[patch 00/20] XEN-paravirt: Xen guest implementation for paravirt_ops interface

[patch 00/20] XEN-paravirt: Xen guest implementation for paravirt_ops interface

[Jeremy Fitzhardinge]

This patch series implements the Linux Xen guest in terms of the
paravirt-ops interface.  The features in implemented this patch series
are:
 * domU only
 * UP only (most code is SMP-safe, but there's no way to create a new vcpu)
 * writable pagetables, with late pinning/early unpinning
   (no shadow pagetable support)
 * supports both PAE and non-PAE modes
 * xen console
 * virtual block device (blockfront)

(Netfront needs a bit of updating, and will be in a separate patch later.)

The patch series is in two parts:

1-11: cleanups to the core kernel, either to fix outright problems,
      or to add appropriate hooks for Xen
12-20: the Xen guest implementation itself

I've tried to make each patch as self-explanatory as possible.  The
series is based on 2.6.20-rc4-mm1.

Thanks,
	J
-- 

[patch 01/20] XEN-paravirt: Fix typo in sync_constant_test_bit()s name.

[Jeremy Fitzhardinge]

[-- Attachment #1: fix-sync-bitops.patch --]
[-- Type: text/plain, Size: 615 bytes --]

Fix typo in sync_constant_test_bit()s name.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>

===================================================================
--- a/include/asm-i386/sync_bitops.h
+++ b/include/asm-i386/sync_bitops.h
@@ -130,7 +130,7 @@ static inline int sync_test_and_change_b
 	return oldbit;
 }
 
-static __always_inline int sync_const_test_bit(int nr, const volatile unsigned long *addr)
+static __always_inline int sync_constant_test_bit(int nr, const volatile unsigned long *addr)
 {
 	return ((1UL << (nr & 31)) &
 		(((const volatile unsigned int *)addr)[nr >> 5])) != 0;

-- 

[patch 02/20] XEN-paravirt: Add a flag to allow the VGA console to be disabled

[Jeremy Fitzhardinge]

[-- Attachment #1: vgacon-enable-flag.patch --]
[-- Type: text/plain, Size: 1374 bytes --]

Add a flag to allow the VGA console to be disabled.  The VGA code will
spin forever if there isn't any real VGA hardware, which will happen
under Xen.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Gerd Hoffmann <kraxel@suse.de>
Cc: Andrew Morton <akpm@osdl.org>

===================================================================
--- a/arch/i386/kernel/setup.c
+++ b/arch/i386/kernel/setup.c
@@ -71,6 +71,9 @@ unsigned long init_pg_tables_end __initd
 
 int disable_pse __devinitdata = 0;
 
+/* use to runtime disable vga_con */
+int vgacon_enabled = 1;
+ 
 /*
  * Machine setup..
  */
@@ -652,7 +655,7 @@ void __init setup_arch(char **cmdline_p)
 
 #ifdef CONFIG_VT
 #if defined(CONFIG_VGA_CONSOLE)
-	if (!efi_enabled || (efi_mem_type(0xa0000) != EFI_CONVENTIONAL_MEMORY))
+	if (vgacon_enabled && (!efi_enabled || (efi_mem_type(0xa0000) != EFI_CONVENTIONAL_MEMORY)))
 		conswitchp = &vga_con;
 #elif defined(CONFIG_DUMMY_CONSOLE)
 	conswitchp = &dummy_con;
===================================================================
--- a/include/asm-i386/setup.h
+++ b/include/asm-i386/setup.h
@@ -77,6 +77,8 @@ void __init add_memory_region(unsigned l
 void __init add_memory_region(unsigned long long start,
 			      unsigned long long size, int type);
 
+extern int vgacon_enabled;
+
 #endif /* __ASSEMBLY__ */
 
 #endif  /*  __KERNEL__  */

-- 

Re: [Xen-devel] [patch 02/20] XEN-paravirt: Add a flag to allow the VGA console to be disabled

[Alan]

On Fri, 12 Jan 2007 17:45:41 -0800
Jeremy Fitzhardinge <jeremy@goop.org> wrote:

> Add a flag to allow the VGA console to be disabled.  The VGA code will
> spin forever if there isn't any real VGA hardware, which will happen
> under Xen.

If it is doing this then the real bug should be fixed so that it doesn't
hang in the same way on a physical system which has no VGA.

[patch 03/20] XEN-paravirt: paravirt: page-table accessors

[Jeremy Fitzhardinge]

[-- Attachment #1: paravirt-pte-accessors.patch --]
[-- Type: text/plain, Size: 14715 bytes --]

Add a set of accessors to pack, unpack and modify page table entries
(at all levels).  This allows a paravirt implementation to control the
contents of pgd/pmd/pte entries.  For example, Xen uses this to
convert the (pseudo-)physical address into a machine address when
populating a pagetable entry, and converting back to pphys address
when an entry is read.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>

--
 arch/i386/kernel/paravirt.c       |  113 +++++++++++++++++++++++++++++++++----
 arch/i386/kernel/vmlinux.lds.S    |    3 
 include/asm-i386/page.h           |   18 ++++-
 include/asm-i386/paravirt.h       |   68 +++++++++++++++++++++-
 include/asm-i386/pgtable-2level.h |    5 -
 include/asm-i386/pgtable-3level.h |   27 ++++----
 6 files changed, 199 insertions(+), 35 deletions(-)

===================================================================
--- a/arch/i386/kernel/paravirt.c
+++ b/arch/i386/kernel/paravirt.c
@@ -34,7 +34,7 @@
 #include <asm/tlbflush.h>
 
 /* nop stub */
-static void native_nop(void)
+void native_nop(void)
 {
 }
 
@@ -400,38 +400,74 @@ static fastcall void native_flush_tlb_si
 }
 
 #ifndef CONFIG_X86_PAE
-static fastcall void native_set_pte(pte_t *ptep, pte_t pteval)
+fastcall void native_set_pte(pte_t *ptep, pte_t pteval)
 {
 	*ptep = pteval;
 }
 
-static fastcall void native_set_pte_at(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pteval)
+fastcall void native_set_pte_at(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pteval)
 {
 	*ptep = pteval;
 }
 
-static fastcall void native_set_pmd(pmd_t *pmdp, pmd_t pmdval)
+fastcall void native_set_pmd(pmd_t *pmdp, pmd_t pmdval)
 {
 	*pmdp = pmdval;
 }
 
+fastcall unsigned long native_pte_val(pte_t pte)
+{
+	return pte.pte_low;
+}
+
+fastcall unsigned long native_pmd_val(pmd_t pmd)
+{
+	BUG();
+	return 0;
+}
+
+fastcall unsigned long native_pgd_val(pgd_t pgd)
+{
+	return pgd.pgd;
+}
+
+fastcall pte_t native_make_pte(unsigned long pte)
+{
+	return (pte_t){ pte };
+}
+
+fastcall pmd_t native_make_pmd(unsigned long pmd)
+{
+	BUG();
+}
+
+fastcall pgd_t native_make_pgd(unsigned long pgd)
+{
+	return (pgd_t){ pgd };
+}
+
+fastcall pte_t native_ptep_get_and_clear(pte_t *ptep)
+{
+	return __pte(xchg(&(ptep)->pte_low, 0));
+}
+
 #else /* CONFIG_X86_PAE */
 
-static fastcall void native_set_pte(pte_t *ptep, pte_t pte)
+fastcall void native_set_pte(pte_t *ptep, pte_t pte)
 {
 	ptep->pte_high = pte.pte_high;
 	smp_wmb();
 	ptep->pte_low = pte.pte_low;
 }
 
-static fastcall void native_set_pte_at(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pte)
+fastcall void native_set_pte_at(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pte)
 {
 	ptep->pte_high = pte.pte_high;
 	smp_wmb();
 	ptep->pte_low = pte.pte_low;
 }
 
-static fastcall void native_set_pte_present(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte)
+fastcall void native_set_pte_present(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pte)
 {
 	ptep->pte_low = 0;
 	smp_wmb();
@@ -440,34 +476,76 @@ static fastcall void native_set_pte_pres
 	ptep->pte_low = pte.pte_low;
 }
 
-static fastcall void native_set_pte_atomic(pte_t *ptep, pte_t pteval)
+fastcall void native_set_pte_atomic(pte_t *ptep, pte_t pteval)
 {
 	set_64bit((unsigned long long *)ptep,pte_val(pteval));
 }
 
-static fastcall void native_set_pmd(pmd_t *pmdp, pmd_t pmdval)
+fastcall void native_set_pmd(pmd_t *pmdp, pmd_t pmdval)
 {
 	set_64bit((unsigned long long *)pmdp,pmd_val(pmdval));
 }
 
-static fastcall void native_set_pud(pud_t *pudp, pud_t pudval)
+fastcall void native_set_pud(pud_t *pudp, pud_t pudval)
 {
 	*pudp = pudval;
 }
 
-static fastcall void native_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
+fastcall void native_pte_clear(struct mm_struct *mm, u32 addr, pte_t *ptep)
 {
 	ptep->pte_low = 0;
 	smp_wmb();
 	ptep->pte_high = 0;
 }
 
-static fastcall void native_pmd_clear(pmd_t *pmd)
+fastcall void native_pmd_clear(pmd_t *pmd)
 {
 	u32 *tmp = (u32 *)pmd;
 	*tmp = 0;
 	smp_wmb();
 	*(tmp + 1) = 0;
+}
+
+fastcall unsigned long long native_pte_val(pte_t pte)
+{
+	return pte.pte_low | ((unsigned long long)pte.pte_high << 32);
+}
+
+fastcall unsigned long long native_pmd_val(pmd_t pmd)
+{
+	return pmd.pmd;
+}
+
+fastcall unsigned long long native_pgd_val(pgd_t pgd)
+{
+	return pgd.pgd;
+}
+
+fastcall pte_t native_make_pte(unsigned long long pte)
+{
+	return (pte_t){ pte };
+}
+
+fastcall pmd_t native_make_pmd(unsigned long long pmd)
+{
+	return (pmd_t){ pmd };
+}
+
+fastcall pgd_t native_make_pgd(unsigned long long pgd)
+{
+	return (pgd_t){ pgd };
+}
+
+fastcall pte_t native_ptep_get_and_clear(pte_t *ptep)
+{
+	pte_t res;
+
+	/* xchg acts as a barrier before the setting of the high bits */
+	res.pte_low = xchg(&ptep->pte_low, 0);
+	res.pte_high = ptep->pte_high;
+	ptep->pte_high = 0;
+
+	return res;
 }
 #endif /* CONFIG_X86_PAE */
 
@@ -564,6 +642,9 @@ struct paravirt_ops paravirt_ops = {
 	.set_pmd = native_set_pmd,
 	.pte_update = (void *)native_nop,
 	.pte_update_defer = (void *)native_nop,
+
+	.ptep_get_and_clear = native_ptep_get_and_clear,
+
 #ifdef CONFIG_X86_PAE
 	.set_pte_atomic = native_set_pte_atomic,
 	.set_pte_present = native_set_pte_present,
@@ -572,6 +653,14 @@ struct paravirt_ops paravirt_ops = {
 	.pmd_clear = native_pmd_clear,
 #endif
 
+	.pte_val = native_pte_val,
+	.pmd_val = native_pmd_val,
+	.pgd_val = native_pgd_val,
+
+	.make_pte = native_make_pte,
+	.make_pmd = native_make_pmd,
+	.make_pgd = native_make_pgd,
+
 	.irq_enable_sysexit = native_irq_enable_sysexit,
 	.iret = native_iret,
 
===================================================================
--- a/arch/i386/kernel/vmlinux.lds.S
+++ b/arch/i386/kernel/vmlinux.lds.S
@@ -21,6 +21,9 @@
 #include <asm/page.h>
 #include <asm/cache.h>
 #include <asm/boot.h>
+
+#undef ENTRY
+#undef ALIGN
 
 OUTPUT_FORMAT("elf32-i386", "elf32-i386", "elf32-i386")
 OUTPUT_ARCH(i386)
===================================================================
--- a/include/asm-i386/page.h
+++ b/include/asm-i386/page.h
@@ -11,7 +11,6 @@
 
 #ifdef __KERNEL__
 #ifndef __ASSEMBLY__
-
 
 #ifdef CONFIG_X86_USE_3DNOW
 
@@ -48,9 +47,11 @@ typedef struct { unsigned long long pmd;
 typedef struct { unsigned long long pmd; } pmd_t;
 typedef struct { unsigned long long pgd; } pgd_t;
 typedef struct { unsigned long long pgprot; } pgprot_t;
+#ifndef CONFIG_PARAVIRT
 #define pmd_val(x)	((x).pmd)
 #define pte_val(x)	((x).pte_low | ((unsigned long long)(x).pte_high << 32))
 #define __pmd(x) ((pmd_t) { (x) } )
+#endif	/* CONFIG_PARAVIRT */
 #define HPAGE_SHIFT	21
 #include <asm-generic/pgtable-nopud.h>
 #else
@@ -58,7 +59,9 @@ typedef struct { unsigned long pgd; } pg
 typedef struct { unsigned long pgd; } pgd_t;
 typedef struct { unsigned long pgprot; } pgprot_t;
 #define boot_pte_t pte_t /* or would you rather have a typedef */
+#ifndef CONFIG_PARAVIRT
 #define pte_val(x)	((x).pte_low)
+#endif
 #define HPAGE_SHIFT	22
 #include <asm-generic/pgtable-nopmd.h>
 #endif
@@ -71,12 +74,14 @@ typedef struct { unsigned long pgprot; }
 #define HAVE_ARCH_HUGETLB_UNMAPPED_AREA
 #endif
 
+#define pgprot_val(x)	((x).pgprot)
+#define __pgprot(x)	((pgprot_t) { (x) } )
+
+#ifndef CONFIG_PARAVIRT
 #define pgd_val(x)	((x).pgd)
-#define pgprot_val(x)	((x).pgprot)
-
 #define __pte(x) ((pte_t) { (x) } )
 #define __pgd(x) ((pgd_t) { (x) } )
-#define __pgprot(x)	((pgprot_t) { (x) } )
+#endif
 
 #endif /* !__ASSEMBLY__ */
 
@@ -143,6 +148,11 @@ extern int page_is_ram(unsigned long pag
 #include <asm-generic/memory_model.h>
 #include <asm-generic/page.h>
 
+#ifdef CONFIG_PARAVIRT
+/* After pte_t, etc, have been defined */
+#include <asm/paravirt.h>
+#endif
+
 #define __HAVE_ARCH_GATE_AREA 1
 #endif /* __KERNEL__ */
 
===================================================================
--- a/include/asm-i386/paravirt.h
+++ b/include/asm-i386/paravirt.h
@@ -25,6 +25,8 @@
 #define CLBR_ANY 0x7
 
 #ifndef __ASSEMBLY__
+#include <linux/types.h>
+
 struct thread_struct;
 struct Xgt_desc_struct;
 struct tss_struct;
@@ -140,12 +142,31 @@ struct paravirt_ops
 	void (fastcall *set_pmd)(pmd_t *pmdp, pmd_t pmdval);
 	void (fastcall *pte_update)(struct mm_struct *mm, u32 addr, pte_t *ptep);
 	void (fastcall *pte_update_defer)(struct mm_struct *mm, u32 addr, pte_t *ptep);
+
+	pte_t (fastcall *ptep_get_and_clear)(pte_t *ptep);
+
 #ifdef CONFIG_X86_PAE
 	void (fastcall *set_pte_atomic)(pte_t *ptep, pte_t pteval);
-	void (fastcall *set_pte_present)(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte);
+	void (fastcall *set_pte_present)(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pte);
 	void (fastcall *set_pud)(pud_t *pudp, pud_t pudval);
-	void (fastcall *pte_clear)(struct mm_struct *mm, unsigned long addr, pte_t *ptep);
+	void (fastcall *pte_clear)(struct mm_struct *mm, u32 addr, pte_t *ptep);
 	void (fastcall *pmd_clear)(pmd_t *pmdp);
+
+	unsigned long long (fastcall *pte_val)(pte_t);
+	unsigned long long (fastcall *pmd_val)(pmd_t);
+	unsigned long long (fastcall *pgd_val)(pgd_t);
+
+	pte_t (fastcall *make_pte)(unsigned long long pte);
+	pmd_t (fastcall *make_pmd)(unsigned long long pmd);
+	pgd_t (fastcall *make_pgd)(unsigned long long pgd);
+#else  /* !CONFIG_X86_PAE */
+	unsigned long (fastcall *pte_val)(pte_t);
+	unsigned long (fastcall *pmd_val)(pmd_t);
+	unsigned long (fastcall *pgd_val)(pgd_t);
+
+	pte_t (fastcall *make_pte)(unsigned long pte);
+	pmd_t (fastcall *make_pmd)(unsigned long pmd);
+	pgd_t (fastcall *make_pgd)(unsigned long pgd);
 #endif
 
 	void (fastcall *set_lazy_mode)(int mode);
@@ -163,6 +184,24 @@ struct paravirt_ops
 		__attribute__((__section__(".paravirtprobe"))) = fn
 
 extern struct paravirt_ops paravirt_ops;
+
+#ifdef CONFIG_X86_PAE
+fastcall unsigned long long native_pte_val(pte_t);
+fastcall unsigned long long native_pmd_val(pmd_t);
+fastcall unsigned long long native_pgd_val(pgd_t);
+
+fastcall pte_t native_make_pte(unsigned long long pte);
+fastcall pmd_t native_make_pmd(unsigned long long pmd);
+fastcall pgd_t native_make_pgd(unsigned long long pgd);
+#else
+fastcall unsigned long native_pte_val(pte_t);
+fastcall unsigned long native_pmd_val(pmd_t);
+fastcall unsigned long native_pgd_val(pgd_t);
+
+fastcall pte_t native_make_pte(unsigned long pte);
+fastcall pmd_t native_make_pmd(unsigned long pmd);
+fastcall pgd_t native_make_pgd(unsigned long pgd);
+#endif
 
 #define paravirt_enabled() (paravirt_ops.paravirt_enabled)
 
@@ -215,6 +254,8 @@ static inline void __cpuid(unsigned int 
 #define read_cr4() paravirt_ops.read_cr4()
 #define read_cr4_safe(x) paravirt_ops.read_cr4_safe()
 #define write_cr4(x) paravirt_ops.write_cr4(x)
+
+#define raw_ptep_get_and_clear(xp)	(paravirt_ops.ptep_get_and_clear(xp))
 
 static inline void raw_safe_halt(void)
 {
@@ -297,6 +338,17 @@ static inline void halt(void)
 	(paravirt_ops.write_idt_entry((dt), (entry), (low), (high)))
 #define set_iopl_mask(mask) (paravirt_ops.set_iopl_mask(mask))
 
+#define __pte(x)	paravirt_ops.make_pte(x)
+#define __pgd(x)	paravirt_ops.make_pgd(x)
+
+#define pte_val(x)	paravirt_ops.pte_val(x)
+#define pgd_val(x)	paravirt_ops.pgd_val(x)
+
+#ifdef CONFIG_X86_PAE
+#define __pmd(x)	paravirt_ops.make_pmd(x)
+#define pmd_val(x)	paravirt_ops.pmd_val(x)
+#endif
+
 /* The paravirtualized I/O functions */
 static inline void slow_down_io(void) {
 	paravirt_ops.io_delay();
@@ -336,6 +388,18 @@ static inline void setup_secondary_clock
 	paravirt_ops.setup_secondary_clock();
 }
 #endif
+
+
+fastcall void native_set_pte(pte_t *ptep, pte_t pteval);
+fastcall void native_set_pte_at(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pteval);
+fastcall void native_set_pmd(pmd_t *pmdp, pmd_t pmdval);
+fastcall void native_set_pte_present(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pte);
+fastcall void native_set_pte_atomic(pte_t *ptep, pte_t pteval);
+fastcall void native_set_pud(pud_t *pudp, pud_t pudval);
+fastcall void native_pte_clear(struct mm_struct *mm, u32 addr, pte_t *ptep);
+fastcall void native_pmd_clear(pmd_t *pmd);
+fastcall pte_t native_ptep_get_and_clear(pte_t *ptep);
+void native_nop(void);
 
 #ifdef CONFIG_SMP
 static inline void startup_ipi_hook(int phys_apicid, unsigned long start_eip,
===================================================================
--- a/include/asm-i386/pgtable-2level.h
+++ b/include/asm-i386/pgtable-2level.h
@@ -15,6 +15,7 @@
 #define set_pte(pteptr, pteval) (*(pteptr) = pteval)
 #define set_pte_at(mm,addr,ptep,pteval) set_pte(ptep,pteval)
 #define set_pmd(pmdptr, pmdval) (*(pmdptr) = (pmdval))
+#define raw_ptep_get_and_clear(xp)	__pte(xchg(&(xp)->pte_low, 0))
 #endif
 
 #define set_pte_atomic(pteptr, pteval) set_pte(pteptr,pteval)
@@ -23,11 +24,9 @@
 #define pte_clear(mm,addr,xp)	do { set_pte_at(mm, addr, xp, __pte(0)); } while (0)
 #define pmd_clear(xp)	do { set_pmd(xp, __pmd(0)); } while (0)
 
-#define raw_ptep_get_and_clear(xp)	__pte(xchg(&(xp)->pte_low, 0))
-
 #define pte_page(x)		pfn_to_page(pte_pfn(x))
 #define pte_none(x)		(!(x).pte_low)
-#define pte_pfn(x)		((unsigned long)(((x).pte_low >> PAGE_SHIFT)))
+#define pte_pfn(x)		(pte_val(x) >> PAGE_SHIFT)
 #define pfn_pte(pfn, prot)	__pte(((pfn) << PAGE_SHIFT) | pgprot_val(prot))
 #define pfn_pmd(pfn, prot)	__pmd(((pfn) << PAGE_SHIFT) | pgprot_val(prot))
 
===================================================================
--- a/include/asm-i386/pgtable-3level.h
+++ b/include/asm-i386/pgtable-3level.h
@@ -98,6 +98,18 @@ static inline void pmd_clear(pmd_t *pmd)
 	smp_wmb();
 	*(tmp + 1) = 0;
 }
+
+static inline pte_t raw_ptep_get_and_clear(pte_t *ptep)
+{
+	pte_t res;
+
+	/* xchg acts as a barrier before the setting of the high bits */
+	res.pte_low = xchg(&ptep->pte_low, 0);
+	res.pte_high = ptep->pte_high;
+	ptep->pte_high = 0;
+
+	return res;
+}
 #endif
 
 /*
@@ -119,18 +131,6 @@ static inline void pud_clear (pud_t * pu
 #define pmd_offset(pud, address) ((pmd_t *) pud_page(*(pud)) + \
 			pmd_index(address))
 
-static inline pte_t raw_ptep_get_and_clear(pte_t *ptep)
-{
-	pte_t res;
-
-	/* xchg acts as a barrier before the setting of the high bits */
-	res.pte_low = xchg(&ptep->pte_low, 0);
-	res.pte_high = ptep->pte_high;
-	ptep->pte_high = 0;
-
-	return res;
-}
-
 #define __HAVE_ARCH_PTE_SAME
 static inline int pte_same(pte_t a, pte_t b)
 {
@@ -146,8 +146,7 @@ static inline int pte_none(pte_t pte)
 
 static inline unsigned long pte_pfn(pte_t pte)
 {
-	return (pte.pte_low >> PAGE_SHIFT) |
-		(pte.pte_high << (32 - PAGE_SHIFT));
+	return pte_val(pte) >> PAGE_SHIFT;
 }
 
 extern unsigned long long __supported_pte_mask;

-- 

[patch 04/20] XEN-paravirt: paravirt pagetable init

[Jeremy Fitzhardinge]

[-- Attachment #1: paravirt-memory-init.patch --]
[-- Type: text/plain, Size: 10884 bytes --]

Add paravirt hooks into the initial pagetable setup.  In the native
case, the kernel builds itself a new initial pagetable from scratch.
In the Xen case, the kernel starts with a pagetable provided by the
hypervisor, which is used as the prototype for the kernel-generated
pagetable.  The hooks added in this patch allow either mode of
operation without having special cases (the main change to the
pagetable construction logic is a testing to make sure a pagetable
slot is actually empty before populating it).

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>

===================================================================
--- a/arch/i386/kernel/paravirt.c +++ b/arch/i386/kernel/paravirt.c @@
-379,6 +379,43 @@ static fastcall void native_io_delay(voi
 {
 	asm volatile("outb %al,$0x80");
 }
+
+void native_pagetable_setup_start(pgd_t *base)
+{
+#ifdef CONFIG_X86_PAE
+	int i;
+
+	/*
+	 * Init entries of the first-level page table to the
+	 * zero page, if they haven't already been set up.
+	 *
+	 * In a normal native boot, we'll be running on a
+	 * pagetable rooted in swapper_pg_dir, but not in PAE
+	 * mode, so this will end up clobbering the mappings
+	 * for the lower 24Mbytes of the address space,
+	 * without affecting the kernel address space.
+	 */
+	for (i = 0; i < USER_PTRS_PER_PGD; i++)
+		set_pgd(&base[i],
+			__pgd(__pa(empty_zero_page) | _PAGE_PRESENT));
+	memset(&base[USER_PTRS_PER_PGD], 0, sizeof(pgd_t));
+#endif
+}
+
+void native_pagetable_setup_done(pgd_t *base)
+{
+#ifdef CONFIG_X86_PAE
+	/*
+	 * Add low memory identity-mappings - SMP needs it when
+	 * starting up on an AP from real-mode. In the non-PAE
+	 * case we already have these mappings through head.S.
+	 * All user-space mappings are explicitly cleared after
+	 * SMP startup.
+	 */
+	set_pgd(&base[0], base[USER_PTRS_PER_PGD]);
+#endif
+}
+
 
 static fastcall void native_flush_tlb(void)
 {
@@ -627,6 +664,9 @@ struct paravirt_ops paravirt_ops = {
 #endif
 	.set_lazy_mode = (void *)native_nop,
 
+	.pagetable_setup_start = native_pagetable_setup_start,
+	.pagetable_setup_done = native_pagetable_setup_done,
+
 	.flush_tlb_user = native_flush_tlb,
 	.flush_tlb_kernel = native_flush_tlb_global,
 	.flush_tlb_single = native_flush_tlb_single,
===================================================================
--- a/arch/i386/mm/init.c
+++ b/arch/i386/mm/init.c
@@ -42,6 +42,7 @@
 #include <asm/tlb.h>
 #include <asm/tlbflush.h>
 #include <asm/sections.h>
+#include <asm/paravirt.h>
 
 unsigned int __VMALLOC_RESERVE = 128 << 20;
 
@@ -62,6 +63,8 @@ static pmd_t * __init one_md_table_init(
 		
 #ifdef CONFIG_X86_PAE
 	pmd_table = (pmd_t *) alloc_bootmem_low_pages(PAGE_SIZE);
+	memset(pmd_table, 0, PAGE_SIZE);
+
 	paravirt_alloc_pd(__pa(pmd_table) >> PAGE_SHIFT);
 	set_pgd(pgd, __pgd(__pa(pmd_table) | _PAGE_PRESENT));
 	pud = pud_offset(pgd, 0);
@@ -83,12 +86,11 @@ static pte_t * __init one_page_table_ini
 {
 	if (pmd_none(*pmd)) {
 		pte_t *page_table = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE);
+		memset(page_table, 0, PAGE_SIZE);
+
 		paravirt_alloc_pt(__pa(page_table) >> PAGE_SHIFT);
 		set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE));
-		if (page_table != pte_offset_kernel(pmd, 0))
-			BUG();	
-
-		return page_table;
+		BUG_ON(page_table != pte_offset_kernel(pmd, 0));
 	}
 	
 	return pte_offset_kernel(pmd, 0);
@@ -119,7 +121,7 @@ static void __init page_table_range_init
 	pgd = pgd_base + pgd_idx;
 
 	for ( ; (pgd_idx < PTRS_PER_PGD) && (vaddr != end); pgd++, pgd_idx++) {
-		if (pgd_none(*pgd)) 
+		if (!(pgd_val(*pgd) & _PAGE_PRESENT)) 
 			one_md_table_init(pgd);
 		pud = pud_offset(pgd, vaddr);
 		pmd = pmd_offset(pud, vaddr);
@@ -158,7 +160,11 @@ static void __init kernel_physical_mappi
 	pfn = 0;
 
 	for (; pgd_idx < PTRS_PER_PGD; pgd++, pgd_idx++) {
-		pmd = one_md_table_init(pgd);
+		if (!(pgd_val(*pgd) & _PAGE_PRESENT))
+			pmd = one_md_table_init(pgd);
+		else
+			pmd = pmd_offset(pud_offset(pgd, PAGE_OFFSET), PAGE_OFFSET);
+
 		if (pfn >= max_low_pfn)
 			continue;
 		for (pmd_idx = 0; pmd_idx < PTRS_PER_PMD && pfn < max_low_pfn; pmd++, pmd_idx++) {
@@ -167,20 +173,26 @@ static void __init kernel_physical_mappi
 			/* Map with big pages if possible, otherwise create normal page tables. */
 			if (cpu_has_pse) {
 				unsigned int address2 = (pfn + PTRS_PER_PTE - 1) * PAGE_SIZE + PAGE_OFFSET + PAGE_SIZE-1;
-
-				if (is_kernel_text(address) || is_kernel_text(address2))
-					set_pmd(pmd, pfn_pmd(pfn, PAGE_KERNEL_LARGE_EXEC));
-				else
-					set_pmd(pmd, pfn_pmd(pfn, PAGE_KERNEL_LARGE));
+				if (!pmd_present(*pmd)) {
+					if (is_kernel_text(address) || is_kernel_text(address2))
+						set_pmd(pmd, pfn_pmd(pfn, PAGE_KERNEL_LARGE_EXEC));
+					else
+						set_pmd(pmd, pfn_pmd(pfn, PAGE_KERNEL_LARGE));
+				}
 				pfn += PTRS_PER_PTE;
 			} else {
 				pte = one_page_table_init(pmd);
 
-				for (pte_ofs = 0; pte_ofs < PTRS_PER_PTE && pfn < max_low_pfn; pte++, pfn++, pte_ofs++) {
-						if (is_kernel_text(address))
-							set_pte(pte, pfn_pte(pfn, PAGE_KERNEL_EXEC));
-						else
-							set_pte(pte, pfn_pte(pfn, PAGE_KERNEL));
+				for (pte_ofs = 0; 
+				     pte_ofs < PTRS_PER_PTE && pfn < max_low_pfn;
+				     pte++, pfn++, pte_ofs++, address += PAGE_SIZE) {
+					if (pte_present(*pte))
+						continue;
+
+					if (is_kernel_text(address))
+						set_pte(pte, pfn_pte(pfn, PAGE_KERNEL_EXEC));
+					else
+						set_pte(pte, pfn_pte(pfn, PAGE_KERNEL));
 				}
 			}
 		}
@@ -337,19 +349,32 @@ extern void __init remap_numa_kva(void);
 #define remap_numa_kva() do {} while (0)
 #endif
 
+/* 
+ * Build a proper pagetable for the kernel mappings.  Up until this
+ * point, we've been running on some set of pagetables constructed by
+ * the boot process.
+ *
+ * If we're booting on native hardware, this will be a pagetable
+ * constructed in arch/i386/kernel/head.S, and not running in PAE mode
+ * (even if we'll end up running in PAE).  The root of the pagetable
+ * will be swapper_pg_dir.
+ *
+ * If we're booting paravirtualized under a hypervisor, then there are
+ * more options: we may already be running PAE, and the pagetable may
+ * or may not be based in swapper_pg_dir.  In any case,
+ * paravirt_pagetable_setup_start() will set up swapper_pg_dir
+ * appropriately for the rest of the initialization to work.
+ *
+ * In general, pagetable_init() assumes that the pagetable may already
+ * be partially populated, and so it avoids stomping on any existing
+ * mappings.
+ */
 static void __init pagetable_init (void)
 {
-	unsigned long vaddr;
+	unsigned long vaddr, end;
 	pgd_t *pgd_base = swapper_pg_dir;
 
-#ifdef CONFIG_X86_PAE
-	int i;
-	/* Init entries of the first-level page table to the zero page */
-	for (i = 0; i < PTRS_PER_PGD; i++)
-		set_pgd(pgd_base + i, __pgd(__pa(empty_zero_page) | _PAGE_PRESENT));
-#else
-	paravirt_alloc_pd(__pa(swapper_pg_dir) >> PAGE_SHIFT);
-#endif
+	paravirt_pagetable_setup_start(pgd_base);
 
 	/* Enable PSE if available */
 	if (cpu_has_pse) {
@@ -371,20 +396,12 @@ static void __init pagetable_init (void)
 	 * created - mappings will be set by set_fixmap():
 	 */
 	vaddr = __fix_to_virt(__end_of_fixed_addresses - 1) & PMD_MASK;
-	page_table_range_init(vaddr, 0, pgd_base);
+	end = (FIXADDR_TOP + PMD_SIZE - 1) & PMD_MASK;
+	page_table_range_init(vaddr, end, pgd_base);
 
 	permanent_kmaps_init(pgd_base);
 
-#ifdef CONFIG_X86_PAE
-	/*
-	 * Add low memory identity-mappings - SMP needs it when
-	 * starting up on an AP from real-mode. In the non-PAE
-	 * case we already have these mappings through head.S.
-	 * All user-space mappings are explicitly cleared after
-	 * SMP startup.
-	 */
-	set_pgd(&pgd_base[0], pgd_base[USER_PTRS_PER_PGD]);
-#endif
+	paravirt_pagetable_setup_done(pgd_base);
 }
 
 #if defined(CONFIG_SOFTWARE_SUSPEND) || defined(CONFIG_ACPI_SLEEP)
===================================================================
--- a/include/asm-i386/paravirt.h
+++ b/include/asm-i386/paravirt.h
@@ -49,6 +49,9 @@ struct paravirt_ops
 	void (*arch_setup)(void);
 	char *(*memory_setup)(void);
 	void (*init_IRQ)(void);
+
+	void (*pagetable_setup_start)(pgd_t *pgd_base);
+	void (*pagetable_setup_done)(pgd_t *pgd_base);
 
 	void (*banner)(void);
 
@@ -185,6 +188,8 @@ struct paravirt_ops
 
 extern struct paravirt_ops paravirt_ops;
 
+void native_pagetable_setup_start(pgd_t *pgd);
+
 #ifdef CONFIG_X86_PAE
 fastcall unsigned long long native_pte_val(pte_t);
 fastcall unsigned long long native_pmd_val(pmd_t);
@@ -389,6 +394,17 @@ static inline void setup_secondary_clock
 }
 #endif
 
+static inline void paravirt_pagetable_setup_start(pgd_t *base)
+{
+	if (paravirt_ops.pagetable_setup_start)
+		(*paravirt_ops.pagetable_setup_start)(base);
+}
+
+static inline void paravirt_pagetable_setup_done(pgd_t *base)
+{
+	if (paravirt_ops.pagetable_setup_done)
+		(*paravirt_ops.pagetable_setup_done)(base);
+}
 
 fastcall void native_set_pte(pte_t *ptep, pte_t pteval);
 fastcall void native_set_pte_at(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pteval);
@@ -615,5 +631,43 @@ 772:;						\
 	call *paravirt_ops+PARAVIRT_read_cr0
 
 #endif /* __ASSEMBLY__ */
+#else  /* !CONFIG_PARAVIRT */
+#include <asm/pgtable.h>
+
+static inline void paravirt_pagetable_setup_start(pgd_t *base)
+{
+#ifdef CONFIG_X86_PAE
+	int i;
+
+	/*
+	 * Init entries of the first-level page table to the
+	 * zero page, if they haven't already been set up.
+	 *
+	 * In a normal native boot, we'll be running on a
+	 * pagetable rooted in swapper_pg_dir, but not in PAE
+	 * mode, so this will end up clobbering the mappings
+	 * for the lower 24Mbytes of the address space,
+	 * without affecting the kernel address space.
+	 */
+	for (i = 0; i < USER_PTRS_PER_PGD; i++)
+		set_pgd(&base[i],
+			__pgd(__pa(empty_zero_page) | _PAGE_PRESENT));
+	memset(&base[USER_PTRS_PER_PGD], 0, sizeof(pgd_t));
+#endif
+}
+
+static inline void paravirt_pagetable_setup_done(pgd_t *base)
+{
+#ifdef CONFIG_X86_PAE
+	/*
+	 * Add low memory identity-mappings - SMP needs it when
+	 * starting up on an AP from real-mode. In the non-PAE
+	 * case we already have these mappings through head.S.
+	 * All user-space mappings are explicitly cleared after
+	 * SMP startup.
+	 */
+	set_pgd(&base[0], base[USER_PTRS_PER_PGD]);
+#endif
+}
 #endif /* CONFIG_PARAVIRT */
 #endif	/* __ASM_PARAVIRT_H */
===================================================================
--- a/include/asm-i386/pgtable.h
+++ b/include/asm-i386/pgtable.h
@@ -15,7 +15,10 @@
 #include <asm/processor.h>
 #include <asm/fixmap.h>
 #include <linux/threads.h>
+
+#ifdef CONFIG_PARAVIRT		/* guarded to prevent cyclic dependency */
 #include <asm/paravirt.h>
+#endif
 
 #ifndef _I386_BITOPS_H
 #include <asm/bitops.h>

-- 

[patch 05/20] XEN-paravirt: paravirt: reserve fixmap slot

[Jeremy Fitzhardinge]

[-- Attachment #1: paravirt-fixmap.patch --]
[-- Type: text/plain, Size: 820 bytes --]

Reserve a new fixmap slot for paravirt backends.  Xen uses this for
mapping the hypervisor shared-info page, which doesn't really exist in
the guest address space.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>

===================================================================
--- a/include/asm-i386/fixmap.h
+++ b/include/asm-i386/fixmap.h
@@ -86,6 +86,9 @@ enum fixed_addresses {
 #ifdef CONFIG_PCI_MMCONFIG
 	FIX_PCIE_MCFG,
 #endif
+#ifdef CONFIG_PARAVIRT
+	FIX_PARAVIRT,
+#endif
 	__end_of_permanent_fixed_addresses,
 	/* temporary boot-time mappings, used before ioremap() is functional */
 #define NR_FIX_BTMAPS	16

-- 

[patch 06/20] XEN-paravirt: remove pgd ctor

[Jeremy Fitzhardinge]

[-- Attachment #1: remove-pgd-ctor.patch --]
[-- Type: text/plain, Size: 3354 bytes --]

Remove the ctor for the pgd cache.  There's no point in having the
cache machinery do this via an indirect call when all pgd are freed in
the one place anyway.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>

--
 arch/i386/mm/init.c        |    8 +++-----
 arch/i386/mm/pgtable.c     |   15 +++++++++++----
 include/asm-i386/pgtable.h |    2 --
 3 files changed, 14 insertions(+), 11 deletions(-)

===================================================================
--- a/arch/i386/mm/init.c
+++ b/arch/i386/mm/init.c
@@ -739,11 +739,9 @@ void __init pgtable_cache_init(void)
 			panic("pgtable_cache_init(): cannot create pmd cache");
 	}
 	pgd_cache = kmem_cache_create("pgd",
-				PTRS_PER_PGD*sizeof(pgd_t),
-				PTRS_PER_PGD*sizeof(pgd_t),
-				0,
-				pgd_ctor,
-				PTRS_PER_PMD == 1 ? pgd_dtor : NULL);
+				      PTRS_PER_PGD*sizeof(pgd_t),
+				      PTRS_PER_PGD*sizeof(pgd_t),
+				      0, NULL, NULL);
 	if (!pgd_cache)
 		panic("pgtable_cache_init(): Cannot create pgd cache");
 }
===================================================================
--- a/arch/i386/mm/pgtable.c
+++ b/arch/i386/mm/pgtable.c
@@ -236,7 +236,7 @@ static inline void pgd_list_del(pgd_t *p
 		set_page_private(next, (unsigned long)pprev);
 }
 
-void pgd_ctor(void *pgd, struct kmem_cache *cache, unsigned long unused)
+static void pgd_ctor(pgd_t *pgd)
 {
 	unsigned long flags;
 
@@ -245,7 +245,7 @@ void pgd_ctor(void *pgd, struct kmem_cac
 		spin_lock_irqsave(&pgd_lock, flags);
 	}
 
-	clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
+	clone_pgd_range(pgd + USER_PTRS_PER_PGD,
 			swapper_pg_dir + USER_PTRS_PER_PGD,
 			KERNEL_PGD_PTRS);
 
@@ -261,10 +261,12 @@ void pgd_ctor(void *pgd, struct kmem_cac
 	spin_unlock_irqrestore(&pgd_lock, flags);
 }
 
-/* never called when PTRS_PER_PMD > 1 */
-void pgd_dtor(void *pgd, struct kmem_cache *cache, unsigned long unused)
+static void pgd_dtor(pgd_t *pgd)
 {
 	unsigned long flags; /* can be called from interrupt context */
+
+	if (PTRS_PER_PMD == 1)
+		return;
 
 	paravirt_release_pd(__pa(pgd) >> PAGE_SHIFT);
 	spin_lock_irqsave(&pgd_lock, flags);
@@ -276,6 +278,9 @@ pgd_t *pgd_alloc(struct mm_struct *mm)
 {
 	int i;
 	pgd_t *pgd = kmem_cache_alloc(pgd_cache, GFP_KERNEL);
+
+	if (pgd)
+		pgd_ctor(pgd);
 
 	if (PTRS_PER_PMD == 1 || !pgd)
 		return pgd;
@@ -296,6 +301,7 @@ out_oom:
 		paravirt_release_pd(__pa(pmd) >> PAGE_SHIFT);
 		kmem_cache_free(pmd_cache, pmd);
 	}
+	pgd_dtor(pgd);
 	kmem_cache_free(pgd_cache, pgd);
 	return NULL;
 }
@@ -313,5 +319,6 @@ void pgd_free(pgd_t *pgd)
 			kmem_cache_free(pmd_cache, pmd);
 		}
 	/* in the non-PAE case, free_pgtables() clears user pgd entries */
+	pgd_dtor(pgd);
 	kmem_cache_free(pgd_cache, pgd);
 }
===================================================================
--- a/include/asm-i386/pgtable.h
+++ b/include/asm-i386/pgtable.h
@@ -41,8 +41,6 @@ extern struct page *pgd_list;
 extern struct page *pgd_list;
 
 void pmd_ctor(void *, struct kmem_cache *, unsigned long);
-void pgd_ctor(void *, struct kmem_cache *, unsigned long);
-void pgd_dtor(void *, struct kmem_cache *, unsigned long);
 void pgtable_cache_init(void);
 void paging_init(void);
 

-- 

[patch 07/20] XEN-paravirt: paravirt shared kernel pmd flag

[Jeremy Fitzhardinge]

[-- Attachment #1: shared-kernel-pmd.patch --]
[-- Type: text/plain, Size: 9852 bytes --]

Xen does not allow guests to have the kernel pmd shared between page
tables, so parameterize pgtable.c to allow both modes of operation.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>

--
 arch/i386/kernel/paravirt.c            |    1 
 arch/i386/mm/fault.c                   |    6 +--
 arch/i386/mm/pageattr.c                |    2 -
 arch/i386/mm/pgtable.c                 |   61 ++++++++++++++++++++------------
 include/asm-i386/page.h                |    7 ++-
 include/asm-i386/paravirt.h            |    1 
 include/asm-i386/pgtable-2level-defs.h |    2 +
 include/asm-i386/pgtable-2level.h      |    2 -
 include/asm-i386/pgtable-3level-defs.h |    6 +++
 include/asm-i386/pgtable-3level.h      |   16 ++------
 include/asm-i386/pgtable.h             |    7 +++
 11 files changed, 68 insertions(+), 43 deletions(-)

===================================================================
--- a/arch/i386/kernel/paravirt.c
+++ b/arch/i386/kernel/paravirt.c
@@ -572,6 +572,7 @@ struct paravirt_ops paravirt_ops = {
 	.name = "bare hardware",
 	.paravirt_enabled = 0,
 	.kernel_rpl = 0,
+	.shared_kernel_pmd = 1,	/* Only used when CONFIG_X86_PAE is set */
 
  	.patch = native_patch,
 	.banner = default_banner,
===================================================================
--- a/arch/i386/mm/fault.c
+++ b/arch/i386/mm/fault.c
@@ -616,8 +616,7 @@ do_sigbus:
 	force_sig_info_fault(SIGBUS, BUS_ADRERR, address, tsk);
 }
 
-#ifndef CONFIG_X86_PAE
-void vmalloc_sync_all(void)
+void _vmalloc_sync_all(void)
 {
 	/*
 	 * Note that races in the updates of insync and start aren't
@@ -628,6 +627,8 @@ void vmalloc_sync_all(void)
 	static DECLARE_BITMAP(insync, PTRS_PER_PGD);
 	static unsigned long start = TASK_SIZE;
 	unsigned long address;
+
+	BUG_ON(SHARED_KERNEL_PMD);
 
 	BUILD_BUG_ON(TASK_SIZE & ~PGDIR_MASK);
 	for (address = start; address >= TASK_SIZE; address += PGDIR_SIZE) {
@@ -651,4 +652,3 @@ void vmalloc_sync_all(void)
 			start = address + PGDIR_SIZE;
 	}
 }
-#endif
===================================================================
--- a/arch/i386/mm/pageattr.c
+++ b/arch/i386/mm/pageattr.c
@@ -91,7 +91,7 @@ static void set_pmd_pte(pte_t *kpte, uns
 	unsigned long flags;
 
 	set_pte_atomic(kpte, pte); 	/* change init_mm */
-	if (PTRS_PER_PMD > 1)
+	if (SHARED_KERNEL_PMD)
 		return;
 
 	spin_lock_irqsave(&pgd_lock, flags);
===================================================================
--- a/arch/i386/mm/pgtable.c
+++ b/arch/i386/mm/pgtable.c
@@ -241,31 +241,42 @@ static void pgd_ctor(pgd_t *pgd)
 	unsigned long flags;
 
 	if (PTRS_PER_PMD == 1) {
+		/* !PAE, no pagetable sharing */
 		memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t));
+
+		clone_pgd_range(pgd + USER_PTRS_PER_PGD,
+				swapper_pg_dir + USER_PTRS_PER_PGD,
+				KERNEL_PGD_PTRS);
+
 		spin_lock_irqsave(&pgd_lock, flags);
-	}
-
-	clone_pgd_range(pgd + USER_PTRS_PER_PGD,
-			swapper_pg_dir + USER_PTRS_PER_PGD,
-			KERNEL_PGD_PTRS);
-
-	if (PTRS_PER_PMD > 1)
-		return;
-
-	/* must happen under lock */
-	paravirt_alloc_pd_clone(__pa(pgd) >> PAGE_SHIFT,
-			__pa(swapper_pg_dir) >> PAGE_SHIFT,
-			USER_PTRS_PER_PGD, PTRS_PER_PGD - USER_PTRS_PER_PGD);
-
-	pgd_list_add(pgd);
-	spin_unlock_irqrestore(&pgd_lock, flags);
+
+		/* must happen under lock */
+		paravirt_alloc_pd_clone(__pa(pgd) >> PAGE_SHIFT,
+					__pa(swapper_pg_dir) >> PAGE_SHIFT,
+					USER_PTRS_PER_PGD,
+					PTRS_PER_PGD - USER_PTRS_PER_PGD);
+
+		pgd_list_add(pgd);
+		spin_unlock_irqrestore(&pgd_lock, flags);
+	} else {
+		/* PAE, PMD may be shared */
+		if (SHARED_KERNEL_PMD) {
+			clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
+					swapper_pg_dir + USER_PTRS_PER_PGD,
+					KERNEL_PGD_PTRS);
+		} else {
+			spin_lock_irqsave(&pgd_lock, flags);
+			pgd_list_add(pgd);
+			spin_unlock_irqrestore(&pgd_lock, flags);
+		}
+	}
 }
 
 static void pgd_dtor(pgd_t *pgd)
 {
 	unsigned long flags; /* can be called from interrupt context */
 
