xen: smattering of dom0 patches

@@ -38,22 +38,30 @@
 extern struct start_info *xen_start_info;
 
 enum xen_domain_type {
-	XEN_NATIVE,
-	XEN_PV_DOMAIN,
-	XEN_HVM_DOMAIN,
+	XEN_NATIVE,		/* running on bare hardware    */
+	XEN_PV_DOMAIN,		/* running in a PV domain      */
+	XEN_HVM_DOMAIN,		/* running in a Xen hvm domain */
 };
 
+#ifdef CONFIG_XEN
 extern enum xen_domain_type xen_domain_type;
-
-#ifdef CONFIG_XEN
-#define xen_domain()		(xen_domain_type != XEN_NATIVE)
 #else
-#define xen_domain()		(0)
+#define xen_domain_type		XEN_NATIVE
 #endif
 
-#define xen_pv_domain()		(xen_domain() && xen_domain_type == XEN_PV_DOMAIN)
-#define xen_hvm_domain()	(xen_domain() && xen_domain_type == XEN_HVM_DOMAIN)
+#define xen_domain()		(xen_domain_type != XEN_NATIVE)
+#define xen_pv_domain()		(xen_domain() &&			\
+				 xen_domain_type == XEN_PV_DOMAIN)
+#define xen_hvm_domain()	(xen_domain() &&			\
+				 xen_domain_type == XEN_HVM_DOMAIN)
 
-#define xen_initial_domain()	(xen_pv_domain() && xen_start_info->flags & SIF_INITDOMAIN)
+#ifdef CONFIG_XEN_DOM0
+#include <xen/interface/xen.h>
+
+#define xen_initial_domain()	(xen_pv_domain() && \
+				 xen_start_info->flags & SIF_INITDOMAIN)
+#else  /* !CONFIG_XEN_DOM0 */
+#define xen_initial_domain()	(0)
+#endif	/* CONFIG_XEN_DOM0 */
 
 #endif /* _ASM_X86_XEN_HYPERVISOR_H */

@@ -55,7 +55,7 @@
 	notify_remote_via_evtchn(xen_start_info->console.domU.evtchn);
 }
 
-static int write_console(uint32_t vtermno, const char *data, int len)
+static int domU_write_console(uint32_t vtermno, const char *data, int len)
 {
 	struct xencons_interface *intf = xencons_interface();
 	XENCONS_RING_IDX cons, prod;

@@ -76,7 +76,7 @@
 	return sent;
 }
 
-static int read_console(uint32_t vtermno, char *buf, int len)
+static int domU_read_console(uint32_t vtermno, char *buf, int len)
 {
 	struct xencons_interface *intf = xencons_interface();
 	XENCONS_RING_IDX cons, prod;

@@ -97,28 +97,62 @@
 	return recv;
 }
 
-static struct hv_ops hvc_ops = {
-	.get_chars = read_console,
-	.put_chars = write_console,
+static struct hv_ops domU_hvc_ops = {
+	.get_chars = domU_read_console,
+	.put_chars = domU_write_console,
 	.notifier_add = notifier_add_irq,
 	.notifier_del = notifier_del_irq,
 	.notifier_hangup = notifier_hangup_irq,
 };
 
-static int __init xen_init(void)
+static int dom0_read_console(uint32_t vtermno, char *buf, int len)
+{
+	return HYPERVISOR_console_io(CONSOLEIO_read, len, buf);
+}
+
+/*
+ * Either for a dom0 to write to the system console, or a domU with a
+ * debug version of Xen
+ */
+static int dom0_write_console(uint32_t vtermno, const char *str, int len)
+{
+	int rc = HYPERVISOR_console_io(CONSOLEIO_write, len, (char *)str);
+	if (rc < 0)
+		return 0;
+
+	return len;
+}
+
+static struct hv_ops dom0_hvc_ops = {
+	.get_chars = dom0_read_console,
+	.put_chars = dom0_write_console,
+	.notifier_add = notifier_add_irq,
+	.notifier_del = notifier_del_irq,
+};
+
+static int __init xen_hvc_init(void)
 {
 	struct hvc_struct *hp;
+	struct hv_ops *ops;
 
