[PATCH 00 of 11] xen: smattering of dom0 patches

[PATCH 00 of 11] xen: smattering of dom0 patches

[Jeremy Fitzhardinge]

Hi Ingo,

Here are some more Xen dom0 patches.  Functionally the patches are all
over the place, and mostly touch pure Xen-only code.  The exception is
the mtrr change, and pv-op for page_is_ram().

Thanks,
    J

[PATCH 01 of 11] xen: short-circuit tests for dom0

[Jeremy Fitzhardinge]

When testing for a dom0/initial/privileged domain, make sure the
predicate evaluates to a compile-time 0 if CONFIG_XEN_DOM0 isn't
enabled.  This will make most of the dom0 code evaporate without
much more effort.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
---
 arch/x86/include/asm/xen/hypervisor.h |   28 ++++++++++++++++++----------
 1 file changed, 18 insertions(+), 10 deletions(-)

diff --git a/arch/x86/include/asm/xen/hypervisor.h b/arch/x86/include/asm/xen/hypervisor.h
--- a/arch/x86/include/asm/xen/hypervisor.h
+++ b/arch/x86/include/asm/xen/hypervisor.h
@@ -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 */

[PATCH 02 of 11] xen dom0: Make hvc_xen console work for dom0

[Jeremy Fitzhardinge]

Use the console hypercalls for dom0 console.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
---
 drivers/char/hvc_xen.c |  100 +++++++++++++++++++++++++++++++-----------------
 drivers/xen/events.c   |    2
 include/xen/events.h   |    2
 3 files changed, 69 insertions(+), 35 deletions(-)

diff --git a/drivers/char/hvc_xen.c b/drivers/char/hvc_xen.c
--- a/drivers/char/hvc_xen.c
+++ b/drivers/char/hvc_xen.c
@@ -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, ...)
diff --git a/drivers/xen/events.c b/drivers/xen/events.c
--- a/drivers/xen/events.c
+++ b/drivers/xen/events.c
@@ -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;
diff --git a/include/xen/events.h b/include/xen/events.h
--- a/include/xen/events.h
+++ b/include/xen/events.h
@@ -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,

[PATCH 03 of 11] xen dom0: Initialize xenbus for dom0

[Jeremy Fitzhardinge]

From: Juan Quintela <quintela@redhat.com>

Do initial xenbus/xenstore setup in dom0.  In dom0 we need to actually
allocate the xenstore resources, rather than being given them from
outside.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
---
 drivers/xen/xenbus/xenbus_probe.c |   30 +++++++++++++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)

diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c
--- a/drivers/xen/xenbus/xenbus_probe.c
+++ b/drivers/xen/xenbus/xenbus_probe.c
@@ -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;
 }

[PATCH 04 of 11] xen dom0: Set up basic IO permissions for dom0

[Jeremy Fitzhardinge]

From: Juan Quintela <quintela@redhat.com>

Add the direct mapping area for ISA bus access, and enable IO space
access for the guest when running as dom0.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
---
 arch/x86/xen/enlighten.c |   32 ++++++++++++++++++++++++++++++++
 arch/x86/xen/setup.c     |    6 +++++-
 2 files changed, 37 insertions(+), 1 deletion(-)

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -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();
 
diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
--- a/arch/x86/xen/setup.c
+++ b/arch/x86/xen/setup.c
@@ -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();
 }

[PATCH 05 of 11] xen-dom0: only selectively disable cpu features

[Jeremy Fitzhardinge]

Dom0 kernels actually want most of the CPU features to be enabled.
Some, like MCA/MCE, are still handled by Xen itself.

