and running.
+ */
+struct shared_info *HYPERVISOR_shared_info = (void *)&dummy_shared_info;
+
+/*
+ * Flag to determine whether vcpu info placement is available on all
+ * VCPUs. We assume it is to start with, and then set it to zero on
+ * the first failure. This is because it can succeed on some VCPUs
+ * and not others, since it can involve hypervisor memory allocation,
+ * or because the guest failed to guarantee all the appropriate
+ * constraints on all VCPUs (ie buffer can't cross a page boundary).
+ *
+ * Note that any particular CPU may be using a placed vcpu structure,
+ * but we can only optimise if the all are.
+ *
+ * 0: not available, 1: available
+ */
+static int have_vcpu_info_placement = 1;
+
+static void __init xen_vcpu_setup(int cpu)
{
+ struct vcpu_register_vcpu_info info;
+ int err;
+ struct vcpu_info *vcpup;
+
per_cpu(xen_vcpu, cpu) = &HYPERVISOR_shared_info->vcpu_info[cpu];
+
+ if (!have_vcpu_info_placement)
+ return; /* already tested, not available */
+
+ vcpup = &per_cpu(xen_vcpu_info, cpu);
+
+ info.mfn = virt_to_mfn(vcpup);
+ info.offset = offset_in_page(vcpup);
+
+ printk(KERN_DEBUG "trying to map vcpu_info %d at %p, mfn %x, offset %d\n",
+ cpu, vcpup, info.mfn, info.offset);
+
+ /* Check to see if the hypervisor will put the vcpu_info
+ structure where we want it, which allows direct access via
+ a percpu-variable. */
+ err = HYPERVISOR_vcpu_op(VCPUOP_register_vcpu_info, cpu, &info);
+
+ if (err) {
+ printk(KERN_DEBUG "register_vcpu_info failed: err=%d\n", err);
+ have_vcpu_info_placement = 0;
+ } else {
+ /* This cpu is using the registered vcpu info, even if
+ later ones fail to. */
+ per_cpu(xen_vcpu, cpu) = vcpup;
+
+ printk(KERN_DEBUG "cpu %d using vcpu_info at %p\n",
+ cpu, vcpup);
+ }
}
static void __init xen_banner(void)
struct vcpu_info *vcpu;
unsigned long flags;
- preempt_disable();
vcpu = x86_read_percpu(xen_vcpu);
+
/* flag has opposite sense of mask */
flags = !vcpu->evtchn_upcall_mask;
- preempt_enable();
/* convert to IF type flag
-0 -> 0x00000000
{
struct vcpu_info *vcpu;
- preempt_disable();
-
/* convert from IF type flag */
flags = !(flags & X86_EFLAGS_IF);
+
+ /* There's a one instruction preempt window here. We need to
+ make sure we're don't switch CPUs between getting the vcpu
+ pointer and updating the mask. */
+ preempt_disable();
vcpu = x86_read_percpu(xen_vcpu);
vcpu->evtchn_upcall_mask = flags;
+ preempt_enable_no_resched();
- if (flags == 0) {
- /* Unmask then check (avoid races). We're only protecting
- against updates by this CPU, so there's no need for
- anything stronger. */
- barrier();
+ /* Doesn't matter if we get preempted here, because any
+ pending event will get dealt with anyway. */
+ if (flags == 0) {
+ preempt_check_resched();
+ barrier(); /* unmask then check (avoid races) */
if (unlikely(vcpu->evtchn_upcall_pending))
force_evtchn_callback();
- preempt_enable();
- } else
- preempt_enable_no_resched();
+ }
}
static void xen_irq_disable(void)
{
- struct vcpu_info *vcpu;
+ /* There's a one instruction preempt window here. We need to
+ make sure we're don't switch CPUs between getting the vcpu
+ pointer and updating the mask. */
preempt_disable();
- vcpu = x86_read_percpu(xen_vcpu);
- vcpu->evtchn_upcall_mask = 1;
+ x86_read_percpu(xen_vcpu)->evtchn_upcall_mask = 1;
preempt_enable_no_resched();
}
{
struct vcpu_info *vcpu;
+ /* There's a one instruction preempt window here. We need to
+ make sure we're don't switch CPUs between getting the vcpu
+ pointer and updating the mask. */
preempt_disable();
vcpu = x86_read_percpu(xen_vcpu);
vcpu->evtchn_upcall_mask = 0;
+ preempt_enable_no_resched();
- /* Unmask then check (avoid races). We're only protecting
- against updates by this CPU, so there's no need for
- anything stronger. */
- barrier();
+ /* Doesn't matter if we get preempted here, because any
+ pending event will get dealt with anyway. */
+ barrier(); /* unmask then check (avoid races) */
if (unlikely(vcpu->evtchn_upcall_pending))
force_evtchn_callback();
- preempt_enable();
}
static void xen_safe_halt(void)
static void xen_set_lazy_mode(enum paravirt_lazy_mode mode)
{
+ BUG_ON(preemptible());
+
switch (mode) {
case PARAVIRT_LAZY_NONE:
BUG_ON(x86_read_percpu(xen_lazy_mode) == PARAVIRT_LAZY_NONE);
load_TLS_descriptor(t, cpu, 2);
xen_mc_issue(PARAVIRT_LAZY_CPU);
+
+ /*
+ * XXX sleazy hack: If we're being called in a lazy-cpu zone,
+ * it means we're in a context switch, and %gs has just been
+ * saved. This means we can zero it out to prevent faults on
+ * exit from the hypervisor if the next process has no %gs.