-	if (!xen_pv_domain() ||
-	    xen_initial_domain() ||
-	    !xen_start_info->console.domU.evtchn)
-		return -ENODEV;
+	if (!xen_pv_domain())
+ 		return -ENODEV;
 
-	xencons_irq = bind_evtchn_to_irq(xen_start_info->console.domU.evtchn);
+	if (xen_initial_domain()) {
+		ops = &dom0_hvc_ops;
+		xencons_irq = bind_virq_to_irq(VIRQ_CONSOLE, 0);
+	} else {
+		if (!xen_start_info->console.domU.evtchn)
+			return -ENODEV;
+
+		ops = &domU_hvc_ops;
+		xencons_irq = bind_evtchn_to_irq(xen_start_info->console.domU.evtchn);
+	}
+
 	if (xencons_irq < 0)
 		xencons_irq = 0; /* NO_IRQ */
 
-	hp = hvc_alloc(HVC_COOKIE, xencons_irq, &hvc_ops, 256);
+	hp = hvc_alloc(HVC_COOKIE, xencons_irq, ops, 256);
 	if (IS_ERR(hp))
 		return PTR_ERR(hp);
 

@@ -135,7 +169,7 @@
 		rebind_evtchn_irq(xen_start_info->console.domU.evtchn, xencons_irq);
 }
 
-static void __exit xen_fini(void)
+static void __exit xen_hvc_fini(void)
 {
 	if (hvc)
 		hvc_remove(hvc);

@@ -143,29 +177,24 @@
 
 static int xen_cons_init(void)
 {
+	struct hv_ops *ops;
+
 	if (!xen_pv_domain())
 		return 0;
 
-	hvc_instantiate(HVC_COOKIE, 0, &hvc_ops);
+	ops = &domU_hvc_ops;
+	if (xen_initial_domain())
+		ops = &dom0_hvc_ops;
+
+	hvc_instantiate(HVC_COOKIE, 0, ops);
+
 	return 0;
 }
 
-module_init(xen_init);
-module_exit(xen_fini);
+module_init(xen_hvc_init);
+module_exit(xen_hvc_fini);
 console_initcall(xen_cons_init);
 
-static void raw_console_write(const char *str, int len)
-{
-	while(len > 0) {
-		int rc = HYPERVISOR_console_io(CONSOLEIO_write, len, (char *)str);
-		if (rc <= 0)
-			break;
-
-		str += rc;
-		len -= rc;
-	}
-}
-
 #ifdef CONFIG_EARLY_PRINTK
 static void xenboot_write_console(struct console *console, const char *string,
 				  unsigned len)

@@ -173,19 +202,22 @@
 	unsigned int linelen, off = 0;
 	const char *pos;
 
-	raw_console_write(string, len);
+	dom0_write_console(0, string, len);
 
-	write_console(0, "(early) ", 8);
+	if (xen_initial_domain())
+		return;
+
+	domU_write_console(0, "(early) ", 8);
 	while (off < len && NULL != (pos = strchr(string+off, '\n'))) {
 		linelen = pos-string+off;
 		if (off + linelen > len)
 			break;
-		write_console(0, string+off, linelen);
-		write_console(0, "\r\n", 2);
+		domU_write_console(0, string+off, linelen);
+		domU_write_console(0, "\r\n", 2);
 		off += linelen + 1;
 	}
 	if (off < len)
-		write_console(0, string+off, len-off);
+		domU_write_console(0, string+off, len-off);
 }
 
 struct console xenboot_console = {

@@ -197,7 +229,7 @@
 
 void xen_raw_console_write(const char *str)
 {
-	raw_console_write(str, strlen(str));
+	dom0_write_console(0, str, strlen(str));
 }
 
 void xen_raw_printk(const char *fmt, ...)