We leave APIC enabled even though we don't really have a functional
local apic so that the ACPI code will parse the corresponding tables
properly.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
---
 arch/x86/xen/enlighten.c |   21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -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)

[PATCH 06 of 11] xen dom0: Add support for the platform_ops hypercall

[Jeremy Fitzhardinge]

From: Stephen Tweedie <sct@redhat.com>

Minimal changes to get platform ops (renamed dom0_ops on pv_ops) working
on pv_ops builds.  Pulls in upstream linux-2.6.18-xen.hg's platform.h

Signed-off-by: Stephen Tweedie <sct@redhat.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
---
 arch/x86/include/asm/xen/hypercall.h |    8 +
 include/xen/interface/platform.h     |  232 ++++++++++++++++++++++++++++++++++
 include/xen/interface/xen.h          |    2
 3 files changed, 242 insertions(+)

diff --git a/arch/x86/include/asm/xen/hypercall.h b/arch/x86/include/asm/xen/hypercall.h
--- a/arch/x86/include/asm/xen/hypercall.h
+++ b/arch/x86/include/asm/xen/hypercall.h
@@ -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);
diff --git a/include/xen/interface/platform.h b/include/xen/interface/platform.h
new file mode 100644
--- /dev/null
+++ b/include/xen/interface/platform.h
@@ -0,0 +1,232 @@
+/******************************************************************************
+ * platform.h
+ *
+ * Hardware platform operations. Intended for use by domain-0 kernel.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Copyright (c) 2002-2006, K Fraser
+ */
+
+#ifndef __XEN_PUBLIC_PLATFORM_H__
+#define __XEN_PUBLIC_PLATFORM_H__
+
+#include "xen.h"
+
+#define XENPF_INTERFACE_VERSION 0x03000001
+
+/*
+ * Set clock such that it would read <secs,nsecs> after 00:00:00 UTC,
+ * 1 January, 1970 if the current system time was <system_time>.
+ */
+#define XENPF_settime             17
+struct xenpf_settime {
+    /* IN variables. */
+    uint32_t secs;
+    uint32_t nsecs;
+    uint64_t system_time;
+};
+typedef struct xenpf_settime xenpf_settime_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_settime_t);
+
+/*
+ * Request memory range (@mfn, @mfn+@nr_mfns-1) to have type @type.
+ * On x86, @type is an architecture-defined MTRR memory type.
+ * On success, returns the MTRR that was used (@reg) and a handle that can
+ * be passed to XENPF_DEL_MEMTYPE to accurately tear down the new setting.
+ * (x86-specific).
+ */
+#define XENPF_add_memtype         31
+struct xenpf_add_memtype {
+    /* IN variables. */
+    unsigned long mfn;
+    uint64_t nr_mfns;
+    uint32_t type;
+    /* OUT variables. */
+    uint32_t handle;
+    uint32_t reg;
+};
+typedef struct xenpf_add_memtype xenpf_add_memtype_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_add_memtype_t);
+
+/*
+ * Tear down an existing memory-range type. If @handle is remembered then it
+ * should be passed in to accurately tear down the correct setting (in case
+ * of overlapping memory regions with differing types). If it is not known
+ * then @handle should be set to zero. In all cases @reg must be set.
+ * (x86-specific).
+ */
+#define XENPF_del_memtype         32
+struct xenpf_del_memtype {
+    /* IN variables. */
+    uint32_t handle;
+    uint32_t reg;
+};
+typedef struct xenpf_del_memtype xenpf_del_memtype_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_del_memtype_t);
+
+/* Read current type of an MTRR (x86-specific). */
+#define XENPF_read_memtype        33
+struct xenpf_read_memtype {
+    /* IN variables. */
+    uint32_t reg;
+    /* OUT variables. */
+    unsigned long mfn;
+    uint64_t nr_mfns;
+    uint32_t type;
+};
+typedef struct xenpf_read_memtype xenpf_read_memtype_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_read_memtype_t);
+
+#define XENPF_microcode_update    35
+struct xenpf_microcode_update {
+    /* IN variables. */
+    GUEST_HANDLE(void) data;          /* Pointer to microcode data */
+    uint32_t length;                  /* Length of microcode data. */
+};
+typedef struct xenpf_microcode_update xenpf_microcode_update_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_microcode_update_t);
+
+#define XENPF_platform_quirk      39
+#define QUIRK_NOIRQBALANCING      1 /* Do not restrict IO-APIC RTE targets */
+#define QUIRK_IOAPIC_BAD_REGSEL   2 /* IO-APIC REGSEL forgets its value    */
+#define QUIRK_IOAPIC_GOOD_REGSEL  3 /* IO-APIC REGSEL behaves properly     */
+struct xenpf_platform_quirk {
+    /* IN variables. */
+    uint32_t quirk_id;
+};
+typedef struct xenpf_platform_quirk xenpf_platform_quirk_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_platform_quirk_t);
+
+#define XENPF_firmware_info       50
+#define XEN_FW_DISK_INFO          1 /* from int 13 AH=08/41/48 */