+ * Either way, it has been saved, and the new value will get
+ * loaded properly. This will go away as soon as Xen has been
+ * modified to not save/restore %gs for normal hypercalls.
+ */
+ if (xen_get_lazy_mode() == PARAVIRT_LAZY_CPU)
+ loadsegment(gs, 0);
}
static void xen_write_ldt_entry(struct desc_struct *dt, int entrynum,
xmaddr_t mach_lp = virt_to_machine(lp);
u64 entry = (u64)high << 32 | low;
+ preempt_disable();
+
xen_mc_flush();
if (HYPERVISOR_update_descriptor(mach_lp.maddr, entry))
BUG();
+
+ preempt_enable();
}
static int cvt_gate_to_trap(int vector, u32 low, u32 high,
static void xen_write_idt_entry(struct desc_struct *dt, int entrynum,
u32 low, u32 high)
{
-
- int cpu = smp_processor_id();
unsigned long p = (unsigned long)&dt[entrynum];
- unsigned long start = per_cpu(idt_desc, cpu).address;
- unsigned long end = start + per_cpu(idt_desc, cpu).size + 1;
+ unsigned long start, end;
+
+ preempt_disable();
+
+ start = __get_cpu_var(idt_desc).address;
+ end = start + __get_cpu_var(idt_desc).size + 1;
xen_mc_flush();
if (HYPERVISOR_set_trap_table(info))
BUG();
}
+
+ preempt_enable();
}
-/* Load a new IDT into Xen. In principle this can be per-CPU, so we
- hold a spinlock to protect the static traps[] array (static because
- it avoids allocation, and saves stack space). */
-static void xen_load_idt(const struct Xgt_desc_struct *desc)
+static void xen_convert_trap_info(const struct Xgt_desc_struct *desc,
+ struct trap_info *traps)
{
- static DEFINE_SPINLOCK(lock);
- static struct trap_info traps[257];
-
- int cpu = smp_processor_id();
unsigned in, out, count;
- per_cpu(idt_desc, cpu) = *desc;
-
count = (desc->size+1) / 8;
BUG_ON(count > 256);
- spin_lock(&lock);
for (in = out = 0; in < count; in++) {
const u32 *entry = (u32 *)(desc->address + in * 8);
out++;
}
traps[out].address = 0;
+}
+
+void xen_copy_trap_info(struct trap_info *traps)
+{
+ const struct Xgt_desc_struct *desc = &__get_cpu_var(idt_desc);
+
+ xen_convert_trap_info(desc, traps);
+}
+
+/* Load a new IDT into Xen. In principle this can be per-CPU, so we
+ hold a spinlock to protect the static traps[] array (static because
+ it avoids allocation, and saves stack space). */
+static void xen_load_idt(const struct Xgt_desc_struct *desc)
+{
+ static DEFINE_SPINLOCK(lock);
+ static struct trap_info traps[257];
+
+ spin_lock(&lock);
+
+ __get_cpu_var(idt_desc) = *desc;
+
+ xen_convert_trap_info(desc, traps);
xen_mc_flush();
if (HYPERVISOR_set_trap_table(traps))
static void xen_write_gdt_entry(struct desc_struct *dt, int entry,
u32 low, u32 high)
{
+ preempt_disable();
+
switch ((high >> 8) & 0xff) {
case DESCTYPE_LDT:
case DESCTYPE_TSS:
}
}
+
+ preempt_enable();
}
static void xen_load_esp0(struct tss_struct *tss,
- struct thread_struct *thread)
+ struct thread_struct *thread)
{
struct multicall_space mcs = xen_mc_entry(0);