@@ -403,7 +403,7 @@
 }
 
 
-static int bind_virq_to_irq(unsigned int virq, unsigned int cpu)
+int bind_virq_to_irq(unsigned int virq, unsigned int cpu)
 {
 	struct evtchn_bind_virq bind_virq;
 	int evtchn, irq;

@@ -12,6 +12,8 @@
 			      irq_handler_t handler,
 			      unsigned long irqflags, const char *devname,
 			      void *dev_id);
+int bind_virq_to_irq(unsigned int virq, unsigned int cpu);
+
 int bind_virq_to_irqhandler(unsigned int virq, unsigned int cpu,
 			    irq_handler_t handler,
 			    unsigned long irqflags, const char *devname,

@@ -818,6 +818,7 @@
 static int __init xenbus_probe_init(void)
 {
 	int err = 0;
+	unsigned long page = 0;
 
 	DPRINTK("");
 

@@ -838,7 +839,31 @@
 	 * Domain0 doesn't have a store_evtchn or store_mfn yet.
 	 */
 	if (xen_initial_domain()) {
-		/* dom0 not yet supported */
+		struct evtchn_alloc_unbound alloc_unbound;
+
+		/* Allocate Xenstore page */
+		page = get_zeroed_page(GFP_KERNEL);
+		if (!page)
+			return -ENOMEM;
+
+		xen_store_mfn = xen_start_info->store_mfn =
+			pfn_to_mfn(virt_to_phys((void *)page) >>
+				   PAGE_SHIFT);
+
+		/* Next allocate a local port which xenstored can bind to */
+		alloc_unbound.dom        = DOMID_SELF;
+		alloc_unbound.remote_dom = 0;
+
+		err = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
+						  &alloc_unbound);
+		if (err == -ENOSYS)
+			goto out_unreg_front;
+
+		BUG_ON(err);
+		xen_store_evtchn = xen_start_info->store_evtchn =
+			alloc_unbound.port;
+
+		xen_store_interface = mfn_to_virt(xen_store_mfn);
 	} else {
 		xenstored_ready = 1;
 		xen_store_evtchn = xen_start_info->store_evtchn;

@@ -874,6 +899,9 @@
 	bus_unregister(&xenbus_frontend.bus);
 
   out_error:
+	if (page != 0)
+		free_page(page);
+
 	return err;
 }

@@ -1437,6 +1437,7 @@
 	return __ka(m2p(maddr));
 }
 
+/* Set the page permissions on an identity-mapped pages */
 static void set_page_prot(void *addr, pgprot_t prot)
 {
 	unsigned long pfn = __pa(addr) >> PAGE_SHIFT;

@@ -1492,6 +1493,29 @@
 	set_page_prot(pmd, PAGE_KERNEL_RO);
 }
 
+static __init void xen_ident_map_ISA(void)
+{
+	unsigned long pa;
+
+	/*
+	 * If we're dom0, then linear map the ISA machine addresses into
+	 * the kernel's address space.
+	 */
+	if (!xen_initial_domain())
+		return;
+
+	xen_raw_printk("Xen: setup ISA identity maps\n");
+
+	for (pa = ISA_START_ADDRESS; pa < ISA_END_ADDRESS; pa += PAGE_SIZE) {
+		pte_t pte = mfn_pte(PFN_DOWN(pa), PAGE_KERNEL_IO);
+
+		if (HYPERVISOR_update_va_mapping(PAGE_OFFSET + pa, pte, 0))
+			BUG();
+	}
+
+	xen_flush_tlb();
+}
+
 #ifdef CONFIG_X86_64
 static void convert_pfn_mfn(void *v)
 {

@@ -1674,6 +1698,7 @@
 
 	xen_raw_console_write("mapping kernel into physical memory\n");
 	pgd = xen_setup_kernel_pagetable(pgd, xen_start_info->nr_pages);
+	xen_ident_map_ISA();
 
 	init_mm.pgd = pgd;
 

@@ -1683,6 +1708,13 @@
 	if (xen_feature(XENFEAT_supervisor_mode_kernel))
 		pv_info.kernel_rpl = 0;
 
+	if (xen_initial_domain()) {
+		struct physdev_set_iopl set_iopl;
+		set_iopl.iopl = 1;
+		if (HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl) == -1)
+			BUG();
+	}
+
 	/* set the limit of our address space */
 	xen_reserve_top();
 