-	if (PTRS_PER_PMD == 1)
+	if (SHARED_KERNEL_PMD)
 		return;
 
 	paravirt_release_pd(__pa(pgd) >> PAGE_SHIFT);
@@ -279,19 +290,25 @@ pgd_t *pgd_alloc(struct mm_struct *mm)
 	int i;
 	pgd_t *pgd = kmem_cache_alloc(pgd_cache, GFP_KERNEL);
 
-	if (pgd)
+	if (likely(pgd))
 		pgd_ctor(pgd);
 
-	if (PTRS_PER_PMD == 1 || !pgd)
+	if (PTRS_PER_PMD == 1 || unlikely(!pgd))
 		return pgd;
 
-	for (i = 0; i < USER_PTRS_PER_PGD; ++i) {
+	for (i = 0; i < (SHARED_KERNEL_PMD ? USER_PTRS_PER_PGD : PTRS_PER_PGD); ++i) {
 		pmd_t *pmd = kmem_cache_alloc(pmd_cache, GFP_KERNEL);
-		if (!pmd)
+		if (unlikely(!pmd))
 			goto out_oom;
+
+		if (i >= USER_PTRS_PER_PGD)
+			memcpy(pmd, (void *)pgd_page_vaddr(swapper_pg_dir[i]),
+			       sizeof(pmd_t) * PTRS_PER_PMD);
+
 		paravirt_alloc_pd(__pa(pmd) >> PAGE_SHIFT);
 		set_pgd(&pgd[i], __pgd(1 + __pa(pmd)));
 	}
+
 	return pgd;
 
 out_oom:
@@ -312,7 +329,7 @@ void pgd_free(pgd_t *pgd)
 
 	/* in the PAE case user pgd entries are overwritten before usage */
 	if (PTRS_PER_PMD > 1)
-		for (i = 0; i < USER_PTRS_PER_PGD; ++i) {
+		for (i = 0; i < (SHARED_KERNEL_PMD ? USER_PTRS_PER_PGD : PTRS_PER_PGD); ++i) {
 			pgd_t pgdent = pgd[i];
 			void* pmd = (void *)__va(pgd_val(pgdent)-1);
 			paravirt_release_pd(__pa(pmd) >> PAGE_SHIFT);
===================================================================
--- a/include/asm-i386/page.h
+++ b/include/asm-i386/page.h
@@ -50,21 +50,23 @@ typedef struct { unsigned long long pgpr
 #ifndef CONFIG_PARAVIRT
 #define pmd_val(x)	((x).pmd)
 #define pte_val(x)	((x).pte_low | ((unsigned long long)(x).pte_high << 32))
+#define __pte(x) ((pte_t) { .pte_low = (x), .pte_high = ((x) >> 32) } )
 #define __pmd(x) ((pmd_t) { (x) } )
 #endif	/* CONFIG_PARAVIRT */
 #define HPAGE_SHIFT	21
 #include <asm-generic/pgtable-nopud.h>
-#else
+#else  /* !CONFIG_X86_PAE */
 typedef struct { unsigned long pte_low; } pte_t;
 typedef struct { unsigned long pgd; } pgd_t;
 typedef struct { unsigned long pgprot; } pgprot_t;
 #define boot_pte_t pte_t /* or would you rather have a typedef */
 #ifndef CONFIG_PARAVIRT
+#define __pte(x) ((pte_t) { (x) })
 #define pte_val(x)	((x).pte_low)
 #endif
 #define HPAGE_SHIFT	22
 #include <asm-generic/pgtable-nopmd.h>
-#endif
+#endif	/* CONFIG_X86_PAE */
 #define PTE_MASK	PAGE_MASK
 
 #ifdef CONFIG_HUGETLB_PAGE
@@ -79,7 +81,6 @@ typedef struct { unsigned long pgprot; }
 
 #ifndef CONFIG_PARAVIRT
 #define pgd_val(x)	((x).pgd)
-#define __pte(x) ((pte_t) { (x) } )
 #define __pgd(x) ((pgd_t) { (x) } )
 #endif
 
===================================================================
--- a/include/asm-i386/paravirt.h
+++ b/include/asm-i386/paravirt.h
@@ -34,6 +34,7 @@ struct paravirt_ops
 struct paravirt_ops
 {
 	unsigned int kernel_rpl;
+	int shared_kernel_pmd;
  	int paravirt_enabled;
 	const char *name;
 
===================================================================
--- a/include/asm-i386/pgtable-2level-defs.h
+++ b/include/asm-i386/pgtable-2level-defs.h
@@ -1,5 +1,7 @@
 #ifndef _I386_PGTABLE_2LEVEL_DEFS_H
 #define _I386_PGTABLE_2LEVEL_DEFS_H
+
+#define SHARED_KERNEL_PMD	0
 
 /*
  * traditional i386 two-level paging structure:
===================================================================
--- a/include/asm-i386/pgtable-2level.h
+++ b/include/asm-i386/pgtable-2level.h
@@ -66,6 +66,4 @@ static inline int pte_exec_kernel(pte_t 
 #define __pte_to_swp_entry(pte)		((swp_entry_t) { (pte).pte_low })
 #define __swp_entry_to_pte(x)		((pte_t) { (x).val })
 
-void vmalloc_sync_all(void);
-
 #endif /* _I386_PGTABLE_2LEVEL_H */
===================================================================
--- a/include/asm-i386/pgtable-3level-defs.h
+++ b/include/asm-i386/pgtable-3level-defs.h
@@ -1,5 +1,11 @@
 #ifndef _I386_PGTABLE_3LEVEL_DEFS_H
 #define _I386_PGTABLE_3LEVEL_DEFS_H
+
+#ifdef CONFIG_PARAVIRT
+#define SHARED_KERNEL_PMD	(paravirt_ops.shared_kernel_pmd)
+#else
+#define SHARED_KERNEL_PMD	1
+#endif
 
 /*
  * PGDIR_SHIFT determines what a top-level page table entry can map
===================================================================
--- a/include/asm-i386/pgtable-3level.h
+++ b/include/asm-i386/pgtable-3level.h
@@ -153,20 +153,14 @@ extern unsigned long long __supported_pt
 
 static inline pte_t pfn_pte(unsigned long page_nr, pgprot_t pgprot)
 {
-	pte_t pte;
-
-	pte.pte_high = (page_nr >> (32 - PAGE_SHIFT)) | \
-					(pgprot_val(pgprot) >> 32);
-	pte.pte_high &= (__supported_pte_mask >> 32);
-	pte.pte_low = ((page_nr << PAGE_SHIFT) | pgprot_val(pgprot)) & \
-							__supported_pte_mask;
-	return pte;
+	return __pte((((unsigned long long)page_nr << PAGE_SHIFT) | 
+		      pgprot_val(pgprot)) & __supported_pte_mask);
 }
 
 static inline pmd_t pfn_pmd(unsigned long page_nr, pgprot_t pgprot)
 {
-	return __pmd((((unsigned long long)page_nr << PAGE_SHIFT) | \
-			pgprot_val(pgprot)) & __supported_pte_mask);
+	return __pmd((((unsigned long long)page_nr << PAGE_SHIFT) |
+		      pgprot_val(pgprot)) & __supported_pte_mask);
 }
 
 /*
@@ -186,6 +180,4 @@ static inline pmd_t pfn_pmd(unsigned lon
 
 #define __pmd_free_tlb(tlb, x)		do { } while (0)
 
-#define vmalloc_sync_all() ((void)0)
-
 #endif /* _I386_PGTABLE_3LEVEL_H */
===================================================================
--- a/include/asm-i386/pgtable.h
+++ b/include/asm-i386/pgtable.h
@@ -239,6 +239,13 @@ static inline pte_t pte_mkwrite(pte_t pt
 static inline pte_t pte_mkwrite(pte_t pte)	{ (pte).pte_low |= _PAGE_RW; return pte; }
 static inline pte_t pte_mkhuge(pte_t pte)	{ (pte).pte_low |= _PAGE_PSE; return pte; }
 
+extern void _vmalloc_sync_all(void);
+static inline void vmalloc_sync_all(void)
+{
+	if (!SHARED_KERNEL_PMD)
+		_vmalloc_sync_all();
+}
+
 #ifdef CONFIG_X86_PAE
 # include <asm/pgtable-3level.h>
 #else

-- 

Re: [Xen-devel] [patch 07/20] XEN-paravirt: paravirt shared kernel pmd flag

[Jan Beulich]

>>> Jeremy Fitzhardinge <jeremy@goop.org> 13.01.07 02:45 >>>
>Xen does not allow guests to have the kernel pmd shared between page
>tables, so parameterize pgtable.c to allow both modes of operation.

I don't think the change to vmalloc_sync_all in this patch is necessary -
the mechanism exists solely for dealing with modular users of the die
notifier in the NMI case. Since there's no NMI visible to guests, I don't
think the mechanism is necessary; it could even be completely
circumvented for paravirtual guests.

Jan

[patch 08/20] XEN-paravirt: paravirt pgd allocation alignment

[Jeremy Fitzhardinge]

[-- Attachment #1: pgd-alignment.patch --]
[-- Type: text/plain, Size: 1982 bytes --]

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>

===================================================================
--- a/arch/i386/kernel/paravirt.c
+++ b/arch/i386/kernel/paravirt.c
@@ -573,6 +573,7 @@ struct paravirt_ops paravirt_ops = {
 	.paravirt_enabled = 0,
 	.kernel_rpl = 0,
 	.shared_kernel_pmd = 1,	/* Only used when CONFIG_X86_PAE is set */
+	.pgd_alignment = sizeof(pgd_t) * PTRS_PER_PGD,
 
  	.patch = native_patch,
 	.banner = default_banner,
===================================================================
--- a/arch/i386/mm/init.c
+++ b/arch/i386/mm/init.c
@@ -745,7 +745,7 @@ void __init pgtable_cache_init(void)
 	}
 	pgd_cache = kmem_cache_create("pgd",
 				      PTRS_PER_PGD*sizeof(pgd_t),
-				      PTRS_PER_PGD*sizeof(pgd_t),
+				      PGD_ALIGNMENT,
 				      0, NULL, NULL);
 	if (!pgd_cache)
 		panic("pgtable_cache_init(): Cannot create pgd cache");
===================================================================
--- a/include/asm-i386/paravirt.h
+++ b/include/asm-i386/paravirt.h
@@ -33,9 +33,12 @@ struct mm_struct;
 struct mm_struct;
 struct paravirt_ops
 {
+ 	int paravirt_enabled;
 	unsigned int kernel_rpl;
+
 	int shared_kernel_pmd;
- 	int paravirt_enabled;
+	int pgd_alignment;
+
 	const char *name;
 
 	/*
===================================================================
--- a/include/asm-i386/pgtable.h
+++ b/include/asm-i386/pgtable.h
@@ -270,6 +270,12 @@ static inline void vmalloc_sync_all(void
 #define pte_update_defer(mm, addr, ptep)	do { } while (0)
 #endif
 
+#ifdef CONFIG_PARAVIRT
+#define PGD_ALIGNMENT	(paravirt_ops.pgd_alignment)
+#else
+#define PGD_ALIGNMENT	(sizeof(pgd_t) * PTRS_PER_PGD)
+#endif
+
 /*
  * We only update the dirty/accessed state if we set
  * the dirty bit by hand in the kernel, since the hardware

-- 

[patch 09/20] XEN-paravirt: dont export paravirt_ops structure, do individual functions

[Jeremy Fitzhardinge]

[-- Attachment #1: no-paravirt_ops-export.patch --]
[-- Type: text/plain, Size: 13310 bytes --]

Wrap the paravirt_ops members we want to export in wrapper functions.
Since we binary-patch the critical ones, this doesn't make a speed
impact.

I moved drm_follow_page into the core, to avoid having to wrap the
various pte ops.  Unlining kernel_fpu_end and using that in the RAID6
code would remove the need to export clts/read_cr0/write_cr0 too.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>

===================================================================
--- a/arch/i386/kernel/paravirt.c
+++ b/arch/i386/kernel/paravirt.c
@@ -596,6 +596,123 @@ static int __init print_banner(void)
 	return 0;
 }
 core_initcall(print_banner);
+
+unsigned long paravirt_save_flags(void)
+{
+	return paravirt_ops.save_fl();
+}
+EXPORT_SYMBOL(paravirt_save_flags);
+
+void paravirt_restore_flags(unsigned long flags)
+{
+	paravirt_ops.restore_fl(flags);
+}
+EXPORT_SYMBOL(paravirt_restore_flags);
+
+void paravirt_irq_disable(void)
+{
+	paravirt_ops.irq_disable();
+}
+EXPORT_SYMBOL(paravirt_irq_disable);
+
+void paravirt_irq_enable(void)
+{
+	paravirt_ops.irq_enable();
+}
+EXPORT_SYMBOL(paravirt_irq_enable);
+
+void paravirt_io_delay(void)
+{
+	paravirt_ops.io_delay();
+}
+EXPORT_SYMBOL(paravirt_io_delay);
+
+void paravirt_const_udelay(unsigned long loops)
+{
+	paravirt_ops.const_udelay(loops);
+}
+EXPORT_SYMBOL(paravirt_const_udelay);
+
+u64 paravirt_read_msr(unsigned int msr, int *err)
+{
+	return paravirt_ops.read_msr(msr, err);
+}
+EXPORT_SYMBOL(paravirt_read_msr);
+
+int paravirt_write_msr(unsigned int msr, u64 val)
+{
+	return paravirt_ops.write_msr(msr, val);
+}
+EXPORT_SYMBOL(paravirt_write_msr);
+
+u64 paravirt_read_tsc(void)
+{
+	return paravirt_ops.read_tsc();
+}
+EXPORT_SYMBOL(paravirt_read_tsc);
+
+int paravirt_enabled(void)
+{
+	return paravirt_ops.paravirt_enabled;
+}
+EXPORT_SYMBOL(paravirt_enabled);
+
+void clts(void)
+{
+	paravirt_ops.clts();
+}
+EXPORT_SYMBOL(clts);
+
+unsigned long read_cr0(void)
+{
+	return paravirt_ops.read_cr0();
+}
+EXPORT_SYMBOL_GPL(read_cr0);
+
+void write_cr0(unsigned long cr0)
+{
+	paravirt_ops.write_cr0(cr0);
+}
+EXPORT_SYMBOL_GPL(write_cr0);
+
+void wbinvd(void)
+{
+	paravirt_ops.wbinvd();
+}
+EXPORT_SYMBOL(wbinvd);
+
+void raw_safe_halt(void)
+{
+	paravirt_ops.safe_halt();
+}
+EXPORT_SYMBOL_GPL(raw_safe_halt);
+
+void halt(void)
+{
+	paravirt_ops.safe_halt();
+}
+EXPORT_SYMBOL_GPL(halt);
+
+#ifdef CONFIG_X86_LOCAL_APIC
+void apic_write(unsigned long reg, unsigned long v)
+{
+	paravirt_ops.apic_write(reg,v);
+}
+EXPORT_SYMBOL_GPL(apic_write);
+
+unsigned long apic_read(unsigned long reg)
+{
+	return paravirt_ops.apic_read(reg);
+}
+EXPORT_SYMBOL_GPL(apic_read);
+#endif
+
+void __cpuid(unsigned int *eax, unsigned int *ebx,
+	     unsigned int *ecx, unsigned int *edx)
+{
+	paravirt_ops.cpuid(eax, ebx, ecx, edx);
+}
+EXPORT_SYMBOL(__cpuid);
 
 /* We simply declare start_kernel to be the paravirt probe of last resort. */
 paravirt_probe(start_kernel);
@@ -712,11 +829,3 @@ struct paravirt_ops paravirt_ops = {
 
 	.startup_ipi_hook = (void *)native_nop,
 };
-
-/*
- * NOTE: CONFIG_PARAVIRT is experimental and the paravirt_ops
- * semantics are subject to change. Hence we only do this
- * internal-only export of this, until it gets sorted out and
- * all lowlevel CPU ops used by modules are separately exported.
- */
-EXPORT_SYMBOL_GPL(paravirt_ops);
===================================================================
--- a/include/asm-i386/delay.h
+++ b/include/asm-i386/delay.h
@@ -17,9 +17,9 @@ extern void __delay(unsigned long loops)
 extern void __delay(unsigned long loops);
 
 #if defined(CONFIG_PARAVIRT) && !defined(USE_REAL_TIME_DELAY)
-#define udelay(n) paravirt_ops.const_udelay((n) * 0x10c7ul)
+#define udelay(n) paravirt_const_udelay((n) * 0x10c7ul)
 
-#define ndelay(n) paravirt_ops.const_udelay((n) * 5ul)
+#define ndelay(n) paravirt_const_udelay((n) * 5ul)
 
 #else /* !PARAVIRT || USE_REAL_TIME_DELAY */
 
===================================================================
--- a/include/asm-i386/paravirt.h
+++ b/include/asm-i386/paravirt.h
@@ -218,8 +218,6 @@ fastcall pgd_t native_make_pgd(unsigned 
 fastcall pgd_t native_make_pgd(unsigned long pgd);
 #endif
 
-#define paravirt_enabled() (paravirt_ops.paravirt_enabled)
-
 static inline void load_esp0(struct tss_struct *tss,
 			     struct thread_struct *thread)
 {
@@ -243,11 +241,8 @@ static inline void do_time_init(void)
 }
 
 /* The paravirtualized CPUID instruction. */
-static inline void __cpuid(unsigned int *eax, unsigned int *ebx,
-			   unsigned int *ecx, unsigned int *edx)
-{
-	paravirt_ops.cpuid(eax, ebx, ecx, edx);
-}
+void __cpuid(unsigned int *eax, unsigned int *ebx,
+	     unsigned int *ecx, unsigned int *edx);
 
 /*
  * These special macros can be used to get or set a debugging register
@@ -255,11 +250,6 @@ static inline void __cpuid(unsigned int 
 #define get_debugreg(var, reg) var = paravirt_ops.get_debugreg(reg)
 #define set_debugreg(val, reg) paravirt_ops.set_debugreg(reg, val)
 
-#define clts() paravirt_ops.clts()
-
-#define read_cr0() paravirt_ops.read_cr0()
-#define write_cr0(x) paravirt_ops.write_cr0(x)
-
 #define read_cr2() paravirt_ops.read_cr2()
 #define write_cr2(x) paravirt_ops.write_cr2(x)
 
@@ -272,62 +262,51 @@ static inline void __cpuid(unsigned int 
 
 #define raw_ptep_get_and_clear(xp)	(paravirt_ops.ptep_get_and_clear(xp))
 
-static inline void raw_safe_halt(void)
-{
-	paravirt_ops.safe_halt();
-}
-
-static inline void halt(void)
-{
-	paravirt_ops.safe_halt();
-}
-#define wbinvd() paravirt_ops.wbinvd()
-
 #define get_kernel_rpl()  (paravirt_ops.kernel_rpl)
 
 #define rdmsr(msr,val1,val2) do {				\
 	int _err;						\
-	u64 _l = paravirt_ops.read_msr(msr,&_err);		\
+	u64 _l = paravirt_read_msr(msr,&_err);			\
 	val1 = (u32)_l;						\
 	val2 = _l >> 32;					\
 } while(0)
 
 #define wrmsr(msr,val1,val2) do {				\
 	u64 _l = ((u64)(val2) << 32) | (val1);			\
-	paravirt_ops.write_msr((msr), _l);			\
+	paravirt_write_msr((msr), _l);				\
 } while(0)
 
 #define rdmsrl(msr,val) do {					\
 	int _err;						\
-	val = paravirt_ops.read_msr((msr),&_err);		\
+	val = paravirt_read_msr((msr),&_err);			\
 } while(0)
 
-#define wrmsrl(msr,val) (paravirt_ops.write_msr((msr),(val)))
+#define wrmsrl(msr,val) (paravirt_write_msr((msr),(val)))
 #define wrmsr_safe(msr,a,b) ({					\
 	u64 _l = ((u64)(b) << 32) | (a);			\
-	paravirt_ops.write_msr((msr),_l);			\
+	paravirt_write_msr((msr),_l);				\
 })
 
 /* rdmsr with exception handling */
 #define rdmsr_safe(msr,a,b) ({					\
 	int _err;						\
-	u64 _l = paravirt_ops.read_msr(msr,&_err);		\
+	u64 _l = paravirt_read_msr(msr,&_err);			\
 	(*a) = (u32)_l;						\
 	(*b) = _l >> 32;					\
 	_err; })
 
 #define rdtsc(low,high) do {					\
-	u64 _l = paravirt_ops.read_tsc();			\
+	u64 _l = paravirt_read_tsc();				\
 	low = (u32)_l;						\
 	high = _l >> 32;					\
 } while(0)
 
 #define rdtscl(low) do {					\
-	u64 _l = paravirt_ops.read_tsc();			\
+	u64 _l = paravirt_read_tsc();				\
 	low = (int)_l;						\
 } while(0)
 
-#define rdtscll(val) (val = paravirt_ops.read_tsc())
+#define rdtscll(val) (val = paravirt_read_tsc())
 
 #define write_tsc(val1,val2) wrmsr(0x10, val1, val2)
 
@@ -364,35 +343,7 @@ static inline void halt(void)
 #define pmd_val(x)	paravirt_ops.pmd_val(x)
 #endif
 
-/* The paravirtualized I/O functions */
-static inline void slow_down_io(void) {
-	paravirt_ops.io_delay();
-#ifdef REALLY_SLOW_IO
-	paravirt_ops.io_delay();
-	paravirt_ops.io_delay();
-	paravirt_ops.io_delay();
-#endif
-}
-
 #ifdef CONFIG_X86_LOCAL_APIC
-/*
- * Basic functions accessing APICs.
- */
-static inline void apic_write(unsigned long reg, unsigned long v)
-{
-	paravirt_ops.apic_write(reg,v);
-}
-
-static inline void apic_write_atomic(unsigned long reg, unsigned long v)
-{
-	paravirt_ops.apic_write_atomic(reg,v);
-}
-
-static inline unsigned long apic_read(unsigned long reg)
-{
-	return paravirt_ops.apic_read(reg);
-}
-
 static inline void setup_boot_clock(void)
 {
 	paravirt_ops.setup_boot_clock();
@@ -489,6 +440,46 @@ static inline void pte_update_defer(stru
 	paravirt_ops.pte_update_defer(mm, addr, ptep);
 }
 
+/* These are the functions exported to modules. */
+int paravirt_enabled(void);
+unsigned long paravirt_save_flags(void);
+void paravirt_restore_flags(unsigned long flags);
+void paravirt_irq_disable(void);
+void paravirt_irq_enable(void);
+void paravirt_const_udelay(unsigned long loops);
+u64 paravirt_read_msr(unsigned int msr, int *err);
+int paravirt_write_msr(unsigned int msr, u64 val);
+u64 paravirt_read_tsc(void);
+void raw_safe_halt(void);
+void halt(void);
+void wbinvd(void);
+void paravirt_io_delay(void);
+static inline void slow_down_io(void) {
+	paravirt_io_delay();
+#ifdef REALLY_SLOW_IO
+	paravirt_io_delay();
+	paravirt_io_delay();
+	paravirt_io_delay();
+#endif
+}
+
+#ifdef CONFIG_X86_LOCAL_APIC
+/*
+ * Basic functions accessing APICs.
+ */
+void apic_write(unsigned long reg, unsigned long v);
+static inline void apic_write_atomic(unsigned long reg, unsigned long v)
+{
+	paravirt_ops.apic_write_atomic(reg,v);
+}
+unsigned long apic_read(unsigned long reg);
+#endif
+
+/* These will be unexported once raid6 is fixed... */
+void clts(void);
+unsigned long read_cr0(void);
+void write_cr0(unsigned long);
+
 #ifdef CONFIG_X86_PAE
 static inline void set_pte_atomic(pte_t *ptep, pte_t pteval)
 {
@@ -551,42 +542,38 @@ static inline unsigned long __raw_local_
 	unsigned long f;
 
 	__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
-					   "call *%1;"
+					   "call paravirt_save_flags;"
 					   "popl %%edx; popl %%ecx",
 					  PARAVIRT_SAVE_FLAGS, CLBR_NONE)
-			     : "=a"(f): "m"(paravirt_ops.save_fl)
-			     : "memory", "cc");
+			     : "=a"(f) : : "memory", "cc");
 	return f;
 }
 
 static inline void raw_local_irq_restore(unsigned long f)
 {
 	__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
-					   "call *%1;"
+					   "call paravirt_restore_flags;"
 					   "popl %%edx; popl %%ecx",
 					  PARAVIRT_RESTORE_FLAGS, CLBR_EAX)
-			     : "=a"(f) : "m" (paravirt_ops.restore_fl), "0"(f)
-			     : "memory", "cc");
+			     : "=a"(f) : "0"(f) : "memory", "cc");
 }
 
 static inline void raw_local_irq_disable(void)
 {
 	__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
-					   "call *%0;"
+					   "call paravirt_irq_disable;"
 					   "popl %%edx; popl %%ecx",
 					  PARAVIRT_IRQ_DISABLE, CLBR_EAX)
-			     : : "m" (paravirt_ops.irq_disable)
-			     : "memory", "eax", "cc");
+			     : : : "memory", "eax", "cc");
 }
 
 static inline void raw_local_irq_enable(void)
 {
 	__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
-					   "call *%0;"
+					   "call paravirt_irq_enable;"
 					   "popl %%edx; popl %%ecx",
 					  PARAVIRT_IRQ_ENABLE, CLBR_EAX)
-			     : : "m" (paravirt_ops.irq_enable)
-			     : "memory", "eax", "cc");
+			     : : : "memory", "eax", "cc");
 }
 
 static inline unsigned long __raw_local_irq_save(void)
@@ -594,15 +581,13 @@ static inline unsigned long __raw_local_
 	unsigned long f;
 
 	__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
-					   "call *%1; pushl %%eax;"
-					   "call *%2; popl %%eax;"
-					   "popl %%edx; popl %%ecx",
+					   "call paravirt_save_flags;"
+					   "pushl %%eax;"
+					   "call paravirt_irq_disable;"
+					   "popl %%eax;popl %%edx; popl %%ecx",
 					  PARAVIRT_SAVE_FLAGS_IRQ_DISABLE,
 					  CLBR_NONE)
-			     : "=a"(f)
-			     : "m" (paravirt_ops.save_fl),
-			       "m" (paravirt_ops.irq_disable)
-			     : "memory", "cc");
+			     : "=a"(f) : : "memory", "cc");
 	return f;
 }
 
===================================================================
--- a/include/linux/irqflags.h
+++ b/include/linux/irqflags.h
@@ -74,11 +74,11 @@
 #endif /* CONFIG_TRACE_IRQFLAGS_SUPPORT */
 
 #ifdef CONFIG_TRACE_IRQFLAGS_SUPPORT
-#define safe_halt()						\
-	do {							\
-		trace_hardirqs_on();				\
-		raw_safe_halt();				\
-	} while (0)
+static inline void safe_halt(void)
+{
+	trace_hardirqs_on();
+	raw_safe_halt();
+}
 
 #define local_save_flags(flags)		raw_local_save_flags(flags)
 
===================================================================
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1174,6 +1174,8 @@ struct page *follow_page(struct vm_area_
 #define FOLL_GET	0x04	/* do get_page on page */
 #define FOLL_ANON	0x08	/* give ZERO_PAGE if no pgtable */
 
+unsigned long __follow_page(void *vaddr);
+
 #ifdef CONFIG_PROC_FS
 void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
 #else
===================================================================
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -976,6 +976,17 @@ no_page_table:
 	return page;
 }
 
+/* You don't want to use this function.  It's for drm_memory.c. */
+unsigned long __follow_page(void *vaddr)
+{
+	pgd_t *pgd = pgd_offset_k((unsigned long)vaddr);
+	pud_t *pud = pud_offset(pgd, (unsigned long)vaddr);
+	pmd_t *pmd = pmd_offset(pud, (unsigned long)vaddr);
+	pte_t *ptep = pte_offset_kernel(pmd, (unsigned long)vaddr);
+	return pte_pfn(*ptep) << PAGE_SHIFT;
+}
+EXPORT_SYMBOL_GPL(__follow_page);
+
 int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
 		unsigned long start, int len, int write, int force,
 		struct page **pages, struct vm_area_struct **vmas)

-- 

Re: [patch 09/20] XEN-paravirt: dont export paravirt_ops structure, do individual functions

[Rusty Russell]

On Fri, 2007-01-12 at 17:45 -0800, Jeremy Fitzhardinge wrote:
> Wrap the paravirt_ops members we want to export in wrapper functions.

Andrew, the removal of paravirt_ops export here will break kvm.  Feel
free to re-add "EXPORT_SYMBOL_GPL(paravirt_ops)" at the bottom of
arch/i386/kernel/paravirt.c; I'm working on a cleaner way for modules
like kvm / lguest (which want to use the native versions directly
anyway).

Thanks,
Rusty.

[patch 10/20] XEN-paravirt: mm lifetime hooks

[Jeremy Fitzhardinge]

[-- Attachment #1: mm-lifetime-hooks.patch --]
[-- Type: text/plain, Size: 4238 bytes --]

Add hooks to allow a paravirt implementation to track the lifetime of
an mm.  Unfortunately dup_mmap and exit_mmap are in generic code, so
we need to #ifdef CONFIG_PARAVIRT.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>

--
 arch/i386/kernel/paravirt.c    |    4 ++++
 include/asm-i386/mmu_context.h |    8 ++++++--
 include/asm-i386/paravirt.h    |   38 ++++++++++++++++++++++++++++++++++++++
 kernel/fork.c                  |    3 +++
 mm/mmap.c                      |    4 ++++
 5 files changed, 55 insertions(+), 2 deletions(-)

===================================================================
--- a/arch/i386/kernel/paravirt.c
+++ b/arch/i386/kernel/paravirt.c
@@ -706,6 +706,10 @@ struct paravirt_ops paravirt_ops = {
 	.irq_enable_sysexit = native_irq_enable_sysexit,
 	.iret = native_iret,
 
+	.dup_mmap = (void *)native_nop,
+	.exit_mmap = (void *)native_nop,
+	.activate_mm = (void *)native_nop,
+
 	.startup_ipi_hook = (void *)native_nop,
 };
 
===================================================================
--- a/include/asm-i386/mmu_context.h
+++ b/include/asm-i386/mmu_context.h
@@ -5,6 +5,7 @@
 #include <asm/atomic.h>
 #include <asm/pgalloc.h>
 #include <asm/tlbflush.h>
+#include <asm/paravirt.h>
 
 /*
  * Used for LDT copy/destruction.
@@ -65,7 +66,10 @@ static inline void switch_mm(struct mm_s
 #define deactivate_mm(tsk, mm)			\
 	asm("movl %0,%%gs": :"r" (0));
 
-#define activate_mm(prev, next) \
-	switch_mm((prev),(next),NULL)
+#define activate_mm(prev, next)				\
+	do {						\
+		paravirt_activate_mm(prev, next);	\
+		switch_mm((prev),(next),NULL);		\
+	} while(0);
 
 #endif
===================================================================
--- a/include/asm-i386/paravirt.h
+++ b/include/asm-i386/paravirt.h
@@ -126,6 +126,12 @@ struct paravirt_ops
 	void (fastcall *io_delay)(void);
 	void (*const_udelay)(unsigned long loops);
 
+	void (fastcall *activate_mm)(struct mm_struct *prev,
+				     struct mm_struct *next);
+	void (fastcall *dup_mmap)(struct mm_struct *oldmm, 
+				  struct mm_struct *mm);
+	void (fastcall *exit_mmap)(struct mm_struct *mm);
+
 #ifdef CONFIG_X86_LOCAL_APIC
 	void (fastcall *apic_write)(unsigned long reg, unsigned long v);
 	void (fastcall *apic_write_atomic)(unsigned long reg, unsigned long v);
@@ -429,6 +435,23 @@ static inline void startup_ipi_hook(int 
 }
 #endif
 
+static inline void paravirt_activate_mm(struct mm_struct *prev,
+					struct mm_struct *next)
+{
+	paravirt_ops.activate_mm(prev, next);
+}
+
+static inline void paravirt_dup_mmap(struct mm_struct *oldmm,
+				     struct mm_struct *mm)
+{
+	paravirt_ops.dup_mmap(oldmm, mm);
+}
+
+static inline void paravirt_exit_mmap(struct mm_struct *mm)
+{
+	paravirt_ops.exit_mmap(mm);
+}
+
 #define __flush_tlb() paravirt_ops.flush_tlb_user()
 #define __flush_tlb_global() paravirt_ops.flush_tlb_kernel()
 #define __flush_tlb_single(addr) paravirt_ops.flush_tlb_single(addr)
@@ -673,5 +696,20 @@ static inline void paravirt_pagetable_se
 	set_pgd(&base[0], base[USER_PTRS_PER_PGD]);
 #endif
 }
+
+static inline void paravirt_activate_mm(struct mm_struct *prev,
+					struct mm_struct *next)
+{
+}
+
+static inline void paravirt_dup_mmap(struct mm_struct *oldmm,
+				     struct mm_struct *mm)
+{
+}
+
+static inline void paravirt_exit_mmap(struct mm_struct *mm)
+{
+}
+
 #endif /* CONFIG_PARAVIRT */
 #endif	/* __ASM_PARAVIRT_H */
===================================================================
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -287,6 +287,9 @@ static inline int dup_mmap(struct mm_str
 		if (retval)
 			goto out;
 	}
+#ifdef CONFIG_PARAVIRT
+	paravirt_dup_mmap(oldmm, mm);
+#endif
 	retval = 0;
 out:
 	up_write(&mm->mmap_sem);
===================================================================
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1969,6 +1969,10 @@ void exit_mmap(struct mm_struct *mm)
 	struct vm_area_struct *vma = mm->mmap;
 	unsigned long nr_accounted = 0;
 	unsigned long end;
+
+#ifdef CONFIG_PARAVIRT
+	paravirt_exit_mmap(mm);
+#endif
 
 	lru_add_drain();
 	flush_cache_mm(mm);

-- 

[patch 11/20] XEN-paravirt: Add apply_to_page_range() which applies a function to a pte range.

[Jeremy Fitzhardinge]

[-- Attachment #1: apply-to-page-range.patch --]
[-- Type: text/plain, Size: 4118 bytes --]

Add a new mm function apply_to_page_range() which applies a given
function to every pte in a given virtual address range in a given mm
structure. This is a generic alternative to cut-and-pasting the Linux
idiomatic pagetable walking code in every place that a sequence of
PTEs must be accessed.

Although this interface is intended to be useful in a wide range of
situations, it is currently used specifically by several Xen
subsystems, for example: to ensure that pagetables have been allocated
for a virtual address range, and to construct batched special
pagetable update requests to map I/O memory (in ioremap()).

Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Christoph Lameter <clameter@sgi.com>

---
 include/linux/mm.h |    5 ++
 mm/memory.c        |   94 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 99 insertions(+)

===================================================================
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1130,6 +1130,11 @@ struct page *follow_page(struct vm_area_
 
 unsigned long __follow_page(void *vaddr);
 
+typedef int (*pte_fn_t)(pte_t *pte, struct page *pmd_page, unsigned long addr,
+			void *data);
+extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
+			       unsigned long size, pte_fn_t fn, void *data);
+
 #ifdef CONFIG_PROC_FS
 void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
 #else
===================================================================
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -1414,6 +1414,100 @@ int remap_pfn_range(struct vm_area_struc
 }
 EXPORT_SYMBOL(remap_pfn_range);
 
+static int apply_to_pte_range(struct mm_struct *mm, pmd_t *pmd,
+				     unsigned long addr, unsigned long end,
+				     pte_fn_t fn, void *data)
+{
+	pte_t *pte;
+	int err;
+	struct page *pmd_page;
+	spinlock_t *ptl;
+
+	pte = (mm == &init_mm) ?
+		pte_alloc_kernel(pmd, addr) :
+		pte_alloc_map_lock(mm, pmd, addr, &ptl);
+	if (!pte)
+		return -ENOMEM;
+
+	BUG_ON(pmd_huge(*pmd));
+
+	pmd_page = pmd_page(*pmd);
+
+	do {
+		err = fn(pte, pmd_page, addr, data);
+		if (err)
+			break;
+	} while (pte++, addr += PAGE_SIZE, addr != end);
+
+	if (mm != &init_mm)
+		pte_unmap_unlock(pte-1, ptl);
+	return err;
+}
+
+static int apply_to_pmd_range(struct mm_struct *mm, pud_t *pud,
+				     unsigned long addr, unsigned long end,
+				     pte_fn_t fn, void *data)
+{
+	pmd_t *pmd;
+	unsigned long next;
+	int err;
+
+	pmd = pmd_alloc(mm, pud, addr);
+	if (!pmd)
+		return -ENOMEM;
+	do {
+		next = pmd_addr_end(addr, end);
+		err = apply_to_pte_range(mm, pmd, addr, next, fn, data);
+		if (err)
+			break;
+	} while (pmd++, addr = next, addr != end);
+	return err;
+}
+
+static int apply_to_pud_range(struct mm_struct *mm, pgd_t *pgd,
+				     unsigned long addr, unsigned long end,
+				     pte_fn_t fn, void *data)
+{
+	pud_t *pud;
+	unsigned long next;
+	int err;
+
+	pud = pud_alloc(mm, pgd, addr);
+	if (!pud)
+		return -ENOMEM;
+	do {
+		next = pud_addr_end(addr, end);
+		err = apply_to_pmd_range(mm, pud, addr, next, fn, data);
+		if (err)
+			break;
+	} while (pud++, addr = next, addr != end);
+	return err;
+}
+
+/*
+ * Scan a region of virtual memory, filling in page tables as necessary
+ * and calling a provided function on each leaf page table.
+ */
+int apply_to_page_range(struct mm_struct *mm, unsigned long addr,
+			unsigned long size, pte_fn_t fn, void *data)
+{
+	pgd_t *pgd;
+	unsigned long next;
+	unsigned long end = addr + size;
+	int err;
+
+	BUG_ON(addr >= end);
+	pgd = pgd_offset(mm, addr);
+	do {
+		next = pgd_addr_end(addr, end);
+		err = apply_to_pud_range(mm, pgd, addr, next, fn, data);
+		if (err)
+			break;
+	} while (pgd++, addr = next, addr != end);
+	return err;
+}
+EXPORT_SYMBOL_GPL(apply_to_page_range);
+
 /*
  * handle_pte_fault chooses page fault handler according to an entry
  * which was read non-atomically.  Before making any commitment, on

-- 

[patch 12/20] XEN-paravirt: Xen: Add nosegneg capability to the vsyscall page notes

[Jeremy Fitzhardinge]

[-- Attachment #1: xen-vsyscall-note.patch --]
[-- Type: text/plain, Size: 2072 bytes --]

Add the "nosegneg" fake capabilty to the vsyscall page notes. This is
used by the runtime linker to select a glibc version which then
disables negative-offset accesses to the thread-local segment via
%gs. These accesses require emulation in Xen (because segments are
truncated to protect the hypervisor address space) and avoiding them
provides a measurable performance boost.

Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
---
 arch/i386/kernel/vsyscall-note.S |   29 +++++++++++++++++++++++++++++
 1 files changed, 29 insertions(+)

===================================================================
--- a/arch/i386/kernel/vsyscall-note.S
+++ b/arch/i386/kernel/vsyscall-note.S
@@ -23,3 +24,31 @@ 3:	.balign 4;		/* pad out section */			 
 	ASM_ELF_NOTE_BEGIN(".note.kernel-version", "a", UTS_SYSNAME, 0)
 	.long LINUX_VERSION_CODE
 	ASM_ELF_NOTE_END
+
+#ifdef CONFIG_XEN
+/*
+ * Add a special note telling glibc's dynamic linker a fake hardware
+ * flavor that it will use to choose the search path for libraries in the
+ * same way it uses real hardware capabilities like "mmx".
+ * We supply "nosegneg" as the fake capability, to indicate that we
+ * do not like negative offsets in instructions using segment overrides,
+ * since we implement those inefficiently.  This makes it possible to
+ * install libraries optimized to avoid those access patterns in someplace
+ * like /lib/i686/tls/nosegneg.  Note that an /etc/ld.so.conf.d/file
+ * corresponding to the bits here is needed to make ldconfig work right.
+ * It should contain:
+ *	hwcap 0 nosegneg
+ * to match the mapping of bit to name that we give here.
+ */
+#define NOTE_KERNELCAP_BEGIN(ncaps, mask) \
+	ASM_ELF_NOTE_BEGIN(".note.kernelcap", "a", "GNU", 2) \
+	.long ncaps, mask
+#define NOTE_KERNELCAP(bit, name) \
+	.byte bit; .asciz name
+#define NOTE_KERNELCAP_END ASM_ELF_NOTE_END
+
+NOTE_KERNELCAP_BEGIN(1, 1)
+NOTE_KERNELCAP(1, "nosegneg")
+NOTE_KERNELCAP_END
+#endif
+

-- 

[patch 13/20] XEN-paravirt: Xen: Add config options and disable unsupported config options.

[Jeremy Fitzhardinge]

[-- Attachment #1: xen-config.patch --]
[-- Type: text/plain, Size: 3758 bytes --]

The XEN config option enables the Xen paravirt_ops interface, which is
installed when the kernel finds itself running under Xen. (By some
as-yet fully defined mechanism, implemented in a future patch.)

Xen is no longer a sub-architecture, so the X86_XEN subarch config
option has gone.

The disabled config options are:
- PREEMPT: Xen doesn't support it
- HZ: set to 100Hz for now, to cut down on VCPU context switch rate.
  This will be adapted to use tickless later.
- kexec: not yet supported

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>

---
 arch/i386/Kconfig              |    8 +++++---
 arch/i386/Kconfig.debug        |    1 +
 arch/i386/paravirt-xen/Kconfig |   17 +++++++++++++++++
 kernel/Kconfig.hz              |    4 ++--
 kernel/Kconfig.preempt         |    1 +
 5 files changed, 26 insertions(+), 5 deletions(-)

===================================================================
--- a/arch/i386/Kconfig
+++ b/arch/i386/Kconfig
@@ -192,6 +192,8 @@ config PARAVIRT
 	  under a hypervisor, improving performance significantly.
 	  However, when run without a hypervisor the kernel is
 	  theoretically slower.  If in doubt, say N.
+
+source "arch/i386/paravirt-xen/Kconfig"
 
 config ACPI_SRAT
 	bool
@@ -298,12 +300,12 @@ config X86_UP_IOAPIC
 
 config X86_LOCAL_APIC
 	bool
-	depends on X86_UP_APIC || ((X86_VISWS || SMP) && !X86_VOYAGER) || X86_GENERICARCH
+	depends on X86_UP_APIC || (((X86_VISWS || SMP) && !X86_VOYAGER) || X86_GENERICARCH)
 	default y
 
 config X86_IO_APIC
 	bool
-	depends on X86_UP_IOAPIC || (SMP && !(X86_VISWS || X86_VOYAGER)) || X86_GENERICARCH
+	depends on X86_UP_IOAPIC || ((SMP && !(X86_VISWS || X86_VOYAGER)) || X86_GENERICARCH)
 	default y
 
 config X86_VISWS_APIC
@@ -743,6 +745,7 @@ source kernel/Kconfig.hz
 
 config KEXEC
 	bool "kexec system call"
+	depends on !XEN
 	help
 	  kexec is a system call that implements the ability to shutdown your
 	  current kernel, and to start another kernel.  It is like a reboot
===================================================================
--- a/arch/i386/Kconfig.debug
+++ b/arch/i386/Kconfig.debug
@@ -79,6 +79,7 @@ config DOUBLEFAULT
 config DOUBLEFAULT
 	default y
 	bool "Enable doublefault exception handler" if EMBEDDED
+	depends on !XEN
 	help
           This option allows trapping of rare doublefault exceptions that
           would otherwise cause a system to silently reboot. Disabling this
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/Kconfig
@@ -0,0 +1,10 @@
+#
+# This Kconfig describes xen options
+#
+
+config XEN
+	bool "Enable support for Xen hypervisor"
+	depends PARAVIRT
+	default y
+	help
+	  This is the Linux Xen port.
===================================================================
--- a/kernel/Kconfig.hz
+++ b/kernel/Kconfig.hz
@@ -3,7 +3,7 @@
 #
 
 choice
-	prompt "Timer frequency"
+	prompt "Timer frequency" if !XEN
 	default HZ_250
 	help
 	 Allows the configuration of the timer frequency. It is customary
@@ -49,7 +49,7 @@ endchoice
 
 config HZ
 	int
-	default 100 if HZ_100
+	default 100 if HZ_100 || XEN
 	default 250 if HZ_250
 	default 300 if HZ_300
 	default 1000 if HZ_1000
===================================================================
--- a/kernel/Kconfig.preempt
+++ b/kernel/Kconfig.preempt
@@ -35,6 +35,7 @@ config PREEMPT_VOLUNTARY
 
 config PREEMPT
 	bool "Preemptible Kernel (Low-Latency Desktop)"
+	depends on !XEN
 	help
 	  This option reduces the latency of the kernel by making
 	  all kernel code (that is not executing in a critical section)

-- 

[patch 15/20] XEN-paravirt: Xen: core paravirt guest changes

[Jeremy Fitzhardinge]

[-- Attachment #1: xen-core.patch --]
[-- Type: text/plain, Size: 90339 bytes --]

This patch adds the core of the Xen implementation of the paravirt_ops
interface.  It consists of the following parts:

* Boot

The kernel provides Xen with a Xen-specific entrypoint.  This
entrypoint simply saves away the Xen boot-info structure pointer
(initially in %esi), and jumps to the common paravirt entrypoint.
This will do the normal paravirt probe; in the Xen case, the probe
simply looks for a non-NULL xen start_info pointer.

The xen_start_kernel() function does the functional equivalent of
head.S, and sets up the basic kernel environment before calling
start_kernel().  This is simpler than the native case because the
hypervisor has already started the kernel with a pagetable setup and
paging enabled.  A number of CPU feature checks are redundant because
Xen's minimum CPU requirements are higher than a native Linux boot.