MULTI_stack_switch(mcs.mc, __KERNEL_DS, thread->esp0);
{
return 0;
}
+
+static void xen_apic_write(unsigned long reg, unsigned long val)
+{
+ /* Warn to see if there's any stray references */
+ WARN_ON(1);
+}
#endif
static void xen_flush_tlb(void)
{
- struct mmuext_op op;
+ struct mmuext_op *op;
+ struct multicall_space mcs = xen_mc_entry(sizeof(*op));
- op.cmd = MMUEXT_TLB_FLUSH_LOCAL;
- if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
- BUG();
+ op = mcs.args;
+ op->cmd = MMUEXT_TLB_FLUSH_LOCAL;
+ MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+ xen_mc_issue(PARAVIRT_LAZY_MMU);
}
static void xen_flush_tlb_single(unsigned long addr)
{
- struct mmuext_op op;
+ struct mmuext_op *op;
+ struct multicall_space mcs = xen_mc_entry(sizeof(*op));
- op.cmd = MMUEXT_INVLPG_LOCAL;
- op.arg1.linear_addr = addr & PAGE_MASK;
- if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
- BUG();
+ op = mcs.args;
+ op->cmd = MMUEXT_INVLPG_LOCAL;
+ op->arg1.linear_addr = addr & PAGE_MASK;
+ MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+ xen_mc_issue(PARAVIRT_LAZY_MMU);
+}
+
+static void xen_flush_tlb_others(const cpumask_t *cpus, struct mm_struct *mm,
+ unsigned long va)
+{
+ struct {
+ struct mmuext_op op;
+ cpumask_t mask;
+ } *args;
+ cpumask_t cpumask = *cpus;
+ struct multicall_space mcs;
+
+ /*
+ * A couple of (to be removed) sanity checks:
+ *
+ * - current CPU must not be in mask
+ * - mask must exist :)
+ */
+ BUG_ON(cpus_empty(cpumask));
+ BUG_ON(cpu_isset(smp_processor_id(), cpumask));
+ BUG_ON(!mm);
+
+ /* If a CPU which we ran on has gone down, OK. */
+ cpus_and(cpumask, cpumask, cpu_online_map);
+ if (cpus_empty(cpumask))
+ return;
+
+ mcs = xen_mc_entry(sizeof(*args));
+ args = mcs.args;
+ args->mask = cpumask;
+ args->op.arg2.vcpumask = &args->mask;
+
+ if (va == TLB_FLUSH_ALL) {
+ args->op.cmd = MMUEXT_TLB_FLUSH_MULTI;
+ } else {
+ args->op.cmd = MMUEXT_INVLPG_MULTI;
+ args->op.arg1.linear_addr = va;
+ }
+
+ MULTI_mmuext_op(mcs.mc, &args->op, 1, NULL, DOMID_SELF);
+
+ xen_mc_issue(PARAVIRT_LAZY_MMU);
+}
+
+static void xen_write_cr2(unsigned long cr2)
+{
+ x86_read_percpu(xen_vcpu)->arch.cr2 = cr2;
}
static unsigned long xen_read_cr2(void)
return x86_read_percpu(xen_vcpu)->arch.cr2;
}
+static unsigned long xen_read_cr2_direct(void)
+{
+ return x86_read_percpu(xen_vcpu_info.arch.cr2);
+}
+
static void xen_write_cr4(unsigned long cr4)
{
/* never allow TSC to be disabled */
native_write_cr4(cr4 & ~X86_CR4_TSD);
}
-/*
- * Page-directory addresses above 4GB do not fit into architectural %cr3.
- * When accessing %cr3, or equivalent field in vcpu_guest_context, guests
- * must use the following accessor macros to pack/unpack valid MFNs.
- *
- * Note that Xen is using the fact that the pagetable base is always
- * page-aligned, and putting the 12 MSB of the address into the 12 LSB
- * of cr3.