@@ -51,6 +51,9 @@
 	 * Even though this is normal, usable memory under Xen, reserve
 	 * ISA memory anyway because too many things think they can poke
 	 * about in there.
+	 *
+	 * In a dom0 kernel, this region is identity mapped with the
+	 * hardware ISA area, so it really is out of bounds.
 	 */
 	e820_add_region(ISA_START_ADDRESS, ISA_END_ADDRESS - ISA_START_ADDRESS,
 			E820_RESERVED);

@@ -188,7 +191,8 @@
 
 	pm_idle = xen_idle;
 
-	paravirt_disable_iospace();
+	if (!xen_initial_domain())
+		paravirt_disable_iospace();
 
 	fiddle_vdso();
 }

@@ -205,18 +205,23 @@
 static void xen_cpuid(unsigned int *ax, unsigned int *bx,
 		      unsigned int *cx, unsigned int *dx)
 {
-	unsigned maskedx = ~0;
+	unsigned maskedx = 0;
 
 	/*
 	 * Mask out inconvenient features, to try and disable as many
 	 * unsupported kernel subsystems as possible.
 	 */
-	if (*ax == 1)
-		maskedx = ~((1 << X86_FEATURE_APIC) |  /* disable APIC */
-			    (1 << X86_FEATURE_ACPI) |  /* disable ACPI */
-			    (1 << X86_FEATURE_MCE)  |  /* disable MCE */
-			    (1 << X86_FEATURE_MCA)  |  /* disable MCA */
-			    (1 << X86_FEATURE_ACC));   /* thermal monitoring */
+	if (*ax == 1) {
+		maskedx =
+			(1 << X86_FEATURE_MCE)  |  /* disable MCE */
+			(1 << X86_FEATURE_MCA)  |  /* disable MCA */
+			(1 << X86_FEATURE_ACC);   /* thermal monitoring */
+
+		if (!xen_initial_domain())
+			maskedx |=
+				(1 << X86_FEATURE_APIC) |  /* disable local APIC */
+				(1 << X86_FEATURE_ACPI);  /* disable ACPI */
+	}
 
 	asm(XEN_EMULATE_PREFIX "cpuid"
 		: "=a" (*ax),

@@ -224,7 +229,7 @@
 		  "=c" (*cx),
 		  "=d" (*dx)
 		: "0" (*ax), "2" (*cx));
-	*dx &= maskedx;
+	*dx &= ~maskedx;
 }
 
 static void xen_set_debugreg(int reg, unsigned long val)

@@ -45,6 +45,7 @@
 #include <xen/interface/xen.h>
 #include <xen/interface/sched.h>
 #include <xen/interface/physdev.h>
+#include <xen/interface/platform.h>
 
 /*
  * The hypercall asms have to meet several constraints:

@@ -282,6 +283,13 @@
 }
 
 static inline int
+HYPERVISOR_dom0_op(struct xen_platform_op *platform_op)
+{
+	platform_op->interface_version = XENPF_INTERFACE_VERSION;
+	return _hypercall1(int, dom0_op, platform_op);
+}
+
+static inline int
 HYPERVISOR_set_debugreg(int reg, unsigned long value)
 {
 	return _hypercall2(int, set_debugreg, reg, value);

232 lines not on this page · raw message

@@ -461,6 +461,8 @@
 #define __mk_unsigned_long(x) x ## UL
 #define mk_unsigned_long(x) __mk_unsigned_long(x)
 
+DEFINE_GUEST_HANDLE(uint64_t);
+
 #else /* __ASSEMBLY__ */
 
 /* In assembly code we cannot use C numeric constant suffixes. */

@@ -108,6 +108,7 @@
 	.get_free_region   = generic_get_free_region,
 	.validate_add_page = amd_validate_add_page,
 	.have_wrcomb       = positive_have_wrcomb,
+	.num_var_ranges	   = common_num_var_ranges,
 };
 
 int __init amd_init_mtrr(void)