Having called start_kernel(), the bulk of the Xen backends
interactions with the rest of the kernel are via the various
paravirt_ops entrypoints.

* Memory management

This Xen guest code only implements direct pagetables (rather than
shadow pagetables), so the guest is responsible for creating a
suitable pagetable which can be directly used by the CPU.

This places a few constraints on how the initial pagetable is
constructed vs native boots.  In particular, if PAE is enabled, then
Xen does not allow the guest to share the kernel PMD between multiple
processes (since Xen has its own per-pagetable mappings in this page
as well).

Also, Xen requires all pagetable pages to be mapped read-only in all
pagetables.  To achieve this, all processes share the kernelk pte
pages; this way marking a kernel page read-only in one process will be
automatically reflected in all other processes.

Since in this mode, all pagetable entries contains machine frame
numbers (MFNs), all pagetable updates/reads must translate PFNs
to/from MFNs.  This is not complex, but there are rather a lot of
little hook functions to perform this translation.

The Xen guest also implements late-pin/early-unpin for pagetables.
This means that when creating a new pagetable (on fork or exec), as
much setup is done as possible while the pages are just normal RW
pages, and when pulling down a pagetable on exit, the pages are make
RW before updating all the entries.  This allows these operations to
be peformed with as little hypervisor interaction as possible.

* Context switches

The Xen guest implements batched hypercalls, so that context switches
can be done with a single hypercall.  A typical context switch will
update the kernel stack pointer, and then write the 3 TLS GDT entries;
some context switches will also need to do an LDT update.

* Interrupts

Xen implements interrupts with general event channels.  The Xen guest
code maps these onto dynamically-allocated IRQs.  When Xen wishes to
raise an event, it makes a callback into the kernel code.  This takes
event number, maps it into an IRQ and then feeds it into the top of
the normal interrupt delivery path.

* Time

The hypervisor maintains a nanosecond timebase.  This is exposed to
guests via the tsc; the guests can use the tsc to extrapolate the
current nanosecond timebase without having to make a hypercall.  This
is exposed to the kernel timecode as a clocksource.

The guest also sets up a periodic timer event to drive the jiffies
tick timebase.  The timer tick is turned off while everything is idle,
and replaced with a tick scheduled for the next timeout (if any).

* SMP

This code is UP-only at present, though almost everything here is
already SMP-safe.  All it really lacks is the means to start/stop new
vcpus.


Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Chris Wright <chris@sous-sol.org>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Ian Pratt <ian.pratt@xensource.com>
Cc: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>

---
 arch/i386/Makefile                  |    3 
 arch/i386/kernel/cpu/common.c       |    3 
 arch/i386/kernel/entry.S            |   77 +++
 arch/i386/kernel/head.S             |   12 
 arch/i386/kernel/paravirt.c         |   48 +-
 arch/i386/kernel/vmlinux.lds.S      |    1 
 arch/i386/mm/pgtable.c              |    1 
 arch/i386/paravirt-xen/Makefile     |    2 
 arch/i386/paravirt-xen/enlighten.c  |  805 +++++++++++++++++++++++++++++++++++
 arch/i386/paravirt-xen/events.c     |  473 ++++++++++++++++++++
 arch/i386/paravirt-xen/events.h     |   28 +
 arch/i386/paravirt-xen/features.c   |   29 +
 arch/i386/paravirt-xen/memory.c     |   50 ++
 arch/i386/paravirt-xen/mmu.c        |  375 ++++++++++++++++
 arch/i386/paravirt-xen/mmu.h        |   51 ++
 arch/i386/paravirt-xen/multicalls.c |   62 ++
 arch/i386/paravirt-xen/multicalls.h |   13 
 arch/i386/paravirt-xen/setup.c      |   96 ++++
 arch/i386/paravirt-xen/time.c       |  446 +++++++++++++++++++
 arch/i386/paravirt-xen/xen-head.S   |   29 +
 arch/i386/paravirt-xen/xen-ops.h    |   20 
 arch/i386/paravirt-xen/xen-page.h   |  175 +++++++
 include/asm-i386/hypercall.h        |   21 
 include/asm-i386/irq.h              |    1 
 include/asm-i386/paravirt.h         |   42 +
 include/asm-i386/pda.h              |   11 
 include/xen/features.h              |   26 +
 27 files changed, 2871 insertions(+), 29 deletions(-)

===================================================================
--- a/arch/i386/Makefile
+++ b/arch/i386/Makefile
@@ -93,6 +93,9 @@ mcore-$(CONFIG_X86_ES7000)	:= mach-defau
 mcore-$(CONFIG_X86_ES7000)	:= mach-default
 core-$(CONFIG_X86_ES7000)	:= arch/i386/mach-es7000/
 
+# Xen paravirtualization support
+core-$(CONFIG_XEN)		+= arch/i386/paravirt-xen/
+
 # default subarch .h files
 mflags-y += -Iinclude/asm-i386/mach-default
 
===================================================================
--- a/arch/i386/kernel/cpu/common.c
+++ b/arch/i386/kernel/cpu/common.c
@@ -19,6 +19,7 @@
 #include <mach_apic.h>
 #endif
 #include <asm/pda.h>
+#include <asm/paravirt.h>
 
 #include "cpu.h"
 
@@ -707,6 +708,8 @@ __cpuinit int init_gdt(int cpu, struct t
 	pda->cpu_number = cpu;
 	pda->pcurrent = idle;
 
+	paravirt_init_pda(pda, cpu);
+
 	return 1;
 }
 
===================================================================
--- a/arch/i386/kernel/entry.S
+++ b/arch/i386/kernel/entry.S
@@ -997,6 +997,83 @@ ENTRY(kernel_thread_helper)
 	CFI_ENDPROC
 ENDPROC(kernel_thread_helper)
 
+#ifdef CONFIG_XEN
+/* Xen only supports sysenter/sysexit in ring0 guests,
+   and only if it the guest asks for it.  So for now,
+   this should never be used. */
+ENTRY(xen_sti_sysexit)
+	CFI_STARTPROC
+	ud2
+	CFI_ENDPROC
+	
+ENTRY(xen_hypervisor_callback)
+	CFI_STARTPROC
+	pushl $0
+	CFI_ADJUST_CFA_OFFSET 4
+	SAVE_ALL
+	mov %esp, %eax
+	call xen_evtchn_do_upcall
+	jmp  ret_from_intr
+	CFI_ENDPROC
+	
+# Hypervisor uses this for application faults while it executes.
+# We get here for two reasons:
+#  1. Fault while reloading DS, ES, FS or GS
+#  2. Fault while executing IRET
+# Category 1 we fix up by reattempting the load, and zeroing the segment
+# register if the load fails.
+# Category 2 we fix up by jumping to do_iret_error. We cannot use the
+# normal Linux return path in this case because if we use the IRET hypercall
+# to pop the stack frame we end up in an infinite loop of failsafe callbacks.
+# We distinguish between categories by maintaining a status value in EAX.
+ENTRY(xen_failsafe_callback)
+	CFI_STARTPROC
+	pushl %eax
+	CFI_ADJUST_CFA_OFFSET 4
+	movl $1,%eax
+1:	mov 4(%esp),%ds
+2:	mov 8(%esp),%es
+3:	mov 12(%esp),%fs
+4:	mov 16(%esp),%gs
+	testl %eax,%eax
+	popl %eax
+	CFI_ADJUST_CFA_OFFSET -4
+	jz 5f
+	addl $16,%esp		# EAX != 0 => Category 2 (Bad IRET)
+	CFI_ADJUST_CFA_OFFSET -16
+	jmp iret_exc
+5:	addl $16,%esp		# EAX == 0 => Category 1 (Bad segment)
+	CFI_ADJUST_CFA_OFFSET -16
+	pushl $0
+	CFI_ADJUST_CFA_OFFSET 4
+	SAVE_ALL
+	jmp ret_from_exception
+	CFI_ENDPROC
+	
+.section .fixup,"ax"
+6:	xorl %eax,%eax
+	movl %eax,4(%esp)
+	jmp 1b
+7:	xorl %eax,%eax
+	movl %eax,8(%esp)
+	jmp 2b
+8:	xorl %eax,%eax
+	movl %eax,12(%esp)
+	jmp 3b
+9:	xorl %eax,%eax
+	movl %eax,16(%esp)
+	jmp 4b
+.previous
+.section __ex_table,"a"
+	.align 4
+	.long 1b,6b
+	.long 2b,7b
+	.long 3b,8b
+	.long 4b,9b
+.previous
+		
+#endif	/* CONFIG_XEN */
+	
 .section .rodata,"a"
 #include "syscall_table.S"
 
===================================================================
--- a/arch/i386/kernel/head.S
+++ b/arch/i386/kernel/head.S
@@ -530,6 +530,10 @@ 1:
 	jmp	1b
 #endif
 
+#ifdef CONFIG_XEN
+#include "../paravirt-xen/xen-head.S"
+#endif
+	
 /*
  * Real beginning of normal "text" segment
  */
@@ -539,7 +543,7 @@ ENTRY(_stext)
 /*
  * BSS section
  */
-.section ".bss.page_aligned","w"
+.section ".bss.page_aligned"
 ENTRY(swapper_pg_dir)
 	.fill 1024,4,0
 ENTRY(empty_zero_page)
@@ -609,7 +613,8 @@ ENTRY(boot_gdt_table)
 /*
  * The Global Descriptor Table contains 28 quadwords, per-CPU.
  */
-	.align L1_CACHE_BYTES
+	.section ".data.page_aligned"
+	.align PAGE_SIZE_asm
 ENTRY(cpu_gdt_table)
 	.quad 0x0000000000000000	/* NULL descriptor */
 	.quad 0x0000000000000000	/* 0x0b reserved */
@@ -658,3 +663,6 @@ ENTRY(cpu_gdt_table)
 	.quad 0x0000000000000000	/* 0xf0 - unused */
 	.quad 0x0000000000000000	/* 0xf8 - GDT entry 31: double-fault TSS */
 
+	/* Be sure this is zeroed to avoid false validations in Xen */
+	.fill PAGE_SIZE_asm / 8 - GDT_ENTRIES,8,0
+	.previous
===================================================================
--- a/arch/i386/kernel/paravirt.c
+++ b/arch/i386/kernel/paravirt.c
@@ -146,55 +146,55 @@ void init_IRQ(void)
 	paravirt_ops.init_IRQ();
 }
 
-static fastcall void native_clts(void)
+fastcall void native_clts(void)
 {
 	asm volatile ("clts");
 }
 
-static fastcall unsigned long native_read_cr0(void)
+fastcall unsigned long native_read_cr0(void)
 {
 	unsigned long val;
 	asm volatile("movl %%cr0,%0\n\t" :"=r" (val));
 	return val;
 }
 
-static fastcall void native_write_cr0(unsigned long val)
+fastcall void native_write_cr0(unsigned long val)
 {
 	asm volatile("movl %0,%%cr0": :"r" (val));
 }
 
-static fastcall unsigned long native_read_cr2(void)
+fastcall unsigned long native_read_cr2(void)
 {
 	unsigned long val;
 	asm volatile("movl %%cr2,%0\n\t" :"=r" (val));
 	return val;
 }
 
-static fastcall void native_write_cr2(unsigned long val)
+fastcall void native_write_cr2(unsigned long val)
 {
 	asm volatile("movl %0,%%cr2": :"r" (val));
 }
 
-static fastcall unsigned long native_read_cr3(void)
+fastcall unsigned long native_read_cr3(void)
 {
 	unsigned long val;
 	asm volatile("movl %%cr3,%0\n\t" :"=r" (val));
 	return val;
 }
 
-static fastcall void native_write_cr3(unsigned long val)
+fastcall void native_write_cr3(unsigned long val)
 {
 	asm volatile("movl %0,%%cr3": :"r" (val));
 }
 
-static fastcall unsigned long native_read_cr4(void)
+fastcall unsigned long native_read_cr4(void)
 {
 	unsigned long val;
 	asm volatile("movl %%cr4,%0\n\t" :"=r" (val));
 	return val;
 }
 
-static fastcall unsigned long native_read_cr4_safe(void)
+fastcall unsigned long native_read_cr4_safe(void)
 {
 	unsigned long val;
 	/* This could fault if %cr4 does not exist */
@@ -207,7 +207,7 @@ static fastcall unsigned long native_rea
 	return val;
 }
 
-static fastcall void native_write_cr4(unsigned long val)
+fastcall void native_write_cr4(unsigned long val)
 {
 	asm volatile("movl %0,%%cr4": :"r" (val));
 }
@@ -246,12 +246,12 @@ static fastcall void native_halt(void)
 	asm volatile("hlt": : :"memory");
 }
 
-static fastcall void native_wbinvd(void)
+fastcall void native_wbinvd(void)
 {
 	asm volatile("wbinvd": : :"memory");
 }
 
-static fastcall unsigned long long native_read_msr(unsigned int msr, int *err)
+fastcall unsigned long long native_read_msr(unsigned int msr, int *err)
 {
 	unsigned long long val;
 
@@ -270,7 +270,7 @@ static fastcall unsigned long long nativ
 	return val;
 }
 
-static fastcall int native_write_msr(unsigned int msr, unsigned long long val)
+fastcall int native_write_msr(unsigned int msr, unsigned long long val)
 {
 	int err;
 	asm volatile("2: wrmsr ; xorl %0,%0\n"
@@ -288,14 +288,14 @@ static fastcall int native_write_msr(uns
 	return err;
 }
 
-static fastcall unsigned long long native_read_tsc(void)
+fastcall unsigned long long native_read_tsc(void)
 {
 	unsigned long long val;
 	asm volatile("rdtsc" : "=A" (val));
 	return val;
 }
 
-static fastcall unsigned long long native_read_pmc(void)
+fastcall unsigned long long native_read_pmc(void)
 {
 	unsigned long long val;
 	asm volatile("rdpmc" : "=A" (val));
@@ -317,17 +317,17 @@ static fastcall void native_load_idt(con
 	asm volatile("lidt %0"::"m" (*dtr));
 }
 
-static fastcall void native_store_gdt(struct Xgt_desc_struct *dtr)
+fastcall void native_store_gdt(struct Xgt_desc_struct *dtr)
 {
 	asm ("sgdt %0":"=m" (*dtr));
 }
 
-static fastcall void native_store_idt(struct Xgt_desc_struct *dtr)
+fastcall void native_store_idt(struct Xgt_desc_struct *dtr)
 {
 	asm ("sidt %0":"=m" (*dtr));
 }
 
-static fastcall unsigned long native_store_tr(void)
+fastcall unsigned long native_store_tr(void)
 {
 	unsigned long tr;
 	asm ("str %0":"=r" (tr));
@@ -336,9 +336,9 @@ static fastcall unsigned long native_sto
 
 static fastcall void native_load_tls(struct thread_struct *t, unsigned int cpu)
 {
-#define C(i) get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + i] = t->tls_array[i]
-	C(0); C(1); C(2);
-#undef C
+	get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + 0] = t->tls_array[0];
+	get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + 1] = t->tls_array[1];
+	get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + 2] = t->tls_array[2];
 }
 
 static inline void native_write_dt_entry(void *dt, int entry, u32 entry_low, u32 entry_high)
@@ -348,17 +348,17 @@ static inline void native_write_dt_entry
 	lp[1] = entry_high;
 }
 
-static fastcall void native_write_ldt_entry(void *dt, int entrynum, u32 low, u32 high)
+fastcall void native_write_ldt_entry(void *dt, int entrynum, u32 low, u32 high)
 {
 	native_write_dt_entry(dt, entrynum, low, high);
 }
 
-static fastcall void native_write_gdt_entry(void *dt, int entrynum, u32 low, u32 high)
+fastcall void native_write_gdt_entry(void *dt, int entrynum, u32 low, u32 high)
 {
 	native_write_dt_entry(dt, entrynum, low, high);
 }
 
-static fastcall void native_write_idt_entry(void *dt, int entrynum, u32 low, u32 high)
+fastcall void native_write_idt_entry(void *dt, int entrynum, u32 low, u32 high)
 {
 	native_write_dt_entry(dt, entrynum, low, high);
 }
===================================================================
--- a/arch/i386/kernel/vmlinux.lds.S
+++ b/arch/i386/kernel/vmlinux.lds.S
@@ -100,6 +100,7 @@ SECTIONS
 
   . = ALIGN(4096);
   .data.page_aligned : AT(ADDR(.data.page_aligned) - LOAD_OFFSET) {
+	*(.data.page_aligned)
 	*(.data.idt)
   }
 