+#define XEN_FW_DISK_MBR_SIGNATURE 2 /* from MBR offset 0x1b8 */
+#define XEN_FW_VBEDDC_INFO        3 /* from int 10 AX=4f15 */
+struct xenpf_firmware_info {
+    /* IN variables. */
+    uint32_t type;
+    uint32_t index;
+    /* OUT variables. */
+    union {
+        struct {
+            /* Int13, Fn48: Check Extensions Present. */
+            uint8_t device;                   /* %dl: bios device number */
+            uint8_t version;                  /* %ah: major version      */
+            uint16_t interface_support;       /* %cx: support bitmap     */
+            /* Int13, Fn08: Legacy Get Device Parameters. */
+            uint16_t legacy_max_cylinder;     /* %cl[7:6]:%ch: max cyl # */
+            uint8_t legacy_max_head;          /* %dh: max head #         */
+            uint8_t legacy_sectors_per_track; /* %cl[5:0]: max sector #  */
+            /* Int13, Fn41: Get Device Parameters (as filled into %ds:%esi). */
+            /* NB. First uint16_t of buffer must be set to buffer size.      */
+            GUEST_HANDLE(void) edd_params;
+        } disk_info; /* XEN_FW_DISK_INFO */
+        struct {
+            uint8_t device;                   /* bios device number  */
+            uint32_t mbr_signature;           /* offset 0x1b8 in mbr */
+        } disk_mbr_signature; /* XEN_FW_DISK_MBR_SIGNATURE */
+        struct {
+            /* Int10, AX=4F15: Get EDID info. */
+            uint8_t capabilities;
+            uint8_t edid_transfer_time;
+            /* must refer to 128-byte buffer */
+            GUEST_HANDLE(uchar) edid;
+        } vbeddc_info; /* XEN_FW_VBEDDC_INFO */
+    } u;
+};
+typedef struct xenpf_firmware_info xenpf_firmware_info_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_firmware_info_t);
+
+#define XENPF_enter_acpi_sleep    51
+struct xenpf_enter_acpi_sleep {
+    /* IN variables */
+    uint16_t pm1a_cnt_val;      /* PM1a control value. */
+    uint16_t pm1b_cnt_val;      /* PM1b control value. */
+    uint32_t sleep_state;       /* Which state to enter (Sn). */
+    uint32_t flags;             /* Must be zero. */
+};
+typedef struct xenpf_enter_acpi_sleep xenpf_enter_acpi_sleep_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_enter_acpi_sleep_t);
+
+#define XENPF_change_freq         52
+struct xenpf_change_freq {
+    /* IN variables */
+    uint32_t flags; /* Must be zero. */
+    uint32_t cpu;   /* Physical cpu. */
+    uint64_t freq;  /* New frequency (Hz). */
+};
+typedef struct xenpf_change_freq xenpf_change_freq_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_change_freq_t);
+
+/*
+ * Get idle times (nanoseconds since boot) for physical CPUs specified in the
+ * @cpumap_bitmap with range [0..@cpumap_nr_cpus-1]. The @idletime array is
+ * indexed by CPU number; only entries with the corresponding @cpumap_bitmap
+ * bit set are written to. On return, @cpumap_bitmap is modified so that any
+ * non-existent CPUs are cleared. Such CPUs have their @idletime array entry
+ * cleared.
+ */
+#define XENPF_getidletime         53
+struct xenpf_getidletime {
+    /* IN/OUT variables */
+    /* IN: CPUs to interrogate; OUT: subset of IN which are present */
+    GUEST_HANDLE(uchar) cpumap_bitmap;
+    /* IN variables */
+    /* Size of cpumap bitmap. */
+    uint32_t cpumap_nr_cpus;
+    /* Must be indexable for every cpu in cpumap_bitmap. */
+    GUEST_HANDLE(uint64_t) idletime;
+    /* OUT variables */
+    /* System time when the idletime snapshots were taken. */
+    uint64_t now;
+};
+typedef struct xenpf_getidletime xenpf_getidletime_t;
+DEFINE_GUEST_HANDLE_STRUCT(xenpf_getidletime_t);
+
+struct xen_platform_op {
+    uint32_t cmd;
+    uint32_t interface_version; /* XENPF_INTERFACE_VERSION */
+    union {
+        struct xenpf_settime           settime;
+        struct xenpf_add_memtype       add_memtype;
+        struct xenpf_del_memtype       del_memtype;
+        struct xenpf_read_memtype      read_memtype;
+        struct xenpf_microcode_update  microcode;
+        struct xenpf_platform_quirk    platform_quirk;
+        struct xenpf_firmware_info     firmware_info;
+        struct xenpf_enter_acpi_sleep  enter_acpi_sleep;
+        struct xenpf_change_freq       change_freq;
+        struct xenpf_getidletime       getidletime;
+        uint8_t                        pad[128];
+    } u;
+};
+typedef struct xen_platform_op xen_platform_op_t;
+DEFINE_GUEST_HANDLE_STRUCT(xen_platform_op_t);
+
+#endif /* __XEN_PUBLIC_PLATFORM_H__ */
+
+/*
+ * Local variables:
+ * mode: C
+ * c-set-style: "BSD"
+ * c-basic-offset: 4
+ * tab-width: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/include/xen/interface/xen.h b/include/xen/interface/xen.h
--- a/include/xen/interface/xen.h
+++ b/include/xen/interface/xen.h
@@ -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. */