@@ -213,6 +213,7 @@
 	.get_free_region   = centaur_get_free_region,
 	.validate_add_page = centaur_validate_add_page,
 	.have_wrcomb       = positive_have_wrcomb,
+	.num_var_ranges	   = common_num_var_ranges,
 };
 
 int __init centaur_init_mtrr(void)

@@ -263,6 +263,7 @@
 	.get_free_region   = cyrix_get_free_region,
 	.validate_add_page = generic_validate_add_page,
 	.have_wrcomb       = positive_have_wrcomb,
+	.num_var_ranges	   = common_num_var_ranges,
 };
 
 int __init cyrix_init_mtrr(void)

@@ -667,4 +667,5 @@
 	.set               = generic_set_mtrr,
 	.validate_add_page = generic_validate_add_page,
 	.have_wrcomb       = generic_have_wrcomb,
+	.num_var_ranges	   = common_num_var_ranges,
 };

@@ -99,7 +99,7 @@
 }
 
 /*  This function returns the number of variable MTRRs  */
-static void __init set_num_var_ranges(void)
+int __init common_num_var_ranges(void)
 {
 	unsigned long config = 0, dummy;
 

@@ -109,7 +109,7 @@
 		config = 2;
 	else if (is_cpu(CYRIX) || is_cpu(CENTAUR))
 		config = 8;
-	num_var_ranges = config & 0xff;
+	return config & 0xff;
 }
 
 static void __init init_table(void)

@@ -1676,12 +1676,17 @@
 void __init mtrr_bp_init(void)
 {
 	u32 phys_addr;
+
 	init_ifs();
 
 	phys_addr = 32;
 
 	if (cpu_has_mtrr) {
 		mtrr_if = &generic_mtrr_ops;
+#ifdef CONFIG_XEN_DOM0
+		xen_init_mtrr();
+#endif
+
 		size_or_mask = 0xff000000;	/* 36 bits */
 		size_and_mask = 0x00f00000;
 		phys_addr = 36;

@@ -1739,7 +1744,7 @@
 	}
 
 	if (mtrr_if) {
-		set_num_var_ranges();
+		num_var_ranges = mtrr_if->num_var_ranges();
 		init_table();
 		if (use_intel()) {
 			get_mtrr_state();

@@ -50,6 +50,8 @@
 	int	(*validate_add_page)(unsigned long base, unsigned long size,
 				     unsigned int type);
 	int	(*have_wrcomb)(void);
+
+	int	(*num_var_ranges)(void);
 };
 
 extern int generic_get_free_region(unsigned long base, unsigned long size,

@@ -61,6 +63,8 @@
 
 extern int positive_have_wrcomb(void);
 
+extern int __init common_num_var_ranges(void);
+
 /* library functions for processor-specific routines */
 struct set_mtrr_context {
 	unsigned long flags;

@@ -104,3 +108,4 @@
 int amd_init_mtrr(void);
 int cyrix_init_mtrr(void);
 int centaur_init_mtrr(void);
+void xen_init_mtrr(void);

59 lines not on this page · raw message

@@ -6,9 +6,11 @@
 #ifdef CONFIG_X86_PAT
 extern int pat_enabled;
 extern void validate_pat_support(struct cpuinfo_x86 *c);
+extern void pat_disable(char *reason);
 #else
 static const int pat_enabled;
 static inline void validate_pat_support(struct cpuinfo_x86 *c) { }
+static inline void pat_disable(char *reason) { }
 #endif
 
 extern void pat_init(void);

@@ -17,6 +19,4 @@
 		unsigned long req_type, unsigned long *ret_type);
 extern int free_memtype(u64 start, u64 end);
 
-extern void pat_disable(char *reason);
-
 #endif /* _ASM_X86_PAT_H */

@@ -48,6 +48,7 @@
 #include <asm/pgtable.h>
 #include <asm/tlbflush.h>
 #include <asm/reboot.h>
+#include <asm/pat.h>
 
 #include "xen-ops.h"
 #include "mmu.h"

@@ -1743,6 +1744,8 @@
 		add_preferred_console("hvc", 0, NULL);
 	}
 
+	pat_disable("PAT disabled on Xen");
+
 	xen_raw_console_write("about to get started...\n");
 