===================================================================
--- a/arch/i386/mm/pgtable.c
+++ b/arch/i386/mm/pgtable.c
@@ -267,6 +267,7 @@ static void pgd_ctor(pgd_t *pgd)
 					swapper_pg_dir + USER_PTRS_PER_PGD,
 					KERNEL_PGD_PTRS);
 		} else {
+			memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t));
 			spin_lock_irqsave(&pgd_lock, flags);
 			pgd_list_add(pgd);
 			spin_unlock_irqrestore(&pgd_lock, flags);
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/Makefile
@@ -0,0 +1,2 @@
+obj-y		:= enlighten.o setup.o events.o memory.o time.o \
+			features.o mmu.o multicalls.o
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/enlighten.c
@@ -0,0 +1,805 @@
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/preempt.h>
+#include <linux/percpu.h>
+#include <linux/delay.h>
+#include <linux/start_kernel.h>
+#include <linux/sched.h>
+#include <linux/bootmem.h>
+
+#include <xen/interface/xen.h>
+#include <xen/features.h>
+
+#include <asm/paravirt.h>
+#include <asm/page.h>
+#include <asm/hypercall.h>
+#include <asm/hypervisor.h>
+#include <asm/fixmap.h>
+#include <asm/processor.h>
+#include <asm/setup.h>
+#include <asm/desc.h>
+#include <asm/pgtable.h>
+
+#include "xen-ops.h"
+#include "xen-page.h"
+#include "mmu.h"
+#include "multicalls.h"
+
+extern struct Xgt_desc_struct cpu_gdt_descr;
+extern struct i386_pda boot_pda;
+extern unsigned long init_pg_tables_end;
+
+static DEFINE_PER_CPU(unsigned, lazy_mode);
+
+/* Code defined in entry.S (not a function) */
+extern const char xen_sti_sysexit[];
+
+struct start_info *xen_start_info;
+
+static unsigned xen_patch(u8 type, u16 clobber, void *firstinsn, unsigned len)
+{
+	/* Xen will require relocations to patch calls and jmps, and
+	   perhaps chunks of inline code */
+	return len;
+}
+
+static void __init xen_banner(void)
+{
+	printk(KERN_INFO "Booting paravirtualized kernel on %s\n",
+	       paravirt_ops.name);
+	printk(KERN_INFO "Hypervisor signature: %s\n", xen_start_info->magic);
+}
+
+static void xen_init_pda(struct i386_pda *pda, int cpu)
+{
+	/* Don't re-init boot CPU; we do it once very early in boot,
+	   and then then cpu_init tries to do it again. If so, just
+	   reuse the stuff we already set up. */
+	if (cpu == 0 && pda != &boot_pda) {
+		BUG_ON(boot_pda.xen.vcpu == NULL);
+		pda->xen = boot_pda.xen;
+		return;
+	}
+
+	pda->xen.vcpu = &HYPERVISOR_shared_info->vcpu_info[cpu];
+	pda->xen.cr3 = 0;
+}
+
+static fastcall void xen_cpuid(unsigned int *eax, unsigned int *ebx,
+			       unsigned int *ecx, unsigned int *edx)
+{
+	unsigned maskedx = ~0;
+	if (*eax == 1)
+		maskedx = ~(1 << X86_FEATURE_APIC);
+
+	asm(XEN_EMULATE_PREFIX "cpuid"
+		: "=a" (*eax),
+		  "=b" (*ebx),
+		  "=c" (*ecx),
+		  "=d" (*edx)
+		: "0" (*eax), "2" (*ecx));
+	*edx &= maskedx;
+}
+
+static fastcall void xen_set_debugreg(int reg, unsigned long val)
+{
+	HYPERVISOR_set_debugreg(reg, val);
+}
+
+static fastcall unsigned long xen_get_debugreg(int reg)
+{
+	return HYPERVISOR_get_debugreg(reg);
+}
+
+static fastcall unsigned long xen_save_fl(void)
+{
+	struct vcpu_info *vcpu;
+	unsigned long flags;
+
+	preempt_disable();
+	vcpu = read_pda(xen.vcpu);
+	/* flag has opposite sense of mask */
+	flags = !vcpu->evtchn_upcall_mask;
+	preempt_enable();
+
+	/* convert to IF type flag 
+	   -0 -> 0x00000000
+	   -1 -> 0xffffffff
+	*/
+	return (-flags) & X86_EFLAGS_IF;
+}
+
+static fastcall void xen_restore_fl(unsigned long flags)
+{
+	struct vcpu_info *vcpu;
+
+	preempt_disable();
+
+	/* convert from IF type flag */
+	flags = !(flags & X86_EFLAGS_IF);
+	vcpu = read_pda(xen.vcpu);
+	vcpu->evtchn_upcall_mask = flags;
+	if (flags == 0) {
+		barrier(); /* unmask then check (avoid races) */
+		if (unlikely(vcpu->evtchn_upcall_pending))
+			force_evtchn_callback();
+		preempt_enable();
+	} else
+		preempt_enable_no_resched();
+}
+
+static fastcall void xen_irq_disable(void)
+{
+	struct vcpu_info *vcpu;
+	preempt_disable();
+	vcpu = read_pda(xen.vcpu);
+	vcpu->evtchn_upcall_mask = 1;
+	preempt_enable_no_resched();
+}
+
+static fastcall void xen_irq_enable(void)
+{
+	struct vcpu_info *vcpu;
+
+	preempt_disable();
+	vcpu = read_pda(xen.vcpu);
+	vcpu->evtchn_upcall_mask = 0;
+	barrier(); /* unmask then check (avoid races) */
+	if (unlikely(vcpu->evtchn_upcall_pending))
+		force_evtchn_callback();
+	preempt_enable();
+}
+
+static fastcall void xen_safe_halt(void)
+{
+	stop_hz_timer();
+	/* Blocking includes an implicit local_irq_enable(). */
+	if (HYPERVISOR_sched_op(SCHEDOP_block, 0) != 0)
+		BUG();
+	start_hz_timer();
+}
+
+static fastcall void xen_halt(void)
+{
+#if 0
+	if (irqs_disabled())
+		HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
+#endif
+}
+
+static void xen_set_lazy_mode(int mode)
+{
+	unsigned *lazy = &get_cpu_var(lazy_mode);
+
+	if (xen_mc_flush())
+		BUG();
+
+	*lazy = mode;
+
+	put_cpu_var(lazy_mode);
+}
+
+static unsigned xen_get_lazy_mode(void)
+{
+	unsigned ret = get_cpu_var(lazy_mode);
+	put_cpu_var(lazy_mode);
+
+	return ret;
+}
+
+static fastcall void xen_load_tr_desc(void)
+{
+	/* do nothing */
+}
+
+static fastcall unsigned long xen_store_tr(void)
+{
+	return 0;
+}
+
+static fastcall void xen_set_ldt(const void *addr, unsigned entries)
+{
+	struct mmuext_op *op;
+	struct multicall_space mcs = xen_mc_entry(sizeof(*op));
+
+	op = mcs.args;
+	op->cmd = MMUEXT_SET_LDT;
+	op->arg1.linear_addr = (unsigned long)addr;
+	if (addr)
+		/* ldt my be vmalloced, use arbitrary_virt_to_machine */
+		op->arg1.linear_addr = arbitrary_virt_to_machine((unsigned long)addr).maddr;
+	op->arg2.nr_ents = entries;
+
+	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+	if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU)
+		xen_mc_flush();
+}
+
+static fastcall void xen_load_gdt(const struct Xgt_desc_struct *dtr)
+{
+        unsigned long va;
+        int f;
+	unsigned size = dtr->size + 1;
+	unsigned long frames[16];
+
+	BUG_ON(size > 16*PAGE_SIZE);
+
+        for (va = dtr->address, f = 0;
+             va < dtr->address + size;
+             va += PAGE_SIZE, f++) {
+                frames[f] = virt_to_mfn(va);
+		make_lowmem_page_readonly((void *)va);
+        }
+
+	/* This is used very early, so we can't rely on per-cpu data
+	   being set up, so no multicalls */
+	if (HYPERVISOR_set_gdt(frames, size/8))
+		BUG();
+}
+
+static void load_TLS_descriptor(struct thread_struct *t,
+				unsigned int cpu, unsigned int i)
+{
+	xmaddr_t maddr = virt_to_machine(&get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN+i]);
+	struct multicall_space mc = xen_mc_entry(0);
+
+	MULTI_update_descriptor(mc.mc, maddr.maddr, t->tls_array[i]);
+}
+
+static fastcall void xen_load_tls(struct thread_struct *t, unsigned int cpu)
+{
+	load_TLS_descriptor(t, cpu, 0);
+	load_TLS_descriptor(t, cpu, 1);
+	load_TLS_descriptor(t, cpu, 2);
+
+	if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU && xen_mc_flush())
+		BUG();
+}
+
+static fastcall void xen_write_ldt_entry(void *dt, int entrynum, u32 low, u32 high)
+{
+        unsigned long lp = (unsigned long)dt + entrynum * 8;
+        xmaddr_t mach_lp = virt_to_machine(lp);
+	u64 entry = (u64)high << 32 | low;
+
+	xen_mc_flush();
+        if (HYPERVISOR_update_descriptor(mach_lp.maddr, entry))
+		BUG();
+}
+
+static int cvt_gate_to_trap(int vector, u32 low, u32 high, struct trap_info *info)
+{
+	u8 type, dpl;
+
+	type = (high >> 8) & 0x1f;
+	dpl = (high >> 13) & 3;
+
+	if (type != 0xf && type != 0xe)
+		return 0;
+
+	info->vector = vector;
+	info->address = (high & 0xffff0000) | (low & 0x0000ffff);
+	info->cs = low >> 16;
+	info->flags = dpl;
+	/* interrupt gates clear IF */
+	if (type == 0xe)
+		info->flags |= 4;
+
+	return 1;
+}
+
+#if 0
+static void unpack_desc(u32 low, u32 high,
+			unsigned long *base, unsigned long *limit,
+			unsigned char *type, unsigned char *flags)
+{
+	*base = (high & 0xff000000) | ((high << 16) & 0x00ff0000) | ((low >> 16) & 0xffff);
+	*limit = (high & 0x000f0000) | (low & 0xffff);
+	*type = (high >> 8) & 0xff;
+	*flags = (high >> 20) & 0xf;
+}
+#endif
+
+/* Locations of each CPU's IDT */
+static DEFINE_PER_CPU(struct Xgt_desc_struct, idt_desc);
+
+/* Set an IDT entry.  If the entry is part of the current IDT, then
+   also update Xen. */
+static fastcall void xen_write_idt_entry(void *dt, int entrynum, u32 low, u32 high)
+{
+
+	int cpu = smp_processor_id();
+	unsigned long p = (unsigned long)dt + entrynum * 8;
+	unsigned long start = per_cpu(idt_desc, cpu).address;
+	unsigned long end = start + per_cpu(idt_desc, cpu).size + 1;
+
+	xen_mc_flush();
+
+	native_write_idt_entry(dt, entrynum, low, high);
+
+	if (p >= start && (p + 8) <= end) {
+		struct trap_info info[2];
+
+		info[1].address = 0;
+
+		if (cvt_gate_to_trap(entrynum, low, high, &info[0]))
+			if (HYPERVISOR_set_trap_table(info))
+				BUG();
+	}
+}
+
+/* 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 fastcall void xen_load_idt(const struct Xgt_desc_struct *desc)
+{
+	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 / 8;
+	BUG_ON(count > 256);
+
+	spin_lock(&lock);
+	for(in = out = 0; in < count; in++) {
+		const u32 *entry = (u32 *)(desc->address + in * 8);
+
+		if (cvt_gate_to_trap(in, entry[0], entry[1], &traps[out]))
+			out++;
+	}
+	traps[out].address = 0;
+
+	xen_mc_flush();
+	if (HYPERVISOR_set_trap_table(traps))
+		BUG();
+
+	spin_unlock(&lock);
+}
+
+/* Write a GDT descriptor entry.  Ignore LDT descriptors, since
+   they're handled differently. */
+static fastcall void xen_write_gdt_entry(void *dt, int entry, u32 low, u32 high)
+{
+	switch ((high >> 8) & 0xff) {
+	case DESCTYPE_LDT:
+	case DESCTYPE_TSS:
+		/* ignore */
+		break;
+
+	default:
+		xen_mc_flush();
+		if (HYPERVISOR_update_descriptor(virt_to_machine(dt + entry*8).maddr,
+						 (u64)high << 32 | low))
+			BUG();
+	}
+}
+
+static fastcall void xen_load_esp0(struct tss_struct *tss,
+				   struct thread_struct *thread)
+{
+	if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU) {
+		if (HYPERVISOR_stack_switch(__KERNEL_DS, thread->esp0))
+			BUG();
+	} else {
+		struct multicall_space mcs = xen_mc_entry(0);
+		MULTI_stack_switch(mcs.mc, __KERNEL_DS, thread->esp0);
+	}
+}
+
+static fastcall void xen_set_iopl_mask(unsigned mask)
+{
+#if 0
+	struct physdev_set_iopl set_iopl;
+
+	/* Force the change at ring 0. */
+	set_iopl.iopl = (mask == 0) ? 1 : (mask >> 12) & 3;
+	HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
+#endif
+}
+
+static fastcall void xen_io_delay(void)
+{
+}
+
+#ifdef CONFIG_X86_LOCAL_APIC
+static fastcall void xen_apic_write(unsigned long reg, unsigned long v)
+{
+}
+
+static fastcall void xen_apic_write_atomic(unsigned long reg, unsigned long v)
+{
+}
+
+static fastcall unsigned long xen_apic_read(unsigned long reg)
+{
+	return 0;
+}
+#endif
+
+static fastcall void xen_flush_tlb(void)
+{
+	struct mmuext_op *op;
+	struct multicall_space mcs = xen_mc_entry(sizeof(*op));
+
+	op = mcs.args;
+	op->cmd = MMUEXT_TLB_FLUSH_LOCAL;
+	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+	if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU && xen_mc_flush())
+		BUG();
+}
+
+static fastcall void xen_flush_tlb_global(void)
+{
+	struct mmuext_op *op;
+	struct multicall_space mcs = xen_mc_entry(sizeof(*op));
+
+	op = mcs.args;
+	op->cmd = MMUEXT_TLB_FLUSH_ALL;
+	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+	if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU && xen_mc_flush())
+		BUG();
+}
+
+static fastcall void xen_flush_tlb_single(u32 addr)
+{
+	struct mmuext_op *op;
+	struct multicall_space mcs = xen_mc_entry(sizeof(*op));
+
+	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);
+
+	if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU && xen_mc_flush())
+		BUG();
+}
+
+static fastcall unsigned long xen_read_cr2(void)
+{
+	return read_pda(xen.vcpu)->arch.cr2;
+}
+
+static fastcall 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.
+ */
+#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 fastcall unsigned long xen_read_cr3(void)
+{
+	return read_pda(xen.cr3);
+}
+
+static fastcall void xen_write_cr3(unsigned long cr3)
+{
+	if (cr3 == read_pda(xen.cr3)) {
+		/* just a simple tlb flush */
+		xen_flush_tlb();
+		return;
+	}
+
+	write_pda(xen.cr3, cr3);
+
+
+	{
+		struct mmuext_op *op;
+		struct multicall_space mcs = xen_mc_entry(sizeof(*op));
+		unsigned long mfn = pfn_to_mfn(PFN_DOWN(cr3));
+
+		op = mcs.args;
+		op->cmd = MMUEXT_NEW_BASEPTR;
+		op->arg1.mfn = mfn;
+
+		MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+		if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU && xen_mc_flush())
+			BUG();
+	}
+}
+
+static fastcall void xen_alloc_pt(u32 pfn)
+{
+	/* XXX pfn isn't necessarily a lowmem page */
+	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
+}
+
+static fastcall void xen_alloc_pd(u32 pfn)
+{
+	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
+}
+
+static fastcall void xen_release_pd(u32 pfn)
+{
+	make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
+	/* make sure next person to allocate this page gets a clean
+	   pmd */
+	clear_page(__va(PFN_PHYS(pfn)));
+}
+
+static fastcall void xen_release_pt(u32 pfn)
+{
+	make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
+}
+
+static fastcall void xen_alloc_pd_clone(u32 pfn, u32 clonepfn,
+					u32 start, u32 count)
+{
+	xen_alloc_pd(pfn);
+}
+
+static __init void xen_pagetable_setup_start(pgd_t *base)
+{
+	pgd_t *xen_pgd = (pgd_t *)xen_start_info->pt_base;
+
+	init_mm.pgd = base;
+
+	/* copy top-level of Xen-supplied pagetable into place.  For
+	   !PAE we can use this as-is, but for PAE it is a stand-in
+	   while we copy the pmd pages. */
+	memcpy(base, xen_pgd, PTRS_PER_PGD * sizeof(pgd_t));
+
+	if (PTRS_PER_PMD > 1) {
+		int i;
+
+		/* For PAE, need to allocate new pmds, rather than
+		   share Xen's, since Xen doesn't like pmd's being
+		   shared between address spaces, even though in this
+		   case they're effectively the same address space. */
+		for(i = 0; i < PTRS_PER_PGD; i++) {
+			if (pgd_val_ma(xen_pgd[i]) & _PAGE_PRESENT) {
+				pmd_t *pmd = (pmd_t *)alloc_bootmem_low_pages(PAGE_SIZE);
+
+				memcpy(pmd, (void *)pgd_page_vaddr(xen_pgd[i]),
+				       PAGE_SIZE);
+
+				xen_alloc_pd(PFN_DOWN(__pa(pmd)));
+
+				set_pgd(&base[i], __pgd(1 + __pa(pmd)));
+			} else
+				pgd_clear(&base[i]);
+		}
+	}
+
+	/* make sure the zero_page is mapped RO so we
+	   can use it in pagetables */
+	make_lowmem_page_readonly(empty_zero_page);
+	make_lowmem_page_readonly(base);
+
+	/* Switch to new pagetable.  This is done before
+	   pagetable_init has done anything so that the new pages
+	   added to the table can be prepared properly for Xen.  */
+	printk("about to switch to new pagetable %p...\n", base);
+	xen_write_cr3(__pa(base));
+	printk("done\n");
+}
+
+static __init void xen_pagetable_setup_done(pgd_t *base)
+{
+	/* init_mm has a new pagetable set up - make sure the GDT page
+	   is still read-only in the new pagetable */
+	xen_load_gdt(&cpu_gdt_descr);
+
+	if (!xen_feature(XENFEAT_writable_page_tables)) {
+		/* Create a mapping for the shared info page.
+		   Should be set_fixmap(), but shared_info is a machine
+		   address with no corresponding pseudo-phys address. */
+		set_pte_mfn(fix_to_virt(FIX_PARAVIRT),
+			    PFN_DOWN(xen_start_info->shared_info),
+			    PAGE_KERNEL);
+		
+		HYPERVISOR_shared_info =
+			(struct shared_info *)fix_to_virt(FIX_PARAVIRT);
+	} else
+		HYPERVISOR_shared_info =
+			(struct shared_info *)__va(xen_start_info->shared_info);
+
+	xen_pgd_pin(base);
+
+	write_pda(xen.vcpu, &HYPERVISOR_shared_info->vcpu_info[smp_processor_id()]);
+}
+
+static const struct paravirt_ops xen_paravirt_ops __initdata = {
+	.paravirt_enabled = 1,
+	.shared_kernel_pmd = 0,
+	.pgd_alignment = PAGE_SIZE,
+
+	.name = "Xen",
+	.banner = xen_banner,
+
+	.patch = xen_patch,
+
+	.memory_setup = xen_memory_setup,
+	.arch_setup = xen_arch_setup,
+	.init_IRQ = xen_init_IRQ,
+	.time_init = xen_time_init,
+	.init_pda = xen_init_pda,
+
+	.cpuid = xen_cpuid,
+
+	.set_debugreg = xen_set_debugreg,
+	.get_debugreg = xen_get_debugreg,
+
+	.clts = native_clts,
+
+	.read_cr0 = native_read_cr0,
+	.write_cr0 = native_write_cr0,
+
+	.read_cr2 = xen_read_cr2,
+	.write_cr2 = native_write_cr2,
+
+	.read_cr3 = xen_read_cr3,
+	.write_cr3 = xen_write_cr3,
+
+	.read_cr4 = native_read_cr4,
+	.read_cr4_safe = native_read_cr4_safe,
+	.write_cr4 = xen_write_cr4,
+
+	.save_fl = xen_save_fl,
+	.restore_fl = xen_restore_fl,
+	.irq_disable = xen_irq_disable,
+	.irq_enable = xen_irq_enable,
+	.safe_halt = xen_safe_halt,
+	.halt = xen_halt,
+	.wbinvd = native_wbinvd,
+
+	.read_msr = native_read_msr,
+	.write_msr = native_write_msr,
+	.read_tsc = native_read_tsc,
+	.read_pmc = native_read_pmc,
+
+	.iret = (void (fastcall *)(void))&hypercall_page[__HYPERVISOR_iret],
+	.irq_enable_sysexit = (void (fastcall *)(void))xen_sti_sysexit,
+
+	.load_tr_desc = xen_load_tr_desc,
+	.set_ldt = xen_set_ldt,
+	.load_gdt = xen_load_gdt,
+	.load_idt = xen_load_idt,
+	.load_tls = xen_load_tls,
+
+	.store_gdt = native_store_gdt,
+	.store_idt = native_store_idt,
+	.store_tr = xen_store_tr,
+
+	.write_ldt_entry = xen_write_ldt_entry,
+	.write_gdt_entry = xen_write_gdt_entry,
+	.write_idt_entry = xen_write_idt_entry,
+	.load_esp0 = xen_load_esp0,
+
+	.set_iopl_mask = xen_set_iopl_mask,
+	.io_delay = xen_io_delay,
+	.const_udelay = __const_udelay,
+	.set_wallclock = xen_set_wallclock,
+	.get_wallclock = xen_get_wallclock,
+
+#ifdef CONFIG_X86_LOCAL_APIC
+	.apic_write = xen_apic_write,
+	.apic_write_atomic = xen_apic_write_atomic,
+	.apic_read = xen_apic_read,
+#endif
+
+	.flush_tlb_user = xen_flush_tlb,
+	.flush_tlb_kernel = xen_flush_tlb_global,
+	.flush_tlb_single = xen_flush_tlb_single,
+
+	.pte_update = (void *)native_nop,
+	.pte_update_defer = (void *)native_nop,
+
+	.pagetable_setup_start = xen_pagetable_setup_start,
+	.pagetable_setup_done = xen_pagetable_setup_done,
+	.activate_mm = xen_activate_mm,
+	.dup_mmap = xen_dup_mmap,
+	.exit_mmap = xen_exit_mmap,
+
+	.set_pte = xen_set_pte,
+	.set_pte_at = xen_set_pte_at,
+	.set_pmd = xen_set_pmd,
+
+	.alloc_pt = xen_alloc_pt,
+	.alloc_pd = xen_alloc_pd,
+	.alloc_pd_clone = xen_alloc_pd_clone,
+	.release_pd = xen_release_pd,
+	.release_pt = xen_release_pt,
+
+	.pte_val = xen_pte_val,
+	.pmd_val = xen_pmd_val,
+	.pgd_val = xen_pgd_val,
+
+	.make_pte = xen_make_pte,
+	.make_pmd = xen_make_pmd,
+	.make_pgd = xen_make_pgd,
+
+	.ptep_get_and_clear = xen_ptep_get_and_clear,
+
+#ifdef CONFIG_X86_PAE
+	.set_pte_atomic = xen_set_pte,
+	.set_pte_present = xen_set_pte_at,
+	.set_pud = xen_set_pud,
+	.pte_clear = xen_pte_clear,
+	.pmd_clear = xen_pmd_clear,
+#endif	/* PAE */
+
+	.set_lazy_mode = xen_set_lazy_mode,
+	.startup_ipi_hook = (void *)native_nop,
+};
+
+/* First C function to be called on Xen boot */
+static asmlinkage void __init xen_start_kernel(void)
+{
+	u32 low, high;
+	pgd_t *pgd;
+
+	if (!xen_start_info)
+		return;
+
+	BUG_ON(memcmp(xen_start_info->magic, "xen-3.0", 7) != 0);
+
+	/* Install Xen paravirt ops */
+	paravirt_ops = xen_paravirt_ops;
+
+	xen_setup_features();
+
+	/* Get mfn list */
+	if (!xen_feature(XENFEAT_auto_translated_physmap))
+		phys_to_machine_mapping = (unsigned long *)xen_start_info->mfn_list;
+
+	pgd = (pgd_t *)xen_start_info->pt_base;
+
+	init_pg_tables_end = __pa(pgd) + xen_start_info->nr_pt_frames*PAGE_SIZE;
+
+	/* set up the boot-time gdt and segments */
+	init_mm.pgd = pgd; /* use the Xen pagetables to start */
+
+	xen_load_gdt(&cpu_gdt_descr);
+
+	/* set up PDA descriptor */
+	pack_descriptor(&low, &high, (unsigned)&boot_pda, sizeof(boot_pda)-1,
+			0x80 | DESCTYPE_S | 0x02, 0);
+
+	/* Use hypercall directly, because xen_write_gdt_entry can't
+	 * be used until batched multicalls work. */
+	if (HYPERVISOR_update_descriptor(virt_to_machine(cpu_gdt_table +
+							 GDT_ENTRY_PDA).maddr,
+					 (u64)high << 32 | low))
+		BUG();
+
+	/* set up %fs and init Xen parts of the PDA */
+	asm volatile("mov %0, %%fs" : : "r" (__KERNEL_PDA) : "memory");
+	xen_init_pda(&boot_pda, 0);
+	boot_pda.xen.cr3 = __pa(pgd);
+
+	paravirt_ops.kernel_rpl = xen_feature(XENFEAT_supervisor_mode_kernel) ? 0 : 1;
+
+	/* set the limit of our address space */
+	reserve_top_address(-HYPERVISOR_VIRT_START + 2 * PAGE_SIZE);
+
+	/* set up basic CPUID stuff */
+	cpu_detect(&new_cpu_data);
+	new_cpu_data.hard_math = 1;
+	identify_cpu(&new_cpu_data);
+
+	/* Poke various useful things into boot_params */
+	LOADER_TYPE = (9 << 4) | 0;
+	INITRD_START = xen_start_info->mod_start ? __pa(xen_start_info->mod_start) : 0;
+	INITRD_SIZE = xen_start_info->mod_len;
+
+	/* Start the world */
+	start_kernel();
+}
+
+paravirt_probe(xen_start_kernel);
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/events.c
@@ -0,0 +1,473 @@
+#include <linux/linkage.h>
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+#include <linux/module.h>
+#include <linux/string.h>
+
+#include <asm/ptrace.h>
+#include <asm/irq.h>
+#include <asm/sync_bitops.h>
+#include <asm/hypercall.h>
+
+#include <xen/interface/xen.h>
+#include <xen/interface/event_channel.h>
+
+#include "xen-ops.h"
+#include "events.h"
+
+/*
+ * This lock protects updates to the following mapping and reference-count
+ * arrays. The lock does not need to be acquired to read the mapping tables.
+ */
+static DEFINE_SPINLOCK(irq_mapping_update_lock);
+
+/* IRQ <-> VIRQ mapping. */
+DEFINE_PER_CPU(int, virq_to_irq[NR_VIRQS]) = {[0 ... NR_VIRQS-1] = -1};
+
+/* Packed IRQ information: binding type, sub-type index, and event channel. */
+static u32 irq_info[NR_IRQS];
+
+/* Binding types. */
+enum { IRQT_UNBOUND, IRQT_PIRQ, IRQT_VIRQ, IRQT_IPI, IRQT_EVTCHN };
+
+/* Convenient shorthand for packed representation of an unbound IRQ. */
+#define IRQ_UNBOUND	mk_irq_info(IRQT_UNBOUND, 0, 0)
+
+static int evtchn_to_irq[NR_EVENT_CHANNELS] = {
+	[0 ... NR_EVENT_CHANNELS-1] = -1
+};
+static unsigned long cpu_evtchn_mask[NR_CPUS][NR_EVENT_CHANNELS/BITS_PER_LONG];
+static u8 cpu_evtchn[NR_EVENT_CHANNELS];
+
+/* Reference counts for bindings to IRQs. */
+static int irq_bindcount[NR_IRQS];
+
+/* Xen will never allocate port zero for any purpose. */
+#define VALID_EVTCHN(chn)	((chn) != 0)
+
+/*
+ * Force a proper event-channel callback from Xen after clearing the
+ * callback mask. We do this in a very simple manner, by making a call
+ * down into Xen. The pending flag will be checked by Xen on return.
+ */
+void force_evtchn_callback(void)
+{
+	(void)HYPERVISOR_xen_version(0, NULL);
+}
+EXPORT_SYMBOL_GPL(force_evtchn_callback);
+
+static struct irq_chip xen_dynamic_chip;
+
+/* Constructor for packed IRQ information. */
+static inline u32 mk_irq_info(u32 type, u32 index, u32 evtchn)
+{
+	return ((type << 24) | (index << 16) | evtchn);
+}
+
+/*
+ * Accessors for packed IRQ information.
+ */
+static inline unsigned int evtchn_from_irq(int irq)
+{
+	return (u16)(irq_info[irq]);
+}
+
+static inline unsigned int index_from_irq(int irq)
+{
+	return (u8)(irq_info[irq] >> 16);
+}
+
+static inline unsigned int type_from_irq(int irq)
+{
+	return (u8)(irq_info[irq] >> 24);
+}
+
+static inline unsigned long active_evtchns(unsigned int cpu,
+					   struct shared_info *sh,
+					   unsigned int idx)
+{
+	return (sh->evtchn_pending[idx] &
+		cpu_evtchn_mask[cpu][idx] &
+		~sh->evtchn_mask[idx]);
+}
+
+static void bind_evtchn_to_cpu(unsigned int chn, unsigned int cpu)
+{
+	int irq = evtchn_to_irq[chn];
+
+	BUG_ON(irq == -1);
+	set_native_irq_info(irq, cpumask_of_cpu(cpu));
+
+	__clear_bit(chn, (unsigned long *)cpu_evtchn_mask[cpu_evtchn[chn]]);
+	__set_bit(chn, (unsigned long *)cpu_evtchn_mask[cpu]);
+
+	cpu_evtchn[chn] = cpu;
+}
+
+static void init_evtchn_cpu_bindings(void)
+{
+	int i;
+
+	/* By default all event channels notify CPU#0. */
+	for (i = 0; i < NR_IRQS; i++)
+		set_native_irq_info(i, cpumask_of_cpu(0));
+
+	memset(cpu_evtchn, 0, sizeof(cpu_evtchn));
+	memset(cpu_evtchn_mask[0], ~0, sizeof(cpu_evtchn_mask[0]));
+}
+
+static inline unsigned int cpu_from_evtchn(unsigned int evtchn)
+{
+	return cpu_evtchn[evtchn];
+}
+
+static inline void clear_evtchn(int port)
+{
+	struct shared_info *s = HYPERVISOR_shared_info;
+	sync_clear_bit(port, &s->evtchn_pending[0]);
+}
+
+static inline void set_evtchn(int port)
+{
+	struct shared_info *s = HYPERVISOR_shared_info;
+	sync_set_bit(port, &s->evtchn_pending[0]);
+}
+
+
+/**
+ * notify_remote_via_irq - send event to remote end of event channel via irq
+ * @irq: irq of event channel to send event to
+ *
+ * Unlike notify_remote_via_evtchn(), this is safe to use across
+ * save/restore. Notifications on a broken connection are silently
+ * dropped.
+ */
+void notify_remote_via_irq(int irq)
+{
+	int evtchn = evtchn_from_irq(irq);
+
+	if (VALID_EVTCHN(evtchn))
+		notify_remote_via_evtchn(evtchn);
+}
+EXPORT_SYMBOL_GPL(notify_remote_via_irq);
+
+void mask_evtchn(int port)
+{
+	struct shared_info *s = HYPERVISOR_shared_info;
+	sync_set_bit(port, &s->evtchn_mask[0]);
+}
+EXPORT_SYMBOL_GPL(mask_evtchn);
+
+void unmask_evtchn(int port)
+{
+	struct shared_info *s = HYPERVISOR_shared_info;
+	unsigned int cpu = smp_processor_id();
+	struct vcpu_info *vcpu_info = read_pda(xen.vcpu);
+
+	BUG_ON(!irqs_disabled());
+
+	/* Slow path (hypercall) if this is a non-local port. */
+	if (unlikely(cpu != cpu_from_evtchn(port))) {
+		struct evtchn_unmask unmask = { .port = port };
+		(void)HYPERVISOR_event_channel_op(EVTCHNOP_unmask, &unmask);
+		return;
+	}
+
+	sync_clear_bit(port, &s->evtchn_mask[0]);
+
+	/*
+	 * The following is basically the equivalent of 'hw_resend_irq'. Just
+	 * like a real IO-APIC we 'lose the interrupt edge' if the channel is
+	 * masked.
+	 */
+	if (sync_test_bit(port, &s->evtchn_pending[0]) &&
+	    !sync_test_and_set_bit(port / BITS_PER_LONG,
+				   &vcpu_info->evtchn_pending_sel))
+		vcpu_info->evtchn_upcall_pending = 1;
+}
+EXPORT_SYMBOL_GPL(unmask_evtchn);
+
+static int find_unbound_irq(void)
+{
+	int irq;
+
+	/* Only allocate from dynirq range */
+	for (irq = 0; irq < NR_IRQS; irq++)
+		if (irq_bindcount[irq] == 0)
+			break;
+
+	if (irq == NR_IRQS)
+		panic("No available IRQ to bind to: increase NR_IRQS!\n");
+
+	return irq;
+}
+
+static int bind_evtchn_to_irq(unsigned int evtchn)
+{
+	int irq;
+
+	spin_lock(&irq_mapping_update_lock);
+
+	irq = evtchn_to_irq[evtchn];
+
+	if (irq == -1) {
+		irq = find_unbound_irq();
+
+		dynamic_irq_init(irq);
+		set_irq_chip_and_handler(irq, &xen_dynamic_chip, handle_level_irq);
+
+		evtchn_to_irq[evtchn] = irq;
+		irq_info[irq] = mk_irq_info(IRQT_EVTCHN, 0, evtchn);
+	}
+
+	irq_bindcount[irq]++;
+
+	spin_unlock(&irq_mapping_update_lock);
+
+	return irq;
+}
+
+static int bind_virq_to_irq(unsigned int virq, unsigned int cpu)
+{
+	struct evtchn_bind_virq bind_virq;
+	int evtchn, irq;
+
+	spin_lock(&irq_mapping_update_lock);
+
+	irq = per_cpu(virq_to_irq, cpu)[virq];
+
+	if (irq == -1) {
+		bind_virq.virq = virq;
+		bind_virq.vcpu = cpu;
+		if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
+						&bind_virq) != 0)
+			BUG();
+		evtchn = bind_virq.port;
+
+		irq = find_unbound_irq();
+
+		dynamic_irq_init(irq);
+		set_irq_chip_and_handler(irq, &xen_dynamic_chip, handle_level_irq);
+
+		evtchn_to_irq[evtchn] = irq;
+		irq_info[irq] = mk_irq_info(IRQT_VIRQ, virq, evtchn);
+
+		per_cpu(virq_to_irq, cpu)[virq] = irq;
+
+		bind_evtchn_to_cpu(evtchn, cpu);
+	}
+
+	irq_bindcount[irq]++;
+
+	spin_unlock(&irq_mapping_update_lock);
+
+	return irq;
+}
+
+static void unbind_from_irq(unsigned int irq)
+{
+	struct evtchn_close close;
+	int evtchn = evtchn_from_irq(irq);
+
+	spin_lock(&irq_mapping_update_lock);
+
+	if (VALID_EVTCHN(evtchn) && (--irq_bindcount[irq] == 0)) {
+		close.port = evtchn;
+		if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
+			BUG();
+
+		switch (type_from_irq(irq)) {
+		case IRQT_VIRQ:
+			per_cpu(virq_to_irq, cpu_from_evtchn(evtchn))
+				[index_from_irq(irq)] = -1;
+			break;
+		default:
+			break;
+		}
+
+		/* Closed ports are implicitly re-bound to VCPU0. */
+		bind_evtchn_to_cpu(evtchn, 0);
+
+		evtchn_to_irq[evtchn] = -1;
+		irq_info[irq] = IRQ_UNBOUND;
+
+		dynamic_irq_init(irq);
+	}
+
+	spin_unlock(&irq_mapping_update_lock);
+}
+
+int bind_evtchn_to_irqhandler(unsigned int evtchn,
+			      irqreturn_t (*handler)(int, void *),
+			      unsigned long irqflags, const char *devname, void *dev_id)
+{
+	unsigned int irq;
+	int retval;
+
+	irq = bind_evtchn_to_irq(evtchn);
+	retval = request_irq(irq, handler, irqflags, devname, dev_id);
+	if (retval != 0) {
+		unbind_from_irq(irq);
+		return retval;
+	}
+
+	return irq;
+}
+EXPORT_SYMBOL_GPL(bind_evtchn_to_irqhandler);
+
+int bind_virq_to_irqhandler(unsigned int virq, unsigned int cpu,
+			    irqreturn_t (*handler)(int, void *),
+			    unsigned long irqflags, const char *devname, void *dev_id)
+{
+	unsigned int irq;
+	int retval;
+
+	irq = bind_virq_to_irq(virq, cpu);
+	retval = request_irq(irq, handler, irqflags, devname, dev_id);
+	if (retval != 0) {
+		unbind_from_irq(irq);
+		return retval;
+	}
+
+	return irq;
+}
+EXPORT_SYMBOL_GPL(bind_virq_to_irqhandler);
+
+void unbind_from_irqhandler(unsigned int irq, void *dev_id)
+{
+	free_irq(irq, dev_id);
+	unbind_from_irq(irq);
+}
+EXPORT_SYMBOL_GPL(unbind_from_irqhandler);
+
+/*
+  Search the CPUs pending events bitmasks.  For each one found, map
+  the event number to an irq, and feed it into do_IRQ() for
+  handling.
+
+  Xen uses a two-level bitmap to speed searching.  The first level is
+  a bitset of words which contain pending event bits.  The second
+  level is a bitset of pending events themselves.
+*/
+asmlinkage fastcall void xen_evtchn_do_upcall(struct pt_regs *regs)
+{
+	int cpu = smp_processor_id();
+	struct shared_info *s = HYPERVISOR_shared_info;
+	struct vcpu_info *vcpu_info = read_pda(xen.vcpu);
+	unsigned long pending_words;
+
+	vcpu_info->evtchn_upcall_pending = 0;
+
+	/* NB. No need for a barrier here -- XCHG is a barrier on x86. */
+	pending_words = xchg(&vcpu_info->evtchn_pending_sel, 0);
+	while (pending_words != 0) {
+		unsigned long pending_bits;
+		int word_idx = __ffs(pending_words);
+		pending_words &= ~(1UL << word_idx);
+
+		while ((pending_bits = active_evtchns(cpu, s, word_idx)) != 0) {
+			int bit_idx = __ffs(pending_bits);
+			int port = (word_idx * BITS_PER_LONG) + bit_idx;
+			int irq = evtchn_to_irq[port];
+
+			if (irq != -1) {
+				regs->orig_eax = ~irq;
+				do_IRQ(regs);
+			}
+		}
+	}
+}
+
+/* Rebind an evtchn so that it gets delivered to a specific cpu */
+static void rebind_irq_to_cpu(unsigned irq, unsigned tcpu)
+{
+	struct evtchn_bind_vcpu bind_vcpu;
+	int evtchn = evtchn_from_irq(irq);
+
+	if (!VALID_EVTCHN(evtchn))
+		return;
+
+	/* Send future instances of this interrupt to other vcpu. */
+	bind_vcpu.port = evtchn;
+	bind_vcpu.vcpu = tcpu;
+
+	/*
+	 * If this fails, it usually just indicates that we're dealing with a 
+	 * virq or IPI channel, which don't actually need to be rebound. Ignore
+	 * it, but don't do the xenlinux-level rebind in that case.
+	 */
+	if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_vcpu, &bind_vcpu) >= 0)
+		bind_evtchn_to_cpu(evtchn, tcpu);
+}
+
+
+static void set_affinity_irq(unsigned irq, cpumask_t dest)
+{
+	unsigned tcpu = first_cpu(dest);
+	rebind_irq_to_cpu(irq, tcpu);
+}
+
+static void enable_dynirq(unsigned int irq)
+{
+	int evtchn = evtchn_from_irq(irq);
+
+	if (VALID_EVTCHN(evtchn))
+		unmask_evtchn(evtchn);
+}
+
+static void disable_dynirq(unsigned int irq)
+{
+	int evtchn = evtchn_from_irq(irq);
+
+	if (VALID_EVTCHN(evtchn))
+		mask_evtchn(evtchn);
+}
+
+static void ack_dynirq(unsigned int irq)
+{
+	int evtchn = evtchn_from_irq(irq);
+
+	move_native_irq(irq);
+
+	if (VALID_EVTCHN(evtchn))
+		clear_evtchn(evtchn);
+}
+
+static int retrigger_dynirq(unsigned int irq)
+{
+	int evtchn = evtchn_from_irq(irq);
+	int ret = 0;
+
+	if (VALID_EVTCHN(evtchn)) {
+		set_evtchn(evtchn);
+		ret = 1;
+	}
+
+	return ret;
+}
+
+static struct irq_chip xen_dynamic_chip __read_mostly = {
+	.name		= "xen-virq",
+	.mask		= disable_dynirq,
+	.unmask		= enable_dynirq,
+	.ack		= ack_dynirq,
+	.set_affinity	= set_affinity_irq,
+	.retrigger	= retrigger_dynirq,
+};
+
+void __init xen_init_IRQ(void)
+{
+	int i;
+
+	init_evtchn_cpu_bindings();
+
+	/* No event channels are 'live' right now. */
+	for (i = 0; i < NR_EVENT_CHANNELS; i++)
+		mask_evtchn(i);
+
+	/* Dynamic IRQ space is currently unbound. Zero the refcnts. */
+	for (i = 0; i < NR_IRQS; i++)
+		irq_bindcount[i] = 0;
+
+	irq_ctx_init(smp_processor_id());
+}
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/events.h
@@ -0,0 +1,28 @@
+#ifndef _XEN_EVENTS_H
+#define _XEN_EVENTS_H
+
+#include <linux/irq.h>
+
+int bind_evtchn_to_irqhandler(unsigned int evtchn,
+			      irqreturn_t (*handler)(int, void *),
+			      unsigned long irqflags, const char *devname,
+			      void *dev_id);
+int bind_virq_to_irqhandler(unsigned int virq, unsigned int cpu,
+			    irqreturn_t (*handler)(int, void *),
+			    unsigned long irqflags, const char *devname, void *dev_id);
+
+/*
+ * Common unbind function for all event sources. Takes IRQ to unbind from.
+ * Automatically closes the underlying event channel (even for bindings
+ * made with bind_evtchn_to_irqhandler()).
+ */
+void unbind_from_irqhandler(unsigned int irq, void *dev_id);
+
+static inline void notify_remote_via_evtchn(int port)
+{
+	struct evtchn_send send = { .port = port };
+	(void)HYPERVISOR_event_channel_op(EVTCHNOP_send, &send);
+}
+
+extern void notify_remote_via_irq(int irq);
+#endif	/* _XEN_EVENTS_H */
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/features.c
@@ -0,0 +1,29 @@
+/******************************************************************************
+ * features.c
+ *
+ * Xen feature flags.
+ *
+ * Copyright (c) 2006, Ian Campbell, XenSource Inc.
+ */
+#include <linux/types.h>
+#include <linux/cache.h>
+#include <linux/module.h>
+#include <asm/hypervisor.h>
+#include <xen/features.h>
+
+u8 xen_features[XENFEAT_NR_SUBMAPS * 32] __read_mostly;
+EXPORT_SYMBOL_GPL(xen_features);
+
+void xen_setup_features(void)
+{
+	struct xen_feature_info fi;
+	int i, j;
+
+	for (i = 0; i < XENFEAT_NR_SUBMAPS; i++) {
+		fi.submap_idx = i;
+		if (HYPERVISOR_xen_version(XENVER_get_features, &fi) < 0)
+			break;
+		for (j=0; j<32; j++)
+			xen_features[i*32+j] = !!(fi.submap & 1<<j);
+	}
+}
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/memory.c
@@ -0,0 +1,50 @@
+#include <asm/desc.h>
+#include <asm/hypervisor.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+
+#include "xen-page.h"
+
+xmaddr_t arbitrary_virt_to_machine(unsigned long address)
+{
+	pte_t *pte = lookup_address(address);
+	unsigned offset = address & PAGE_MASK;
+
+	BUG_ON(pte == NULL);
+
+	return XMADDR((pte_mfn(*pte) << PAGE_SHIFT) + offset);
+}
+
+void make_lowmem_page_readonly(void *vaddr)
+{
+	pte_t *pte, ptev;
+	unsigned long address = (unsigned long)vaddr;
+
+	pte = lookup_address(address);
+	BUG_ON(pte == NULL);
+
+	ptev = pte_wrprotect(*pte);
+
+	if (xen_feature(XENFEAT_writable_page_tables))
+		*pte = ptev;
+	else
+		if(HYPERVISOR_update_va_mapping(address, ptev, 0))
+			BUG();
+}
+
+void make_lowmem_page_readwrite(void *vaddr)
+{
+	pte_t *pte, ptev;
+	unsigned long address = (unsigned long)vaddr;
+
+	pte = lookup_address(address);
+	BUG_ON(pte == NULL);
+
+	ptev = pte_mkwrite(*pte);
+
+	if (xen_feature(XENFEAT_writable_page_tables))
+		*pte = ptev;
+	else
+		if(HYPERVISOR_update_va_mapping(address, ptev, 0))
+			BUG();
+}
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/mmu.c
@@ -0,0 +1,375 @@
+//#include <linux/bug.h>
+#include <asm/bug.h>
+
+#include <asm/pgtable.h>
+#include <asm/tlbflush.h>
+#include <asm/mmu_context.h>
+
+#include <asm/hypercall.h>
+#include <asm/paravirt.h>
+
+#include <xen/interface/xen.h>
+
+#include "xen-page.h"
+
+fastcall void xen_set_pte(pte_t *ptep, pte_t pte)
+{
+#if 1
+	struct mmu_update u;
+
+	u.ptr = virt_to_machine(ptep).maddr;
+	u.val = pte_val_ma(pte);
+	if (HYPERVISOR_mmu_update(&u, 1, NULL, DOMID_SELF) < 0)
+		BUG();
+#else
+	ptep->pte_high = pte.pte_high;
+	smp_wmb();
+	ptep->pte_low = pte.pte_low;
+#endif
+}
+
+fastcall void xen_set_pmd(pmd_t *ptr, pmd_t val)
+{
+	struct mmu_update u;
+
+	u.ptr = virt_to_machine(ptr).maddr;
+	u.val = pmd_val_ma(val);
+	if (HYPERVISOR_mmu_update(&u, 1, NULL, DOMID_SELF) < 0)
+		BUG();
+}
+
+#ifdef CONFIG_X86_PAE
+fastcall void xen_set_pud(pmd_t *ptr, pud_t val)
+{
+	struct mmu_update u;
+
+	u.ptr = virt_to_machine(ptr).maddr;
+	u.val = pud_val_ma(val);
+	if (HYPERVISOR_mmu_update(&u, 1, NULL, DOMID_SELF) < 0)
+		BUG();
+}
+#endif
+
+/*
+ * Associate a virtual page frame with a given physical page frame 
+ * and protection flags for that frame.
+ */ 
+void set_pte_mfn(unsigned long vaddr, unsigned long mfn, pgprot_t flags)
+{
+	pgd_t *pgd;
+	pud_t *pud;
+	pmd_t *pmd;
+	pte_t *pte;
+
+	pgd = swapper_pg_dir + pgd_index(vaddr);
+	if (pgd_none(*pgd)) {
+		BUG();
+		return;
+	}
+	pud = pud_offset(pgd, vaddr);
+	if (pud_none(*pud)) {
+		BUG();
+		return;
+	}
+	pmd = pmd_offset(pud, vaddr);
+	if (pmd_none(*pmd)) {
+		BUG();
+		return;
+	}
+	pte = pte_offset_kernel(pmd, vaddr);
+	/* <mfn,flags> stored as-is, to permit clearing entries */
+	xen_set_pte(pte, mfn_pte(mfn, flags));
+
+	/*
+	 * It's enough to flush this one mapping.
+	 * (PGE mappings get flushed as well)
+	 */
+	__flush_tlb_one(vaddr);
+}
+
+void fastcall xen_set_pte_at(struct mm_struct *mm, u32 addr,
+			     pte_t *ptep, pte_t pteval)
+{
+	if ((mm != current->mm && mm != &init_mm) ||
+	    HYPERVISOR_update_va_mapping(addr, pteval, 0) != 0)
+		xen_set_pte(ptep, pteval);
+}
+
+void fastcall xen_pte_update(struct mm_struct *mm, u32 addr, pte_t *ptep)
+{
+}
+
+void fastcall xen_pte_update_defer(struct mm_struct *mm, u32 addr, pte_t *ptep)
+{
+}
+
+#ifdef CONFIG_X86_PAE
+void fastcall xen_set_pte_atomic(pte_t *ptep, pte_t pte)
+{
+	set_64bit((u64 *)ptep, pte_val_ma(pte));
+}
+
+void fastcall xen_pte_clear(struct mm_struct *mm, u32 addr,pte_t *ptep)
+{
+#if 1
+	ptep->pte_low = 0;
+	smp_wmb();
+	ptep->pte_high = 0;	
+#else
+	set_64bit((u64 *)ptep, 0);
+#endif
+}
+
+void fastcall xen_pmd_clear(pmd_t *pmdp)
+{
+	xen_set_pmd(pmdp, __pmd(0));
+}
+
+fastcall unsigned long long xen_pte_val(pte_t pte)
+{
+	unsigned long long ret = 0;
+
+	if (pte.pte_low) {
+		ret = ((unsigned long long)pte.pte_high << 32) | pte.pte_low;
+		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
+	}
+
+	return ret;
+}
+
+fastcall unsigned long long xen_pmd_val(pmd_t pmd)
+{
+	unsigned long long ret = pmd.pmd;
+	if (ret)
+		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
+	return ret;
+}
+
+fastcall unsigned long long xen_pgd_val(pgd_t pgd)
+{
+	unsigned long long ret = pgd.pgd;
+	if (ret)
+		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
+	return ret;
+}
+
+fastcall pte_t xen_make_pte(unsigned long long pte)
+{
+	if (pte & 1)
+		pte = phys_to_machine(XPADDR(pte)).maddr;
+
+	return (pte_t){ pte, pte >> 32 };
+}
+
+fastcall pmd_t xen_make_pmd(unsigned long long pmd)
+{
+	if (pmd & 1)
+		pmd = phys_to_machine(XPADDR(pmd)).maddr;
+
+	return (pmd_t){ pmd };
+}
+
+fastcall pgd_t xen_make_pgd(unsigned long long pgd)
+{
+	if (pgd & _PAGE_PRESENT)
+		pgd = phys_to_machine(XPADDR(pgd)).maddr;
+
+	return (pgd_t){ pgd };
+}
+
+fastcall pte_t xen_ptep_get_and_clear(pte_t *ptep)
+{
+	pte_t res;
+
+	/* xchg acts as a barrier before the setting of the high bits */
+	res.pte_low = xchg(&ptep->pte_low, 0);
+	res.pte_high = ptep->pte_high;
+	ptep->pte_high = 0;
+
+	return res;
+}
+#else  /* !PAE */
+fastcall unsigned long xen_pte_val(pte_t pte)
+{
+	unsigned long ret = pte.pte_low;
+
+	if (ret & _PAGE_PRESENT)
+		ret = machine_to_phys(XMADDR(ret)).paddr;
+
+	return ret;
+}
+
+fastcall unsigned long xen_pmd_val(pmd_t pmd)
+{
+	BUG();
+	return 0;
+}
+
+fastcall unsigned long xen_pgd_val(pgd_t pgd)
+{
+	unsigned long ret = pgd.pgd;
+	if (ret)
+		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
+	return ret;
+}
+
+fastcall pte_t xen_make_pte(unsigned long pte)
+{
+	if (pte & _PAGE_PRESENT)
+		pte = phys_to_machine(XPADDR(pte)).maddr;
+
+	return (pte_t){ pte };
+}
+
+fastcall pmd_t xen_make_pmd(unsigned long pmd)
+{
+	BUG();
+	return __pmd(0);
+}
+
+fastcall pgd_t xen_make_pgd(unsigned long pgd)
+{
+	if (pgd & _PAGE_PRESENT)
+		pgd = phys_to_machine(XPADDR(pgd)).maddr;
+
+	return (pgd_t){ pgd };
+}
+
+fastcall pte_t xen_ptep_get_and_clear(pte_t *ptep)
+{
+	return __pte_ma(xchg(&(ptep)->pte_low, 0));
+}
+#endif	/* CONFIG_X86_PAE */
+
+
+
+static void pgd_walk_set_prot(void *pt, pgprot_t flags)
+{
+	unsigned long pfn = PFN_DOWN(__pa(pt));
+
+	if (HYPERVISOR_update_va_mapping((unsigned long)pt,
+					 pfn_pte(pfn, flags), 0) < 0)
+		BUG();
+}
+
+static void pgd_walk(pgd_t *pgd_base, pgprot_t flags)
+{
+	pgd_t *pgd = pgd_base;
+	pud_t *pud;
+	pmd_t *pmd;
+	pte_t *pte;
+	int    g, u, m;
+
+	if (xen_feature(XENFEAT_auto_translated_physmap))
+		return;
+
+	for (g = 0; g < USER_PTRS_PER_PGD; g++, pgd++) {
+		if (pgd_none(*pgd))
+			continue;
+		pud = pud_offset(pgd, 0);
+
+		if (PTRS_PER_PUD > 1) /* not folded */
+			pgd_walk_set_prot(pud,flags);
+
+		for (u = 0; u < PTRS_PER_PUD; u++, pud++) {
+			if (pud_none(*pud))
+				continue;
+			pmd = pmd_offset(pud, 0);
+
+			if (PTRS_PER_PMD > 1) /* not folded */
+				pgd_walk_set_prot(pmd,flags);
+
+			for (m = 0; m < PTRS_PER_PMD; m++, pmd++) {
+				if (pmd_none(*pmd))
+					continue;
+
+				/* This can get called before mem_map
+				   is set up, so we assume nothing is
+				   highmem at that point. */
+				if (mem_map == NULL ||
+				    !PageHighMem(pmd_page(*pmd))) {
+					pte = pte_offset_kernel(pmd,0);
+					pgd_walk_set_prot(pte,flags);
+				}
+			}
+		}
+	}
+
+	if (HYPERVISOR_update_va_mapping((unsigned long)pgd_base,
+					 pfn_pte(PFN_DOWN(__pa(pgd_base)),
+						 flags),
+					 UVMF_TLB_FLUSH) < 0)
+		BUG();
+}
+
+
+/* This is called just after a mm has been duplicated from its parent,
+   but it has not been used yet.  We need to make sure that its
+   pagetable is all read-only, and can be pinned. The pagetable itself
+   needs to map itself as RO; it doesn't matter what the state its in
+   with respect to any other pagetable. */
+void xen_pgd_pin(pgd_t *pgd)
+{
+	struct mmuext_op op;
+
+	pgd_walk(pgd, PAGE_KERNEL_RO);
+
+#if defined(CONFIG_X86_PAE)
+	op.cmd = MMUEXT_PIN_L3_TABLE;
+#else
+	op.cmd = MMUEXT_PIN_L2_TABLE;
+#endif
+	op.arg1.mfn = pfn_to_mfn(PFN_DOWN(__pa(pgd)));
+	if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF) < 0)
+		BUG();
+}
+
+/* Release a pagetables pages back as normal RW */
+void xen_pgd_unpin(pgd_t *pgd)
+{
+	struct mmuext_op op;
+
+	op.cmd = MMUEXT_UNPIN_TABLE;
+	op.arg1.mfn = pfn_to_mfn(PFN_DOWN(__pa(pgd)));
+
+	if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF) < 0)
+		BUG();
+
+	pgd_walk(pgd, PAGE_KERNEL);
+}
+
+
+fastcall void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next)
+{
+	xen_pgd_pin(next->pgd);
+}
+
+fastcall void xen_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
+{
+	xen_pgd_pin(mm->pgd);	
+}
+
+fastcall void xen_exit_mmap(struct mm_struct *mm)
+{
+	struct task_struct *tsk = current;
+
+	task_lock(tsk);
+
+	/*
+	 * We aggressively remove defunct pgd from cr3. We execute unmap_vmas()
+	 * *much* faster this way, as no tlb flushes means bigger wrpt batches.
+	 */
+	if (tsk->active_mm == mm) {
+		tsk->active_mm = &init_mm;
+		atomic_inc(&init_mm.mm_count);
+
+		switch_mm(mm, &init_mm, tsk);
+
+		atomic_dec(&mm->mm_count);
+		BUG_ON(atomic_read(&mm->mm_count) == 0);
+	}
+
+	task_unlock(tsk);
+
+	xen_pgd_unpin(mm->pgd);
+}
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/mmu.h
@@ -0,0 +1,51 @@
+#ifndef _XEN_MMU_H
+
+#include <linux/linkage.h>
+#include <asm/page.h>
+
+void set_pte_mfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags);
+
+void fastcall xen_set_pte(pte_t *ptep, pte_t pteval);
+void fastcall xen_set_pte_at(struct mm_struct *mm, u32 addr,
+			     pte_t *ptep, pte_t pteval);
+void fastcall xen_set_pmd(pmd_t *pmdp, pmd_t pmdval);
+void fastcall xen_pte_update(struct mm_struct *mm, u32 addr, pte_t *ptep);
+void fastcall xen_pte_update_defer(struct mm_struct *mm, u32 addr, pte_t *ptep);
+
+fastcall void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next);
+fastcall void xen_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm);
+fastcall void xen_exit_mmap(struct mm_struct *mm);
+
+fastcall pte_t xen_ptep_get_and_clear(pte_t *ptep);
+
+void xen_pgd_pin(pgd_t *pgd);
+void xen_pgd_unpin(pgd_t *pgd);
+
+#ifdef CONFIG_X86_PAE
+fastcall unsigned long long xen_pte_val(pte_t);
+fastcall unsigned long long xen_pmd_val(pmd_t);
+fastcall unsigned long long xen_pgd_val(pgd_t);
+
+fastcall pte_t xen_make_pte(unsigned long long);
+fastcall pmd_t xen_make_pmd(unsigned long long);
+fastcall pgd_t xen_make_pgd(unsigned long long);
+
+fastcall void xen_set_pte_at(struct mm_struct *mm, u32 addr,
+			     pte_t *ptep, pte_t pteval);
+fastcall void xen_set_pte_atomic(pte_t *ptep, pte_t pte);
+fastcall void xen_set_pud(pud_t *ptr, pud_t val);
+fastcall void xen_pte_clear(struct mm_struct *mm, u32 addr,pte_t *ptep);
+fastcall void xen_pmd_clear(pmd_t *pmdp);
+
+
+#else
+fastcall unsigned long xen_pte_val(pte_t);
+fastcall unsigned long xen_pmd_val(pmd_t);
+fastcall unsigned long xen_pgd_val(pgd_t);
+
+fastcall pte_t xen_make_pte(unsigned long);
+fastcall pmd_t xen_make_pmd(unsigned long);
+fastcall pgd_t xen_make_pgd(unsigned long);
+#endif
+
+#endif	/* _XEN_MMU_H */
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/multicalls.c
@@ -0,0 +1,62 @@
+#include <linux/percpu.h>
+
+#include <asm/hypercall.h>
+
+#include "multicalls.h"
+
+#define MC_BATCH	8
+#define MC_ARGS		(MC_BATCH * 32 / sizeof(u64))
+
+struct mc_buffer {
+	struct multicall_entry entries[MC_BATCH];
+	u64 args[MC_ARGS];
+	unsigned mcidx, argidx;
+};
+
+static DEFINE_PER_CPU(struct mc_buffer, mc_buffer);
+
+int xen_mc_flush(void)
+{
+	struct mc_buffer *b = &get_cpu_var(mc_buffer);
+	int ret = 0;
+
+	if (b->mcidx) {
+		int i;
+
+		if (HYPERVISOR_multicall(b->entries, b->mcidx) != 0)
+			BUG();
+		for(i = 0; i < b->mcidx; i++)
+			if (b->entries[i].result < 0)
+				ret++;
+		b->mcidx = 0;
+		b->argidx = 0;
+	} else
+		BUG_ON(b->argidx != 0);
+
+	put_cpu_var(mc_buffer);
+
+	return ret;
+}
+
+struct multicall_space xen_mc_entry(size_t args)
+{
+	struct mc_buffer *b = &get_cpu_var(mc_buffer);
+	struct multicall_space ret;
+	unsigned argspace = (args + sizeof(u64) - 1) / sizeof(u64);
+
+	BUG_ON(argspace > MC_ARGS);
+
+	if (b->mcidx == MC_BATCH ||
+	    (b->argidx + argspace) > MC_ARGS)
+		if (xen_mc_flush())
+			BUG();
+
+	ret.mc = &b->entries[b->mcidx];
+	b->mcidx++;
+	ret.args = &b->args[b->argidx];
+	b->argidx += argspace;
+
+	put_cpu_var(mc_buffer);
+
+	return ret;
+}
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/multicalls.h
@@ -0,0 +1,13 @@
+#ifndef _XEN_MULTICALLS_H
+#define _XEN_MULTICALLS_H
+
+struct multicall_space
+{
+	struct multicall_entry *mc;
+	void *args;
+};
+
+struct multicall_space xen_mc_entry(size_t args);
+int xen_mc_flush(void);
+
+#endif /* _XEN_MULTICALLS_H */
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/setup.c
@@ -0,0 +1,96 @@
+/*
+ *	Machine specific setup for xen
+ */
+
+#include <linux/module.h>
+#include <linux/mm.h>
+
+#include <asm/e820.h>
+#include <asm/setup.h>
+#include <asm/hypervisor.h>
+#include <asm/hypercall.h>
+#include <asm/pda.h>
+
+#include <xen/interface/physdev.h>
+#include <xen/features.h>
+
+/* These are code, but not functions.  Defined in entry.S */
+extern const char xen_hypervisor_callback[];
+extern const char xen_failsafe_callback[];
+
+static __initdata struct shared_info init_shared;
+
+/*
+ * 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 = &init_shared;
+EXPORT_SYMBOL(HYPERVISOR_shared_info);
+
+unsigned long *phys_to_machine_mapping;
+unsigned long *pfn_to_mfn_frame_list_list, *pfn_to_mfn_frame_list[16];
+EXPORT_SYMBOL(phys_to_machine_mapping);
+
+/**
+ * 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;
+
+	e820.nr_map = 0;
+	add_memory_region(0, PFN_PHYS(max_pfn), E820_RAM);
+
+	return "Xen";
+}
+
+void xen_idle(void)
+{
+	local_irq_disable();
+
+	if (need_resched())
+		local_irq_enable();
+	else {
+		current_thread_info()->status &= ~TS_POLLING;
+		smp_mb__after_clear_bit();
+		safe_halt();
+		current_thread_info()->status |= TS_POLLING;
+	}
+}
+
+void __init xen_arch_setup(void)
+{
+	struct physdevop_set_iopl set_iopl;
+	int rc;
+
+	HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_4gb_segments);
+	HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_writable_pagetables);
+
+	if (!xen_feature(XENFEAT_auto_translated_physmap))
+		HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_pae_extended_cr3);
+
+	HYPERVISOR_set_callbacks(__KERNEL_CS, (unsigned long)xen_hypervisor_callback,
+				 __KERNEL_CS, (unsigned long)xen_failsafe_callback);
+
+	set_iopl.iopl = 1;
+	rc = HYPERVISOR_physdev_op(PHYSDEVOP_SET_IOPL, &set_iopl);
+	if (rc != 0)
+		printk(KERN_INFO "physdev_op failed %d\n", rc);
+
+#ifdef CONFIG_ACPI
+	if (!(xen_start_info->flags & SIF_INITDOMAIN)) {
+		printk(KERN_INFO "ACPI in unprivileged domain disabled\n");
+		acpi_disabled = 1;
+		acpi_ht = 0;
+	}
+#endif
+
+	memcpy(saved_command_line, xen_start_info->cmd_line,
+	       MAX_GUEST_CMDLINE > COMMAND_LINE_SIZE ?
+	       COMMAND_LINE_SIZE : MAX_GUEST_CMDLINE);
+
+	pm_idle = xen_idle;
+
+	vdso_enabled = 1;	/* enable by default */
+}
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/time.c
@@ -0,0 +1,446 @@
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/kernel_stat.h>
+#include <linux/clocksource.h>
+
+#include <asm/hypercall.h>
+#include <asm/arch_hooks.h>
+
+#include <xen/interface/xen.h>
+#include <xen/interface/vcpu.h>
+
+#include "xen-ops.h"
+#include "events.h"
+
+#define XEN_SHIFT 22
+
+/* Permitted clock jitter, in nsecs, beyond which a warning will be printed. */
+static unsigned long permitted_clock_jitter = 10000000UL; /* 10ms */
+static int __init __permitted_clock_jitter(char *str)
+{
+	permitted_clock_jitter = simple_strtoul(str, NULL, 0);
+	return 1;
+}
+__setup("permitted_clock_jitter=", __permitted_clock_jitter);
+
+
+/* These are perodically updated in shared_info, and then copied here. */
+struct shadow_time_info {
+	u64 tsc_timestamp;     /* TSC at last update of time vals.  */
+	u64 system_timestamp;  /* Time, in nanosecs, since boot.    */
+	u32 tsc_to_nsec_mul;
+	int tsc_shift;
+	u32 version;
+};
+
+static DEFINE_PER_CPU(struct shadow_time_info, shadow_time);
+
+/* Keep track of last time we did processing/updating of jiffies and xtime. */
+static u64 processed_system_time;   /* System time (ns) at last processing. */
+static DEFINE_PER_CPU(u64, processed_system_time);
+
+/* How much CPU time was spent blocked and how much was 'stolen'? */
+static DEFINE_PER_CPU(u64, processed_stolen_time);
+static DEFINE_PER_CPU(u64, processed_blocked_time);
+
+/* Current runstate of each CPU (updated automatically by the hypervisor). */
+static DEFINE_PER_CPU(struct vcpu_runstate_info, runstate);
+
+/* Must be signed, as it's compared with s64 quantities which can be -ve. */
+#define NS_PER_TICK (1000000000LL/HZ)
+
+/*
+ * Reads a consistent set of time-base values from Xen, into a shadow data
+ * area.
+ */
+static void get_time_values_from_xen(void)
+{
+	struct vcpu_time_info   *src;
+	struct shadow_time_info *dst;
+
+	src = &read_pda(xen.vcpu)->time;
+	dst = &get_cpu_var(shadow_time);
+
+	do {
+		dst->version = src->version;
+		rmb();
+		dst->tsc_timestamp     = src->tsc_timestamp;
+		dst->system_timestamp  = src->system_time;
+		dst->tsc_to_nsec_mul   = src->tsc_to_system_mul;
+		dst->tsc_shift         = src->tsc_shift;
+		rmb();
+	} while ((src->version & 1) | (dst->version ^ src->version));
+
+	put_cpu_var(shadow_time);
+}
+
+static inline int time_values_up_to_date(void)
+{
+	struct vcpu_time_info   *src;
+	unsigned dstversion;
+
+	src = &read_pda(xen.vcpu)->time;
+	dstversion = get_cpu_var(shadow_time).version;
+	put_cpu_var(shadow_time);
+
+	rmb();
+	return (dstversion == src->version);
+}
+
+/*
+ * Scale a 64-bit delta by scaling and multiplying by a 32-bit fraction,
+ * yielding a 64-bit result.
+ */
+static inline u64 scale_delta(u64 delta, u32 mul_frac, int shift)
+{
+	u64 product;
+#ifdef __i386__
+	u32 tmp1, tmp2;
+#endif
+
+	if (shift < 0)
+		delta >>= -shift;
+	else
+		delta <<= shift;
+
+#ifdef __i386__
+	__asm__ (
+		"mul  %5       ; "
+		"mov  %4,%%eax ; "
+		"mov  %%edx,%4 ; "
+		"mul  %5       ; "
+		"xor  %5,%5    ; "
+		"add  %4,%%eax ; "
+		"adc  %5,%%edx ; "
+		: "=A" (product), "=r" (tmp1), "=r" (tmp2)
+		: "a" ((u32)delta), "1" ((u32)(delta >> 32)), "2" (mul_frac) );
+#elif __x86_64__
+	__asm__ (
+		"mul %%rdx ; shrd $32,%%rdx,%%rax"
+		: "=a" (product) : "0" (delta), "d" ((u64)mul_frac) );
+#else
+#error implement me!
+#endif
+
+	return product;
+}
+
+static u64 get_nsec_offset(struct shadow_time_info *shadow)
+{
+	u64 now, delta;
+	rdtscll(now);
+	delta = now - shadow->tsc_timestamp;
+	return scale_delta(delta, shadow->tsc_to_nsec_mul, shadow->tsc_shift);
+}
+
+
+static void xen_timer_interrupt_hook(void)
+{
+	s64 delta, delta_cpu, stolen, blocked;
+	u64 sched_time;
+	int i, cpu = smp_processor_id();
+	unsigned long ticks;
+	struct shadow_time_info *shadow = &__get_cpu_var(shadow_time);
+	struct vcpu_runstate_info *runstate = &__get_cpu_var(runstate);
+
+	do {
+		get_time_values_from_xen();
+
+		/* Obtain a consistent snapshot of elapsed wallclock cycles. */
+		delta = delta_cpu =
+			shadow->system_timestamp + get_nsec_offset(shadow);
+		if (0)
+			printk("tsc_timestamp=%llu system_timestamp=%llu tsc_to_nsec=%u tsc_shift=%d, version=%u, delta=%lld processed_system_time=%lld\n",
+			       shadow->tsc_timestamp, shadow->system_timestamp,
+			       shadow->tsc_to_nsec_mul, shadow->tsc_shift,
+			       shadow->version, delta, processed_system_time);
+
+		delta     -= processed_system_time;
+		delta_cpu -= __get_cpu_var(processed_system_time);
+
+		/*
+		 * Obtain a consistent snapshot of stolen/blocked cycles. We
+		 * can use state_entry_time to detect if we get preempted here.
+		 */
+		do {
+			sched_time = runstate->state_entry_time;
+			barrier();
+			stolen = runstate->time[RUNSTATE_runnable] +
+				runstate->time[RUNSTATE_offline] -
+				__get_cpu_var(processed_stolen_time);
+			blocked = runstate->time[RUNSTATE_blocked] -
+				__get_cpu_var(processed_blocked_time);
+			barrier();
+		} while (sched_time != runstate->state_entry_time);
+	} while (!time_values_up_to_date());
+
+	if ((unlikely(delta < -(s64)permitted_clock_jitter) ||
+	     unlikely(delta_cpu < -(s64)permitted_clock_jitter))
+	    && printk_ratelimit()) {
+		printk("Timer ISR/%d: Time went backwards: "
+		       "delta=%lld delta_cpu=%lld shadow=%lld "
+		       "off=%lld processed=%lld cpu_processed=%lld\n",
+		       cpu, delta, delta_cpu, shadow->system_timestamp,
+		       (s64)get_nsec_offset(shadow),
+		       processed_system_time,
+		       __get_cpu_var(processed_system_time));
+		for (i = 0; i < num_online_cpus(); i++)
+			printk(" %d: %lld\n", i,
+			       per_cpu(processed_system_time, i));
+	}
+
+	/* System-wide jiffy work. */
+	ticks = 0;
+	while(delta > NS_PER_TICK) {
+		delta -= NS_PER_TICK;
+		processed_system_time += NS_PER_TICK;
+		ticks++;
+	}
+	do_timer(ticks);
+
+	/*
+	 * Account stolen ticks.
+	 * HACK: Passing NULL to account_steal_time()
+	 * ensures that the ticks are accounted as stolen.
+	 */
+	if ((stolen > 0) && (delta_cpu > 0)) {
+		delta_cpu -= stolen;
+		if (unlikely(delta_cpu < 0))
+			stolen += delta_cpu; /* clamp local-time progress */
+		do_div(stolen, NS_PER_TICK);
+		__get_cpu_var(processed_stolen_time) += stolen * NS_PER_TICK;
+		__get_cpu_var(processed_system_time) += stolen * NS_PER_TICK;
+		account_steal_time(NULL, (cputime_t)stolen);
+	}
+
+	/*
+	 * Account blocked ticks.
+	 * HACK: Passing idle_task to account_steal_time()
+	 * ensures that the ticks are accounted as idle/wait.
+	 */
+	if ((blocked > 0) && (delta_cpu > 0)) {
+		delta_cpu -= blocked;
+		if (unlikely(delta_cpu < 0))
+			blocked += delta_cpu; /* clamp local-time progress */
+		do_div(blocked, NS_PER_TICK);
+		__get_cpu_var(processed_blocked_time) += blocked * NS_PER_TICK;
+		__get_cpu_var(processed_system_time)  += blocked * NS_PER_TICK;
+		account_steal_time(idle_task(cpu), (cputime_t)blocked);
+	}
+
+	update_process_times(user_mode_vm(get_irq_regs()));
+}
+
+static cycle_t xen_clocksource_read(void)
+{
+	struct shadow_time_info *shadow = &get_cpu_var(shadow_time);
+	cycle_t ret;
+
+	get_time_values_from_xen();
+
+	ret = shadow->system_timestamp + get_nsec_offset(shadow);
+
+	put_cpu_var(shadow_time);
+
+	return ret;
+}
+
+static void xen_read_wallclock(struct timespec *ts)
+{
+	const struct shared_info *s = HYPERVISOR_shared_info;
+	u32 version;
+	u64 delta;
+	struct timespec now;
+
+	/* get wallclock at system boot */
+	do {
+		version = s->wc_version;
+		rmb();
+		now.tv_sec  = s->wc_sec;
+		now.tv_nsec = s->wc_nsec;
+		rmb();
+	} while ((s->wc_version & 1) | (version ^ s->wc_version));
+
+	delta = xen_clocksource_read();	/* time since system boot */
+	delta += now.tv_sec * (u64)NSEC_PER_SEC + now.tv_nsec;
+
+	now.tv_nsec = do_div(delta, NSEC_PER_SEC);
+	now.tv_sec = delta;
+
+	set_normalized_timespec(ts, now.tv_sec, now.tv_nsec);
+}
+
+unsigned long xen_get_wallclock(void)
+{
+	struct timespec ts;
+
+	xen_read_wallclock(&ts);
+
+	return ts.tv_sec;
+}
+
+int xen_set_wallclock(unsigned long now)
+{
+	/* do nothing for domU */
+	return -1;
+}
+
+static void init_cpu_khz(void)
+{
+	u64 __cpu_khz = 1000000ULL << 32;
+	struct vcpu_time_info *info;
+	info = &HYPERVISOR_shared_info->vcpu_info[0].time;
+	do_div(__cpu_khz, info->tsc_to_system_mul);
+	if (info->tsc_shift < 0)
+		cpu_khz = __cpu_khz << -info->tsc_shift;
+	else
+		cpu_khz = __cpu_khz >> info->tsc_shift;
+}
+
+static struct clocksource xen_clocksource = {
+	.name = "xen",
+	.rating = 400,
+	.read = xen_clocksource_read,
+	.mask = ~0,
+	.mult = 1<<XEN_SHIFT,		/* time directly in nanoseconds */
+	.shift = XEN_SHIFT,
+	.is_continuous = 1
+};
+
+static void init_missing_ticks_accounting(int cpu)
+{
+	struct vcpu_register_runstate_memory_area area;
+	struct vcpu_runstate_info *runstate = &per_cpu(runstate, cpu);
+
+	memset(runstate, 0, sizeof(*runstate));
+
+	area.addr.v = runstate;
+	HYPERVISOR_vcpu_op(VCPUOP_register_runstate_memory_area, cpu, &area);
+
+	per_cpu(processed_blocked_time, cpu) =
+		runstate->time[RUNSTATE_blocked];
+	per_cpu(processed_stolen_time, cpu) =
+		runstate->time[RUNSTATE_runnable] +
+		runstate->time[RUNSTATE_offline];
+}
+
+static irqreturn_t xen_timer_interrupt(int irq, void *dev_id)
+{
+	/*
+	 * Here we are in the timer irq handler. We just have irqs locally
+	 * disabled but we don't know if the timer_bh is running on the other
+	 * CPU. We need to avoid to SMP race with it. NOTE: we don' t need
+	 * the irq version of write_lock because as just said we have irq
+	 * locally disabled. -arca
+	 */
+	write_seqlock(&xtime_lock);
+
+	xen_timer_interrupt_hook();
+
+	write_sequnlock(&xtime_lock);
+
+	return IRQ_HANDLED;
+}
+
+static void setup_cpu0_timer_irq(void)
+{
+	printk(KERN_DEBUG "installing Xen timer for CPU 0\n");
+
+	bind_virq_to_irqhandler(
+		VIRQ_TIMER,
+		0,
+		xen_timer_interrupt,
+		SA_INTERRUPT,
+		"timer0",
+		NULL);
+}
+
+__init void xen_time_init(void)
+{
+	get_time_values_from_xen();
+
+	processed_system_time = per_cpu(shadow_time, 0).system_timestamp;
+	per_cpu(processed_system_time, 0) = processed_system_time;
+
+	init_cpu_khz();
+	printk(KERN_INFO "Xen reported: %u.%03u MHz processor.\n",
+	       cpu_khz / 1000, cpu_khz % 1000);
+
+	setup_cpu0_timer_irq();
+
+	init_missing_ticks_accounting(0);
+
+	clocksource_register(&xen_clocksource);
+
+	/* Set initial system time with full resolution */
+	xen_read_wallclock(&xtime);
+	set_normalized_timespec(&wall_to_monotonic,
+				-xtime.tv_sec, -xtime.tv_nsec);
+
+	tsc_disable = 0;
+}
+
+/* Convert jiffies to system time. */
+static u64 jiffies_to_st(unsigned long j)
+{
+	unsigned long seq;
+	long delta;
+	u64 st;
+
+	do {
+		seq = read_seqbegin(&xtime_lock);
+		delta = j - jiffies;
+		if (delta < 1) {
+			/* Triggers in some wrap-around cases, but that's okay:
+			 * we just end up with a shorter timeout. */
+			st = processed_system_time + NS_PER_TICK;
+		} else if (((unsigned long)delta >> (BITS_PER_LONG-3)) != 0) {
+			/* Very long timeout means there is no pending timer.
+			 * We indicate this to Xen by passing zero timeout. */
+			st = 0;
+		} else {
+			st = processed_system_time + delta * (u64)NS_PER_TICK;
+		}
+	} while (read_seqretry(&xtime_lock, seq));
+
+	return st;
+}
+
+/*
+ * stop_hz_timer / start_hz_timer - enter/exit 'tickless mode' on an idle cpu
+ * These functions are based on implementations from arch/s390/kernel/time.c
+ */
+void stop_hz_timer(void)
+{
+	unsigned int cpu = smp_processor_id();
+	unsigned long j;
+
+	cpu_set(cpu, nohz_cpu_mask);
+
+	/* 
+	 * See matching smp_mb in rcu_start_batch in rcupdate.c.  These mbs 
+	 * ensure that if __rcu_pending (nested in rcu_needs_cpu) fetches a
+	 * value of rcp->cur that matches rdp->quiescbatch and allows us to
+	 * stop the hz timer then the cpumasks created for subsequent values
+	 * of cur in rcu_start_batch are guaranteed to pick up the updated
+	 * nohz_cpu_mask and so will not depend on this cpu.
+	 */
+
+	smp_mb();
+
+	/* Leave ourselves in tick mode if rcu or softirq or timer pending. */
+	if (rcu_needs_cpu(cpu) || local_softirq_pending() ||
+	    (j = next_timer_interrupt(), time_before_eq(j, jiffies))) {
+		cpu_clear(cpu, nohz_cpu_mask);
+		j = jiffies + 1;
+	}
+
+	if (HYPERVISOR_set_timer_op(jiffies_to_st(j)) != 0)
+		BUG();
+}
+
+void start_hz_timer(void)
+{
+	cpu_clear(smp_processor_id(), nohz_cpu_mask);
+}
+
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/xen-head.S
@@ -0,0 +1,29 @@
+/* Xen-specific pieces of head.S, intended to be included in the right
+	place in head.S */
+
+#include <linux/elfnote.h>
+#include <asm/boot.h>
+#include <xen/interface/elfnote.h>
+
+ENTRY(startup_xen)
+	movl %esi,xen_start_info
+	jmp startup_paravirt
+	
+.pushsection ".bss.page_aligned"
+ENTRY(hypercall_page)
+	.skip 0x1000
+.popsection
+
+	ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS,       .asciz, "linux")
+	ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION,  .asciz, "2.6")
+	ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION,    .asciz, "xen-3.0")
+	ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE,      .long,  __PAGE_OFFSET)
+	ELFNOTE(Xen, XEN_ELFNOTE_ENTRY,          .long,  startup_xen)
+	ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .long,  hypercall_page)
+	ELFNOTE(Xen, XEN_ELFNOTE_FEATURES,       .asciz, "!writable_page_tables|pae_pgdir_above_4gb")
+#ifdef CONFIG_X86_PAE
+	ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE,       .asciz, "yes")
+#else
+	ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE,       .asciz, "no")
+#endif
+	ELFNOTE(Xen, XEN_ELFNOTE_LOADER,         .asciz, "generic")
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/xen-ops.h
@@ -0,0 +1,20 @@
+#ifndef XEN_OPS_H
+#define XEN_OPS_H
+
+#include <linux/init.h>
+
+extern struct start_info *xen_start_info;
+extern struct shared_info *HYPERVISOR_shared_info;
+
+char * __init xen_memory_setup(void);
+void __init xen_arch_setup(void);
+void __init xen_init_IRQ(void);
+
+void __init xen_time_init(void);
+unsigned long xen_get_wallclock(void);
+int xen_set_wallclock(unsigned long time);
+
+void stop_hz_timer(void);
+void start_hz_timer(void);
+
+#endif /* XEN_OPS_H */
===================================================================
--- /dev/null
+++ b/arch/i386/paravirt-xen/xen-page.h
@@ -0,0 +1,175 @@
+#ifndef __XEN_PAGE_H
+#define __XEN_PAGE_H
+
+#include <linux/pfn.h>
+
+#include <asm/uaccess.h>
+
+#include <xen/features.h>
+
+#ifdef CONFIG_X86_PAE
+/* Xen machine address */
+typedef struct xmaddr {
+	unsigned long long maddr;
+} xmaddr_t;
+
+/* Xen pseudo-physical address */
+typedef struct xpaddr {
+	unsigned long long paddr;
+} xpaddr_t;
+#else
+/* Xen machine address */
+typedef struct xmaddr {
+	unsigned long maddr;
+} xmaddr_t;
+
+/* Xen pseudo-physical address */
+typedef struct xpaddr {
+	unsigned long paddr;
+} xpaddr_t;
+#endif
+
+#define XMADDR(x)	((xmaddr_t) { .maddr = (x) })
+#define XPADDR(x)	((xpaddr_t) { .paddr = (x) })
+
+/**** MACHINE <-> PHYSICAL CONVERSION MACROS ****/
+#define INVALID_P2M_ENTRY	(~0UL)
+#define FOREIGN_FRAME_BIT	(1UL<<31)
+#define FOREIGN_FRAME(m)	((m) | FOREIGN_FRAME_BIT)
+
+extern unsigned long *phys_to_machine_mapping;
+
+static inline unsigned long pfn_to_mfn(unsigned long pfn)
+{
+	if (xen_feature(XENFEAT_auto_translated_physmap))
+		return pfn;
+
+	return phys_to_machine_mapping[(unsigned int)(pfn)] &
+		~FOREIGN_FRAME_BIT;
+}
+
+static inline int phys_to_machine_mapping_valid(unsigned long pfn)
+{
+	if (xen_feature(XENFEAT_auto_translated_physmap))
+		return 1;
+
+	return (phys_to_machine_mapping[pfn] != INVALID_P2M_ENTRY);
+}
+
+static inline unsigned long mfn_to_pfn(unsigned long mfn)
+{
+	unsigned long pfn;
+
+	if (xen_feature(XENFEAT_auto_translated_physmap))
+		return mfn;
+
+#if 0
+	if (unlikely((mfn >> machine_to_phys_order) != 0))
+		return max_mapnr;
+#endif
+
+	pfn = 0;
+	/*
+	 * The array access can fail (e.g., device space beyond end of RAM).
+	 * In such cases it doesn't matter what we return (we return garbage),
+	 * but we must handle the fault without crashing!
+	 */
+	__get_user(pfn, &machine_to_phys_mapping[mfn]);
+
+	return pfn;
+}
+
+static inline xmaddr_t phys_to_machine(xpaddr_t phys)
+{
+	unsigned offset = phys.paddr & ~PAGE_MASK;
+	return XMADDR(PFN_PHYS(pfn_to_mfn(PFN_DOWN(phys.paddr))) | offset);
+}
+
+static inline xpaddr_t machine_to_phys(xmaddr_t machine)
+{
+	unsigned offset = machine.maddr & ~PAGE_MASK;
+	return XPADDR(PFN_PHYS(mfn_to_pfn(PFN_DOWN(machine.maddr))) | offset);
+}
+
+/*
+ * We detect special mappings in one of two ways:
+ *  1. If the MFN is an I/O page then Xen will set the m2p entry
+ *     to be outside our maximum possible pseudophys range.
+ *  2. If the MFN belongs to a different domain then we will certainly
+ *     not have MFN in our p2m table. Conversely, if the page is ours,
+ *     then we'll have p2m(m2p(MFN))==MFN.
+ * If we detect a special mapping then it doesn't have a 'struct page'.
+ * We force !pfn_valid() by returning an out-of-range pointer.
+ *
+ * NB. These checks require that, for any MFN that is not in our reservation,
+ * there is no PFN such that p2m(PFN) == MFN. Otherwise we can get confused if
+ * we are foreign-mapping the MFN, and the other domain as m2p(MFN) == PFN.
+ * Yikes! Various places must poke in INVALID_P2M_ENTRY for safety.
+ *
+ * NB2. When deliberately mapping foreign pages into the p2m table, you *must*
+ *      use FOREIGN_FRAME(). This will cause pte_pfn() to choke on it, as we
+ *      require. In all the cases we care about, the FOREIGN_FRAME bit is
+ *      masked (e.g., pfn_to_mfn()) so behaviour there is correct.
+ */
+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)
+	    && (phys_to_machine_mapping[pfn] != mfn))
+		return max_mapnr; /* force !pfn_valid() */
+	return pfn;
+}
+
+static inline void set_phys_to_machine(unsigned long pfn, unsigned long mfn)
+{
+	if (xen_feature(XENFEAT_auto_translated_physmap)) {
+		BUG_ON(pfn != mfn && mfn != INVALID_P2M_ENTRY);
+		return;
+	}
+	phys_to_machine_mapping[pfn] = mfn;
+}
+
+/* VIRT <-> MACHINE conversion */
+#define virt_to_machine(v)	(phys_to_machine(XPADDR(__pa(v))))
+#define virt_to_mfn(v)		(pfn_to_mfn(PFN_DOWN(__pa(v))))
+#define mfn_to_virt(m)		(__va(mfn_to_pfn(m) << PAGE_SHIFT))
+
+#ifdef CONFIG_X86_PAE
+#define pte_mfn(_pte) (((_pte).pte_low >> PAGE_SHIFT) |\
+                       (((_pte).pte_high & 0xfff) << (32-PAGE_SHIFT)))
+
+static inline pte_t mfn_pte(unsigned long page_nr, pgprot_t pgprot)
+{
+	pte_t pte;
+
+	pte.pte_high = (page_nr >> (32 - PAGE_SHIFT)) | (pgprot_val(pgprot) >> 32);
+	pte.pte_high &= (__supported_pte_mask >> 32);
+	pte.pte_low = ((page_nr << PAGE_SHIFT) | pgprot_val(pgprot));
+	pte.pte_low &= __supported_pte_mask;
+
+	return pte;
+}
+
+static inline unsigned long long pte_val_ma(pte_t x)
+{
+	return ((unsigned long long)x.pte_high << 32) | x.pte_low;
+}
+#define pmd_val_ma(v) ((v).pmd)
+#define pud_val_ma(v) ((v).pgd.pgd)
+#else  /* !X86_PAE */
+#define pte_mfn(_pte) ((_pte).pte_low >> PAGE_SHIFT)
+#define mfn_pte(pfn, prot)	__pte_ma(((pfn) << PAGE_SHIFT) | pgprot_val(prot))
+#define pte_val_ma(x)	((x).pte_low)
+#define pmd_val_ma(v)	((v).pud.pgd.pgd)
+#endif	/* CONFIG_X86_PAE */
+#define pgd_val_ma(x)	((x).pgd)
+
+#define __pte_ma(x)	((pte_t) { (x) } )
+
+xmaddr_t arbitrary_virt_to_machine(unsigned long address);
+void make_lowmem_page_readonly(void *vaddr);
+void make_lowmem_page_readwrite(void *vaddr);
+
+#endif /* __XEN_PAGE_H */
===================================================================
--- a/include/asm-i386/hypercall.h
+++ b/include/asm-i386/hypercall.h
@@ -39,9 +39,6 @@
 #include <xen/interface/xen.h>
 #include <xen/interface/sched.h>
 #include <xen/interface/physdev.h>
-
-#define __STR(x) #x
-#define STR(x) __STR(x)
 