[PATCH 07 of 11] xen mtrr: Add mtrr_ops support for Xen mtrr

[Jeremy Fitzhardinge]

From: Stephen Tweedie <sct@redhat.com>

Add a Xen mtrr type, and reorganise mtrr initialisation slightly to allow the
mtrr driver to set up num_var_ranges (Xen needs to do this by querying the
hypervisor itself.)

Only the boot path is handled for now: we set up a xen-specific mtrr_if
and set up the mtrr tables based on hypervisor information, but we don't
yet handle mtrr entry add/delete.

Signed-off-by: Stephen Tweedie <sct@redhat.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
---
 arch/x86/kernel/cpu/mtrr/Makefile  |    1
 arch/x86/kernel/cpu/mtrr/amd.c     |    1
 arch/x86/kernel/cpu/mtrr/centaur.c |    1
 arch/x86/kernel/cpu/mtrr/cyrix.c   |    1
 arch/x86/kernel/cpu/mtrr/generic.c |    1
 arch/x86/kernel/cpu/mtrr/main.c    |   11 ++++--
 arch/x86/kernel/cpu/mtrr/mtrr.h    |    5 +++
 arch/x86/kernel/cpu/mtrr/xen.c     |   59 ++++++++++++++++++++++++++++++++++++
 8 files changed, 77 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kernel/cpu/mtrr/Makefile b/arch/x86/kernel/cpu/mtrr/Makefile
--- a/arch/x86/kernel/cpu/mtrr/Makefile
+++ b/arch/x86/kernel/cpu/mtrr/Makefile
@@ -1,3 +1,4 @@
 obj-y		:= main.o if.o generic.o state.o
 obj-$(CONFIG_X86_32) += amd.o cyrix.o centaur.o
+obj-$(CONFIG_XEN_DOM0) += xen.o
 