 	/* Start the world */

@@ -112,13 +112,9 @@
  */
 static inline unsigned long mfn_to_local_pfn(unsigned long mfn)
 {
-	extern unsigned long max_mapnr;
 	unsigned long pfn = mfn_to_pfn(mfn);
-	if ((pfn < max_mapnr)
-	    && !xen_feature(XENFEAT_auto_translated_physmap)
-	    && (get_phys_to_machine(pfn) != mfn))
-		return max_mapnr; /* force !pfn_valid() */
-	/* XXX fixme; not true with sparsemem */
+	if (get_phys_to_machine(pfn) != mfn)
+		return -1; /* force !pfn_valid() */
 	return pfn;
 }
 

@@ -1162,11 +1162,19 @@
 #ifdef CONFIG_X86_LOCAL_APIC
 	case FIX_APIC_BASE:	/* maps dummy local APIC */
 #endif
+		/* All local page mappings */
 		pte = pfn_pte(phys, prot);
 		break;
 
+	case FIX_PARAVIRT_BOOTMAP:
+		/* This is an MFN, but it isn't an IO mapping from the
+		   IO domain */
+		pte = mfn_pte(phys, prot);
+		break;
+
 	default:
-		pte = mfn_pte(phys, prot);
+		/* By default, set_fixmap is used for hardware mappings */
+		pte = mfn_pte(phys, __pgprot(pgprot_val(prot) | _PAGE_IOMAP));
 		break;
 	}
 

@@ -1695,7 +1703,9 @@
 
 	/* Prevent unwanted bits from being set in PTEs. */
 	__supported_pte_mask &= ~_PAGE_GLOBAL;
-	if (!xen_initial_domain())
+	if (xen_initial_domain())
+		__supported_pte_mask |= _PAGE_IOMAP;
+	else
 		__supported_pte_mask &= ~(_PAGE_PWT | _PAGE_PCD);
 
 	/* Don't do the full vcpu_info placement stuff until we have a

97 lines not on this page · raw message

@@ -56,7 +56,14 @@
 typedef struct { pgdval_t pgd; } pgd_t;
 typedef struct { pgprotval_t pgprot; } pgprot_t;
 
-extern int page_is_ram(unsigned long pagenr);
+extern int native_page_is_ram(unsigned long pagenr);
+#ifndef CONFIG_PARAVIRT
+static inline int page_is_ram(unsigned long pagenr)
+{
+	return native_page_is_ram(pagenr);
+}
+#endif
+
 extern int pagerange_is_ram(unsigned long start, unsigned long end);
 extern int devmem_is_allowed(unsigned long pagenr);
 extern void map_devmem(unsigned long pfn, unsigned long size,

@@ -321,6 +321,8 @@
 	   an mfn.  We can tell which is which from the index. */
 	void (*set_fixmap)(unsigned /* enum fixed_addresses */ idx,
 			   unsigned long phys, pgprot_t flags);
+
+	int (*page_is_ram)(unsigned long pfn);
 };
 
 struct raw_spinlock;

@@ -1386,6 +1388,11 @@
 	pv_mmu_ops.set_fixmap(idx, phys, flags);
 }
 
+static inline int page_is_ram(unsigned long pfn)
+{
+	return PVOP_CALL1(int, pv_mmu_ops.page_is_ram, pfn);
+}
+
 void _paravirt_nop(void);
 #define paravirt_nop	((void *)_paravirt_nop)
 

@@ -451,6 +451,7 @@
 	},
 
 	.set_fixmap = native_set_fixmap,
+	.page_is_ram = native_page_is_ram,
 };
 
 EXPORT_SYMBOL_GPL(pv_time_ops);

@@ -97,7 +97,7 @@
 
 #endif
 
-int page_is_ram(unsigned long pagenr)
+int native_page_is_ram(unsigned long pagenr)
 {
 	resource_size_t addr, end;
 	int i;