 extern struct { char _entry[32]; } hypercall_page[];
 
@@ -413,4 +410,22 @@ MULTI_mmuext_op(struct multicall_entry *
 	mcl->args[2] = (unsigned long)success_count;
 	mcl->args[3] = domid;
 }
+
+static inline void
+MULTI_set_gdt(struct multicall_entry *mcl, unsigned long *frames, int entries)
+{
+	mcl->op = __HYPERVISOR_set_gdt;
+	mcl->args[0] = (unsigned long)frames;
+	mcl->args[1] = entries;
+}
+
+static inline void
+MULTI_stack_switch(struct multicall_entry *mcl, 
+		   unsigned long ss, unsigned long esp)
+{
+	mcl->op = __HYPERVISOR_stack_switch;
+	mcl->args[0] = ss;
+	mcl->args[1] = esp;
+}
+
 #endif /* __HYPERCALL_H__ */
===================================================================
--- a/include/asm-i386/irq.h
+++ b/include/asm-i386/irq.h
@@ -41,6 +41,7 @@ extern void fixup_irqs(cpumask_t map);
 extern void fixup_irqs(cpumask_t map);
 #endif
 
+fastcall unsigned int do_IRQ(struct pt_regs *regs);
 void init_IRQ(void);
 void __init native_init_IRQ(void);
 
===================================================================
--- a/include/asm-i386/paravirt.h
+++ b/include/asm-i386/paravirt.h
@@ -31,6 +31,7 @@ struct Xgt_desc_struct;
 struct Xgt_desc_struct;
 struct tss_struct;
 struct mm_struct;
+struct i386_pda;
 struct paravirt_ops
 {
  	int paravirt_enabled;
@@ -53,6 +54,7 @@ struct paravirt_ops
 	void (*arch_setup)(void);
 	char *(*memory_setup)(void);
 	void (*init_IRQ)(void);
+	void (*init_pda)(struct i386_pda *, int cpu);
 
 	void (*pagetable_setup_start)(pgd_t *pgd_base);
 	void (*pagetable_setup_done)(pgd_t *pgd_base);
@@ -200,6 +202,30 @@ extern struct paravirt_ops paravirt_ops;
 
 void native_pagetable_setup_start(pgd_t *pgd);
 
+/* Non-paravirtualized implementations of various operations for
+   back-ends which don't need their own version. */
+fastcall void native_clts(void);
+
+fastcall unsigned long native_read_cr0(void);
+fastcall void native_write_cr0(unsigned long val);
+
+fastcall unsigned long native_read_cr2(void);
+fastcall void native_write_cr2(unsigned long val);
+
+fastcall unsigned long native_read_cr3(void);
+fastcall void native_write_cr3(unsigned long val);
+
+fastcall unsigned long native_read_cr4(void);
+fastcall unsigned long native_read_cr4_safe(void);
+fastcall void native_write_cr4(unsigned long val);
+
+fastcall void native_wbinvd(void);
+
+fastcall unsigned long long native_read_msr(unsigned int msr, int *err);
+fastcall int native_write_msr(unsigned int msr, unsigned long long val);
+fastcall unsigned long long native_read_tsc(void);
+fastcall unsigned long long native_read_pmc(void);
+
 #ifdef CONFIG_X86_PAE
 fastcall unsigned long long native_pte_val(pte_t);
 fastcall unsigned long long native_pmd_val(pmd_t);
@@ -402,6 +428,19 @@ static inline void paravirt_exit_mmap(st
 {
 	paravirt_ops.exit_mmap(mm);
 }
+
+static inline void paravirt_init_pda(struct i386_pda *pda, int cpu)
+{
+	if (paravirt_ops.init_pda)
+		(*paravirt_ops.init_pda)(pda, cpu);
+}
+
+fastcall void native_write_ldt_entry(void *dt, int entrynum, u32 low, u32 high);
+fastcall void native_write_gdt_entry(void *dt, int entrynum, u32 low, u32 high);
+fastcall void native_write_idt_entry(void *dt, int entrynum, u32 low, u32 high);
+fastcall void native_store_gdt(struct Xgt_desc_struct *dtr);
+fastcall void native_store_idt(struct Xgt_desc_struct *dtr);
+fastcall unsigned long native_store_tr(void);
 
 #define __flush_tlb() paravirt_ops.flush_tlb_user()
 #define __flush_tlb_global() paravirt_ops.flush_tlb_kernel()
@@ -696,5 +735,8 @@ static inline void paravirt_exit_mmap(st
 {
 }
 
+static inline void paravirt_init_pda(struct i386_pda *pda, int cpu)
+{
+}
 #endif /* CONFIG_PARAVIRT */
 #endif	/* __ASM_PARAVIRT_H */
===================================================================
--- a/include/asm-i386/pda.h
+++ b/include/asm-i386/pda.h
@@ -16,6 +16,17 @@ struct i386_pda
 	int cpu_number;
 	struct task_struct *pcurrent;	/* current process */
 	struct pt_regs *irq_regs;
+
+#ifdef CONFIG_PARAVIRT
+	union {
+#ifdef CONFIG_XEN
+		struct {
+			struct vcpu_info *vcpu;
+			unsigned long cr3;
+		} xen;
+#endif	/* CONFIG_XEN */
+	};
+#endif	/* CONFIG_PARAVIRT */
 };
 
 extern struct i386_pda *_cpu_pda[];
===================================================================
--- /dev/null
+++ b/include/xen/features.h
@@ -0,0 +1,26 @@
+/******************************************************************************
+ * features.h
+ *
+ * Query the features reported by Xen.
+ *
+ * Copyright (c) 2006, Ian Campbell
+ */
+
+#ifndef __XEN_FEATURES_H__
+#define __XEN_FEATURES_H__
+
+#include <xen/interface/features.h>
+
+void xen_setup_features(void);
+
+extern u8 xen_features[XENFEAT_NR_SUBMAPS * 32];
+
+static inline int xen_feature(int flag)
+{
+	switch(flag) {
+	}
+
+	return xen_features[flag];
+}
+
+#endif /* __ASM_XEN_FEATURES_H__ */

-- 

[patch 16/20] XEN-paravirt: Add the Xen virtual console driver.

[Jeremy Fitzhardinge]

[-- Attachment #1: xen-console.patch --]
[-- Type: text/plain, Size: 21909 bytes --]

This provides a bootstrap and ongoing emergency console which is
intended to be available from very early during boot and at all times
thereafter, in contrast with alternatives such as UDP-based syslogd,
or logging in via ssh. The protocol is based on a simple shared-memory
ring buffer.

Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>

---
 arch/i386/kernel/early_printk.c    |    2 
 arch/i386/paravirt-xen/enlighten.c |    3 
 drivers/Makefile                   |    3 
 drivers/xen/Makefile               |    1 
 drivers/xen/console/Makefile       |    2 
 drivers/xen/console/console.c      |  588 ++++++++++++++++++++++++++++++++++++
 drivers/xen/console/xencons_ring.c |  144 ++++++++
 include/xen/xencons.h              |   14 
 init/main.c                        |    2 
 9 files changed, 759 insertions(+)

===================================================================
--- a/arch/i386/kernel/early_printk.c
+++ b/arch/i386/kernel/early_printk.c
@@ -1,2 +1,4 @@
 
+#ifndef CONFIG_XEN
 #include "../../x86_64/kernel/early_printk.c"
+#endif
===================================================================
--- a/arch/i386/paravirt-xen/enlighten.c
+++ b/arch/i386/paravirt-xen/enlighten.c
@@ -798,6 +798,9 @@ static asmlinkage void __init xen_start_
 	INITRD_START = xen_start_info->mod_start ? __pa(xen_start_info->mod_start) : 0;
 	INITRD_SIZE = xen_start_info->mod_len;
 
+	/* use Xen console */
+	vgacon_enabled = 0;
+
 	/* Start the world */
 	start_kernel();
 }
===================================================================
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -14,6 +14,9 @@ obj-$(CONFIG_ACPI)		+= acpi/
 # was used and do nothing if so
 obj-$(CONFIG_PNP)		+= pnp/
 obj-$(CONFIG_ARM_AMBA)		+= amba/
+
+# Xen is the default console when running as a guest
+obj-$(CONFIG_XEN)		+= xen/
 
 # char/ comes before serial/ etc so that the VT console is the boot-time
 # default.
===================================================================
--- /dev/null
+++ b/drivers/xen/Makefile
@@ -0,0 +1,1 @@
+obj-y	+= console/
===================================================================
--- /dev/null
+++ b/drivers/xen/console/Makefile
@@ -0,0 +1,2 @@
+
+obj-y	:= console.o xencons_ring.o
===================================================================
--- /dev/null
+++ b/drivers/xen/console/console.c
@@ -0,0 +1,588 @@
+/******************************************************************************
+ * console.c
+ * 
+ * Virtual console driver.
+ * 
+ * Copyright (c) 2002-2004, K A Fraser.
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/interrupt.h>
+#include <linux/tty.h>
+#include <linux/tty_flip.h>
+#include <linux/serial.h>
+#include <linux/major.h>
+#include <linux/ptrace.h>
+#include <linux/ioport.h>
+#include <linux/mm.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+#include <linux/console.h>
+#include <linux/bootmem.h>
+#include <linux/sysrq.h>
+#include <linux/vt.h>
+
+#include <asm/io.h>
+#include <asm/irq.h>
+#include <asm/uaccess.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/event_channel.h>
+#include <asm/hypervisor.h>
+#include "../../../arch/i386/paravirt-xen/events.h"
+#include <xen/xencons.h>
+
+MODULE_LICENSE("Dual BSD/GPL");
+
+static int xc_disabled = 0;
+static int xc_num = -1;
+
+/* /dev/xvc0 device number allocated by lanana.org. */
+#define XEN_XVC_MAJOR 204
+#define XEN_XVC_MINOR 191
+
+#ifdef CONFIG_MAGIC_SYSRQ
+static unsigned long sysrq_requested;
+#endif
+
+static int __init xencons_setup(char *str)
+{
+	if (!strcmp(str, "off"))
+		xc_disabled = 1;
+	return 1;
+}
+__setup("xencons=", xencons_setup);
+
+/* The kernel and user-land drivers share a common transmit buffer. */
+static unsigned int wbuf_size = 4096;
+#define WBUF_MASK(_i) ((_i)&(wbuf_size-1))
+static char *wbuf;
+static unsigned int wc, wp; /* write_cons, write_prod */
+
+static int __init xencons_bufsz_setup(char *str)
+{
+	unsigned int goal;
+	goal = simple_strtoul(str, NULL, 0);
+	if (goal) {
+		goal = roundup_pow_of_two(goal);
+		if (wbuf_size < goal)
+			wbuf_size = goal;
+	}
+	return 1;
+}
+__setup("xencons_bufsz=", xencons_bufsz_setup);
+
+/* This lock protects accesses to the common transmit buffer. */
+static DEFINE_SPINLOCK(xencons_lock);
+
+/* Common transmit-kick routine. */
+static void __xencons_tx_flush(void);
+
+static struct tty_driver *xencons_driver;
+
+/******************** Kernel console driver ********************************/
+
+static void kcons_write(struct console *c, const char *s, unsigned int count)
+{
+	int           i = 0;
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+
+	while (i < count) {
+		for (; i < count; i++) {
+			if ((wp - wc) >= (wbuf_size - 1))
+				break;
+			if ((wbuf[WBUF_MASK(wp++)] = s[i]) == '\n')
+				wbuf[WBUF_MASK(wp++)] = '\r';
+		}
+
+		__xencons_tx_flush();
+	}
+
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+static void kcons_write_dom0(struct console *c, const char *s, unsigned int count)
+{
+	while (count > 0) {
+		int rc;
+		rc = HYPERVISOR_console_io(CONSOLEIO_write, count, (char *)s);
+		if (rc <= 0)
+			break;
+		count -= rc;
+		s += rc;
+	}
+}
+
+void early_printk(const char *fmt, ...)
+{
+	va_list args;
+	int printk_len;
+	static char printk_buf[1024];
+
+	/* Emit the output into the temporary buffer */
+	va_start(args, fmt);
+	printk_len = vsnprintf(printk_buf, sizeof(printk_buf), fmt, args);
+	va_end(args);
+
+	/* Send the processed output directly to Xen. */
+	kcons_write_dom0(NULL, printk_buf, printk_len);
+}
+
+static struct console early_xen_console = {
+	.name =		"earlyxen",
+	.write =	kcons_write_dom0,
+	.flags =	CON_PRINTBUFFER,
+	.index =	-1,
+};
+
+static int early_console_initialized = 0;
+
+int __init setup_early_printk(char *opt)
+{
+	early_console_initialized = 1;
+	register_console(&early_xen_console);
+
+	return 0;
+}
+
+void __init disable_early_printk(void)
+{
+	if (!early_console_initialized)
+		return;
+	printk("disabling early console\n");
+	unregister_console(&early_xen_console);
+	early_console_initialized = 0;
+}
+
+static struct tty_driver *kcons_device(struct console *c, int *index)
+{
+	*index = 0;
+	return xencons_driver;
+}
+
+static struct console kcons_info = {
+	.name   = "xvc",
+	.device	= kcons_device,
+	.flags	= CON_PRINTBUFFER | CON_ENABLED,
+	.index	= -1,
+};
+
+static int __init xen_console_init(void)
+{
+	if (!is_running_on_xen())
+		goto out;
+	if (xc_disabled)
+		goto out;
+
+	if (is_initial_xendomain()) {
+		kcons_info.write = kcons_write_dom0;
+	} else {
+		if (!xen_start_info->console.domU.evtchn)
+			goto out;
+		kcons_info.write = kcons_write;
+	}
+	if (xc_num == -1)
+		xc_num = 0;
+
+	wbuf = alloc_bootmem(wbuf_size);
+
+	register_console(&kcons_info);
+
+out:
+	return 0;
+}
+console_initcall(xen_console_init);
+
+/*** Forcibly flush console data before dying. ***/
+void xencons_force_flush(void)
+{
+	int sz;
+
+	/* Emergency console is synchronous, so there's nothing to flush. */
+	if (!is_running_on_xen() ||
+	    is_initial_xendomain() ||
+	    !xen_start_info->console.domU.evtchn)
+		return;
+
+	/* Spin until console data is flushed through to the daemon. */
+	while (wc != wp) {
+		int sent = 0;
+		if ((sz = wp - wc) == 0)
+			continue;
+		sent = xencons_ring_send(&wbuf[WBUF_MASK(wc)], sz);
+		if (sent > 0)
+			wc += sent;
+	}
+}
+
+
+/******************** User-space console driver (/dev/console) ************/
+
+#define DRV(_d)         (_d)
+
+static struct ktermios *xencons_termios[1];
+static struct ktermios *xencons_termios_locked[1];
+static struct tty_struct *xencons_tty;
+static int xencons_priv_irq;
+static char x_char;
+
+void xencons_rx(char *buf, unsigned len)
+{
+	int           i;
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+	if (xencons_tty == NULL)
+		goto out;
+
+	for (i = 0; i < len; i++) {
+#ifdef CONFIG_MAGIC_SYSRQ
+		if (sysrq_on()) {
+			if (buf[i] == '\x0f') { /* ^O */
+				sysrq_requested = jiffies;
+				continue; /* don't print the sysrq key */
+			} else if (sysrq_requested) {
+				unsigned long sysrq_timeout =
+					sysrq_requested + HZ*2;
+				sysrq_requested = 0;
+				if (time_before(jiffies, sysrq_timeout)) {
+					spin_unlock_irqrestore(
+						&xencons_lock, flags);
+					handle_sysrq(
+						buf[i], xencons_tty);
+					spin_lock_irqsave(
+						&xencons_lock, flags);
+					continue;
+				}
+			}
+		}
+#endif
+		tty_insert_flip_char(xencons_tty, buf[i], 0);
+	}
+	tty_flip_buffer_push(xencons_tty);
+
+ out:
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+static void __xencons_tx_flush(void)
+{
+	int sent, sz, work_done = 0;
+
+	if (x_char) {
+		if (is_initial_xendomain())
+			kcons_write_dom0(NULL, &x_char, 1);
+		else
+			while (x_char)
+				if (xencons_ring_send(&x_char, 1) == 1)
+					break;
+		x_char = 0;
+		work_done = 1;
+	}
+
+	while (wc != wp) {
+		sz = wp - wc;
+		if (sz > (wbuf_size - WBUF_MASK(wc)))
+			sz = wbuf_size - WBUF_MASK(wc);
+		if (is_initial_xendomain()) {
+			kcons_write_dom0(NULL, &wbuf[WBUF_MASK(wc)], sz);
+			wc += sz;
+		} else {
+			sent = xencons_ring_send(&wbuf[WBUF_MASK(wc)], sz);
+			if (sent == 0)
+				break;
+			wc += sent;
+		}
+		work_done = 1;
+	}
+
+	if (work_done && (xencons_tty != NULL)) {
+		wake_up_interruptible(&xencons_tty->write_wait);
+		if ((xencons_tty->flags & (1 << TTY_DO_WRITE_WAKEUP)) &&
+		    (xencons_tty->ldisc.write_wakeup != NULL))
+			(xencons_tty->ldisc.write_wakeup)(xencons_tty);
+	}
+}
+
+void xencons_tx(void)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+	__xencons_tx_flush();
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+/* Privileged receive callback and transmit kicker. */
+static irqreturn_t xencons_priv_interrupt(int irq, void *dev_id)
+{
+	static char rbuf[16];
+	int         l;
+
+	while ((l = HYPERVISOR_console_io(CONSOLEIO_read, 16, rbuf)) > 0)
+		xencons_rx(rbuf, l);
+
+	xencons_tx();
+
+	return IRQ_HANDLED;
+}
+
+static int xencons_write_room(struct tty_struct *tty)
+{
+	return wbuf_size - (wp - wc);
+}
+
+static int xencons_chars_in_buffer(struct tty_struct *tty)
+{
+	return wp - wc;
+}
+
+static void xencons_send_xchar(struct tty_struct *tty, char ch)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+	x_char = ch;
+	__xencons_tx_flush();
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+static void xencons_throttle(struct tty_struct *tty)
+{
+	if (I_IXOFF(tty))
+		xencons_send_xchar(tty, STOP_CHAR(tty));
+}
+
+static void xencons_unthrottle(struct tty_struct *tty)
+{
+	if (I_IXOFF(tty)) {
+		if (x_char != 0)
+			x_char = 0;
+		else
+			xencons_send_xchar(tty, START_CHAR(tty));
+	}
+}
+
+static void xencons_flush_buffer(struct tty_struct *tty)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+	wc = wp = 0;
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+static inline int __xencons_put_char(int ch)
+{
+	char _ch = (char)ch;
+	if ((wp - wc) == wbuf_size)
+		return 0;
+	wbuf[WBUF_MASK(wp++)] = _ch;
+	return 1;
+}
+
+static int xencons_write(
+	struct tty_struct *tty,
+	const unsigned char *buf,
+	int count)
+{
+	int i;
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+
+	for (i = 0; i < count; i++)
+		if (!__xencons_put_char(buf[i]))
+			break;
+
+	if (i != 0)
+		__xencons_tx_flush();
+
+	spin_unlock_irqrestore(&xencons_lock, flags);
+
+	return i;
+}
+
+static void xencons_put_char(struct tty_struct *tty, u_char ch)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+	(void)__xencons_put_char(ch);
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+static void xencons_flush_chars(struct tty_struct *tty)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+	__xencons_tx_flush();
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+static void xencons_wait_until_sent(struct tty_struct *tty, int timeout)
+{
+	unsigned long orig_jiffies = jiffies;
+
+	while (DRV(tty->driver)->chars_in_buffer(tty)) {
+		set_current_state(TASK_INTERRUPTIBLE);
+		schedule_timeout(1);
+		if (signal_pending(current))
+			break;
+		if (timeout && time_after(jiffies, orig_jiffies + timeout))
+			break;
+	}
+
+	set_current_state(TASK_RUNNING);
+}
+
+static int xencons_open(struct tty_struct *tty, struct file *filp)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&xencons_lock, flags);
+	tty->driver_data = NULL;
+	if (xencons_tty == NULL)
+		xencons_tty = tty;
+	__xencons_tx_flush();
+	spin_unlock_irqrestore(&xencons_lock, flags);
+
+	return 0;
+}
+
+static void xencons_close(struct tty_struct *tty, struct file *filp)
+{
+	unsigned long flags;
+
+	mutex_lock(&tty_mutex);
+
+	if (tty->count != 1) {
+		mutex_unlock(&tty_mutex);
+		return;
+	}
+
+	/* Prevent other threads from re-opening this tty. */
+	set_bit(TTY_CLOSING, &tty->flags);
+	mutex_unlock(&tty_mutex);
+
+	tty->closing = 1;
+	tty_wait_until_sent(tty, 0);
+	if (DRV(tty->driver)->flush_buffer != NULL)
+		DRV(tty->driver)->flush_buffer(tty);
+	if (tty->ldisc.flush_buffer != NULL)
+		tty->ldisc.flush_buffer(tty);
+	tty->closing = 0;
+	spin_lock_irqsave(&xencons_lock, flags);
+	xencons_tty = NULL;
+	spin_unlock_irqrestore(&xencons_lock, flags);
+}
+
+static struct tty_operations xencons_ops = {
+	.open = xencons_open,
+	.close = xencons_close,
+	.write = xencons_write,
+	.write_room = xencons_write_room,
+	.put_char = xencons_put_char,
+	.flush_chars = xencons_flush_chars,
+	.chars_in_buffer = xencons_chars_in_buffer,
+	.send_xchar = xencons_send_xchar,
+	.flush_buffer = xencons_flush_buffer,
+	.throttle = xencons_throttle,
+	.unthrottle = xencons_unthrottle,
+	.wait_until_sent = xencons_wait_until_sent,
+};
+
+static int __init xencons_init(void)
+{
+	int rc;
+
+	if (!is_running_on_xen())
+		return -ENODEV;
+
+	if (xc_disabled)
+		return 0;
+
+	if (!is_initial_xendomain()) {
+		rc = xencons_ring_init();
+		if (rc)
+			return rc;
+	}
+
+	xencons_driver = alloc_tty_driver(1);
+	if (xencons_driver == NULL)
+		return -ENOMEM;
+
+	DRV(xencons_driver)->name            = "xvc";
+	DRV(xencons_driver)->major           = XEN_XVC_MAJOR;
+	DRV(xencons_driver)->minor_start     = XEN_XVC_MINOR;
+	DRV(xencons_driver)->name_base       = xc_num;
+	DRV(xencons_driver)->type            = TTY_DRIVER_TYPE_SERIAL;
+	DRV(xencons_driver)->subtype         = SERIAL_TYPE_NORMAL;
+	DRV(xencons_driver)->init_termios    = tty_std_termios;
+	DRV(xencons_driver)->flags           =
+		TTY_DRIVER_REAL_RAW |
+		TTY_DRIVER_RESET_TERMIOS |
+		TTY_DRIVER_DYNAMIC_DEV;
+	DRV(xencons_driver)->termios         = xencons_termios;
+	DRV(xencons_driver)->termios_locked  = xencons_termios_locked;
+
+	tty_set_operations(xencons_driver, &xencons_ops);
+
+	if ((rc = tty_register_driver(DRV(xencons_driver))) != 0) {
+		printk("WARNING: Failed to register Xen virtual "
+		       "console driver as '%s%d'\n",
+		       DRV(xencons_driver)->name,
+		       DRV(xencons_driver)->name_base);
+		put_tty_driver(xencons_driver);
+		xencons_driver = NULL;
+		return rc;
+	}
+
+	if (is_initial_xendomain()) {
+		xencons_priv_irq = bind_virq_to_irqhandler(
+			VIRQ_CONSOLE,
+			0,
+			xencons_priv_interrupt,
+			0,
+			"console",
+			NULL);
+		BUG_ON(xencons_priv_irq < 0);
+	}
+
+	printk("Xen virtual console successfully installed as %s%d\n",
+	       DRV(xencons_driver)->name, xc_num);
+
+	return 0;
+}
+
+module_init(xencons_init);
===================================================================
--- /dev/null
+++ b/drivers/xen/console/xencons_ring.c
@@ -0,0 +1,144 @@
+/* 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/interrupt.h>
+#include <linux/tty.h>
+#include <linux/tty_flip.h>
+#include <linux/serial.h>
+#include <linux/major.h>
+#include <linux/ptrace.h>
+#include <linux/ioport.h>
+#include <linux/mm.h>
+#include <linux/slab.h>
+
+#include <asm/hypervisor.h>
+#include "../../../arch/i386/paravirt-xen/events.h"
+#include <xen/xencons.h>
+#include <linux/wait.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/err.h>
+#include <xen/interface/io/console.h>
+#include "../../../arch/i386/paravirt-xen/xen-page.h"
+
+static int xencons_irq;
+
+static inline struct xencons_interface *xencons_interface(void)
+{
+	return mfn_to_virt(xen_start_info->console.domU.mfn);
+}
+
+static inline void notify_daemon(void)
+{
+	/* Use evtchn: this is called early, before irq is set up. */
+	notify_remote_via_evtchn(xen_start_info->console.domU.evtchn);
+}
+
+int xencons_ring_send(const char *data, unsigned len)
+{
+	int sent = 0;
+	struct xencons_interface *intf = xencons_interface();
+	XENCONS_RING_IDX cons, prod;
+
+	cons = intf->out_cons;
+	prod = intf->out_prod;
+	mb();
+	BUG_ON((prod - cons) > sizeof(intf->out));
+
+	while ((sent < len) && ((prod - cons) < sizeof(intf->out)))
+		intf->out[MASK_XENCONS_IDX(prod++, intf->out)] = data[sent++];
+
+	wmb();
+	intf->out_prod = prod;
+
+	notify_daemon();
+
+	return sent;
+}
+
+static irqreturn_t handle_input(int irq, void *unused)
+{
+	struct xencons_interface *intf = xencons_interface();
+	XENCONS_RING_IDX cons, prod;
+
+	cons = intf->in_cons;
+	prod = intf->in_prod;
+	mb();
+	BUG_ON((prod - cons) > sizeof(intf->in));
+
+	while (cons != prod) {
+		xencons_rx(intf->in+MASK_XENCONS_IDX(cons,intf->in), 1);
+		cons++;
+	}
+
+	mb();
+	intf->in_cons = cons;
+
+	notify_daemon();
+
+	xencons_tx();
+
+	return IRQ_HANDLED;
+}
+
+int xencons_ring_init(void)
+{
+	int irq;
+
+	if (xencons_irq)
+		unbind_from_irqhandler(xencons_irq, NULL);
+	xencons_irq = 0;
+
+	if (!is_running_on_xen() ||
+	    is_initial_xendomain() ||
+	    !xen_start_info->console.domU.evtchn)
+		return -ENODEV;
+
+	irq = bind_evtchn_to_irqhandler(
+		xen_start_info->console.domU.evtchn,
+		handle_input, 0, "xencons", NULL);
+	if (irq < 0) {
+		printk(KERN_ERR "XEN console request irq failed %i\n", irq);
+		return irq;
+	}
+
+	xencons_irq = irq;
+
+	/* In case we have in-flight data after save/restore... */
+	notify_daemon();
+
+	return 0;
+}
+
+void xencons_resume(void)
+{
+	(void)xencons_ring_init();
+}
===================================================================
--- /dev/null
+++ b/include/xen/xencons.h
@@ -0,0 +1,14 @@
+#ifndef __ASM_XENCONS_H__
+#define __ASM_XENCONS_H__
+
+void xencons_force_flush(void);
+void xencons_resume(void);
+
+/* Interrupt work hooks. Receive data, or kick data out. */
+void xencons_rx(char *buf, unsigned len);
+void xencons_tx(void);
+
+int xencons_ring_init(void);
+int xencons_ring_send(const char *data, unsigned len);
+
+#endif /* __ASM_XENCONS_H__ */
===================================================================
--- a/init/main.c
+++ b/init/main.c
@@ -495,6 +495,8 @@ asmlinkage void __init start_kernel(void
 	unwind_init();
 	lockdep_init();
 
+	//setup_early_printk(NULL);
+
 	local_irq_disable();
 	early_boot_irqs_off();
 	early_init_irq_lock_class();

-- 

Re: [patch 16/20] XEN-paravirt: Add the Xen virtual console driver.

[Alan]

> +#endif
> +		tty_insert_flip_char(xencons_tty, buf[i], 0);

Please use the defines like TTY_NORMAL not just 0. 

> +		if ((xencons_tty->flags & (1 << TTY_DO_WRITE_WAKEUP)) &&
> +		    (xencons_tty->ldisc.write_wakeup != NULL))
> +			(xencons_tty->ldisc.write_wakeup)(xencons_tty);
> +	}

You are't allowed to derefence and call ldisc methods without
holding the lock. You can replace that chunk with a call to the helper
function tty_wakeup(tty). Small but real race condition otherwise as you
xencons_tty->ldisc may be changing as you call it.

> +static inline int __xencons_put_char(int ch)
> +{
> +	char _ch = (char)ch;
> +	if ((wp - wc) == wbuf_size)
> +		return 0;
> +	wbuf[WBUF_MASK(wp++)] = _ch;
> +	return 1;
> +}

A lot of very confusing sign stuff here - you turn an int into a char and
put it into a uchar array

> +	for (i = 0; i < count; i++)
> +		if (!__xencons_put_char(buf[i]))
> +			break;

The int coming from a uchar array

Don't think its wrong - just acutely weird and perhaps could be tidier

> +static void xencons_close(struct tty_struct *tty, struct file *filp)
> +{
> +	unsigned long flags;
> +
> +	mutex_lock(&tty_mutex);

It would be good in future if you could avoid using tty_mutex and use a
private lock. At the moment vt "borrows" it and there are a couple of
incestuous spots but the plan is to eventually fix them and make it
private to tty_io.


Andrew: No objection to this tty stuff being merged provided the bugs
noted above (not worried about the sign stuff) are fixed before it goes
on to Linus.

Alan

Re: [patch 16/20] XEN-paravirt: Add the Xen virtual console driver.

[Jeremy Fitzhardinge]

Alan wrote:
> Andrew: No objection to this tty stuff being merged provided the bugs
> noted above (not worried about the sign stuff) are fixed before it goes
> on to Linus.
>   

Thanks for the comments.  I'll see if I can put together a fixup patch
before LCA, but possibly not.

    J

Re: [patch 16/20] XEN-paravirt: Add the Xen virtual console driver.

[Pavel Machek]

Hi!

> This provides a bootstrap and ongoing emergency console which is
> intended to be available from very early during boot and at all times
> thereafter, in contrast with alternatives such as UDP-based syslogd,
> or logging in via ssh. The protocol is based on a simple shared-memory
> ring buffer.
> 
> Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
> Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
> 
> ---
>  arch/i386/kernel/early_printk.c    |    2 
>  arch/i386/paravirt-xen/enlighten.c |    3 
>  drivers/Makefile                   |    3 
>  drivers/xen/Makefile               |    1 

You have drivers/xen... so maybe arch/i386/xen is easier to type?
arch/i386/paravirt/xen would make some sense, too, but it looks too
deep to me.

							Pavel
-- 
Thanks for all the (sleeping) penguins.

Re: [patch 16/20] XEN-paravirt: Add the Xen virtual console driver.

[Jeremy Fitzhardinge]

Pavel Machek wrote:
> You have drivers/xen... so maybe arch/i386/xen is easier to type?
> arch/i386/paravirt/xen would make some sense, too, but it looks too
> deep to me.

I think the exact placement of these files needs a bit of work.  I don't
much care about xen/ vs paravirt/xen vs paravirt-xen/, but placement of
headers needs a bit more thought.  lhype, er, lguest and vmi are in
kernel/, but I think it would be nice to put them near Xen.

    J

[patch 17/20] XEN-paravirt: Add Xen grant table support

[Jeremy Fitzhardinge]

[-- Attachment #1: xen-grant-table.patch --]
[-- Type: text/plain, Size: 17329 bytes --]

Add Xen 'grant table' driver which allows granting of access to
selected local memory pages by other virtual machines and,
symmetrically, the mapping of remote memory pages which other virtual
machines have granted access to.

This driver is a prerequisite for many of the Xen virtual device
drivers, which grant the 'device driver domain' restricted and
temporary access to only those memory pages that are currently
involved in I/O operations.

Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
---
 drivers/xen/core/Makefile |    2
 drivers/xen/core/gnttab.c |  422 ++++++++++++++++++++++++++++++++++++++++++++++
 include/xen/gnttab.h      |  105 +++++++++++
 3 files changed, 528 insertions(+), 1 deletion(-)

===================================================================
--- a/drivers/xen/Makefile
+++ b/drivers/xen/Makefile
@@ -1,1 +1,2 @@ obj-y	+= console/
+obj-y	+= core/
 obj-y	+= console/
===================================================================
--- /dev/null
+++ b/drivers/xen/core/Makefile
@@ -0,0 +1,1 @@
+obj-y	+= grant_table.o
===================================================================
--- /dev/null
+++ b/drivers/xen/core/grant_table.c
@@ -0,0 +1,445 @@
+/******************************************************************************
+ * grant_table.c
+ *
+ * Granting foreign access to our memory reservation.
+ *
+ * Copyright (c) 2005, Christopher Clark
+ * Copyright (c) 2004-2005, K A Fraser
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/vmalloc.h>
+#include <xen/interface/xen.h>
+#include <xen/grant_table.h>
+#include <asm/pgtable.h>
+#include <asm/uaccess.h>
+#include <asm/sync_bitops.h>
+
+#include "../../../arch/i386/paravirt-xen/xen-page.h"
+
+/* External tools reserve first few grant table entries. */
+#define NR_RESERVED_ENTRIES 8
+
+#define NR_GRANT_ENTRIES \
+	(NR_GRANT_FRAMES * PAGE_SIZE / sizeof(struct grant_entry))
+#define GNTTAB_LIST_END (NR_GRANT_ENTRIES + 1)
+
+static grant_ref_t gnttab_list[NR_GRANT_ENTRIES];
+static int gnttab_free_count;
+static grant_ref_t gnttab_free_head;
+static DEFINE_SPINLOCK(gnttab_list_lock);
+
+static struct grant_entry *shared;
+
+static struct gnttab_free_callback *gnttab_free_callback_list;
+
+static int get_free_entries(int count)
+{
+	unsigned long flags;
+	int ref;
+	grant_ref_t head;
+	spin_lock_irqsave(&gnttab_list_lock, flags);
+	if (gnttab_free_count < count) {
+		spin_unlock_irqrestore(&gnttab_list_lock, flags);
+		return -1;
+	}
+	ref = head = gnttab_free_head;
+	gnttab_free_count -= count;
+	while (count-- > 1)
+		head = gnttab_list[head];
+	gnttab_free_head = gnttab_list[head];
+	gnttab_list[head] = GNTTAB_LIST_END;
+	spin_unlock_irqrestore(&gnttab_list_lock, flags);
+	return ref;
+}
+
+#define get_free_entry() get_free_entries(1)
+
+static void do_free_callbacks(void)
+{
+	struct gnttab_free_callback *callback, *next;
+
+	callback = gnttab_free_callback_list;
+	gnttab_free_callback_list = NULL;
+
+	while (callback != NULL) {
+		next = callback->next;
+		if (gnttab_free_count >= callback->count) {
+			callback->next = NULL;
+			callback->fn(callback->arg);
+		} else {
+			callback->next = gnttab_free_callback_list;
+			gnttab_free_callback_list = callback;
+		}
+		callback = next;
+	}
+}
+
+static inline void check_free_callbacks(void)
+{
+	if (unlikely(gnttab_free_callback_list))
+		do_free_callbacks();
+}
+
+static void put_free_entry(grant_ref_t ref)
+{
+	unsigned long flags;
+	spin_lock_irqsave(&gnttab_list_lock, flags);
+	gnttab_list[ref] = gnttab_free_head;
+	gnttab_free_head = ref;
+	gnttab_free_count++;
+	check_free_callbacks();
+	spin_unlock_irqrestore(&gnttab_list_lock, flags);
+}
+
+/*
+ * Public grant-issuing interface functions
+ */
+
+int gnttab_grant_foreign_access(domid_t domid, unsigned long frame,
+				int readonly)
+{
+	int ref;
+
+	if (unlikely((ref = get_free_entry()) == -1))
+		return -ENOSPC;
+
+	shared[ref].frame = frame;
+	shared[ref].domid = domid;
+	wmb();
+	shared[ref].flags = GTF_permit_access | (readonly ? GTF_readonly : 0);
+
+	return ref;
+}
+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_access);
+
+void gnttab_grant_foreign_access_ref(grant_ref_t ref, domid_t domid,
+				     unsigned long frame, int readonly)
+{
+	shared[ref].frame = frame;
+	shared[ref].domid = domid;
+	wmb();
+	shared[ref].flags = GTF_permit_access | (readonly ? GTF_readonly : 0);
+}
+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_access_ref);
+
+
+int gnttab_query_foreign_access(grant_ref_t ref)
+{
+	u16 nflags;
+
+	nflags = shared[ref].flags;
+
+	return (nflags & (GTF_reading|GTF_writing));
+}
+EXPORT_SYMBOL_GPL(gnttab_query_foreign_access);
+
+int gnttab_end_foreign_access_ref(grant_ref_t ref, int readonly)
+{
+	u16 flags, nflags;
+
+	nflags = shared[ref].flags;
+	do {
+		if ((flags = nflags) & (GTF_reading|GTF_writing)) {
+			printk(KERN_ALERT "WARNING: g.e. still in use!\n");
+			return 0;
+		}
+	} while ((nflags = sync_cmpxchg(&shared[ref].flags, flags, 0)) !=
+		 flags);
+
+	return 1;
+}
+EXPORT_SYMBOL_GPL(gnttab_end_foreign_access_ref);
+
+void gnttab_end_foreign_access(grant_ref_t ref, int readonly,
+			       unsigned long page)
+{
+	if (gnttab_end_foreign_access_ref(ref, readonly)) {
+		put_free_entry(ref);
+		if (page != 0)
+			free_page(page);
+	} else {
+		/* XXX This needs to be fixed so that the ref and page are
+		   placed on a list to be freed up later. */
+		printk(KERN_WARNING
+		       "WARNING: leaking g.e. and page still in use!\n");
+	}
+}
+EXPORT_SYMBOL_GPL(gnttab_end_foreign_access);
+
+int gnttab_grant_foreign_transfer(domid_t domid, unsigned long pfn)
+{
+	int ref;
+
+	if (unlikely((ref = get_free_entry()) == -1))
+		return -ENOSPC;
+	gnttab_grant_foreign_transfer_ref(ref, domid, pfn);
+
+	return ref;
+}
+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_transfer);
+
+void gnttab_grant_foreign_transfer_ref(grant_ref_t ref, domid_t domid,
+				       unsigned long pfn)
+{
+	shared[ref].frame = pfn;
+	shared[ref].domid = domid;
+	wmb();
+	shared[ref].flags = GTF_accept_transfer;
+}
+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_transfer_ref);
+
+unsigned long gnttab_end_foreign_transfer_ref(grant_ref_t ref)
+{
+	unsigned long frame;
+	u16           flags;
+
+	/*
+	 * If a transfer is not even yet started, try to reclaim the grant
+	 * reference and return failure (== 0).
+	 */
+	while (!((flags = shared[ref].flags) & GTF_transfer_committed)) {
+		if (sync_cmpxchg(&shared[ref].flags, flags, 0) == flags)
+			return 0;
+		cpu_relax();
+	}
+
+	/* If a transfer is in progress then wait until it is completed. */
+	while (!(flags & GTF_transfer_completed)) {
+		flags = shared[ref].flags;
+		cpu_relax();
+	}
+
+	/* Read the frame number /after/ reading completion status. */
+	rmb();
+	frame = shared[ref].frame;
+	BUG_ON(frame == 0);
+
+	return frame;
+}
+EXPORT_SYMBOL_GPL(gnttab_end_foreign_transfer_ref);
+
+unsigned long gnttab_end_foreign_transfer(grant_ref_t ref)
+{
+	unsigned long frame = gnttab_end_foreign_transfer_ref(ref);
+	put_free_entry(ref);
+	return frame;
+}
+EXPORT_SYMBOL_GPL(gnttab_end_foreign_transfer);
+
+void gnttab_free_grant_reference(grant_ref_t ref)
+{
+	put_free_entry(ref);
+}
+EXPORT_SYMBOL_GPL(gnttab_free_grant_reference);
+
+void gnttab_free_grant_references(grant_ref_t head)
+{
+	grant_ref_t ref;
+	unsigned long flags;
+	int count = 1;
+	if (head == GNTTAB_LIST_END)
+		return;
+	spin_lock_irqsave(&gnttab_list_lock, flags);
+	ref = head;
+	while (gnttab_list[ref] != GNTTAB_LIST_END) {
+		ref = gnttab_list[ref];
+		count++;
+	}
+	gnttab_list[ref] = gnttab_free_head;
+	gnttab_free_head = head;
+	gnttab_free_count += count;
+	check_free_callbacks();
+	spin_unlock_irqrestore(&gnttab_list_lock, flags);
+}
+EXPORT_SYMBOL_GPL(gnttab_free_grant_references);
+
+int gnttab_alloc_grant_references(u16 count, grant_ref_t *head)
+{
+	int h = get_free_entries(count);
+
+	if (h == -1)
+		return -ENOSPC;
+
+	*head = h;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(gnttab_alloc_grant_references);
+
+int gnttab_empty_grant_references(const grant_ref_t *private_head)
+{
+	return (*private_head == GNTTAB_LIST_END);
+}
+EXPORT_SYMBOL_GPL(gnttab_empty_grant_references);
+
+int gnttab_claim_grant_reference(grant_ref_t *private_head)
+{
+	grant_ref_t g = *private_head;
+	if (unlikely(g == GNTTAB_LIST_END))
+		return -ENOSPC;
+	*private_head = gnttab_list[g];
+	return g;
+}
+EXPORT_SYMBOL_GPL(gnttab_claim_grant_reference);
+
+void gnttab_release_grant_reference(grant_ref_t *private_head,
+				    grant_ref_t release)
+{
+	gnttab_list[release] = *private_head;
+	*private_head = release;
+}
+EXPORT_SYMBOL_GPL(gnttab_release_grant_reference);
+
+void gnttab_request_free_callback(struct gnttab_free_callback *callback,
+				  void (*fn)(void *), void *arg, u16 count)
+{
+	unsigned long flags;
+	spin_lock_irqsave(&gnttab_list_lock, flags);
+	if (callback->next)
+		goto out;
+	callback->fn = fn;
+	callback->arg = arg;
+	callback->count = count;
+	callback->next = gnttab_free_callback_list;
+	gnttab_free_callback_list = callback;
+	check_free_callbacks();
+out:
+	spin_unlock_irqrestore(&gnttab_list_lock, flags);
+}
+EXPORT_SYMBOL_GPL(gnttab_request_free_callback);
+
+void gnttab_cancel_free_callback(struct gnttab_free_callback *callback)
+{
+	struct gnttab_free_callback **pcb;
+	unsigned long flags;
+
+	spin_lock_irqsave(&gnttab_list_lock, flags);
+	for (pcb = &gnttab_free_callback_list; *pcb; pcb = &(*pcb)->next) {
+		if (*pcb == callback) {
+			*pcb = callback->next;
+			break;
+		}
+	}
+	spin_unlock_irqrestore(&gnttab_list_lock, flags);
+}
+EXPORT_SYMBOL_GPL(gnttab_cancel_free_callback);
+
+#ifndef __ia64__
+static int map_pte_fn(pte_t *pte, struct page *pmd_page,
+		      unsigned long addr, void *data)
+{
+	unsigned long **frames = (unsigned long **)data;
+
+	set_pte_at(&init_mm, addr, pte, mfn_pte((*frames)[0], PAGE_KERNEL));
+	(*frames)++;
+	return 0;
+}
+
+static int unmap_pte_fn(pte_t *pte, struct page *pmd_page,
+			unsigned long addr, void *data)
+{
+
+	set_pte_at(&init_mm, addr, pte, __pte(0));
+	return 0;
+}
+#endif
+
+int gnttab_resume(void)
+{
+	struct gnttab_setup_table setup;
+	unsigned long frames[NR_GRANT_FRAMES];
+	int rc;
+
+	setup.dom        = DOMID_SELF;
+	setup.nr_frames  = NR_GRANT_FRAMES;
+	setup.frame_list = frames;
+
+	rc = HYPERVISOR_grant_table_op(GNTTABOP_setup_table, &setup, 1);
+	if (rc == -ENOSYS)
+		return -ENOSYS;
+
+	BUG_ON(rc || setup.status);
+
+#ifndef __ia64__
+	{
+		void *pframes = frames;
+		struct vm_struct *area;
+
+		if (shared == NULL) {
+			area = get_vm_area(PAGE_SIZE * NR_GRANT_FRAMES,
+					   VM_IOREMAP);
+			BUG_ON(area == NULL);
+			shared = area->addr;
+		}
+		rc = apply_to_page_range(&init_mm, (unsigned long)shared,
+					 PAGE_SIZE * NR_GRANT_FRAMES,
+					 map_pte_fn, &pframes);
+		BUG_ON(rc);
+	}
+#else
+	shared = __va(frames[0] << PAGE_SHIFT);
+	printk("grant table at %p\n", shared);
+#endif
+
+	return 0;
+}
+
+int gnttab_suspend(void)
+{
+
+#ifndef __ia64__
+	apply_to_page_range(&init_mm, (unsigned long)shared,
+			    PAGE_SIZE * NR_GRANT_FRAMES,
+			    unmap_pte_fn, NULL);
+#endif
+
+	return 0;
+}
+
+static int __init gnttab_init(void)
+{
+	int i;
+
+	if (!is_running_on_xen())
+		return -ENODEV;
+
+	if (gnttab_resume() < 0)
+		return -ENODEV;
+
+	for (i = NR_RESERVED_ENTRIES; i < NR_GRANT_ENTRIES; i++)
+		gnttab_list[i] = i + 1;
+	gnttab_free_count = NR_GRANT_ENTRIES - NR_RESERVED_ENTRIES;
+	gnttab_free_head  = NR_RESERVED_ENTRIES;
+
+	printk("Grant table initialized\n");
+	return 0;
+}
+
+core_initcall(gnttab_init);
===================================================================
--- /dev/null
+++ b/include/xen/grant_table.h
@@ -0,0 +1,107 @@
+/******************************************************************************
+ * grant_table.h
+ * 
+ * Two sets of functionality:
+ * 1. Granting foreign access to our memory reservation.
+ * 2. Accessing others' memory reservations via grant references.
+ * (i.e., mechanisms for both sender and recipient of grant references)
+ * 
+ * Copyright (c) 2004-2005, K A Fraser
+ * Copyright (c) 2005, Christopher Clark
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#ifndef __ASM_GNTTAB_H__
+#define __ASM_GNTTAB_H__
+
+#include <asm/hypervisor.h>
+#include <xen/interface/grant_table.h>
+
+/* NR_GRANT_FRAMES must be less than or equal to that configured in Xen */
+#define NR_GRANT_FRAMES 4
+
+struct gnttab_free_callback {
+	struct gnttab_free_callback *next;
+	void (*fn)(void *);
+	void *arg;
+	u16 count;
+};
+
+int gnttab_grant_foreign_access(domid_t domid, unsigned long frame,
+				int readonly);
+
+/*
+ * End access through the given grant reference, iff the grant entry is no
+ * longer in use.  Return 1 if the grant entry was freed, 0 if it is still in
+ * use.
+ */
+int gnttab_end_foreign_access_ref(grant_ref_t ref, int readonly);
+
+/*
+ * Eventually end access through the given grant reference, and once that
+ * access has been ended, free the given page too.  Access will be ended
+ * immediately iff the grant entry is not in use, otherwise it will happen
+ * some time later.  page may be 0, in which case no freeing will occur.
+ */
+void gnttab_end_foreign_access(grant_ref_t ref, int readonly,
+			       unsigned long page);
+
+int gnttab_grant_foreign_transfer(domid_t domid, unsigned long pfn);
+
+unsigned long gnttab_end_foreign_transfer_ref(grant_ref_t ref);
+unsigned long gnttab_end_foreign_transfer(grant_ref_t ref);
+
+int gnttab_query_foreign_access(grant_ref_t ref);
+
+/*
+ * operations on reserved batches of grant references
+ */
+int gnttab_alloc_grant_references(u16 count, grant_ref_t *pprivate_head);
+
+void gnttab_free_grant_reference(grant_ref_t ref);
+
+void gnttab_free_grant_references(grant_ref_t head);
+
+int gnttab_empty_grant_references(const grant_ref_t *pprivate_head);
+
+int gnttab_claim_grant_reference(grant_ref_t *pprivate_head);
+
+void gnttab_release_grant_reference(grant_ref_t *private_head,
+				    grant_ref_t release);
+
+void gnttab_request_free_callback(struct gnttab_free_callback *callback,
+				  void (*fn)(void *), void *arg, u16 count);
+void gnttab_cancel_free_callback(struct gnttab_free_callback *callback);
+
+void gnttab_grant_foreign_access_ref(grant_ref_t ref, domid_t domid,
+				     unsigned long frame, int readonly);
+
+void gnttab_grant_foreign_transfer_ref(grant_ref_t, domid_t domid,
+				       unsigned long pfn);
+
+#define gnttab_map_vaddr(map) ((void *)(map.host_virt_addr))
+
+#endif /* __ASM_GNTTAB_H__ */

-- 

Re: [patch 17/20] XEN-paravirt: Add Xen grant table support

[Pavel Machek]

Hi!

>  drivers/xen/core/Makefile |    2
>  drivers/xen/core/gnttab.c |  422 ++++++++++++++++++++++++++++++++++++++++++++++
>  include/xen/gnttab.h      |  105 +++++++++++
>  3 files changed, 528 insertions(+), 1 deletion(-)
...
> ===================================================================
> --- /dev/null
> +++ b/drivers/xen/core/grant_table.c

What is going on? Diffstat does not seem to match the diff.


						Pavel
-- 
Thanks for all the (sleeping) penguins.

Re: [patch 17/20] XEN-paravirt: Add Xen grant table support

[Jeremy Fitzhardinge]

Pavel Machek wrote:
> What is going on? Diffstat does not seem to match the diff.
Sorry, forgot to update the diffstat for this one:

 drivers/xen/Makefile           |    1 
 drivers/xen/core/Makefile      |    1 
 drivers/xen/core/grant_table.c |  445 ++++++++++++++++++++++++++++++++++++++++
 include/xen/grant_table.h      |  107 +++++++++
 4 files changed, 554 insertions(+)


    J

[patch 18/20] XEN-paravirt: Add Xen driver utility functions.

[Jeremy Fitzhardinge]

[-- Attachment #1: xen-driver-util.patch --]
[-- Type: text/plain, Size: 3565 bytes --]

Allocate/destroy a 'vmalloc' VM area: alloc_vm_area and free_vm_area
The alloc function ensures that page tables are constructed for the
region of kernel virtual address space and mapped into init_mm.

Lock an area so that PTEs are accessible in the current address space:
lock_vm_area and unlock_vm_area
The lock function prevents context switches to a lazy mm that doesn't
have the area mapped into its page tables.  It also ensures that the
page tables are mapped into the current mm by causing the page fault
handler to copy the page directory pointers from init_mm into the
current mm.

Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Cc: "Jan Beulich" <JBeulich@novell.com>
---

 drivers/xen/Makefile      |    2 +
 drivers/xen/util.c        |   70 ++++++++++++++++++++++++++++++++++++++++++++++
 include/xen/driver_util.h |   16 ++++++++++
 3 files changed, 88 insertions(+)

===================================================================
--- a/drivers/xen/Makefile
+++ b/drivers/xen/Makefile
@@ -1,2 +1,4 @@ obj-y	+= core/
+obj-y	+= util.o
+
 obj-y	+= core/
 obj-y	+= console/
===================================================================
--- /dev/null
+++ b/drivers/xen/util.c
@@ -0,0 +1,69 @@
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <asm/uaccess.h>
+#include <xen/driver_util.h>
+
+static int f(pte_t *pte, struct page *pmd_page, unsigned long addr, void *data)
+{
+	/* apply_to_page_range() does all the hard work. */
+	return 0;
+}
+
+struct vm_struct *alloc_vm_area(unsigned long size)
+{
+	struct vm_struct *area;
+
+	area = get_vm_area(size, VM_IOREMAP);
+	if (area == NULL)
+		return NULL;
+
+	/*
+	 * This ensures that page tables are constructed for this region
+	 * of kernel virtual address space and mapped into init_mm.
+	 */
+	if (apply_to_page_range(&init_mm, (unsigned long)area->addr,
+				area->size, f, NULL)) {
+		free_vm_area(area);
+		return NULL;
+	}
+
+	return area;
+}
+EXPORT_SYMBOL_GPL(alloc_vm_area);
+
+void free_vm_area(struct vm_struct *area)
+{
+	struct vm_struct *ret;
+	ret = remove_vm_area(area->addr);
+	BUG_ON(ret != area);
+	kfree(area);
+}
+EXPORT_SYMBOL_GPL(free_vm_area);
+
+void lock_vm_area(struct vm_struct *area)
+{
+	unsigned long i;
+	char c;
+
+	/*
+	 * Prevent context switch to a lazy mm that doesn't have this area
+	 * mapped into its page tables.
+	 */
+	preempt_disable();
+
+	/*
+	 * Ensure that the page tables are mapped into the current mm. The
+	 * page-fault path will copy the page directory pointers from init_mm.
+	 */
+	for (i = 0; i < area->size; i += PAGE_SIZE)
+		(void)__get_user(c, (char __user *)area->addr + i);
+}
+EXPORT_SYMBOL_GPL(lock_vm_area);
+
+void unlock_vm_area(struct vm_struct *area)
+{
+	preempt_enable();
+}
+EXPORT_SYMBOL_GPL(unlock_vm_area);
===================================================================
--- /dev/null
+++ b/include/xen/driver_util.h
@@ -0,0 +1,15 @@
+
+#ifndef __ASM_XEN_DRIVER_UTIL_H__
+#define __ASM_XEN_DRIVER_UTIL_H__
+
+#include <linux/vmalloc.h>
+
+/* Allocate/destroy a 'vmalloc' VM area. */
+extern struct vm_struct *alloc_vm_area(unsigned long size);
+extern void free_vm_area(struct vm_struct *area);
+
+/* Lock an area so that PTEs are accessible in the current address space. */
+extern void lock_vm_area(struct vm_struct *area);
+extern void unlock_vm_area(struct vm_struct *area);
+
+#endif /* __ASM_XEN_DRIVER_UTIL_H__ */

-- 

Re: [patch 18/20] XEN-paravirt: Add Xen driver utility functions.

[Greg KH]

On Fri, Jan 12, 2007 at 05:45:57PM -0800, Jeremy Fitzhardinge wrote:
> Allocate/destroy a 'vmalloc' VM area: alloc_vm_area and free_vm_area
> The alloc function ensures that page tables are constructed for the
> region of kernel virtual address space and mapped into init_mm.

Shouldn't these functions go into the core mm/ kernel code if they are
needed, instead of living in the xen directories?

thanks,

greg k-h

[patch 19/20] XEN-paravirt: Add the Xenbus sysfs and virtual device hotplug driver.

[Jeremy Fitzhardinge]

[-- Attachment #1: xenbus.patch --]
[-- Type: text/plain, Size: 89063 bytes --]

This communicates with the machine control software via a registry
residing in a controlling virtual machine. This allows dynamic
creation, destruction and modification of virtual device
configurations (network devices, block devices and CPUS, to name some
examples).

Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
---
 drivers/xen/Makefile               |    1
 drivers/xen/xenbus/Makefile        |    7
 drivers/xen/xenbus/xenbus_client.c |  528 ++++++++++++++++++++
 drivers/xen/xenbus/xenbus_comms.c  |  222 ++++++++
 drivers/xen/xenbus/xenbus_comms.h  |   43 +
 drivers/xen/xenbus/xenbus_probe.c  |  954 +++++++++++++++++++++++++++++++++++++
 drivers/xen/xenbus/xenbus_xs.c     |  828 ++++++++++++++++++++++++++++++++
 include/xen/xenbus.h               |  199 +++++++
 8 files changed, 2782 insertions(+)

===================================================================
--- a/arch/i386/paravirt-xen/Kconfig
+++ b/arch/i386/paravirt-xen/Kconfig
@@ -8,3 +8,8 @@ config XEN
 	default y
 	help
 	  This is the Linux Xen port.
+
+config XEN_BACKEND
+	bool
+	depends XEN
+	default n
===================================================================
--- a/arch/i386/paravirt-xen/events.h
+++ b/arch/i386/paravirt-xen/events.h
@@ -2,6 +2,8 @@
 #define _XEN_EVENTS_H
 
 #include <linux/irq.h>
+
+#include <asm/hypercall.h>
 
 int bind_evtchn_to_irqhandler(unsigned int evtchn,
 			      irqreturn_t (*handler)(int, void *),
===================================================================
--- a/drivers/xen/Makefile
+++ b/drivers/xen/Makefile
@@ -2,3 +2,4 @@ obj-y	+= util.o
 