- */
-#define xen_pfn_to_cr3(pfn) (((unsigned)(pfn) << 12) | ((unsigned)(pfn) >> 20))
-#define xen_cr3_to_pfn(cr3) (((unsigned)(cr3) >> 12) | ((unsigned)(cr3) << 20))
-
static unsigned long xen_read_cr3(void)
{
return x86_read_percpu(xen_cr3);
static void xen_write_cr3(unsigned long cr3)
{
+ BUG_ON(preemptible());
+
if (cr3 == x86_read_percpu(xen_cr3)) {
/* just a simple tlb flush */
xen_flush_tlb();
/* Early in boot, while setting up the initial pagetable, assume
everything is pinned. */
-static void xen_alloc_pt_init(struct mm_struct *mm, u32 pfn)
+static __init void xen_alloc_pt_init(struct mm_struct *mm, u32 pfn)
{
BUG_ON(mem_map); /* should only be used early */
make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
}
#endif
+static __init pte_t mask_rw_pte(pte_t *ptep, pte_t pte)
+{
+ /* If there's an existing pte, then don't allow _PAGE_RW to be set */
+ if (pte_val_ma(*ptep) & _PAGE_PRESENT)
+ pte = __pte_ma(((pte_val_ma(*ptep) & _PAGE_RW) | ~_PAGE_RW) &
+ pte_val_ma(pte));
+
+ return pte;
+}
+
+/* Init-time set_pte while constructing initial pagetables, which
+ doesn't allow RO pagetable pages to be remapped RW */
+static __init void xen_set_pte_init(pte_t *ptep, pte_t pte)
+{
+ pte = mask_rw_pte(ptep, pte);
+
+ xen_set_pte(ptep, pte);
+}
+
static __init void xen_pagetable_setup_start(pgd_t *base)
{
pgd_t *xen_pgd = (pgd_t *)xen_start_info->pt_base;
+ /* special set_pte for pagetable initialization */
+ paravirt_ops.set_pte = xen_set_pte_init;
+
init_mm.pgd = base;
/*
* copy top-level of Xen-supplied pagetable into place. For
/* This will work as long as patching hasn't happened yet
(which it hasn't) */
paravirt_ops.alloc_pt = xen_alloc_pt;
+ paravirt_ops.set_pte = xen_set_pte;
if (!xen_feature(XENFEAT_auto_translated_physmap)) {
/*
if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
BUG();
}
+}
- xen_vcpu_setup(smp_processor_id());
+/* This is called once we have the cpu_possible_map */
+void __init xen_setup_vcpu_info_placement(void)
+{
+ int cpu;
+
+ for_each_possible_cpu(cpu)
+ xen_vcpu_setup(cpu);
+
+ /* xen_vcpu_setup managed to place the vcpu_info within the
+ percpu area for all cpus, so make use of it */
+ if (have_vcpu_info_placement) {
+ printk(KERN_INFO "Xen: using vcpu_info placement\n");
+
+ paravirt_ops.save_fl = xen_save_fl_direct;
+ paravirt_ops.restore_fl = xen_restore_fl_direct;
+ paravirt_ops.irq_disable = xen_irq_disable_direct;
+ paravirt_ops.irq_enable = xen_irq_enable_direct;
+ paravirt_ops.read_cr2 = xen_read_cr2_direct;
+ paravirt_ops.iret = xen_iret_direct;
+ }
+}
+
+static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
+ unsigned long addr, unsigned len)
+{
+ char *start, *end, *reloc;
+ unsigned ret;
+
+ start = end = reloc = NULL;
+
+#define SITE(x) \
+ case PARAVIRT_PATCH(x): \
+ if (have_vcpu_info_placement) { \
+ start = (char *)xen_##x##_direct; \
+ end = xen_##x##_direct_end; \
+ reloc = xen_##x##_direct_reloc; \
+ } \
+ goto patch_site
+
+ switch (type) {
+ SITE(irq_enable);
+ SITE(irq_disable);
+ SITE(save_fl);
+ SITE(restore_fl);
+#undef SITE
+
+ patch_site:
+ if (start == NULL || (end-start) > len)
+ goto default_patch;
+
+ ret = paravirt_patch_insns(insnbuf, len, start, end);
+
+ /* Note: because reloc is assigned from something that
+ appears to be an array, gcc assumes it's non-null,
+ but doesn't know its relationship with start and
+ end. */
+ if (reloc > start && reloc < end) {
+ int reloc_off = reloc - start;
+ long *relocp = (long *)(insnbuf + reloc_off);