@@ -1190,6 +1190,16 @@
 #endif
 }
 
+static int xen_page_is_ram(unsigned long pfn)
+{
+	/* Granted pages are not RAM.  They will not have a proper
+	   identity pfn<->mfn translation. */
+	if (mfn_to_local_pfn(pfn_to_mfn(pfn)) != pfn)
+		return 0;
+
+	return native_page_is_ram(pfn);
+}
+
 static const struct pv_info xen_info __initdata = {
 	.paravirt_enabled = 1,
 	.shared_kernel_pmd = 0,

@@ -1364,6 +1374,7 @@
 	},
 
 	.set_fixmap = xen_set_fixmap,
+	.page_is_ram = xen_page_is_ram,
 };
 
 static void xen_reboot(int reason)

@@ -19,6 +19,7 @@
 
 #include <xen/page.h>
 #include <xen/interface/callback.h>
+#include <xen/interface/memory.h>
 #include <xen/interface/physdev.h>
 #include <xen/features.h>
 

@@ -36,16 +37,54 @@
 /**
  * machine_specific_memory_setup - Hook for machine specific memory setup.
  **/
-
 char * __init xen_memory_setup(void)
 {
 	unsigned long max_pfn = xen_start_info->nr_pages;
+	unsigned long long mem_end;
+	int rc;
+	struct xen_memory_map memmap;
+	/*
+	 * This is rather large for a stack variable but this early in
+	 * the boot process we know we have plenty slack space.
+	 */
+	struct e820entry map[E820MAX];
+	int op = xen_initial_domain() ?
+		XENMEM_machine_memory_map :
+		XENMEM_memory_map;
+	int i;
 
 	max_pfn = min(MAX_DOMAIN_PAGES, max_pfn);
+	mem_end = PFN_PHYS((u64)max_pfn);
+
+	memmap.nr_entries = E820MAX;
+	set_xen_guest_handle(memmap.buffer, map);
+
+	rc = HYPERVISOR_memory_op(op, &memmap);
+	if (rc == -ENOSYS) {
+		memmap.nr_entries = 1;
+		map[0].addr = 0ULL;
+		map[0].size = mem_end;
+		/* 8MB slack (to balance backend allocations). */
+		map[0].size += 8ULL << 20;
+		map[0].type = E820_RAM;
+		rc = 0;
+	}
+	BUG_ON(rc);
 
 	e820.nr_map = 0;
-
-	e820_add_region(0, PFN_PHYS((u64)max_pfn), E820_RAM);
+	for (i = 0; i < memmap.nr_entries; i++) {
+		unsigned long long end = map[i].addr + map[i].size;
+		if (map[i].type == E820_RAM) {
+			if (map[i].addr > mem_end)
+				continue;
+			if (end > mem_end) {
+				/* Truncate region to max_mem. */
+				map[i].size -= end - mem_end;
+			}
+		}
+		if (map[i].size > 0)
+			e820_add_region(map[i].addr, map[i].size, map[i].type);
+	}
 
 	/*
 	 * Even though this is normal, usable memory under Xen, reserve

@@ -142,4 +142,33 @@
 };
 DEFINE_GUEST_HANDLE_STRUCT(xen_translate_gpfn_list);
 
+/*
+ * Returns the pseudo-physical memory map as it was when the domain
+ * was started (specified by XENMEM_set_memory_map).
+ * arg == addr of struct xen_memory_map.
+ */
+#define XENMEM_memory_map           9
+struct xen_memory_map {
+    /*
+     * On call the number of entries which can be stored in buffer. On
+     * return the number of entries which have been stored in
+     * buffer.
+     */
+    unsigned int nr_entries;
+
+    /*
+     * Entries in the buffer are in the same format as returned by the
+     * BIOS INT 0x15 EAX=0xE820 call.
+     */
+    GUEST_HANDLE(void) buffer;
+};
+DEFINE_GUEST_HANDLE_STRUCT(xen_memory_map);
+
+/*
+ * Returns the real physical memory map. Passes the same structure as
+ * XENMEM_memory_map.
+ * arg == addr of struct xen_memory_map.
+ */
+#define XENMEM_machine_memory_map   10
+
 #endif /* __XEN_PUBLIC_MEMORY_H__ */