 obj-y	+= core/
 obj-y	+= console/
+obj-y	+= xenbus/
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/Makefile
@@ -0,0 +1,9 @@
+obj-y	+= xenbus.o
+
+xenbus-objs =
+xenbus-objs += xenbus_client.o
+xenbus-objs += xenbus_comms.o
+xenbus-objs += xenbus_xs.o
+xenbus-objs += xenbus_probe.o
+
+obj-$(CONFIG_XEN_BACKEND) += xenbus_probe_backend.o
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_client.c
@@ -0,0 +1,565 @@
+/******************************************************************************
+ * Client-facing interface for the Xenbus driver.  In other words, the
+ * interface between the Xenbus and the device-specific code, be it the
+ * frontend or the backend of that driver.
+ *
+ * Copyright (C) 2005 XenSource Ltd
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include <linux/types.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/event_channel.h>
+#include "../../../arch/i386/paravirt-xen/events.h"
+#include <xen/grant_table.h>
+#include <xen/xenbus.h>
+#include <xen/driver_util.h>
+
+char *xenbus_strstate(enum xenbus_state state)
+{
+	static char *name[] = {
+		[ XenbusStateUnknown      ] = "Unknown",
+		[ XenbusStateInitialising ] = "Initialising",
+		[ XenbusStateInitWait     ] = "InitWait",
+		[ XenbusStateInitialised  ] = "Initialised",
+		[ XenbusStateConnected    ] = "Connected",
+		[ XenbusStateClosing      ] = "Closing",
+		[ XenbusStateClosed	  ] = "Closed",
+	};
+	return (state < ARRAY_SIZE(name)) ? name[state] : "INVALID";
+}
+
+/**
+ * xenbus_watch_path - register a watch
+ * @dev: xenbus device
+ * @path: path to watch
+ * @watch: watch to register
+ * @callback: callback to register
+ *
+ * Register a @watch on the given path, using the given xenbus_watch structure
+ * for storage, and the given @callback function as the callback.  Return 0 on
+ * success, or -errno on error.  On success, the given @path will be saved as
+ * @watch->node, and remains the caller's to free.  On error, @watch->node will
+ * be NULL, the device will switch to %XenbusStateClosing, and the error will
+ * be saved in the store.
+ */
+int xenbus_watch_path(struct xenbus_device *dev, const char *path,
+		      struct xenbus_watch *watch,
+		      void (*callback)(struct xenbus_watch *,
+				       const char **, unsigned int))
+{
+	int err;
+
+	watch->node = path;
+	watch->callback = callback;
+
+	err = register_xenbus_watch(watch);
+
+	if (err) {
+		watch->node = NULL;
+		watch->callback = NULL;
+		xenbus_dev_fatal(dev, err, "adding watch on %s", path);
+	}
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(xenbus_watch_path);
+
+
+/**
+ * xenbus_watch_path2 - register a watch on path/path2
+ * @dev: xenbus device
+ * @path: first half of path to watch
+ * @path2: second half of path to watch
+ * @watch: watch to register
+ * @callback: callback to register
+ *
+ * Register a watch on the given @path/@path2, using the given xenbus_watch
+ * structure for storage, and the given @callback function as the callback.
+ * Return 0 on success, or -errno on error.  On success, the watched path
+ * (@path/@path2) will be saved as @watch->node, and becomes the caller's to
+ * kfree().  On error, watch->node will be NULL, so the caller has nothing to
+ * free, the device will switch to %XenbusStateClosing, and the error will be
+ * saved in the store.
+ */
+int xenbus_watch_path2(struct xenbus_device *dev, const char *path,
+		       const char *path2, struct xenbus_watch *watch,
+		       void (*callback)(struct xenbus_watch *,
+					const char **, unsigned int))
+{
+	int err;
+	char *state = kasprintf(GFP_KERNEL, "%s/%s", path, path2);
+	if (!state) {
+		xenbus_dev_fatal(dev, -ENOMEM, "allocating path for watch");
+		return -ENOMEM;
+	}
+	err = xenbus_watch_path(dev, state, watch, callback);
+
+	if (err)
+		kfree(state);
+	return err;
+}
+EXPORT_SYMBOL_GPL(xenbus_watch_path2);
+
+
+/**
+ * xenbus_switch_state
+ * @dev: xenbus device
+ * @xbt: transaction handle
+ * @state: new state
+ *
+ * Advertise in the store a change of the given driver to the given new_state.
+ * Return 0 on success, or -errno on error.  On error, the device will switch
+ * to XenbusStateClosing, and the error will be saved in the store.
+ */
+int xenbus_switch_state(struct xenbus_device *dev, enum xenbus_state state)
+{
+	/* We check whether the state is currently set to the given value, and
+	   if not, then the state is set.  We don't want to unconditionally
+	   write the given state, because we don't want to fire watches
+	   unnecessarily.  Furthermore, if the node has gone, we don't write
+	   to it, as the device will be tearing down, and we don't want to
+	   resurrect that directory.
+
+	   Note that, because of this cached value of our state, this function
+	   will not work inside a Xenstore transaction (something it was
+	   trying to in the past) because dev->state would not get reset if
+	   the transaction was aborted.
+
+	 */
+
+	int current_state;
+	int err;
+
+	if (state == dev->state)
+		return 0;
+
+	err = xenbus_scanf(XBT_NIL, dev->nodename, "state", "%d",
+			   &current_state);
+	if (err != 1)
+		return 0;
+
+	err = xenbus_printf(XBT_NIL, dev->nodename, "state", "%d", state);
+	if (err) {
+		if (state != XenbusStateClosing) /* Avoid looping */
+			xenbus_dev_fatal(dev, err, "writing new state");
+		return err;
+	}
+
+	dev->state = state;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xenbus_switch_state);
+
+int xenbus_frontend_closed(struct xenbus_device *dev)
+{
+	xenbus_switch_state(dev, XenbusStateClosed);
+	complete(&dev->down);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xenbus_frontend_closed);
+
+/**
+ * Return the path to the error node for the given device, or NULL on failure.
+ * If the value returned is non-NULL, then it is the caller's to kfree.
+ */
+static char *error_path(struct xenbus_device *dev)
+{
+	return kasprintf(GFP_KERNEL, "error/%s", dev->nodename);
+}
+
+
+static void xenbus_va_dev_error(struct xenbus_device *dev, int err,
+				const char *fmt, va_list ap)
+{
+	int ret;
+	unsigned int len;
+	char *printf_buffer = NULL;
+	char *path_buffer = NULL;
+
+#define PRINTF_BUFFER_SIZE 4096
+	printf_buffer = kmalloc(PRINTF_BUFFER_SIZE, GFP_KERNEL);
+	if (printf_buffer == NULL)
+		goto fail;
+
+	len = sprintf(printf_buffer, "%i ", -err);
+	ret = vsnprintf(printf_buffer+len, PRINTF_BUFFER_SIZE-len, fmt, ap);
+
+	BUG_ON(len + ret > PRINTF_BUFFER_SIZE-1);
+
+	dev_err(&dev->dev, "%s\n", printf_buffer);
+
+	path_buffer = error_path(dev);
+
+	if (path_buffer == NULL) {
+		dev_err(&dev->dev, "failed to write error node for %s (%s)\n",
+		       dev->nodename, printf_buffer);
+		goto fail;
+	}
+
+	if (xenbus_write(XBT_NIL, path_buffer, "error", printf_buffer) != 0) {
+		dev_err(&dev->dev, "failed to write error node for %s (%s)\n",
+		       dev->nodename, printf_buffer);
+		goto fail;
+	}
+
+fail:
+	kfree(printf_buffer);
+	kfree(path_buffer);
+}
+
+
+/**
+ * xenbus_dev_error
+ * @dev: xenbus device
+ * @err: error to report
+ * @fmt: error message format
+ *
+ * Report the given negative errno into the store, along with the given
+ * formatted message.
+ */
+void xenbus_dev_error(struct xenbus_device *dev, int err, const char *fmt, ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	xenbus_va_dev_error(dev, err, fmt, ap);
+	va_end(ap);
+}
+EXPORT_SYMBOL_GPL(xenbus_dev_error);
+
+/**
+ * xenbus_dev_fatal
+ * @dev: xenbus device
+ * @err: error to report
+ * @fmt: error message format
+ *
+ * Equivalent to xenbus_dev_error(dev, err, fmt, args), followed by
+ * xenbus_switch_state(dev, NULL, XenbusStateClosing) to schedule an orderly
+ * closedown of this driver and its peer.
+ */
+
+void xenbus_dev_fatal(struct xenbus_device *dev, int err, const char *fmt, ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	xenbus_va_dev_error(dev, err, fmt, ap);
+	va_end(ap);
+
+	xenbus_switch_state(dev, XenbusStateClosing);
+}
+EXPORT_SYMBOL_GPL(xenbus_dev_fatal);
+
+/**
+ * xenbus_grant_ring
+ * @dev: xenbus device
+ * @ring_mfn: mfn of ring to grant
+
+ * Grant access to the given @ring_mfn to the peer of the given device.  Return
+ * 0 on success, or -errno on error.  On error, the device will switch to
+ * XenbusStateClosing, and the error will be saved in the store.
+ */
+int xenbus_grant_ring(struct xenbus_device *dev, unsigned long ring_mfn)
+{
+	int err = gnttab_grant_foreign_access(dev->otherend_id, ring_mfn, 0);
+	if (err < 0)
+		xenbus_dev_fatal(dev, err, "granting access to ring page");
+	return err;
+}
+EXPORT_SYMBOL_GPL(xenbus_grant_ring);
+
+
+/**
+ * Allocate an event channel for the given xenbus_device, assigning the newly
+ * created local port to *port.  Return 0 on success, or -errno on error.  On
+ * error, the device will switch to XenbusStateClosing, and the error will be
+ * saved in the store.
+ */
+int xenbus_alloc_evtchn(struct xenbus_device *dev, int *port)
+{
+	struct evtchn_alloc_unbound alloc_unbound;
+	int err;
+
+	alloc_unbound.dom = DOMID_SELF;
+	alloc_unbound.remote_dom = dev->otherend_id;
+
+	err = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
+					  &alloc_unbound);
+	if (err)
+		xenbus_dev_fatal(dev, err, "allocating event channel");
+	else
+		*port = alloc_unbound.port;
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(xenbus_alloc_evtchn);
+
+
+/**
+ * Bind to an existing interdomain event channel in another domain. Returns 0
+ * on success and stores the local port in *port. On error, returns -errno,
+ * switches the device to XenbusStateClosing, and saves the error in XenStore.
+ */
+int xenbus_bind_evtchn(struct xenbus_device *dev, int remote_port, int *port)
+{
+	struct evtchn_bind_interdomain bind_interdomain;
+	int err;
+
+	bind_interdomain.remote_dom = dev->otherend_id;
+	bind_interdomain.remote_port = remote_port;
+
+	err = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
+					  &bind_interdomain);
+	if (err)
+		xenbus_dev_fatal(dev, err,
+				 "binding to event channel %d from domain %d",
+				 remote_port, dev->otherend_id);
+	else
+		*port = bind_interdomain.local_port;
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(xenbus_bind_evtchn);
+
+
+/**
+ * Free an existing event channel. Returns 0 on success or -errno on error.
+ */
+int xenbus_free_evtchn(struct xenbus_device *dev, int port)
+{
+	struct evtchn_close close;
+	int err;
+
+	close.port = port;
+
+	err = HYPERVISOR_event_channel_op(EVTCHNOP_close, &close);
+	if (err)
+		xenbus_dev_error(dev, err, "freeing event channel %d", port);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(xenbus_free_evtchn);
+
+
+/**
+ * xenbus_map_ring_valloc
+ * @dev: xenbus device
+ * @gnt_ref: grant reference
+ * @vaddr: pointer to address to be filled out by mapping
+ *
+ * Based on Rusty Russell's skeleton driver's map_page.
+ * Map a page of memory into this domain from another domain's grant table.
+ * xenbus_map_ring_valloc allocates a page of virtual address space, maps the
+ * page to that address, and sets *vaddr to that address.
+ * Returns 0 on success, and GNTST_* (see xen/include/interface/grant_table.h)
+ * or -ENOMEM on error. If an error is returned, device will switch to
+ * XenbusStateClosing and the error message will be saved in XenStore.
+ */
+int xenbus_map_ring_valloc(struct xenbus_device *dev, int gnt_ref, void **vaddr)
+{
+	struct gnttab_map_grant_ref op = {
+		.flags = GNTMAP_host_map,
+		.ref   = gnt_ref,
+		.dom   = dev->otherend_id,
+	};
+	struct vm_struct *area;
+
+	*vaddr = NULL;
+
+	area = alloc_vm_area(PAGE_SIZE);
+	if (!area)
+		return -ENOMEM;
+
+	op.host_addr = (unsigned long)area->addr;
+
+	lock_vm_area(area);
+	if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
+		BUG();
+	unlock_vm_area(area);
+
+	if (op.status != GNTST_okay) {
+		free_vm_area(area);
+		xenbus_dev_fatal(dev, op.status,
+				 "mapping in shared page %d from domain %d",
+				 gnt_ref, dev->otherend_id);
+		return op.status;
+	}
+
+	/* Stuff the handle in an unused field */
+	area->phys_addr = (unsigned long)op.handle;
+
+	*vaddr = area->addr;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xenbus_map_ring_valloc);
+
+
+/**
+ * xenbus_map_ring
+ * @dev: xenbus device
+ * @gnt_ref: grant reference
+ * @handle: pointer to grant handle to be filled
+ * @vaddr: address to be mapped to
+ *
+ * Map a page of memory into this domain from another domain's grant table.
+ * xenbus_map_ring does not allocate the virtual address space (you must do
+ * this yourself!). It only maps in the page to the specified address.
+ * Returns 0 on success, and GNTST_* (see xen/include/interface/grant_table.h)
+ * or -ENOMEM on error. If an error is returned, device will switch to
+ * XenbusStateClosing and the error message will be saved in XenStore.
+ */
+int xenbus_map_ring(struct xenbus_device *dev, int gnt_ref,
+		    grant_handle_t *handle, void *vaddr)
+{
+	struct gnttab_map_grant_ref op = {
+		.host_addr = (unsigned long)vaddr,
+		.flags     = GNTMAP_host_map,
+		.ref       = gnt_ref,
+		.dom       = dev->otherend_id,
+	};
+
+	if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
+		BUG();
+
+	if (op.status != GNTST_okay) {
+		xenbus_dev_fatal(dev, op.status,
+				 "mapping in shared page %d from domain %d",
+				 gnt_ref, dev->otherend_id);
+	} else
+		*handle = op.handle;
+
+	return op.status;
+}
+EXPORT_SYMBOL_GPL(xenbus_map_ring);
+
+
+/**
+ * xenbus_unmap_ring_vfree
+ * @dev: xenbus device
+ * @vaddr: addr to unmap
+ *
+ * Based on Rusty Russell's skeleton driver's unmap_page.
+ * Unmap a page of memory in this domain that was imported from another domain.
+ * Use xenbus_unmap_ring_vfree if you mapped in your memory with
+ * xenbus_map_ring_valloc (it will free the virtual address space).
+ * Returns 0 on success and returns GNTST_* on error
+ * (see xen/include/interface/grant_table.h).
+ */
+int xenbus_unmap_ring_vfree(struct xenbus_device *dev, void *vaddr)
+{
+	struct vm_struct *area;
+	struct gnttab_unmap_grant_ref op = {
+		.host_addr = (unsigned long)vaddr,
+	};
+
+	/* It'd be nice if linux/vmalloc.h provided a find_vm_area(void *addr)
+	 * method so that we don't have to muck with vmalloc internals here.
+	 * We could force the user to hang on to their struct vm_struct from
+	 * xenbus_map_ring_valloc, but these 6 lines considerably simplify
+	 * this API.
+	 */
+	read_lock(&vmlist_lock);
+	for (area = vmlist; area != NULL; area = area->next) {
+		if (area->addr == vaddr)
+			break;
+	}
+	read_unlock(&vmlist_lock);
+
+	if (!area) {
+		xenbus_dev_error(dev, -ENOENT,
+				 "can't find mapped virtual address %p", vaddr);
+		return GNTST_bad_virt_addr;
+	}
+
+	op.handle = (grant_handle_t)area->phys_addr;
+
+	lock_vm_area(area);
+	if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
+		BUG();
+	unlock_vm_area(area);
+
+	if (op.status == GNTST_okay)
+		free_vm_area(area);
+	else
+		xenbus_dev_error(dev, op.status,
+				 "unmapping page at handle %d error %d",
+				 (int16_t)area->phys_addr, op.status);
+
+	return op.status;
+}
+EXPORT_SYMBOL_GPL(xenbus_unmap_ring_vfree);
+
+
+/**
+ * xenbus_unmap_ring
+ * @dev: xenbus device
+ * @handle: grant handle
+ * @vaddr: addr to unmap
+ *
+ * Unmap a page of memory in this domain that was imported from another domain.
+ * Returns 0 on success and returns GNTST_* on error
+ * (see xen/include/interface/grant_table.h).
+ */
+int xenbus_unmap_ring(struct xenbus_device *dev,
+		      grant_handle_t handle, void *vaddr)
+{
+	struct gnttab_unmap_grant_ref op = {
+		.host_addr = (unsigned long)vaddr,
+		.handle    = handle,
+	};
+
+	if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
+		BUG();
+
+	if (op.status != GNTST_okay)
+		xenbus_dev_error(dev, op.status,
+				 "unmapping page at handle %d error %d",
+				 handle, op.status);
+
+	return op.status;
+}
+EXPORT_SYMBOL_GPL(xenbus_unmap_ring);
+
+
+/**
+ * xenbus_read_driver_state
+ * @path: path for driver
+ *
+ * Return the state of the driver rooted at the given store path, or
+ * XenbusStateUnknown if no state can be read.
+ */
+enum xenbus_state xenbus_read_driver_state(const char *path)
+{
+	enum xenbus_state result;
+	int err = xenbus_gather(XBT_NIL, path, "state", "%d", &result, NULL);
+	if (err)
+		result = XenbusStateUnknown;
+
+	return result;
+}
+EXPORT_SYMBOL_GPL(xenbus_read_driver_state);
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_comms.c
@@ -0,0 +1,218 @@
+/******************************************************************************
+ * xenbus_comms.c
+ *
+ * Low level code to talks to Xen Store: ringbuffer and event channel.
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include <linux/wait.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/err.h>
+#include <xen/xenbus.h>
+#include <asm/hypervisor.h>
+#include "../../../arch/i386/paravirt-xen/events.h"
+#include "../../../arch/i386/paravirt-xen/xen-page.h"
+#include "xenbus_comms.h"
+
+static int xenbus_irq;
+
+static DECLARE_WORK(probe_work, xenbus_probe);
+
+static DECLARE_WAIT_QUEUE_HEAD(xb_waitq);
+
+static irqreturn_t wake_waiting(int irq, void *unused)
+{
+	if (unlikely(xenstored_ready == 0)) {
+		xenstored_ready = 1;
+		schedule_work(&probe_work);
+	}
+
+	wake_up(&xb_waitq);
+	return IRQ_HANDLED;
+}
+
+static int check_indexes(XENSTORE_RING_IDX cons, XENSTORE_RING_IDX prod)
+{
+	return ((prod - cons) <= XENSTORE_RING_SIZE);
+}
+
+static void *get_output_chunk(XENSTORE_RING_IDX cons,
+			      XENSTORE_RING_IDX prod,
+			      char *buf, uint32_t *len)
+{
+	*len = XENSTORE_RING_SIZE - MASK_XENSTORE_IDX(prod);
+	if ((XENSTORE_RING_SIZE - (prod - cons)) < *len)
+		*len = XENSTORE_RING_SIZE - (prod - cons);
+	return buf + MASK_XENSTORE_IDX(prod);
+}
+
+static const void *get_input_chunk(XENSTORE_RING_IDX cons,
+				   XENSTORE_RING_IDX prod,
+				   const char *buf, uint32_t *len)
+{
+	*len = XENSTORE_RING_SIZE - MASK_XENSTORE_IDX(cons);
+	if ((prod - cons) < *len)
+		*len = prod - cons;
+	return buf + MASK_XENSTORE_IDX(cons);
+}
+
+/**
+ * xb_write - low level write
+ * @data: buffer to send
+ * @len: length of buffer
+ *
+ * Returns 0 on success, error otherwise.
+ */
+int xb_write(const void *data, unsigned len)
+{
+	struct xenstore_domain_interface *intf = xen_store_interface;
+	XENSTORE_RING_IDX cons, prod;
+	int rc;
+
+	while (len != 0) {
+		void *dst;
+		unsigned int avail;
+
+		rc = wait_event_interruptible(
+			xb_waitq,
+			(intf->req_prod - intf->req_cons) !=
+			XENSTORE_RING_SIZE);
+		if (rc < 0)
+			return rc;
+
+		/* Read indexes, then verify. */
+		cons = intf->req_cons;
+		prod = intf->req_prod;
+		mb();
+		if (!check_indexes(cons, prod)) {
+			intf->req_cons = intf->req_prod = 0;
+			return -EIO;
+		}
+
+		dst = get_output_chunk(cons, prod, intf->req, &avail);
+		if (avail == 0)
+			continue;
+		if (avail > len)
+			avail = len;
+
+		memcpy(dst, data, avail);
+		data += avail;
+		len -= avail;
+
+		/* Other side must not see new header until data is there. */
+		wmb();
+		intf->req_prod += avail;
+
+		/* This implies mb() before other side sees interrupt. */
+		notify_remote_via_evtchn(xen_store_evtchn);
+	}
+
+	return 0;
+}
+
+/**
+ * xb_read - low level read
+ * @data: buffer to fill
+ * @len: length of buffer
+ *
+ * Returns 0 on success, error otherwise.
+ */
+int xb_read(void *data, unsigned len)
+{
+	struct xenstore_domain_interface *intf = xen_store_interface;
+	XENSTORE_RING_IDX cons, prod;
+	int rc;
+
+	while (len != 0) {
+		unsigned int avail;
+		const char *src;
+
+		rc = wait_event_interruptible(
+			xb_waitq,
+			intf->rsp_cons != intf->rsp_prod);
+		if (rc < 0)
+			return rc;
+
+		/* Read indexes, then verify. */
+		cons = intf->rsp_cons;
+		prod = intf->rsp_prod;
+		mb();
+		if (!check_indexes(cons, prod)) {
+			intf->rsp_cons = intf->rsp_prod = 0;
+			return -EIO;
+		}
+
+		src = get_input_chunk(cons, prod, intf->rsp, &avail);
+		if (avail == 0)
+			continue;
+		if (avail > len)
+			avail = len;
+
+		/* We must read header before we read data. */
+		rmb();
+
+		memcpy(data, src, avail);
+		data += avail;
+		len -= avail;
+
+		/* Other side must not see free space until we've copied out */
+		mb();
+		intf->rsp_cons += avail;
+
+		pr_debug("Finished read of %i bytes (%i to go)\n", avail, len);
+
+		/* Implies mb(): they will see new header. */
+		notify_remote_via_evtchn(xen_store_evtchn);
+	}
+
+	return 0;
+}
+
+/**
+ * xb_init_comms - Set up interrupt handler off store event channel.
+ */
+int xb_init_comms(void)
+{
+	int err;
+
+	if (xenbus_irq)
+		unbind_from_irqhandler(xenbus_irq, &xb_waitq);
+
+	err = bind_evtchn_to_irqhandler(
+		xen_store_evtchn, wake_waiting,
+		0, "xenbus", &xb_waitq);
+	if (err <= 0) {
+		printk(KERN_ERR "XENBUS request irq failed %i\n", err);
+		return err;
+	}
+
+	xenbus_irq = err;
+
+	return 0;
+}
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_comms.h
@@ -0,0 +1,44 @@
+/*
+ * Private include for xenbus communications.
+ * 
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#ifndef _XENBUS_COMMS_H
+#define _XENBUS_COMMS_H
+
+int xs_init(void);
+int xb_init_comms(void);
+
+/* Low level routines. */
+int xb_write(const void *data, unsigned len);
+int xb_read(void *data, unsigned len);
+int xs_input_avail(void);
+extern struct xenstore_domain_interface *xen_store_interface;
+extern int xen_store_evtchn;
+
+#endif /* _XENBUS_COMMS_H */
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_probe.c
@@ -0,0 +1,891 @@
+/******************************************************************************
+ * Talks to Xen Store to figure out what devices we have.
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * Copyright (C) 2005 Mike Wray, Hewlett-Packard
+ * Copyright (C) 2005, 2006 XenSource Ltd
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#define DPRINTK(fmt, args...)				\
+	pr_debug("xenbus_probe (%s:%d) " fmt ".\n",	\
+		 __FUNCTION__, __LINE__, ##args)
+
+#include <linux/kernel.h>
+#include <linux/err.h>
+#include <linux/string.h>
+#include <linux/ctype.h>
+#include <linux/fcntl.h>
+#include <linux/mm.h>
+#include <linux/notifier.h>
+#include <linux/kthread.h>
+#include <linux/mutex.h>
+
+#include <asm/io.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/hypervisor.h>
+#include <xen/xenbus.h>
+
+#include "xenbus_comms.h"
+#include "xenbus_probe.h"
+#include "../../../arch/i386/paravirt-xen/events.h"
+#include "../../../arch/i386/paravirt-xen/xen-page.h"
+
+int xen_store_evtchn;
+struct xenstore_domain_interface *xen_store_interface;
+static unsigned long xen_store_mfn;
+
+static BLOCKING_NOTIFIER_HEAD(xenstore_chain);
+
+static void wait_for_devices(struct xenbus_driver *xendrv);
+
+static int xenbus_probe_frontend(const char *type, const char *name);
+
+static void xenbus_dev_shutdown(struct device *_dev);
+
+/* If something in array of ids matches this device, return it. */
+static const struct xenbus_device_id *
+match_device(const struct xenbus_device_id *arr, struct xenbus_device *dev)
+{
+	for (; *arr->devicetype != '\0'; arr++) {
+		if (!strcmp(arr->devicetype, dev->devicetype))
+			return arr;
+	}
+	return NULL;
+}
+
+int xenbus_match(struct device *_dev, struct device_driver *_drv)
+{
+	struct xenbus_driver *drv = to_xenbus_driver(_drv);
+
+	if (!drv->ids)
+		return 0;
+
+	return match_device(drv->ids, to_xenbus_device(_dev)) != NULL;
+}
+
+/* device/<type>/<id> => <type>-<id> */
+static int frontend_bus_id(char bus_id[BUS_ID_SIZE], const char *nodename)
+{
+	nodename = strchr(nodename, '/');
+	if (!nodename || strlen(nodename + 1) >= BUS_ID_SIZE) {
+		printk(KERN_WARNING "XENBUS: bad frontend %s\n", nodename);
+		return -EINVAL;
+	}
+
+	strlcpy(bus_id, nodename + 1, BUS_ID_SIZE);
+	if (!strchr(bus_id, '/')) {
+		printk(KERN_WARNING "XENBUS: bus_id %s no slash\n", bus_id);
+		return -EINVAL;
+	}
+	*strchr(bus_id, '/') = '-';
+	return 0;
+}
+
+
+static void free_otherend_details(struct xenbus_device *dev)
+{
+	kfree(dev->otherend);
+	dev->otherend = NULL;
+}
+
+
+static void free_otherend_watch(struct xenbus_device *dev)
+{
+	if (dev->otherend_watch.node) {
+		unregister_xenbus_watch(&dev->otherend_watch);
+		kfree(dev->otherend_watch.node);
+		dev->otherend_watch.node = NULL;
+	}
+}
+
+
+int read_otherend_details(struct xenbus_device *xendev,
+				 char *id_node, char *path_node)
+{
+	int err = xenbus_gather(XBT_NIL, xendev->nodename,
+				id_node, "%i", &xendev->otherend_id,
+				path_node, NULL, &xendev->otherend,
+				NULL);
+	if (err) {
+		xenbus_dev_fatal(xendev, err,
+				 "reading other end details from %s",
+				 xendev->nodename);
+		return err;
+	}
+	if (strlen(xendev->otherend) == 0 ||
+	    !xenbus_exists(XBT_NIL, xendev->otherend, "")) {
+		xenbus_dev_fatal(xendev, -ENOENT,
+				 "unable to read other end from %s.  "
+				 "missing or inaccessible.",
+				 xendev->nodename);
+		free_otherend_details(xendev);
+		return -ENOENT;
+	}
+
+	return 0;
+}
+
+
+static int read_backend_details(struct xenbus_device *xendev)
+{
+	return read_otherend_details(xendev, "backend-id", "backend");
+}
+
+
+/* Bus type for frontend drivers. */
+static struct xen_bus_type xenbus_frontend = {
+	.root = "device",
+	.levels = 2, 		/* device/type/<id> */
+	.get_bus_id = frontend_bus_id,
+	.probe = xenbus_probe_frontend,
+	.bus = {
+		.name     = "xen",
+		.match    = xenbus_match,
+		.probe    = xenbus_dev_probe,
+		.remove   = xenbus_dev_remove,
+		.shutdown = xenbus_dev_shutdown,
+	},
+};
+
+static void otherend_changed(struct xenbus_watch *watch,
+			     const char **vec, unsigned int len)
+{
+	struct xenbus_device *dev =
+		container_of(watch, struct xenbus_device, otherend_watch);
+	struct xenbus_driver *drv = to_xenbus_driver(dev->dev.driver);
+	enum xenbus_state state;
+
+	/* Protect us against watches firing on old details when the otherend
+	   details change, say immediately after a resume. */
+	if (!dev->otherend ||
+	    strncmp(dev->otherend, vec[XS_WATCH_PATH],
+		    strlen(dev->otherend))) {
+		dev_dbg(&dev->dev, "Ignoring watch at %s", vec[XS_WATCH_PATH]);
+		return;
+	}
+
+	state = xenbus_read_driver_state(dev->otherend);
+
+	dev_dbg(&dev->dev, "state is %d, (%s), %s, %s",
+		state, xenbus_strstate(state), dev->otherend_watch.node,
+		vec[XS_WATCH_PATH]);
+
+	/*
+	 * Ignore xenbus transitions during shutdown. This prevents us doing
+	 * work that can fail e.g., when the rootfs is gone.
+	 */
+	if (system_state > SYSTEM_RUNNING) {
+		struct xen_bus_type *bus = bus;
+		bus = container_of(dev->dev.bus, struct xen_bus_type, bus);
+		/* If we're frontend, drive the state machine to Closed. */
+		/* This should cause the backend to release our resources. */
+		if ((bus == &xenbus_frontend) && (state == XenbusStateClosing))
+			xenbus_frontend_closed(dev);
+		return;
+	}
+
+	if (drv->otherend_changed)
+		drv->otherend_changed(dev, state);
+}
+
+
+static int talk_to_otherend(struct xenbus_device *dev)
+{
+	struct xenbus_driver *drv = to_xenbus_driver(dev->dev.driver);
+
+	free_otherend_watch(dev);
+	free_otherend_details(dev);
+
+	return drv->read_otherend_details(dev);
+}
+
+
+static int watch_otherend(struct xenbus_device *dev)
+{
+	return xenbus_watch_path2(dev, dev->otherend, "state",
+				  &dev->otherend_watch, otherend_changed);
+}
+
+
+int xenbus_dev_probe(struct device *_dev)
+{
+	struct xenbus_device *dev = to_xenbus_device(_dev);
+	struct xenbus_driver *drv = to_xenbus_driver(_dev->driver);
+	const struct xenbus_device_id *id;
+	int err;
+
+	DPRINTK("%s", dev->nodename);
+
+	if (!drv->probe) {
+		err = -ENODEV;
+		goto fail;
+	}
+
+	id = match_device(drv->ids, dev);
+	if (!id) {
+		err = -ENODEV;
+		goto fail;
+	}
+
+	err = talk_to_otherend(dev);
+	if (err) {
+		dev_warn(&dev->dev, "talk_to_otherend on %s failed.\n",
+			 dev->nodename);
+		return err;
+	}
+
+	err = drv->probe(dev, id);
+	if (err)
+		goto fail;
+
+	err = watch_otherend(dev);
+	if (err) {
+		dev_warn(&dev->dev, "watch_otherend on %s failed.\n",
+		       dev->nodename);
+		return err;
+	}
+
+	return 0;
+fail:
+	xenbus_dev_error(dev, err, "xenbus_dev_probe on %s", dev->nodename);
+	xenbus_switch_state(dev, XenbusStateClosed);
+	return -ENODEV;
+}
+
+int xenbus_dev_remove(struct device *_dev)
+{
+	struct xenbus_device *dev = to_xenbus_device(_dev);
+	struct xenbus_driver *drv = to_xenbus_driver(_dev->driver);
+
+	DPRINTK("%s", dev->nodename);
+
+	free_otherend_watch(dev);
+	free_otherend_details(dev);
+
+	if (drv->remove)
+		drv->remove(dev);
+
+	xenbus_switch_state(dev, XenbusStateClosed);
+	return 0;
+}
+
+static void xenbus_dev_shutdown(struct device *_dev)
+{
+	struct xenbus_device *dev = to_xenbus_device(_dev);
+	unsigned long timeout = 5*HZ;
+
+	DPRINTK("%s", dev->nodename);
+
+	get_device(&dev->dev);
+	if (dev->state != XenbusStateConnected) {
+		printk("%s: %s: %s != Connected, skipping\n", __FUNCTION__,
+		       dev->nodename, xenbus_strstate(dev->state));
+		goto out;
+	}
+	xenbus_switch_state(dev, XenbusStateClosing);
+	timeout = wait_for_completion_timeout(&dev->down, timeout);
+	if (!timeout)
+		printk("%s: %s timeout closing device\n", __FUNCTION__, dev->nodename);
+ out:
+	put_device(&dev->dev);
+}
+
+int xenbus_register_driver_common(struct xenbus_driver *drv,
+				  struct xen_bus_type *bus)
+{
+	drv->driver.name = drv->name;
+	drv->driver.bus = &bus->bus;
+	drv->driver.owner = drv->owner;
+
+	return driver_register(&drv->driver);
+}
+
+int xenbus_register_frontend(struct xenbus_driver *drv)
+{
+	int ret;
+
+	drv->read_otherend_details = read_backend_details;
+
+	ret = xenbus_register_driver_common(drv, &xenbus_frontend);
+	if (ret)
+		return ret;
+
+	/* If this driver is loaded as a module wait for devices to attach. */
+	wait_for_devices(drv);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xenbus_register_frontend);
+
+void xenbus_unregister_driver(struct xenbus_driver *drv)
+{
+	driver_unregister(&drv->driver);
+}
+EXPORT_SYMBOL_GPL(xenbus_unregister_driver);
+
+struct xb_find_info
+{
+	struct xenbus_device *dev;
+	const char *nodename;
+};
+
+static int cmp_dev(struct device *dev, void *data)
+{
+	struct xenbus_device *xendev = to_xenbus_device(dev);
+	struct xb_find_info *info = data;
+
+	if (!strcmp(xendev->nodename, info->nodename)) {
+		info->dev = xendev;
+		get_device(dev);
+		return 1;
+	}
+	return 0;
+}
+
+struct xenbus_device *xenbus_device_find(const char *nodename,
+					 struct bus_type *bus)
+{
+	struct xb_find_info info = { .dev = NULL, .nodename = nodename };
+
+	bus_for_each_dev(bus, NULL, &info, cmp_dev);
+	return info.dev;
+}
+
+static int cleanup_dev(struct device *dev, void *data)
+{
+	struct xenbus_device *xendev = to_xenbus_device(dev);
+	struct xb_find_info *info = data;
+	int len = strlen(info->nodename);
+
+	DPRINTK("%s", info->nodename);
+
+	/* Match the info->nodename path, or any subdirectory of that path. */
+	if (strncmp(xendev->nodename, info->nodename, len))
+		return 0;
+
+	/* If the node name is longer, ensure it really is a subdirectory. */
+	if ((strlen(xendev->nodename) > len) && (xendev->nodename[len] != '/'))
+		return 0;
+
+	info->dev = xendev;
+	get_device(dev);
+	return 1;
+}
+
+static void xenbus_cleanup_devices(const char *path, struct bus_type *bus)
+{
+	struct xb_find_info info = { .nodename = path };
+
+	do {
+		info.dev = NULL;
+		bus_for_each_dev(bus, NULL, &info, cleanup_dev);
+		if (info.dev) {
+			device_unregister(&info.dev->dev);
+			put_device(&info.dev->dev);
+		}
+	} while (info.dev);
+}
+
+static void xenbus_dev_release(struct device *dev)
+{
+	if (dev)
+		kfree(to_xenbus_device(dev));
+}
+
+static ssize_t xendev_show_nodename(struct device *dev,
+				    struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%s\n", to_xenbus_device(dev)->nodename);
+}
+DEVICE_ATTR(nodename, S_IRUSR | S_IRGRP | S_IROTH, xendev_show_nodename, NULL);
+
+static ssize_t xendev_show_devtype(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%s\n", to_xenbus_device(dev)->devicetype);
+}
+DEVICE_ATTR(devtype, S_IRUSR | S_IRGRP | S_IROTH, xendev_show_devtype, NULL);
+
+
+int xenbus_probe_node(struct xen_bus_type *bus,
+		      const char *type,
+		      const char *nodename)
+{
+	int err;
+	struct xenbus_device *xendev;
+	size_t stringlen;
+	char *tmpstring;
+
+	enum xenbus_state state = xenbus_read_driver_state(nodename);
+
+	if (state != XenbusStateInitialising) {
+		/* Device is not new, so ignore it.  This can happen if a
+		   device is going away after switching to Closed.  */
+		return 0;
+	}
+
+	stringlen = strlen(nodename) + 1 + strlen(type) + 1;
+	xendev = kzalloc(sizeof(*xendev) + stringlen, GFP_KERNEL);
+	if (!xendev)
+		return -ENOMEM;
+
+	xendev->state = XenbusStateInitialising;
+
+	/* Copy the strings into the extra space. */
+
+	tmpstring = (char *)(xendev + 1);
+	strcpy(tmpstring, nodename);
+	xendev->nodename = tmpstring;
+
+	tmpstring += strlen(tmpstring) + 1;
+	strcpy(tmpstring, type);
+	xendev->devicetype = tmpstring;
+	init_completion(&xendev->down);
+
+	xendev->dev.bus = &bus->bus;
+	xendev->dev.release = xenbus_dev_release;
+
+	err = bus->get_bus_id(xendev->dev.bus_id, xendev->nodename);
+	if (err)
+		goto fail;
+
+	/* Register with generic device framework. */
+	err = device_register(&xendev->dev);
+	if (err)
+		goto fail;
+
+	device_create_file(&xendev->dev, &dev_attr_nodename);
+	device_create_file(&xendev->dev, &dev_attr_devtype);
+
+	return 0;
+fail:
+	kfree(xendev);
+	return err;
+}
+
+/* device/<typename>/<name> */
+static int xenbus_probe_frontend(const char *type, const char *name)
+{
+	char *nodename;
+	int err;
+
+	nodename = kasprintf(GFP_KERNEL, "%s/%s/%s", xenbus_frontend.root, type, name);
+	if (!nodename)
+		return -ENOMEM;
+
+	DPRINTK("%s", nodename);
+
+	err = xenbus_probe_node(&xenbus_frontend, type, nodename);
+	kfree(nodename);
+	return err;
+}
+
+static int xenbus_probe_device_type(struct xen_bus_type *bus, const char *type)
+{
+	int err = 0;
+	char **dir;
+	unsigned int dir_n = 0;
+	int i;
+
+	dir = xenbus_directory(XBT_NIL, bus->root, type, &dir_n);
+	if (IS_ERR(dir))
+		return PTR_ERR(dir);
+
+	for (i = 0; i < dir_n; i++) {
+		err = bus->probe(type, dir[i]);
+		if (err)
+			break;
+	}
+	kfree(dir);
+	return err;
+}
+
+int xenbus_probe_devices(struct xen_bus_type *bus)
+{
+	int err = 0;
+	char **dir;
+	unsigned int i, dir_n;
+
+	dir = xenbus_directory(XBT_NIL, bus->root, "", &dir_n);
+	if (IS_ERR(dir))
+		return PTR_ERR(dir);
+
+	for (i = 0; i < dir_n; i++) {
+		err = xenbus_probe_device_type(bus, dir[i]);
+		if (err)
+			break;
+	}
+	kfree(dir);
+	return err;
+}
+
+static unsigned int char_count(const char *str, char c)
+{
+	unsigned int i, ret = 0;
+
+	for (i = 0; str[i]; i++)
+		if (str[i] == c)
+			ret++;
+	return ret;
+}
+
+static int strsep_len(const char *str, char c, unsigned int len)
+{
+	unsigned int i;
+
+	for (i = 0; str[i]; i++)
+		if (str[i] == c) {
+			if (len == 0)
+				return i;
+			len--;
+		}
+	return (len == 0) ? i : -ERANGE;
+}
+
+void dev_changed(const char *node, struct xen_bus_type *bus)
+{
+	int exists, rootlen;
+	struct xenbus_device *dev;
+	char type[BUS_ID_SIZE];
+	const char *p, *root;
+
+	if (char_count(node, '/') < 2)
+ 		return;
+
+	exists = xenbus_exists(XBT_NIL, node, "");
+	if (!exists) {
+		xenbus_cleanup_devices(node, &bus->bus);
+		return;
+	}
+
+	/* backend/<type>/... or device/<type>/... */
+	p = strchr(node, '/') + 1;
+	snprintf(type, BUS_ID_SIZE, "%.*s", (int)strcspn(p, "/"), p);
+	type[BUS_ID_SIZE-1] = '\0';
+
+	rootlen = strsep_len(node, '/', bus->levels);
+	if (rootlen < 0)
+		return;
+	root = kasprintf(GFP_KERNEL, "%.*s", rootlen, node);
+	if (!root)
+		return;
+
+	dev = xenbus_device_find(root, &bus->bus);
+	if (!dev)
+		xenbus_probe_node(bus, type, root);
+	else
+		put_device(&dev->dev);
+
+	kfree(root);
+}
+
+static void frontend_changed(struct xenbus_watch *watch,
+			     const char **vec, unsigned int len)
+{
+	DPRINTK("");
+
+	dev_changed(vec[XS_WATCH_PATH], &xenbus_frontend);
+}
+
+/* We watch for devices appearing and vanishing. */
+static struct xenbus_watch fe_watch = {
+	.node = "device",
+	.callback = frontend_changed,
+};
+
+static int suspend_dev(struct device *dev, void *data)
+{
+	int err = 0;
+	struct xenbus_driver *drv;
+	struct xenbus_device *xdev;
+
+	DPRINTK("");
+
+	if (dev->driver == NULL)
+		return 0;
+	drv = to_xenbus_driver(dev->driver);
+	xdev = container_of(dev, struct xenbus_device, dev);
+	if (drv->suspend)
+		err = drv->suspend(xdev);
+	if (err)
+		printk(KERN_WARNING
+		       "xenbus: suspend %s failed: %i\n", dev->bus_id, err);
+	return 0;
+}
+
+static int resume_dev(struct device *dev, void *data)
+{
+	int err;
+	struct xenbus_driver *drv;
+	struct xenbus_device *xdev;
+
+	DPRINTK("");
+
+	if (dev->driver == NULL)
+		return 0;
+
+	drv = to_xenbus_driver(dev->driver);
+	xdev = container_of(dev, struct xenbus_device, dev);
+
+	err = talk_to_otherend(xdev);
+	if (err) {
+		printk(KERN_WARNING
+		       "xenbus: resume (talk_to_otherend) %s failed: %i\n",
+		       dev->bus_id, err);
+		return err;
+	}
+
+	xdev->state = XenbusStateInitialising;
+
+	if (drv->resume) {
+		err = drv->resume(xdev);
+		if (err) { 
+			printk(KERN_WARNING
+			       "xenbus: resume %s failed: %i\n", 
+			       dev->bus_id, err);
+			return err;
+		}
+	}
+
+	err = watch_otherend(xdev);
+	if (err) {
+		printk(KERN_WARNING
+		       "xenbus_probe: resume (watch_otherend) %s failed: "
+		       "%d.\n", dev->bus_id, err);
+		return err;
+	}
+
+	return 0;
+}
+
+void xenbus_suspend(void)
+{
+	DPRINTK("");
+
+	bus_for_each_dev(&xenbus_frontend.bus, NULL, NULL, suspend_dev);
+	xenbus_backend_suspend(suspend_dev);
+	xs_suspend();
+}
+EXPORT_SYMBOL_GPL(xenbus_suspend);
+
+void xenbus_resume(void)
+{
+	xb_init_comms();
+	xs_resume();
+	bus_for_each_dev(&xenbus_frontend.bus, NULL, NULL, resume_dev);
+	xenbus_backend_resume(resume_dev);
+}
+EXPORT_SYMBOL_GPL(xenbus_resume);
+
+
+/* A flag to determine if xenstored is 'ready' (i.e. has started) */
+int xenstored_ready = 0;
+
+
+int register_xenstore_notifier(struct notifier_block *nb)
+{
+	int ret = 0;
+
+	if (xenstored_ready > 0)
+		ret = nb->notifier_call(nb, 0, NULL);
+	else
+		blocking_notifier_chain_register(&xenstore_chain, nb);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(register_xenstore_notifier);
+
+void unregister_xenstore_notifier(struct notifier_block *nb)
+{
+	blocking_notifier_chain_unregister(&xenstore_chain, nb);
+}
+EXPORT_SYMBOL_GPL(unregister_xenstore_notifier);
+
+void xenbus_probe(struct work_struct *unused)
+{
+	BUG_ON((xenstored_ready <= 0));
+
+	/* Enumerate devices in xenstore and watch for changes. */
+	xenbus_probe_devices(&xenbus_frontend);
+	register_xenbus_watch(&fe_watch);
+	xenbus_backend_probe_and_watch();
+
+	/* Notify others that xenstore is up */
+	blocking_notifier_call_chain(&xenstore_chain, 0, NULL);
+}
+
+static int __init xenbus_probe_init(void)
+{
+	int err = 0;
+
+	DPRINTK("");
+
+	err = -ENODEV;
+	if (!is_running_on_xen())
+		goto out_error;
+
+	/* Register ourselves with the kernel bus subsystem */
+	err = bus_register(&xenbus_frontend.bus);
+	if (err)
+		goto out_error;
+
+	err = xenbus_backend_bus_register();
+	if (err)
+		goto out_unreg_front;
+
+	/*
+	 * Domain0 doesn't have a store_evtchn or store_mfn yet.
+	 */
+	if (is_initial_xendomain()) {
+		/* dom0 not yet supported */
+	} else {
+		xenstored_ready = 1;
+		xen_store_evtchn = xen_start_info->store_evtchn;
+		xen_store_mfn = xen_start_info->store_mfn;
+	}
+	xen_store_interface = mfn_to_virt(xen_store_mfn);
+
+	/* Initialize the interface to xenstore. */
+	err = xs_init();
+	if (err) {
+		printk(KERN_WARNING
+		       "XENBUS: Error initializing xenstore comms: %i\n", err);
+		goto out_unreg_back;
+	}
+
+	if (!is_initial_xendomain())
+		xenbus_probe(NULL);
+
+	return 0;
+
+  out_unreg_back:
+	xenbus_backend_bus_unregister();
+
+  out_unreg_front:
+	bus_unregister(&xenbus_frontend.bus);
+
+  out_error:
+	return err;
+}
+
+postcore_initcall(xenbus_probe_init);
+
+MODULE_LICENSE("Dual BSD/GPL");
+
+static int is_disconnected_device(struct device *dev, void *data)
+{
+	struct xenbus_device *xendev = to_xenbus_device(dev);
+	struct device_driver *drv = data;
+
+	/*
+	 * A device with no driver will never connect. We care only about
+	 * devices which should currently be in the process of connecting.
+	 */
+	if (!dev->driver)
+		return 0;
+
+	/* Is this search limited to a particular driver? */
+	if (drv && (dev->driver != drv))
+		return 0;
+
+	return (xendev->state != XenbusStateConnected);
+}
+
+static int exists_disconnected_device(struct device_driver *drv)
+{
+	return bus_for_each_dev(&xenbus_frontend.bus, NULL, drv,
+				is_disconnected_device);
+}
+
+static int print_device_status(struct device *dev, void *data)
+{
+	struct xenbus_device *xendev = to_xenbus_device(dev);
+	struct device_driver *drv = data;
+
+	/* Is this operation limited to a particular driver? */
+	if (drv && (dev->driver != drv))
+		return 0;
+
+	if (!dev->driver) {
+		/* Information only: is this too noisy? */
+		printk(KERN_INFO "XENBUS: Device with no driver: %s\n",
+		       xendev->nodename);
+	} else if (xendev->state != XenbusStateConnected) {
+		printk(KERN_WARNING "XENBUS: Timeout connecting "
+		       "to device: %s (state %d)\n",
+		       xendev->nodename, xendev->state);
+	}
+
+	return 0;
+}
+
+/* We only wait for device setup after most initcalls have run. */
+static int ready_to_wait_for_devices;
+
+/*
+ * On a 10 second timeout, wait for all devices currently configured.  We need
+ * to do this to guarantee that the filesystems and / or network devices
+ * needed for boot are available, before we can allow the boot to proceed.
+ *
+ * This needs to be on a late_initcall, to happen after the frontend device
+ * drivers have been initialised, but before the root fs is mounted.
+ *
+ * A possible improvement here would be to have the tools add a per-device
+ * flag to the store entry, indicating whether it is needed at boot time.
+ * This would allow people who knew what they were doing to accelerate their
+ * boot slightly, but of course needs tools or manual intervention to set up
+ * those flags correctly.
+ */
+static void wait_for_devices(struct xenbus_driver *xendrv)
+{
+	unsigned long timeout = jiffies + 10*HZ;
+	struct device_driver *drv = xendrv ? &xendrv->driver : NULL;
+
+	if (!ready_to_wait_for_devices || !is_running_on_xen())
+		return;
+
+	while (exists_disconnected_device(drv)) {
+		if (time_after(jiffies, timeout))
+			break;
+		schedule_timeout_interruptible(HZ/10);
+	}
+
+	bus_for_each_dev(&xenbus_frontend.bus, NULL, drv,
+			 print_device_status);
+}
+
+#ifndef MODULE
+static int __init boot_wait_for_devices(void)
+{
+	ready_to_wait_for_devices = 1;
+	wait_for_devices(NULL);
+	return 0;
+}
+
+late_initcall(boot_wait_for_devices);
+#endif
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_probe.h
@@ -0,0 +1,72 @@
+/******************************************************************************
+ * xenbus_probe.h
+ *
+ * Talks to Xen Store to figure out what devices we have.
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * Copyright (C) 2005 XenSource Ltd.
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#ifndef _XENBUS_PROBE_H
+#define _XENBUS_PROBE_H
+
+#ifdef CONFIG_XEN_BACKEND
+extern void xenbus_backend_suspend(int (*fn)(struct device *, void *));
+extern void xenbus_backend_resume(int (*fn)(struct device *, void *));
+extern void xenbus_backend_probe_and_watch(void);
+extern int xenbus_backend_bus_register(void);
+extern void xenbus_backend_bus_unregister(void);
+#else
+static inline void xenbus_backend_suspend(int (*fn)(struct device *, void *)) {}
+static inline void xenbus_backend_resume(int (*fn)(struct device *, void *)) {}
+static inline void xenbus_backend_probe_and_watch(void) {}
+static inline int xenbus_backend_bus_register(void) {return 0;}
+static inline void xenbus_backend_bus_unregister(void) {}
+#endif
+
+struct xen_bus_type
+{
+	char *root;
+	unsigned int levels;
+	int (*get_bus_id)(char bus_id[BUS_ID_SIZE], const char *nodename);
+	int (*probe)(const char *type, const char *dir);
+	struct bus_type bus;
+};
+
+extern int xenbus_match(struct device *_dev, struct device_driver *_drv);
+extern int xenbus_dev_probe(struct device *_dev);
+extern int xenbus_dev_remove(struct device *_dev);
+extern int xenbus_register_driver_common(struct xenbus_driver *drv,
+					 struct xen_bus_type *bus);
+extern int xenbus_probe_node(struct xen_bus_type *bus,
+			     const char *type,
+			     const char *nodename);
+extern int xenbus_probe_devices(struct xen_bus_type *bus);
+
+extern void dev_changed(const char *node, struct xen_bus_type *bus);
+
+#endif
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_probe_backend.c
@@ -0,0 +1,269 @@
+/******************************************************************************
+ * Talks to Xen Store to figure out what devices we have (backend half).
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * Copyright (C) 2005 Mike Wray, Hewlett-Packard
+ * Copyright (C) 2005, 2006 XenSource Ltd
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#define DPRINTK(fmt, args...)				\
+	pr_debug("xenbus_probe (%s:%d) " fmt ".\n",	\
+		 __FUNCTION__, __LINE__, ##args)
+
+#include <linux/kernel.h>
+#include <linux/err.h>
+#include <linux/string.h>
+#include <linux/ctype.h>
+#include <linux/fcntl.h>
+#include <linux/mm.h>
+#include <linux/notifier.h>
+#include <linux/kthread.h>
+
+#include <asm/io.h>
+#include <asm/page.h>
+//#include <asm/maddr.h>
+#include <asm/pgtable.h>
+#include <asm/hypervisor.h>
+#include <xen/xenbus.h>
+//#include <xen/xen_proc.h>
+//#include <xen/evtchn.h>
+#include <xen/features.h>
+//#include <xen/hvm.h>
+
+#include "xenbus_comms.h"
+#include "xenbus_probe.h"
+
+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
+#include <xen/platform-compat.h>
+#endif
+
+static int xenbus_uevent_backend(struct device *dev, char **envp,
+				 int num_envp, char *buffer, int buffer_size);
+static int xenbus_probe_backend(const char *type, const char *domid);
+
+extern int read_otherend_details(struct xenbus_device *xendev,
+				 char *id_node, char *path_node);
+
+static int read_frontend_details(struct xenbus_device *xendev)
+{
+	return read_otherend_details(xendev, "frontend-id", "frontend");
+}
+
+/* backend/<type>/<fe-uuid>/<id> => <type>-<fe-domid>-<id> */
+static int backend_bus_id(char bus_id[BUS_ID_SIZE], const char *nodename)
+{
+	int domid, err;
+	const char *devid, *type, *frontend;
+	unsigned int typelen;
+
+	type = strchr(nodename, '/');
+	if (!type)
+		return -EINVAL;
+	type++;
+	typelen = strcspn(type, "/");
+	if (!typelen || type[typelen] != '/')
+		return -EINVAL;
+
+	devid = strrchr(nodename, '/') + 1;
+
+	err = xenbus_gather(XBT_NIL, nodename, "frontend-id", "%i", &domid,
+			    "frontend", NULL, &frontend,
+			    NULL);
+	if (err)
+		return err;
+	if (strlen(frontend) == 0)
+		err = -ERANGE;
+	if (!err && !xenbus_exists(XBT_NIL, frontend, ""))
+		err = -ENOENT;
+
+	kfree(frontend);
+
+	if (err)
+		return err;
+
+	if (snprintf(bus_id, BUS_ID_SIZE,
+		     "%.*s-%i-%s", typelen, type, domid, devid) >= BUS_ID_SIZE)
+		return -ENOSPC;
+	return 0;
+}
+
+static struct xen_bus_type xenbus_backend = {
+	.root = "backend",
+	.levels = 3, 		/* backend/type/<frontend>/<id> */
+	.get_bus_id = backend_bus_id,
+	.probe = xenbus_probe_backend,
+	.bus = {
+		.name     = "xen-backend",
+		.match    = xenbus_match,
+		.probe    = xenbus_dev_probe,
+		.remove   = xenbus_dev_remove,
+//		.shutdown = xenbus_dev_shutdown,
+		.uevent   = xenbus_uevent_backend,
+	},
+};
+
+static int xenbus_uevent_backend(struct device *dev, char **envp,
+				 int num_envp, char *buffer, int buffer_size)
+{
+	struct xenbus_device *xdev;
+	struct xenbus_driver *drv;
+	int i = 0;
+	int length = 0;
+
+	DPRINTK("");
+
+	if (dev == NULL)
+		return -ENODEV;
+
+	xdev = to_xenbus_device(dev);
+	if (xdev == NULL)
+		return -ENODEV;
+
+	/* stuff we want to pass to /sbin/hotplug */
+	add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
+		       "XENBUS_TYPE=%s", xdev->devicetype);
+
+	add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
+		       "XENBUS_PATH=%s", xdev->nodename);
+
+	add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
+		       "XENBUS_BASE_PATH=%s", xenbus_backend.root);
+
+	/* terminate, set to next free slot, shrink available space */
+	envp[i] = NULL;
+	envp = &envp[i];
+	num_envp -= i;
+	buffer = &buffer[length];
+	buffer_size -= length;
+
+	if (dev->driver) {
+		drv = to_xenbus_driver(dev->driver);
+		if (drv && drv->uevent)
+			return drv->uevent(xdev, envp, num_envp, buffer,
+					   buffer_size);
+	}
+
+	return 0;
+}
+
+int xenbus_register_backend(struct xenbus_driver *drv)
+{
+	drv->read_otherend_details = read_frontend_details;
+
+	return xenbus_register_driver_common(drv, &xenbus_backend);
+}
+EXPORT_SYMBOL_GPL(xenbus_register_backend);
+
+/* backend/<typename>/<frontend-uuid>/<name> */
+static int xenbus_probe_backend_unit(const char *dir,
+				     const char *type,
+				     const char *name)
+{
+	char *nodename;
+	int err;
+
+	nodename = kasprintf(GFP_KERNEL, "%s/%s", dir, name);
+	if (!nodename)
+		return -ENOMEM;
+
+	DPRINTK("%s\n", nodename);
+
+	err = xenbus_probe_node(&xenbus_backend, type, nodename);
+	kfree(nodename);
+	return err;
+}
+
+/* backend/<typename>/<frontend-domid> */
+static int xenbus_probe_backend(const char *type, const char *domid)
+{
+	char *nodename;
+	int err = 0;
+	char **dir;
+	unsigned int i, dir_n = 0;
+
+	DPRINTK("");
+
+	nodename = kasprintf(GFP_KERNEL, "%s/%s/%s", xenbus_backend.root, type, domid);
+	if (!nodename)
+		return -ENOMEM;
+
+	dir = xenbus_directory(XBT_NIL, nodename, "", &dir_n);
+	if (IS_ERR(dir)) {
+		kfree(nodename);
+		return PTR_ERR(dir);
+	}
+
+	for (i = 0; i < dir_n; i++) {
+		err = xenbus_probe_backend_unit(nodename, type, dir[i]);
+		if (err)
+			break;
+	}
+	kfree(dir);
+	kfree(nodename);
+	return err;
+}
+
+static void backend_changed(struct xenbus_watch *watch,
+			    const char **vec, unsigned int len)
+{
+	DPRINTK("");
+
+	dev_changed(vec[XS_WATCH_PATH], &xenbus_backend);
+}
+
+static struct xenbus_watch be_watch = {
+	.node = "backend",
+	.callback = backend_changed,
+};
+
+void xenbus_backend_suspend(int (*fn)(struct device *, void *))
+{
+	DPRINTK("");
+	bus_for_each_dev(&xenbus_backend.bus, NULL, NULL, fn);
+}
+
+void xenbus_backend_resume(int (*fn)(struct device *, void *))
+{
+	DPRINTK("");
+	bus_for_each_dev(&xenbus_backend.bus, NULL, NULL, fn);
+}
+
+void xenbus_backend_probe_and_watch(void)
+{
+	xenbus_probe_devices(&xenbus_backend);
+	register_xenbus_watch(&be_watch);
+}
+
+int xenbus_backend_bus_register(void)
+{
+	return bus_register(&xenbus_backend.bus);
+}
+
+void xenbus_backend_bus_unregister(void)
+{
+	bus_unregister(&xenbus_backend.bus);
+}
===================================================================
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_xs.c
@@ -0,0 +1,840 @@
+/******************************************************************************
+ * xenbus_xs.c
+ *
+ * This is the kernel equivalent of the "xs" library.  We don't need everything
+ * and we use xenbus_comms for communication.
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include <linux/unistd.h>
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/uio.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/err.h>
+#include <linux/slab.h>
+#include <linux/fcntl.h>
+#include <linux/kthread.h>
+#include <linux/rwsem.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <xen/xenbus.h>
+#include "xenbus_comms.h"
+
+struct xs_stored_msg {
+	struct list_head list;
+
+	struct xsd_sockmsg hdr;
+
+	union {
+		/* Queued replies. */
+		struct {
+			char *body;
+		} reply;
+
+		/* Queued watch events. */
+		struct {
+			struct xenbus_watch *handle;
+			char **vec;
+			unsigned int vec_size;
+		} watch;
+	} u;
+};
+
+struct xs_handle {
+	/* A list of replies. Currently only one will ever be outstanding. */
+	struct list_head reply_list;
+	spinlock_t reply_lock;
+	wait_queue_head_t reply_waitq;
+
+	/* One request at a time. */
+	struct mutex request_mutex;
+
+	/* Protect transactions against save/restore. */
+	struct rw_semaphore suspend_mutex;
+};
+
+static struct xs_handle xs_state;
+
+/* List of registered watches, and a lock to protect it. */
+static LIST_HEAD(watches);
+static DEFINE_SPINLOCK(watches_lock);
+
+/* List of pending watch callback events, and a lock to protect it. */
+static LIST_HEAD(watch_events);
+static DEFINE_SPINLOCK(watch_events_lock);
+
+/*
+ * Details of the xenwatch callback kernel thread. The thread waits on the
+ * watch_events_waitq for work to do (queued on watch_events list). When it
+ * wakes up it acquires the xenwatch_mutex before reading the list and
+ * carrying out work.
+ */
+static pid_t xenwatch_pid;
+static DEFINE_MUTEX(xenwatch_mutex);
+static DECLARE_WAIT_QUEUE_HEAD(watch_events_waitq);
+
+static int get_error(const char *errorstring)
+{
+	unsigned int i;
+
+	for (i = 0; strcmp(errorstring, xsd_errors[i].errstring) != 0; i++) {
+		if (i == ARRAY_SIZE(xsd_errors) - 1) {
+			printk(KERN_WARNING
+			       "XENBUS xen store gave: unknown error %s",
+			       errorstring);
+			return EINVAL;
+		}
+	}
+	return xsd_errors[i].errnum;
+}
+
+static void *read_reply(enum xsd_sockmsg_type *type, unsigned int *len)
+{
+	struct xs_stored_msg *msg;
+	char *body;
+
+	spin_lock(&xs_state.reply_lock);
+
+	while (list_empty(&xs_state.reply_list)) {
+		spin_unlock(&xs_state.reply_lock);
+		/* XXX FIXME: Avoid synchronous wait for response here. */
+		wait_event(xs_state.reply_waitq,
+			   !list_empty(&xs_state.reply_list));
+		spin_lock(&xs_state.reply_lock);
+	}
+
+	msg = list_entry(xs_state.reply_list.next,
+			 struct xs_stored_msg, list);
+	list_del(&msg->list);
+
+	spin_unlock(&xs_state.reply_lock);
+
+	*type = msg->hdr.type;
+	if (len)
+		*len = msg->hdr.len;
+	body = msg->u.reply.body;
+
+	kfree(msg);
+
+	return body;
+}
+
+/* Emergency write. */
+void xenbus_debug_write(const char *str, unsigned int count)
+{
+	struct xsd_sockmsg msg = { 0 };
+
+	msg.type = XS_DEBUG;
+	msg.len = sizeof("print") + count + 1;
+
+	mutex_lock(&xs_state.request_mutex);
+	xb_write(&msg, sizeof(msg));
+	xb_write("print", sizeof("print"));
+	xb_write(str, count);
+	xb_write("", 1);
+	mutex_unlock(&xs_state.request_mutex);
+}
+
+void *xenbus_dev_request_and_reply(struct xsd_sockmsg *msg)
+{
+	void *ret;
+	struct xsd_sockmsg req_msg = *msg;
+	int err;
+
+	if (req_msg.type == XS_TRANSACTION_START)
+		down_read(&xs_state.suspend_mutex);
+
+	mutex_lock(&xs_state.request_mutex);
+
+	err = xb_write(msg, sizeof(*msg) + msg->len);
+	if (err) {
+		msg->type = XS_ERROR;
+		ret = ERR_PTR(err);
+	} else
+		ret = read_reply(&msg->type, &msg->len);
+
+	mutex_unlock(&xs_state.request_mutex);
+
+	if ((msg->type == XS_TRANSACTION_END) ||
+	    ((req_msg.type == XS_TRANSACTION_START) &&
+	     (msg->type == XS_ERROR)))
+		up_read(&xs_state.suspend_mutex);
+
+	return ret;
+}
+
+/* Send message to xs, get kmalloc'ed reply.  ERR_PTR() on error. */
+static void *xs_talkv(struct xenbus_transaction t,
+		      enum xsd_sockmsg_type type,
+		      const struct kvec *iovec,
+		      unsigned int num_vecs,
+		      unsigned int *len)
+{
+	struct xsd_sockmsg msg;
+	void *ret = NULL;
+	unsigned int i;
+	int err;
+
+	msg.tx_id = t.id;
+	msg.req_id = 0;
+	msg.type = type;
+	msg.len = 0;
+	for (i = 0; i < num_vecs; i++)
+		msg.len += iovec[i].iov_len;
+
+	mutex_lock(&xs_state.request_mutex);
+
+	err = xb_write(&msg, sizeof(msg));
+	if (err) {
+		mutex_unlock(&xs_state.request_mutex);
+		return ERR_PTR(err);
+	}
+
+	for (i = 0; i < num_vecs; i++) {
+		err = xb_write(iovec[i].iov_base, iovec[i].iov_len);;
+		if (err) {
+			mutex_unlock(&xs_state.request_mutex);
+			return ERR_PTR(err);
+		}
+	}
+
+	ret = read_reply(&msg.type, len);
+
+	mutex_unlock(&xs_state.request_mutex);
+
+	if (IS_ERR(ret))
+		return ret;
+
+	if (msg.type == XS_ERROR) {
+		err = get_error(ret);
+		kfree(ret);
+		return ERR_PTR(-err);
+	}
+
+	if (msg.type != type) {
+		if (printk_ratelimit())
+			printk(KERN_WARNING
+			       "XENBUS unexpected type [%d], expected [%d]\n",
+			       msg.type, type);
+		kfree(ret);
+		return ERR_PTR(-EINVAL);
+	}
+	return ret;
+}
+
+/* Simplified version of xs_talkv: single message. */
+static void *xs_single(struct xenbus_transaction t,
+		       enum xsd_sockmsg_type type,
+		       const char *string,
+		       unsigned int *len)
+{
+	struct kvec iovec;
+
+	iovec.iov_base = (void *)string;
+	iovec.iov_len = strlen(string) + 1;
+	return xs_talkv(t, type, &iovec, 1, len);
+}
+
+/* Many commands only need an ack, don't care what it says. */
+static int xs_error(char *reply)
+{
+	if (IS_ERR(reply))
+		return PTR_ERR(reply);
+	kfree(reply);
+	return 0;
+}
+
+static unsigned int count_strings(const char *strings, unsigned int len)
+{
+	unsigned int num;
+	const char *p;
+
+	for (p = strings, num = 0; p < strings + len; p += strlen(p) + 1)
+		num++;
+
+	return num;
+}
+
+/* Return the path to dir with /name appended. Buffer must be kfree()'ed. */
+static char *join(const char *dir, const char *name)
+{
+	char *buffer;
+
+	if (strlen(name) == 0)
+		buffer = kasprintf(GFP_KERNEL, "%s", dir);
+	else
+		buffer = kasprintf(GFP_KERNEL, "%s/%s", dir, name);
+	return (!buffer) ? ERR_PTR(-ENOMEM) : buffer;
+}
+
+static char **split(char *strings, unsigned int len, unsigned int *num)
+{
+	char *p, **ret;
+
+	/* Count the strings. */
+	*num = count_strings(strings, len);
+
+	/* Transfer to one big alloc for easy freeing. */
+	ret = kmalloc(*num * sizeof(char *) + len, GFP_KERNEL);
+	if (!ret) {
+		kfree(strings);
+		return ERR_PTR(-ENOMEM);
+	}
+	memcpy(&ret[*num], strings, len);
+	kfree(strings);
+
+	strings = (char *)&ret[*num];
+	for (p = strings, *num = 0; p < strings + len; p += strlen(p) + 1)
+		ret[(*num)++] = p;
+
+	return ret;
+}
+
+char **xenbus_directory(struct xenbus_transaction t,
+			const char *dir, const char *node, unsigned int *num)
+{
+	char *strings, *path;
+	unsigned int len;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return (char **)path;
+
+	strings = xs_single(t, XS_DIRECTORY, path, &len);
+	kfree(path);
+	if (IS_ERR(strings))
+		return (char **)strings;
+
+	return split(strings, len, num);
+}
+EXPORT_SYMBOL_GPL(xenbus_directory);
+
+/* Check if a path exists. Return 1 if it does. */
+int xenbus_exists(struct xenbus_transaction t,
+		  const char *dir, const char *node)
+{
+	char **d;
+	int dir_n;
+
+	d = xenbus_directory(t, dir, node, &dir_n);
+	if (IS_ERR(d))
+		return 0;
+	kfree(d);
+	return 1;
+}
+EXPORT_SYMBOL_GPL(xenbus_exists);
+
+/* Get the value of a single file.
+ * Returns a kmalloced value: call free() on it after use.
+ * len indicates length in bytes.
+ */
+void *xenbus_read(struct xenbus_transaction t,
+		  const char *dir, const char *node, unsigned int *len)
+{
+	char *path;
+	void *ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return (void *)path;
+
+	ret = xs_single(t, XS_READ, path, len);
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(xenbus_read);
+
+/* Write the value of a single file.
+ * Returns -err on failure.
+ */
+int xenbus_write(struct xenbus_transaction t,
+		 const char *dir, const char *node, const char *string)
+{
+	const char *path;
+	struct kvec iovec[2];
+	int ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+
+	iovec[0].iov_base = (void *)path;
+	iovec[0].iov_len = strlen(path) + 1;
+	iovec[1].iov_base = (void *)string;
+	iovec[1].iov_len = strlen(string);
+
+	ret = xs_error(xs_talkv(t, XS_WRITE, iovec, ARRAY_SIZE(iovec), NULL));
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(xenbus_write);
+
+/* Create a new directory. */
+int xenbus_mkdir(struct xenbus_transaction t,
+		 const char *dir, const char *node)
+{
+	char *path;
+	int ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+
+	ret = xs_error(xs_single(t, XS_MKDIR, path, NULL));
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(xenbus_mkdir);
+
+/* Destroy a file or directory (directories must be empty). */
+int xenbus_rm(struct xenbus_transaction t, const char *dir, const char *node)
+{
+	char *path;
+	int ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+
+	ret = xs_error(xs_single(t, XS_RM, path, NULL));
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(xenbus_rm);
+
+/* Start a transaction: changes by others will not be seen during this
+ * transaction, and changes will not be visible to others until end.
+ */
+int xenbus_transaction_start(struct xenbus_transaction *t)
+{
+	char *id_str;
+
+	down_read(&xs_state.suspend_mutex);
+
+	id_str = xs_single(XBT_NIL, XS_TRANSACTION_START, "", NULL);
+	if (IS_ERR(id_str)) {
+		up_read(&xs_state.suspend_mutex);
+		return PTR_ERR(id_str);
+	}
+
+	t->id = simple_strtoul(id_str, NULL, 0);
+	kfree(id_str);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xenbus_transaction_start);
+
+/* End a transaction.
+ * If abandon is true, transaction is discarded instead of committed.
+ */
+int xenbus_transaction_end(struct xenbus_transaction t, int abort)
+{
+	char abortstr[2];
+	int err;
+
+	if (abort)
+		strcpy(abortstr, "F");
+	else
+		strcpy(abortstr, "T");
+
+	err = xs_error(xs_single(t, XS_TRANSACTION_END, abortstr, NULL));
+
+	up_read(&xs_state.suspend_mutex);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(xenbus_transaction_end);
+
+/* Single read and scanf: returns -errno or num scanned. */
+int xenbus_scanf(struct xenbus_transaction t,
+		 const char *dir, const char *node, const char *fmt, ...)
+{
+	va_list ap;
+	int ret;
+	char *val;
+
+	val = xenbus_read(t, dir, node, NULL);
+	if (IS_ERR(val))
+		return PTR_ERR(val);
+
+	va_start(ap, fmt);
+	ret = vsscanf(val, fmt, ap);
+	va_end(ap);
+	kfree(val);
+	/* Distinctive errno. */
+	if (ret == 0)
+		return -ERANGE;
+	return ret;
+}
+EXPORT_SYMBOL_GPL(xenbus_scanf);
+
+/* Single printf and write: returns -errno or 0. */
+int xenbus_printf(struct xenbus_transaction t,
+		  const char *dir, const char *node, const char *fmt, ...)
+{
+	va_list ap;
+	int ret;
+#define PRINTF_BUFFER_SIZE 4096
+	char *printf_buffer;
+
+	printf_buffer = kmalloc(PRINTF_BUFFER_SIZE, GFP_KERNEL);
+	if (printf_buffer == NULL)
+		return -ENOMEM;
+
+	va_start(ap, fmt);
+	ret = vsnprintf(printf_buffer, PRINTF_BUFFER_SIZE, fmt, ap);
+	va_end(ap);
+
+	BUG_ON(ret > PRINTF_BUFFER_SIZE-1);
+	ret = xenbus_write(t, dir, node, printf_buffer);
+
+	kfree(printf_buffer);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(xenbus_printf);
+
+/* Takes tuples of names, scanf-style args, and void **, NULL terminated. */
+int xenbus_gather(struct xenbus_transaction t, const char *dir, ...)
+{
+	va_list ap;
+	const char *name;
+	int ret = 0;
+
+	va_start(ap, dir);
+	while (ret == 0 && (name = va_arg(ap, char *)) != NULL) {
+		const char *fmt = va_arg(ap, char *);
+		void *result = va_arg(ap, void *);
+		char *p;
+
+		p = xenbus_read(t, dir, name, NULL);
+		if (IS_ERR(p)) {
+			ret = PTR_ERR(p);
+			break;
+		}
+		if (fmt) {
+			if (sscanf(p, fmt, result) == 0)
+				ret = -EINVAL;
+			kfree(p);
+		} else
+			*(char **)result = p;
+	}
+	va_end(ap);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(xenbus_gather);
+
+static int xs_watch(const char *path, const char *token)
+{
+	struct kvec iov[2];
+
+	iov[0].iov_base = (void *)path;
+	iov[0].iov_len = strlen(path) + 1;
+	iov[1].iov_base = (void *)token;
+	iov[1].iov_len = strlen(token) + 1;
+
+	return xs_error(xs_talkv(XBT_NIL, XS_WATCH, iov,
+				 ARRAY_SIZE(iov), NULL));
+}
+
+static int xs_unwatch(const char *path, const char *token)
+{
+	struct kvec iov[2];
+
+	iov[0].iov_base = (char *)path;
+	iov[0].iov_len = strlen(path) + 1;
+	iov[1].iov_base = (char *)token;
+	iov[1].iov_len = strlen(token) + 1;
+
+	return xs_error(xs_talkv(XBT_NIL, XS_UNWATCH, iov,
+				 ARRAY_SIZE(iov), NULL));
+}
+
+static struct xenbus_watch *find_watch(const char *token)
+{
+	struct xenbus_watch *i, *cmp;
+
+	cmp = (void *)simple_strtoul(token, NULL, 16);
+
+	list_for_each_entry(i, &watches, list)
+		if (i == cmp)
+			return i;
+
+	return NULL;
+}
+
+/* Register callback to watch this node. */
+int register_xenbus_watch(struct xenbus_watch *watch)
+{
+	/* Pointer in ascii is the token. */
+	char token[sizeof(watch) * 2 + 1];
+	int err;
+
+	sprintf(token, "%lX", (long)watch);
+
+	down_read(&xs_state.suspend_mutex);
+
+	spin_lock(&watches_lock);
+	BUG_ON(find_watch(token));
+	list_add(&watch->list, &watches);
+	spin_unlock(&watches_lock);
+
+	err = xs_watch(watch->node, token);
+
+	/* Ignore errors due to multiple registration. */
+	if ((err != 0) && (err != -EEXIST)) {
+		spin_lock(&watches_lock);
+		list_del(&watch->list);
+		spin_unlock(&watches_lock);
+	}
+
+	up_read(&xs_state.suspend_mutex);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(register_xenbus_watch);
+
+void unregister_xenbus_watch(struct xenbus_watch *watch)
+{
+	struct xs_stored_msg *msg, *tmp;
+	char token[sizeof(watch) * 2 + 1];
+	int err;
+
+	sprintf(token, "%lX", (long)watch);
+
+	down_read(&xs_state.suspend_mutex);
+
+	spin_lock(&watches_lock);
+	BUG_ON(!find_watch(token));
+	list_del(&watch->list);
+	spin_unlock(&watches_lock);
+
+	err = xs_unwatch(watch->node, token);
+	if (err)
+		printk(KERN_WARNING
+		       "XENBUS Failed to release watch %s: %i\n",
+		       watch->node, err);
+
+	up_read(&xs_state.suspend_mutex);
+
+	/* Make sure there are no callbacks running currently (unless
+	   its us) */
+	if (current->pid != xenwatch_pid)
+		mutex_lock(&xenwatch_mutex);
+
+	/* Cancel pending watch events. */
+	spin_lock(&watch_events_lock);
+	list_for_each_entry_safe(msg, tmp, &watch_events, list) {
+		if (msg->u.watch.handle != watch)
+			continue;
+		list_del(&msg->list);
+		kfree(msg->u.watch.vec);
+		kfree(msg);
+	}
+	spin_unlock(&watch_events_lock);
+
+	if (current->pid != xenwatch_pid)
+		mutex_unlock(&xenwatch_mutex);
+}
+EXPORT_SYMBOL_GPL(unregister_xenbus_watch);
+
+void xs_suspend(void)
+{
+	struct xenbus_watch *watch;
+	char token[sizeof(watch) * 2 + 1];
+
+	down_write(&xs_state.suspend_mutex);
+
+	/* No need for watches_lock: the suspend_mutex is sufficient. */
+	list_for_each_entry(watch, &watches, list) {
+		sprintf(token, "%lX", (long)watch);
+		xs_unwatch(watch->node, token);
+	}
+
+	mutex_lock(&xs_state.request_mutex);
+}
+
+void xs_resume(void)
+{
+	struct xenbus_watch *watch;
+	char token[sizeof(watch) * 2 + 1];
+
+	mutex_unlock(&xs_state.request_mutex);
+
+	/* No need for watches_lock: the suspend_mutex is sufficient. */
+	list_for_each_entry(watch, &watches, list) {
+		sprintf(token, "%lX", (long)watch);
+		xs_watch(watch->node, token);
+	}
+
+	up_write(&xs_state.suspend_mutex);
+}
+
+static int xenwatch_thread(void *unused)
+{
+	struct list_head *ent;
+	struct xs_stored_msg *msg;
+
+	for (;;) {
+		wait_event_interruptible(watch_events_waitq,
+					 !list_empty(&watch_events));
+
+		if (kthread_should_stop())
+			break;
+
+		mutex_lock(&xenwatch_mutex);
+
+		spin_lock(&watch_events_lock);
+		ent = watch_events.next;
+		if (ent != &watch_events)
+			list_del(ent);
+		spin_unlock(&watch_events_lock);
+
+		if (ent != &watch_events) {
+			msg = list_entry(ent, struct xs_stored_msg, list);
+			msg->u.watch.handle->callback(
+				msg->u.watch.handle,
+				(const char **)msg->u.watch.vec,
+				msg->u.watch.vec_size);
+			kfree(msg->u.watch.vec);
+			kfree(msg);
+		}
+
+		mutex_unlock(&xenwatch_mutex);
+	}
+
+	return 0;
+}
+
+static int process_msg(void)
+{
+	struct xs_stored_msg *msg;
+	char *body;
+	int err;
+
+	msg = kmalloc(sizeof(*msg), GFP_KERNEL);
+	if (msg == NULL)
+		return -ENOMEM;
+
+	err = xb_read(&msg->hdr, sizeof(msg->hdr));
+	if (err) {
+		kfree(msg);
+		return err;
+	}
+
+	body = kmalloc(msg->hdr.len + 1, GFP_KERNEL);
+	if (body == NULL) {
+		kfree(msg);
+		return -ENOMEM;
+	}
+
+	err = xb_read(body, msg->hdr.len);
+	if (err) {
+		kfree(body);
+		kfree(msg);
+		return err;
+	}
+	body[msg->hdr.len] = '\0';
+
+	if (msg->hdr.type == XS_WATCH_EVENT) {
+		msg->u.watch.vec = split(body, msg->hdr.len,
+					 &msg->u.watch.vec_size);
+		if (IS_ERR(msg->u.watch.vec)) {
+			kfree(msg);
+			return PTR_ERR(msg->u.watch.vec);
+		}
+
+		spin_lock(&watches_lock);
+		msg->u.watch.handle = find_watch(
+			msg->u.watch.vec[XS_WATCH_TOKEN]);
+		if (msg->u.watch.handle != NULL) {
+			spin_lock(&watch_events_lock);
+			list_add_tail(&msg->list, &watch_events);
+			wake_up(&watch_events_waitq);
+			spin_unlock(&watch_events_lock);
+		} else {
+			kfree(msg->u.watch.vec);
+			kfree(msg);
+		}
+		spin_unlock(&watches_lock);
+	} else {
+		msg->u.reply.body = body;
+		spin_lock(&xs_state.reply_lock);
+		list_add_tail(&msg->list, &xs_state.reply_list);
+		spin_unlock(&xs_state.reply_lock);
+		wake_up(&xs_state.reply_waitq);
+	}
+
+	return 0;
+}
+
+static int xenbus_thread(void *unused)
+{
+	int err;
+
+	for (;;) {
+		err = process_msg();
+		if (err)
+			printk(KERN_WARNING "XENBUS error %d while reading "
+			       "message\n", err);
+		if (kthread_should_stop())
+			break;
+	}
+
+	return 0;
+}
+
+int xs_init(void)
+{
+	int err;
+	struct task_struct *task;
+
+	INIT_LIST_HEAD(&xs_state.reply_list);
+	spin_lock_init(&xs_state.reply_lock);
+	init_waitqueue_head(&xs_state.reply_waitq);
+
+	mutex_init(&xs_state.request_mutex);
+	init_rwsem(&xs_state.suspend_mutex);
+
+	/* Initialize the shared memory rings to talk to xenstored */
+	err = xb_init_comms();
+	if (err)
+		return err;
+
+	task = kthread_run(xenwatch_thread, NULL, "xenwatch");
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	xenwatch_pid = task->pid;
+
+	task = kthread_run(xenbus_thread, NULL, "xenbus");
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+
+	return 0;
+}
===================================================================
--- /dev/null
+++ b/include/xen/xenbus.h
@@ -0,0 +1,208 @@
+/******************************************************************************
+ * xenbus.h
+ *
+ * Talks to Xen Store to figure out what devices we have.
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * Copyright (C) 2005 XenSource Ltd.
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#ifndef _XEN_XENBUS_H
+#define _XEN_XENBUS_H
+
+#include <linux/device.h>
+#include <linux/notifier.h>
+#include <linux/mutex.h>
+#include <linux/completion.h>
+#include <linux/init.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/grant_table.h>
+#include <xen/interface/io/xenbus.h>
+#include <xen/interface/io/xs_wire.h>
+
+/* Register callback to watch this node. */
+struct xenbus_watch
+{
+	struct list_head list;
+
+	/* Path being watched. */
+	const char *node;
+
+	/* Callback (executed in a process context with no locks held). */
+	void (*callback)(struct xenbus_watch *,
+			 const char **vec, unsigned int len);
+};
+
+
+/* A xenbus device. */
+struct xenbus_device {
+	const char *devicetype;
+	const char *nodename;
+	const char *otherend;
+	int otherend_id;
+	struct xenbus_watch otherend_watch;
+	struct device dev;
+	enum xenbus_state state;
+	struct completion down;
+};
+
+static inline struct xenbus_device *to_xenbus_device(struct device *dev)
+{
+	return container_of(dev, struct xenbus_device, dev);
+}
+
+struct xenbus_device_id
+{
+	/* .../device/<device_type>/<identifier> */
+	char devicetype[32]; 	/* General class of device. */
+};
+
+/* A xenbus driver. */
+struct xenbus_driver {
+	char *name;
+	struct module *owner;
+	const struct xenbus_device_id *ids;
+	int (*probe)(struct xenbus_device *dev,
+		     const struct xenbus_device_id *id);
+	void (*otherend_changed)(struct xenbus_device *dev,
+				 enum xenbus_state backend_state);
+	int (*remove)(struct xenbus_device *dev);
+	int (*suspend)(struct xenbus_device *dev);
+	int (*resume)(struct xenbus_device *dev);
+	int (*uevent)(struct xenbus_device *, char **, int, char *, int);
+	struct device_driver driver;
+	int (*read_otherend_details)(struct xenbus_device *dev);
+};
+
+static inline struct xenbus_driver *to_xenbus_driver(struct device_driver *drv)
+{
+	return container_of(drv, struct xenbus_driver, driver);
+}
+
+int xenbus_register_frontend(struct xenbus_driver *drv);
+int xenbus_register_backend(struct xenbus_driver *drv);
+void xenbus_unregister_driver(struct xenbus_driver *drv);
+
+struct xenbus_transaction
+{
+	u32 id;
+};
+
+/* Nil transaction ID. */
+#define XBT_NIL ((struct xenbus_transaction) { 0 })
+
+int __init xenbus_dev_init(void);
+
+char **xenbus_directory(struct xenbus_transaction t,
+			const char *dir, const char *node, unsigned int *num);
+void *xenbus_read(struct xenbus_transaction t,
+		  const char *dir, const char *node, unsigned int *len);
+int xenbus_write(struct xenbus_transaction t,
+		 const char *dir, const char *node, const char *string);
+int xenbus_mkdir(struct xenbus_transaction t,
+		 const char *dir, const char *node);
+int xenbus_exists(struct xenbus_transaction t,
+		  const char *dir, const char *node);
+int xenbus_rm(struct xenbus_transaction t, const char *dir, const char *node);
+int xenbus_transaction_start(struct xenbus_transaction *t);
+int xenbus_transaction_end(struct xenbus_transaction t, int abort);
+
+/* Single read and scanf: returns -errno or num scanned if > 0. */
+int xenbus_scanf(struct xenbus_transaction t,
+		 const char *dir, const char *node, const char *fmt, ...)
+	__attribute__((format(scanf, 4, 5)));
+
+/* Single printf and write: returns -errno or 0. */
+int xenbus_printf(struct xenbus_transaction t,
+		  const char *dir, const char *node, const char *fmt, ...)
+	__attribute__((format(printf, 4, 5)));
+
+/* Generic read function: NULL-terminated triples of name,
+ * sprintf-style type string, and pointer. Returns 0 or errno.*/
+int xenbus_gather(struct xenbus_transaction t, const char *dir, ...);
+
+/* notifer routines for when the xenstore comes up */
+extern int xenstored_ready;
+int register_xenstore_notifier(struct notifier_block *nb);
+void unregister_xenstore_notifier(struct notifier_block *nb);
+
+int register_xenbus_watch(struct xenbus_watch *watch);
+void unregister_xenbus_watch(struct xenbus_watch *watch);
+void xs_suspend(void);
+void xs_resume(void);
+
+/* Used by xenbus_dev to borrow kernel's store connection. */
+void *xenbus_dev_request_and_reply(struct xsd_sockmsg *msg);
+
+/* Called from xen core code. */
+void xenbus_suspend(void);
+void xenbus_resume(void);
+void xenbus_probe(struct work_struct *);
+
+#define XENBUS_IS_ERR_READ(str) ({			\
+	if (!IS_ERR(str) && strlen(str) == 0) {		\
+		kfree(str);				\
+		str = ERR_PTR(-ERANGE);			\
+	}						\
+	IS_ERR(str);					\
+})
+
+#define XENBUS_EXIST_ERR(err) ((err) == -ENOENT || (err) == -ERANGE)
+
+int xenbus_watch_path(struct xenbus_device *dev, const char *path,
+		      struct xenbus_watch *watch,
+		      void (*callback)(struct xenbus_watch *,
+				       const char **, unsigned int));
+int xenbus_watch_path2(struct xenbus_device *dev, const char *path,
+		       const char *path2, struct xenbus_watch *watch,
+		       void (*callback)(struct xenbus_watch *,
+					const char **, unsigned int));
+int xenbus_switch_state(struct xenbus_device *dev, enum xenbus_state new_state);
+int xenbus_grant_ring(struct xenbus_device *dev, unsigned long ring_mfn);
+int xenbus_map_ring_valloc(struct xenbus_device *dev,
+			   int gnt_ref, void **vaddr);
+int xenbus_map_ring(struct xenbus_device *dev, int gnt_ref,
+			   grant_handle_t *handle, void *vaddr);
+
+int xenbus_unmap_ring_vfree(struct xenbus_device *dev, void *vaddr);
+int xenbus_unmap_ring(struct xenbus_device *dev,
+		      grant_handle_t handle, void *vaddr);
+
+int xenbus_alloc_evtchn(struct xenbus_device *dev, int *port);
+int xenbus_bind_evtchn(struct xenbus_device *dev, int remote_port, int *port);
+int xenbus_free_evtchn(struct xenbus_device *dev, int port);
+
+enum xenbus_state xenbus_read_driver_state(const char *path);
+
+void xenbus_dev_error(struct xenbus_device *dev, int err, const char *fmt, ...);
+void xenbus_dev_fatal(struct xenbus_device *dev, int err, const char *fmt, ...);
+
+char *xenbus_strstate(enum xenbus_state state);
+int xenbus_dev_is_online(struct xenbus_device *dev);
+int xenbus_frontend_closed(struct xenbus_device *dev);
+
+#endif /* _XEN_XENBUS_H */