+ long delta = start - (char *)addr;
+
+ *relocp += delta;
+ }
+ break;
+
+ default_patch:
+ default:
+ ret = paravirt_patch_default(type, clobbers, insnbuf,
+ addr, len);
+ break;
+ }
+
+ return ret;
}
static const struct paravirt_ops xen_paravirt_ops __initdata = {
.name = "Xen",
.banner = xen_banner,
- .patch = paravirt_patch_default,
+ .patch = xen_patch,
.memory_setup = xen_memory_setup,
.arch_setup = xen_arch_setup,
.set_wallclock = xen_set_wallclock,
.get_wallclock = xen_get_wallclock,
.get_cpu_khz = xen_cpu_khz,
- .sched_clock = xen_clocksource_read,
+ .sched_clock = xen_sched_clock,
.cpuid = xen_cpuid,
.write_cr0 = native_write_cr0,
.read_cr2 = xen_read_cr2,
- .write_cr2 = native_write_cr2,
+ .write_cr2 = xen_write_cr2,
.read_cr3 = xen_read_cr3,
.write_cr3 = xen_write_cr3,
.io_delay = xen_io_delay,
#ifdef CONFIG_X86_LOCAL_APIC
- .apic_write = paravirt_nop,
- .apic_write_atomic = paravirt_nop,
+ .apic_write = xen_apic_write,
+ .apic_write_atomic = xen_apic_write,
.apic_read = xen_apic_read,
.setup_boot_clock = paravirt_nop,
.setup_secondary_clock = paravirt_nop,
.flush_tlb_user = xen_flush_tlb,
.flush_tlb_kernel = xen_flush_tlb,
.flush_tlb_single = xen_flush_tlb_single,
+ .flush_tlb_others = xen_flush_tlb_others,
.pte_update = paravirt_nop,
.pte_update_defer = paravirt_nop,
.kmap_atomic_pte = xen_kmap_atomic_pte,
#endif
- .set_pte = xen_set_pte,
+ .set_pte = NULL, /* see xen_pagetable_setup_* */
.set_pte_at = xen_set_pte_at,
.set_pmd = xen_set_pmd,
.set_lazy_mode = xen_set_lazy_mode,
};
+#ifdef CONFIG_SMP
+static const struct smp_ops xen_smp_ops __initdata = {
+ .smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
+ .smp_prepare_cpus = xen_smp_prepare_cpus,
+ .cpu_up = xen_cpu_up,
+ .smp_cpus_done = xen_smp_cpus_done,
+
+ .smp_send_stop = xen_smp_send_stop,
+ .smp_send_reschedule = xen_smp_send_reschedule,
+ .smp_call_function_mask = xen_smp_call_function_mask,
+};
+#endif /* CONFIG_SMP */
+
+static void xen_reboot(int reason)
+{
+#ifdef CONFIG_SMP
+ smp_send_stop();
+#endif
+
+ if (HYPERVISOR_sched_op(SCHEDOP_shutdown, reason))
+ BUG();
+}
+
+static void xen_restart(char *msg)
+{
+ xen_reboot(SHUTDOWN_reboot);
+}
+
+static void xen_emergency_restart(void)
+{
+ xen_reboot(SHUTDOWN_reboot);
+}
+
+static void xen_machine_halt(void)
+{
+ xen_reboot(SHUTDOWN_poweroff);
+}
+
+static void xen_crash_shutdown(struct pt_regs *regs)
+{
+ xen_reboot(SHUTDOWN_crash);
+}
+
+static const struct machine_ops __initdata xen_machine_ops = {
+ .restart = xen_restart,
+ .halt = xen_machine_halt,
+ .power_off = xen_machine_halt,
+ .shutdown = xen_machine_halt,
+ .crash_shutdown = xen_crash_shutdown,
+ .emergency_restart = xen_emergency_restart,
+};
+
+
/* First C function to be called on Xen boot */
asmlinkage void __init xen_start_kernel(void)
{
/* Install Xen paravirt ops */
paravirt_ops = xen_paravirt_ops;
+ machine_ops = xen_machine_ops;
+
+#ifdef CONFIG_SMP
+ smp_ops = xen_smp_ops;
+#endif
xen_setup_features();
/* keep using Xen gdt for now; no urgent need to change it */
x86_write_percpu(xen_cr3, __pa(pgd));
- xen_vcpu_setup(0);
+
+#ifdef CONFIG_SMP
+ /* Don't do the full vcpu_info placement stuff until we have a
+ possible map. */
+ per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0];
+#else
+ /* May as well do it now, since there's no good time to call
+ it later on UP. */
+ xen_setup_vcpu_info_placement();
+#endif
paravirt_ops.kernel_rpl = 1;
if (xen_feature(XENFEAT_supervisor_mode_kernel))