diff --git a/arch/x86/kernel/cpu/mtrr/amd.c b/arch/x86/kernel/cpu/mtrr/amd.c
--- a/arch/x86/kernel/cpu/mtrr/amd.c
+++ b/arch/x86/kernel/cpu/mtrr/amd.c
@@ -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)
diff --git a/arch/x86/kernel/cpu/mtrr/centaur.c b/arch/x86/kernel/cpu/mtrr/centaur.c
--- a/arch/x86/kernel/cpu/mtrr/centaur.c
+++ b/arch/x86/kernel/cpu/mtrr/centaur.c
@@ -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)
diff --git a/arch/x86/kernel/cpu/mtrr/cyrix.c b/arch/x86/kernel/cpu/mtrr/cyrix.c
--- a/arch/x86/kernel/cpu/mtrr/cyrix.c
+++ b/arch/x86/kernel/cpu/mtrr/cyrix.c
@@ -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)
diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c
--- a/arch/x86/kernel/cpu/mtrr/generic.c
+++ b/arch/x86/kernel/cpu/mtrr/generic.c
@@ -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,
 };
diff --git a/arch/x86/kernel/cpu/mtrr/main.c b/arch/x86/kernel/cpu/mtrr/main.c
--- a/arch/x86/kernel/cpu/mtrr/main.c
+++ b/arch/x86/kernel/cpu/mtrr/main.c
@@ -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();
diff --git a/arch/x86/kernel/cpu/mtrr/mtrr.h b/arch/x86/kernel/cpu/mtrr/mtrr.h
--- a/arch/x86/kernel/cpu/mtrr/mtrr.h
+++ b/arch/x86/kernel/cpu/mtrr/mtrr.h
@@ -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);
diff --git a/arch/x86/kernel/cpu/mtrr/xen.c b/arch/x86/kernel/cpu/mtrr/xen.c
new file mode 100644
--- /dev/null
+++ b/arch/x86/kernel/cpu/mtrr/xen.c
@@ -0,0 +1,59 @@
+#include <linux/init.h>
+#include <linux/proc_fs.h>
+#include <linux/ctype.h>
+#include <linux/module.h>
+#include <linux/seq_file.h>
+#include <asm/uaccess.h>
+#include <linux/mutex.h>
+
+#include <asm/mtrr.h>
+#include "mtrr.h"
+
+#include <xen/interface/platform.h>
+#include <asm/xen/hypervisor.h>
+#include <asm/xen/hypercall.h>
+
+static int __init xen_num_var_ranges(void);
+
+/* DOM0 TODO: Need to fill in the remaining mtrr methods to have full
+ * working userland mtrr support. */
+static struct mtrr_ops xen_mtrr_ops = {
+	.vendor            = X86_VENDOR_UNKNOWN,
+//	.set               = xen_set_mtrr,
+//	.get               = xen_get_mtrr,
+	.get_free_region   = generic_get_free_region,
+//	.validate_add_page = xen_validate_add_page,
+	.have_wrcomb       = positive_have_wrcomb,
+	.use_intel_if	   = 0,
+	.num_var_ranges	   = xen_num_var_ranges,
+};
+
+static int __init xen_num_var_ranges(void)
+{
+	int ranges;
+	struct xen_platform_op op;
+
+	for (ranges = 0; ; ranges++) {
+		op.cmd = XENPF_read_memtype;
+		op.u.read_memtype.reg = ranges;
+		if (HYPERVISOR_dom0_op(&op) != 0)
+			break;
+	}
+	return ranges;
+}
+
+void __init xen_init_mtrr(void)
+{
+	struct cpuinfo_x86 *c = &boot_cpu_data;
+
+	if (!xen_initial_domain())
+		return;
+
+	if ((!cpu_has(c, X86_FEATURE_MTRR)) &&
+	    (!cpu_has(c, X86_FEATURE_K6_MTRR)) &&
+	    (!cpu_has(c, X86_FEATURE_CYRIX_ARR)) &&
+	    (!cpu_has(c, X86_FEATURE_CENTAUR_MCR)))
+		return;
+
+	mtrr_if = &xen_mtrr_ops;
+}

[PATCH 08 of 11] xen: forcibly disable PAT support

[Jeremy Fitzhardinge]

From: Ian Campbell <ian.campbell@citrix.com>

Xen imposes a particular PAT layout on all paravirtual guests which
does not match the layout Linux would like to use.

Force PAT to be disabled until this is resolved.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
---
 arch/x86/include/asm/pat.h |    4 ++--
 arch/x86/xen/enlighten.c   |    3 +++
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/arch/x86/include/asm/pat.h b/arch/x86/include/asm/pat.h
--- a/arch/x86/include/asm/pat.h
+++ b/arch/x86/include/asm/pat.h
@@ -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 */
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -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 */

[PATCH 09 of 11] xen/dom0: use _PAGE_IOMAP in ioremap to do machine mappings

[Jeremy Fitzhardinge]

In a Xen domain, ioremap operates on machine addresses, not
pseudo-physical addresses.  We use _PAGE_IOMAP to determine whether a
mapping is intended for machine addresses.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
---
 arch/x86/include/asm/xen/page.h |    8 +----
 arch/x86/xen/enlighten.c        |   14 +++++++--
 arch/x86/xen/mmu.c              |   60 ++++++++++++++++++++++++++++++++++++++-
 3 files changed, 73 insertions(+), 9 deletions(-)

diff --git a/arch/x86/include/asm/xen/page.h b/arch/x86/include/asm/xen/page.h
--- a/arch/x86/include/asm/xen/page.h
+++ b/arch/x86/include/asm/xen/page.h
@@ -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;
 }
 
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -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
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -302,6 +302,28 @@
 	return PagePinned(page);
 }
 
+static bool xen_iomap_pte(pte_t pte)
+{
+	return xen_initial_domain() && (pte_flags(pte) & _PAGE_IOMAP);
+}
+
+static void xen_set_iomap_pte(pte_t *ptep, pte_t pteval)
+{
+	struct multicall_space mcs;
+	struct mmu_update *u;
+
+	mcs = xen_mc_entry(sizeof(*u));
+	u = mcs.args;
+
+	/* ptep might be kmapped when using 32-bit HIGHPTE */
+	u->ptr = arbitrary_virt_to_machine(ptep).maddr;
+	u->val = pte_val_ma(pteval);
+
+	MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, DOMID_IO);
+
+	xen_mc_issue(PARAVIRT_LAZY_MMU);
+}
+
 static void xen_extend_mmu_update(const struct mmu_update *update)
 {
 	struct multicall_space mcs;
@@ -382,6 +404,11 @@
 	if (mm == &init_mm)
 		preempt_disable();
 
+	if (xen_iomap_pte(pteval)) {
+		xen_set_iomap_pte(ptep, pteval);
+		goto out;
+	}
+
 	ADD_STATS(set_pte_at, 1);
 //	ADD_STATS(set_pte_at_pinned, xen_page_pinned(ptep));
 	ADD_STATS(set_pte_at_current, mm == current->mm);
@@ -454,8 +481,25 @@
 	return val;
 }
 
+static pteval_t iomap_pte(pteval_t val)
+{
+	if (val & _PAGE_PRESENT) {
+		unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
+		pteval_t flags = val & PTE_FLAGS_MASK;
+
+		/* We assume the pte frame number is a MFN, so
+		   just use it as-is. */
+		val = ((pteval_t)pfn << PAGE_SHIFT) | flags;
+	}
+
+	return val;
+}
+
 pteval_t xen_pte_val(pte_t pte)
 {
+	if (xen_initial_domain() && (pte.pte & _PAGE_IOMAP))
+		return pte.pte;
+
 	return pte_mfn_to_pfn(pte.pte);
 }
 
@@ -466,7 +510,11 @@
 
 pte_t xen_make_pte(pteval_t pte)
 {
-	pte = pte_pfn_to_mfn(pte);
+	if (unlikely(xen_initial_domain() && (pte & _PAGE_IOMAP)))
+		pte = iomap_pte(pte);
+	else
+		pte = pte_pfn_to_mfn(pte);
+
 	return native_make_pte(pte);
 }
 
@@ -519,6 +567,11 @@
 
 void xen_set_pte(pte_t *ptep, pte_t pte)
 {
+	if (xen_iomap_pte(pte)) {
+		xen_set_iomap_pte(ptep, pte);
+		return;
+	}
+
 	ADD_STATS(pte_update, 1);
 //	ADD_STATS(pte_update_pinned, xen_page_pinned(ptep));
 	ADD_STATS(pte_update_batched, paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU);
@@ -535,6 +588,11 @@
 #ifdef CONFIG_X86_PAE
 void xen_set_pte_atomic(pte_t *ptep, pte_t pte)
 {
+	if (xen_iomap_pte(pte)) {
+		xen_set_iomap_pte(ptep, pte);
+		return;
+	}
+
 	set_64bit((u64 *)ptep, native_pte_val(pte));
 }

[PATCH 10 of 11] paravirt/xen: add pvop for page_is_ram

[Jeremy Fitzhardinge]

A guest domain may have external pages mapped into its address space,
in order to share memory with other domains.  These shared pages are
more akin to io mappings than real RAM, and should not pass the
page_is_ram test.  Add a paravirt op for this so that a hypervisor
backend can validate whether a page should be considered ram or not.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
---
 arch/x86/include/asm/page.h     |    9 ++++++++-
 arch/x86/include/asm/paravirt.h |    7 +++++++
 arch/x86/kernel/paravirt.c      |    1 +
 arch/x86/mm/ioremap.c           |    2 +-
 arch/x86/xen/enlighten.c        |   11 +++++++++++
 5 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/arch/x86/include/asm/page.h b/arch/x86/include/asm/page.h
--- a/arch/x86/include/asm/page.h
+++ b/arch/x86/include/asm/page.h
@@ -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,
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -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)
 
diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c
--- a/arch/x86/kernel/paravirt.c
+++ b/arch/x86/kernel/paravirt.c
@@ -451,6 +451,7 @@
 	},
 
 	.set_fixmap = native_set_fixmap,
+	.page_is_ram = native_page_is_ram,
 };
 
 EXPORT_SYMBOL_GPL(pv_time_ops);
diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c
--- a/arch/x86/mm/ioremap.c
+++ b/arch/x86/mm/ioremap.c
@@ -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;
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -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)

[PATCH 11 of 11] xen: use machine memory map in domain 0

[Jeremy Fitzhardinge]

Unlike the non-paravirt Xen port we do not have distinct
psuedo-physical and I/O memory resource-spaces and therefore resources
in the two can clash. Fix this by registering a memory map which
matches the underlying I/O map. Currently this wastes the memory in
the reserved regions. Eventually we should remap this memory to the
end of the address space.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
---
 arch/x86/xen/setup.c           |   45 +++++++++++++++++++++++++++++++++++++---
 include/xen/interface/memory.h |   29 +++++++++++++++++++++++++
 2 files changed, 71 insertions(+), 3 deletions(-)

diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
--- a/arch/x86/xen/setup.c
+++ b/arch/x86/xen/setup.c
@@ -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
diff --git a/include/xen/interface/memory.h b/include/xen/interface/memory.h
--- a/include/xen/interface/memory.h
+++ b/include/xen/interface/memory.h
@@ -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__ */