-- 

Re: [patch 19/20] XEN-paravirt: Add the Xenbus sysfs and virtual device hotplug driver.

[Pavel Machek]

Hi!

> This communicates with the machine control software via a registry
> residing in a controlling virtual machine. This allows dynamic
> creation, destruction and modification of virtual device
> configurations (network devices, block devices and CPUS, to name some
> examples).
> 
> Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
> Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
> Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
> ---
>  drivers/xen/Makefile               |    1
>  drivers/xen/xenbus/Makefile        |    7
>  drivers/xen/xenbus/xenbus_client.c |  528 ++++++++++++++++++++
>  drivers/xen/xenbus/xenbus_comms.c  |  222 ++++++++
>  drivers/xen/xenbus/xenbus_comms.h  |   43 +
>  drivers/xen/xenbus/xenbus_probe.c  |  954 +++++++++++++++++++++++++++++++++++++
>  drivers/xen/xenbus/xenbus_xs.c     |  828 ++++++++++++++++++++++++++++++++

Hmm, you have word 'xen' three times in this pathname, and then the
file is name 'xs'. Trying to do xen advertising in path names? :-)

> @@ -0,0 +1,565 @@
> +/******************************************************************************
> + * Client-facing interface for the Xenbus driver.  In other words, the
> + * interface between the Xenbus and the device-specific code, be it the
> + * frontend or the backend of that driver.
> + *
> + * Copyright (C) 2005 XenSource Ltd
> + * 
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public License version 2
> + * as published by the Free Software Foundation; or, when distributed
> + * separately from the Linux kernel or incorporated into other
> + * software packages, subject to the following license:
> + * 
> + * Permission is hereby granted, free of charge, to any person obtaining a copy
> + * of this source file (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.

So this is dual BSD/GPL?

> +#include <linux/types.h>
> +#include <xen/interface/xen.h>
> +#include <xen/interface/event_channel.h>
> +#include "../../../arch/i386/paravirt-xen/events.h"
> +#include <xen/grant_table.h>
> +#include <xen/xenbus.h>
> +#include <xen/driver_util.h>
> +
> +char *xenbus_strstate(enum xenbus_state state)
> +{
> +	static char *name[] = {
> +		[ XenbusStateUnknown      ] = "Unknown",

Constants are normally XENBUS_STATE_UNKNOWN in linux...

Can we get description of its kernel-user interface in Documentation/
somewhere?


> +#include <asm/io.h>
> +#include <asm/page.h>
> +//#include <asm/maddr.h>
> +#include <asm/pgtable.h>
> +#include <asm/hypervisor.h>
> +#include <xen/xenbus.h>
> +//#include <xen/xen_proc.h>
> +//#include <xen/evtchn.h>
> +#include <xen/features.h>
> +//#include <xen/hvm.h>
> +

Just delete the lines.

> +#ifdef HAVE_XEN_PLATFORM_COMPAT_H
> +#include <xen/platform-compat.h>
> +#endif

Is this really neccessary?

> +	.bus = {
> +		.name     = "xen-backend",
> +		.match    = xenbus_match,
> +		.probe    = xenbus_dev_probe,
> +		.remove   = xenbus_dev_remove,
> +//		.shutdown = xenbus_dev_shutdown,

Delete.

> --- /dev/null
> +++ b/drivers/xen/xenbus/xenbus_xs.c
> @@ -0,0 +1,840 @@
> +/******************************************************************************
> + * xenbus_xs.c
> + *
> + * This is the kernel equivalent of the "xs" library.  We don't need everything
> + * and we use xenbus_comms for communication.

One line explanation what xs is would be nice.

						Pavel
-- 
Thanks for all the (sleeping) penguins.

[patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Jeremy Fitzhardinge]

[-- Attachment #1: xen-blockfront.patch --]
[-- Type: text/plain, Size: 37121 bytes --]

The block device frontend driver allows the kernel to access block
devices exported exported by a virtual machine containing a physical
block device driver.

Signed-off-by: Ian Pratt <ian.pratt@xensource.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
---
 drivers/block/Kconfig        |    1
 drivers/block/Makefile       |    1
 drivers/block/xen/Kconfig    |   14
 drivers/block/xen/Makefile   |    5
 drivers/block/xen/blkfront.c |  812 +++++++++++++++++++++++++++++++++++++++++++ drivers/block/xen/block.h    |  155 ++++++++
 drivers/block/xen/vbd.c      |  214 +++++++++++
 7 files changed, 1202 insertions(+)

===================================================================
--- a/drivers/block/Kconfig
+++ b/drivers/block/Kconfig
@@ -461,6 +461,7 @@ config CDROM_PKTCDVD_WCACHE
 	  don't do deferred write error handling yet.
 
 source "drivers/s390/block/Kconfig"
+source "drivers/block/xen/Kconfig"
 
 config ATA_OVER_ETH
 	tristate "ATA over Ethernet support"
===================================================================
--- a/drivers/block/Makefile
+++ b/drivers/block/Makefile
@@ -30,3 +30,4 @@ obj-$(CONFIG_BLK_DEV_SX8)	+= sx8.o
 obj-$(CONFIG_BLK_DEV_SX8)	+= sx8.o
 obj-$(CONFIG_BLK_DEV_UB)	+= ub.o
 
+obj-$(CONFIG_XEN)		+= xen/
===================================================================
--- a/include/linux/major.h
+++ b/include/linux/major.h
@@ -156,6 +156,8 @@
 #define VXSPEC_MAJOR		200	/* VERITAS volume config driver */
 #define VXDMP_MAJOR		201	/* VERITAS volume multipath driver */
 
+#define XENVBD_MAJOR		202	/* Xen virtual block device */
+
 #define MSR_MAJOR		202
 #define CPUID_MAJOR		203
 
===================================================================
--- /dev/null
+++ b/drivers/block/xen/Kconfig
@@ -0,0 +1,14 @@
+menu "Xen block device drivers"
+        depends on XEN
+
+config XEN_BLKDEV_FRONTEND
+	tristate "Block device frontend driver"
+	depends on XEN
+	default y
+	help
+	  The block device frontend driver allows the kernel to access block
+	  devices exported from a device driver virtual machine. Unless you
+	  are building a dedicated device driver virtual machine, then you
+	  almost certainly want to say Y here.
+
+endmenu
===================================================================
--- /dev/null
+++ b/drivers/block/xen/Makefile
@@ -0,0 +1,5 @@
+
+obj-$(CONFIG_XEN_BLKDEV_FRONTEND)	:= xenblk.o
+
+xenblk-objs := blkfront.o vbd.o
+
===================================================================
--- /dev/null
+++ b/drivers/block/xen/blkfront.c
@@ -0,0 +1,870 @@
+/******************************************************************************
+ * blkfront.c
+ * 
+ * XenLinux virtual block device driver.
+ * 
+ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
+ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
+ * Copyright (c) 2004, Christian Limpach
+ * Copyright (c) 2004, Andrew Warfield
+ * Copyright (c) 2005, Christopher Clark
+ * Copyright (c) 2005, XenSource Ltd
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include <linux/version.h>
+#include "block.h"
+#include <linux/cdrom.h>
+#include <linux/sched.h>
+#include <linux/interrupt.h>
+#include <scsi/scsi.h>
+#include <xen/xenbus.h>
+#include <xen/interface/grant_table.h>
+#include <xen/grant_table.h>
+#include "../../../arch/i386/paravirt-xen/events.h"
+#include "../../../arch/i386/paravirt-xen/xen-page.h"
+#include <asm/hypervisor.h>
+
+#define BLKIF_STATE_DISCONNECTED 0
+#define BLKIF_STATE_CONNECTED    1
+#define BLKIF_STATE_SUSPENDED    2
+
+#define MAXIMUM_OUTSTANDING_BLOCK_REQS \
+    (BLKIF_MAX_SEGMENTS_PER_REQUEST * BLK_RING_SIZE)
+#define GRANT_INVALID_REF	0
+
+static void connect(struct blkfront_info *);
+static void blkfront_closing(struct xenbus_device *);
+static int blkfront_remove(struct xenbus_device *);
+static int talk_to_backend(struct xenbus_device *, struct blkfront_info *);
+static int setup_blkring(struct xenbus_device *, struct blkfront_info *);
+
+static void kick_pending_request_queues(struct blkfront_info *);
+
+static irqreturn_t blkif_int(int irq, void *dev_id);
+static void blkif_restart_queue(struct work_struct *work);
+static void blkif_recover(struct blkfront_info *);
+static void blkif_completion(struct blk_shadow *);
+static void blkif_free(struct blkfront_info *, int);
+
+
+/**
+ * Entry point to this code when a new device is created.  Allocate the basic
+ * structures and the ring buffer for communication with the backend, and
+ * inform the backend of the appropriate details for those.  Switch to
+ * Initialised state.
+ */
+static int blkfront_probe(struct xenbus_device *dev,
+			  const struct xenbus_device_id *id)
+{
+	int err, vdevice, i;
+	struct blkfront_info *info;
+
+	/* FIXME: Use dynamic device id if this is not set. */
+	err = xenbus_scanf(XBT_NIL, dev->nodename,
+			   "virtual-device", "%i", &vdevice);
+	if (err != 1) {
+		xenbus_dev_fatal(dev, err, "reading virtual-device");
+		return err;
+	}
+
+	info = kzalloc(sizeof(*info), GFP_KERNEL);
+	if (!info) {
+		xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
+		return -ENOMEM;
+	}
+
+	info->xbdev = dev;
+	info->vdevice = vdevice;
+	info->connected = BLKIF_STATE_DISCONNECTED;
+	INIT_WORK(&info->work, blkif_restart_queue);
+
+	for (i = 0; i < BLK_RING_SIZE; i++)
+		info->shadow[i].req.id = i+1;
+	info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
+
+	/* Front end dir is a number, which is used as the id. */
+	info->handle = simple_strtoul(strrchr(dev->nodename,'/')+1, NULL, 0);
+	dev->dev.driver_data = info;
+
+	err = talk_to_backend(dev, info);
+	if (err) {
+		kfree(info);
+		dev->dev.driver_data = NULL;
+		return err;
+	}
+
+	return 0;
+}
+
+
+/**
+ * We are reconnecting to the backend, due to a suspend/resume, or a backend
+ * driver restart.  We tear down our blkif structure and recreate it, but
+ * leave the device-layer structures intact so that this is transparent to the
+ * rest of the kernel.
+ */
+static int blkfront_resume(struct xenbus_device *dev)
+{
+	struct blkfront_info *info = dev->dev.driver_data;
+	int err;
+
+	DPRINTK("blkfront_resume: %s\n", dev->nodename);
+
+	blkif_free(info, info->connected == BLKIF_STATE_CONNECTED);
+
+	err = talk_to_backend(dev, info);
+	if (info->connected == BLKIF_STATE_SUSPENDED && !err)
+		blkif_recover(info);
+
+	return err;
+}
+
+
+/* Common code used when first setting up, and when resuming. */
+static int talk_to_backend(struct xenbus_device *dev,
+			   struct blkfront_info *info)
+{
+	const char *message = NULL;
+	struct xenbus_transaction xbt;
+	int err;
+
+	/* Create shared ring, alloc event channel. */
+	err = setup_blkring(dev, info);
+	if (err)
+		goto out;
+
+again:
+	err = xenbus_transaction_start(&xbt);
+	if (err) {
+		xenbus_dev_fatal(dev, err, "starting transaction");
+		goto destroy_blkring;
+	}
+
+	err = xenbus_printf(xbt, dev->nodename,
+			    "ring-ref","%u", info->ring_ref);
+	if (err) {
+		message = "writing ring-ref";
+		goto abort_transaction;
+	}
+	err = xenbus_printf(xbt, dev->nodename,
+			    "event-channel", "%u", info->evtchn);
+	if (err) {
+		message = "writing event-channel";
+		goto abort_transaction;
+	}
+
+	err = xenbus_transaction_end(xbt, 0);
+	if (err) {
+		if (err == -EAGAIN)
+			goto again;
+		xenbus_dev_fatal(dev, err, "completing transaction");
+		goto destroy_blkring;
+	}
+
+	xenbus_switch_state(dev, XenbusStateInitialised);
+
+	return 0;
+
+ abort_transaction:
+	xenbus_transaction_end(xbt, 1);
+	if (message)
+		xenbus_dev_fatal(dev, err, "%s", message);
+ destroy_blkring:
+	blkif_free(info, 0);
+ out:
+	return err;
+}
+
+
+static int setup_blkring(struct xenbus_device *dev,
+			 struct blkfront_info *info)
+{
+	struct blkif_sring *sring;
+	int err;
+
+	info->ring_ref = GRANT_INVALID_REF;
+
+	sring = (struct blkif_sring *)__get_free_page(GFP_KERNEL);
+	if (!sring) {
+		xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring");
+		return -ENOMEM;
+	}
+	SHARED_RING_INIT(sring);
+	FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
+
+	err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
+	if (err < 0) {
+		free_page((unsigned long)sring);
+		info->ring.sring = NULL;
+		goto fail;
+	}
+	info->ring_ref = err;
+
+	err = xenbus_alloc_evtchn(dev, &info->evtchn);
+	if (err)
+		goto fail;
+
+	err = bind_evtchn_to_irqhandler(
+		info->evtchn, blkif_int, SA_SAMPLE_RANDOM, "blkif", info);
+	if (err <= 0) {
+		xenbus_dev_fatal(dev, err,
+				 "bind_evtchn_to_irqhandler failed");
+		goto fail;
+	}
+	info->irq = err;
+
+	return 0;
+fail:
+	blkif_free(info, 0);
+	return err;
+}
+
+
+/**
+ * Callback received when the backend's state changes.
+ */
+static void backend_changed(struct xenbus_device *dev,
+			    enum xenbus_state backend_state)
+{
+	struct blkfront_info *info = dev->dev.driver_data;
+	struct block_device *bd;
+
+	DPRINTK("blkfront:backend_changed.\n");
+
+	switch (backend_state) {
+	case XenbusStateInitialising:
+	case XenbusStateInitWait:
+	case XenbusStateInitialised:
+	case XenbusStateUnknown:
+	case XenbusStateClosed:
+		break;
+
+	case XenbusStateConnected:
+		connect(info);
+		break;
+
+	case XenbusStateClosing:
+		bd = bdget(info->dev);
+		if (bd == NULL)
+			xenbus_dev_fatal(dev, -ENODEV, "bdget failed");
+
+		mutex_lock(&bd->bd_mutex);
+		if (info->users > 0)
+			xenbus_dev_error(dev, -EBUSY,
+					 "Device in use; refusing to close");
+		else
+			blkfront_closing(dev);
+		mutex_unlock(&bd->bd_mutex);
+		bdput(bd);
+		break;
+	}
+}
+
+
+/* ** Connection ** */
+
+
+/*
+ * Invoked when the backend is finally 'ready' (and has told produced
+ * the details about the physical device - #sectors, size, etc).
+ */
+static void connect(struct blkfront_info *info)
+{
+	unsigned long long sectors;
+	unsigned long sector_size;
+	unsigned int binfo;
+	int err;
+
+	if ((info->connected == BLKIF_STATE_CONNECTED) ||
+	    (info->connected == BLKIF_STATE_SUSPENDED) )
+		return;
+
+	DPRINTK("blkfront.c:connect:%s.\n", info->xbdev->otherend);
+
+	err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
+			    "sectors", "%llu", &sectors,
+			    "info", "%u", &binfo,
+			    "sector-size", "%lu", &sector_size,
+			    NULL);
+	if (err) {
+		xenbus_dev_fatal(info->xbdev, err,
+				 "reading backend fields at %s",
+				 info->xbdev->otherend);
+		return;
+	}
+
+	err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
+			    "feature-barrier", "%lu", &info->feature_barrier,
+			    NULL);
+	if (err)
+		info->feature_barrier = 0;
+
+	err = xlvbd_add(sectors, info->vdevice, binfo, sector_size, info);
+	if (err) {
+		xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
+				 info->xbdev->otherend);
+		return;
+	}
+
+	(void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
+
+	/* Kick pending requests. */
+	spin_lock_irq(&blkif_io_lock);
+	info->connected = BLKIF_STATE_CONNECTED;
+	kick_pending_request_queues(info);
+	spin_unlock_irq(&blkif_io_lock);
+
+	add_disk(info->gd);
+}
+
+/**
+ * Handle the change of state of the backend to Closing.  We must delete our
+ * device-layer structures now, to ensure that writes are flushed through to
+ * the backend.  Once is this done, we can switch to Closed in
+ * acknowledgement.
+ */
+static void blkfront_closing(struct xenbus_device *dev)
+{
+	struct blkfront_info *info = dev->dev.driver_data;
+	unsigned long flags;
+
+	DPRINTK("blkfront_closing: %s removed\n", dev->nodename);
+
+	if (info->rq == NULL)
+		goto out;
+
+	spin_lock_irqsave(&blkif_io_lock, flags);
+	/* No more blkif_request(). */
+	blk_stop_queue(info->rq);
+	/* No more gnttab callback work. */
+	gnttab_cancel_free_callback(&info->callback);
+	spin_unlock_irqrestore(&blkif_io_lock, flags);
+
+	/* Flush gnttab callback work. Must be done with no locks held. */
+	flush_scheduled_work();
+
+	xlvbd_del(info);
+
+ out:
+	xenbus_frontend_closed(dev);
+}
+
+
+static int blkfront_remove(struct xenbus_device *dev)
+{
+	struct blkfront_info *info = dev->dev.driver_data;
+
+	DPRINTK("blkfront_remove: %s removed\n", dev->nodename);
+
+	blkif_free(info, 0);
+
+	kfree(info);
+
+	return 0;
+}
+
+
+static inline int GET_ID_FROM_FREELIST(
+	struct blkfront_info *info)
+{
+	unsigned long free = info->shadow_free;
+	BUG_ON(free > BLK_RING_SIZE);
+	info->shadow_free = info->shadow[free].req.id;
+	info->shadow[free].req.id = 0x0fffffee; /* debug */
+	return free;
+}
+
+static inline void ADD_ID_TO_FREELIST(
+	struct blkfront_info *info, unsigned long id)
+{
+	info->shadow[id].req.id  = info->shadow_free;
+	info->shadow[id].request = 0;
+	info->shadow_free = id;
+}
+
+static inline void flush_requests(struct blkfront_info *info)
+{
+	int notify;
+
+	RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&info->ring, notify);
+
+	if (notify)
+		notify_remote_via_irq(info->irq);
+}
+
+static void kick_pending_request_queues(struct blkfront_info *info)
+{
+	if (!RING_FULL(&info->ring)) {
+		/* Re-enable calldowns. */
+		blk_start_queue(info->rq);
+		/* Kick things off immediately. */
+		do_blkif_request(info->rq);
+	}
+}
+
+static void blkif_restart_queue(struct work_struct *work)
+{
+	struct blkfront_info *info = container_of(work, struct blkfront_info, work);
+
+	spin_lock_irq(&blkif_io_lock);
+	if (info->connected == BLKIF_STATE_CONNECTED)
+		kick_pending_request_queues(info);
+	spin_unlock_irq(&blkif_io_lock);
+}
+
+static void blkif_restart_queue_callback(void *arg)
+{
+	struct blkfront_info *info = (struct blkfront_info *)arg;
+	schedule_work(&info->work);
+}
+
+int blkif_open(struct inode *inode, struct file *filep)
+{
+	struct blkfront_info *info = inode->i_bdev->bd_disk->private_data;
+	info->users++;
+	return 0;
+}
+
+
+int blkif_release(struct inode *inode, struct file *filep)
+{
+	struct blkfront_info *info = inode->i_bdev->bd_disk->private_data;
+	info->users--;
+	if (info->users == 0) {
+		/* Check whether we have been instructed to close.  We will
+		   have ignored this request initially, as the device was
+		   still mounted. */
+		struct xenbus_device * dev = info->xbdev;
+		enum xenbus_state state = xenbus_read_driver_state(dev->otherend);
+
+		if (state == XenbusStateClosing)
+			blkfront_closing(dev);
+	}
+	return 0;
+}
+
+
+int blkif_ioctl(struct inode *inode, struct file *filep,
+		unsigned command, unsigned long argument)
+{
+	int i;
+
+	DPRINTK_IOCTL("command: 0x%x, argument: 0x%lx, dev: 0x%04x\n",
+		      command, (long)argument, inode->i_rdev);
+
+	switch (command) {
+	case CDROMMULTISESSION:
+		DPRINTK("FIXME: support multisession CDs later\n");
+		for (i = 0; i < sizeof(struct cdrom_multisession); i++)
+			if (put_user(0, (char __user *)(argument + i)))
+				return -EFAULT;
+		return 0;
+
+	default:
+		/*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
+		  command);*/
+		return -EINVAL; /* same return as native Linux */
+	}
+
+	return 0;
+}
+
+
+int blkif_getgeo(struct block_device *bd, struct hd_geometry *hg)
+{
+	/* We don't have real geometry info, but let's at least return
+	   values consistent with the size of the device */
+	sector_t nsect = get_capacity(bd->bd_disk);
+	sector_t cylinders = nsect;
+
+	hg->heads = 0xff;
+	hg->sectors = 0x3f;
+	sector_div(cylinders, hg->heads * hg->sectors);
+	hg->cylinders = cylinders;
+	if ((sector_t)(hg->cylinders + 1) * hg->heads * hg->sectors < nsect)
+		hg->cylinders = 0xffff;
+	return 0;
+}
+
+/*
+ * blkif_queue_request
+ *
+ * request block io
+ *
+ * id: for guest use only.
+ * operation: BLKIF_OP_{READ,WRITE,PROBE}
+ * buffer: buffer to read/write into. this should be a
+ *   virtual address in the guest os.
+ */
+static int blkif_queue_request(struct request *req)
+{
+	struct blkfront_info *info = req->rq_disk->private_data;
+	unsigned long buffer_mfn;
+	struct blkif_request *ring_req;
+	struct bio *bio;
+	struct bio_vec *bvec;
+	int idx;
+	unsigned long id;
+	unsigned int fsect, lsect;
+	int ref;
+	grant_ref_t gref_head;
+
+	if (unlikely(info->connected != BLKIF_STATE_CONNECTED))
+		return 1;
+
+	if (gnttab_alloc_grant_references(
+		BLKIF_MAX_SEGMENTS_PER_REQUEST, &gref_head) < 0) {
+		gnttab_request_free_callback(
+			&info->callback,
+			blkif_restart_queue_callback,
+			info,
+			BLKIF_MAX_SEGMENTS_PER_REQUEST);
+		return 1;
+	}
+
+	/* Fill out a communications ring structure. */
+	ring_req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
+	id = GET_ID_FROM_FREELIST(info);
+	info->shadow[id].request = (unsigned long)req;
+
+	ring_req->id = id;
+	ring_req->sector_number = (blkif_sector_t)req->sector;
+	ring_req->handle = info->handle;
+
+	ring_req->operation = rq_data_dir(req) ?
+		BLKIF_OP_WRITE : BLKIF_OP_READ;
+	if (blk_barrier_rq(req))
+		ring_req->operation = BLKIF_OP_WRITE_BARRIER;
+
+	ring_req->nr_segments = 0;
+	rq_for_each_bio (bio, req) {
+		bio_for_each_segment (bvec, bio, idx) {
+			BUG_ON(ring_req->nr_segments
+			       == BLKIF_MAX_SEGMENTS_PER_REQUEST);
+			buffer_mfn = pfn_to_mfn(page_to_pfn(bvec->bv_page));
+			fsect = bvec->bv_offset >> 9;
+			lsect = fsect + (bvec->bv_len >> 9) - 1;
+			/* install a grant reference. */
+			ref = gnttab_claim_grant_reference(&gref_head);
+			BUG_ON(ref == -ENOSPC);
+
+			gnttab_grant_foreign_access_ref(
+				ref,
+				info->xbdev->otherend_id,
+				buffer_mfn,
+				rq_data_dir(req) );
+
+			info->shadow[id].frame[ring_req->nr_segments] =
+				mfn_to_pfn(buffer_mfn);
+
+			ring_req->seg[ring_req->nr_segments] =
+				(struct blkif_request_segment) {
+					.gref       = ref,
+					.first_sect = fsect,
+					.last_sect  = lsect };
+
+			ring_req->nr_segments++;
+		}
+	}
+
+	info->ring.req_prod_pvt++;
+
+	/* Keep a private copy so we can reissue requests when recovering. */
+	info->shadow[id].req = *ring_req;
+
+	gnttab_free_grant_references(gref_head);
+
+	return 0;
+}
+
+/*
+ * do_blkif_request
+ *  read a block; request is in a request queue
+ */
+void do_blkif_request(request_queue_t *rq)
+{
+	struct blkfront_info *info = NULL;
+	struct request *req;
+	int queued;
+
+	DPRINTK("Entered do_blkif_request\n");
+
+	queued = 0;
+
+	while ((req = elv_next_request(rq)) != NULL) {
+		info = req->rq_disk->private_data;
+		if (!blk_fs_request(req)) {
+			end_request(req, 0);
+			continue;
+		}
+
+		if (RING_FULL(&info->ring))
+			goto wait;
+
+		DPRINTK("do_blk_req %p: cmd %p, sec %lx, "
+			"(%u/%li) buffer:%p [%s]\n",
+			req, req->cmd, (unsigned long)req->sector,
+			req->current_nr_sectors,
+			req->nr_sectors, req->buffer,
+			rq_data_dir(req) ? "write" : "read");
+
+
+		blkdev_dequeue_request(req);
+		if (blkif_queue_request(req)) {
+			blk_requeue_request(rq, req);
+		wait:
+			/* Avoid pointless unplugs. */
+			blk_stop_queue(rq);
+			break;
+		}
+
+		queued++;
+	}
+
+	if (queued != 0)
+		flush_requests(info);
+}
+
+
+static irqreturn_t blkif_int(int irq, void *dev_id)
+{
+	struct request *req;
+	struct blkif_response *bret;
+	RING_IDX i, rp;
+	unsigned long flags;
+	struct blkfront_info *info = (struct blkfront_info *)dev_id;
+	int uptodate;
+
+	spin_lock_irqsave(&blkif_io_lock, flags);
+
+	if (unlikely(info->connected != BLKIF_STATE_CONNECTED)) {
+		spin_unlock_irqrestore(&blkif_io_lock, flags);
+		return IRQ_HANDLED;
+	}
+
+ again:
+	rp = info->ring.sring->rsp_prod;
+	rmb(); /* Ensure we see queued responses up to 'rp'. */
+
+	for (i = info->ring.rsp_cons; i != rp; i++) {
+		unsigned long id;
+		int ret;
+
+		bret = RING_GET_RESPONSE(&info->ring, i);
+		id   = bret->id;
+		req  = (struct request *)info->shadow[id].request;
+
+		blkif_completion(&info->shadow[id]);
+
+		ADD_ID_TO_FREELIST(info, id);
+
+		uptodate = (bret->status == BLKIF_RSP_OKAY);
+		switch (bret->operation) {
+		case BLKIF_OP_WRITE_BARRIER:
+			if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
+				printk("blkfront: %s: write barrier op failed\n",
+				       info->gd->disk_name);
+				uptodate = -EOPNOTSUPP;
+				info->feature_barrier = 0;
+			        xlvbd_barrier(info);
+			}
+			/* fall through */
+		case BLKIF_OP_READ:
+		case BLKIF_OP_WRITE:
+			if (unlikely(bret->status != BLKIF_RSP_OKAY))
+				DPRINTK("Bad return from blkdev data "
+					"request: %x\n", bret->status);
+
+			ret = end_that_request_first(req, uptodate,
+				req->hard_nr_sectors);
+			BUG_ON(ret);
+			end_that_request_last(req, uptodate);
+			break;
+		default:
+			BUG();
+		}
+	}
+
+	info->ring.rsp_cons = i;
+
+	if (i != info->ring.req_prod_pvt) {
+		int more_to_do;
+		RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
+		if (more_to_do)
+			goto again;
+	} else
+		info->ring.sring->rsp_event = i + 1;
+
+	kick_pending_request_queues(info);
+
+	spin_unlock_irqrestore(&blkif_io_lock, flags);
+
+	return IRQ_HANDLED;
+}
+
+static void blkif_free(struct blkfront_info *info, int suspend)
+{
+	/* Prevent new requests being issued until we fix things up. */
+	spin_lock_irq(&blkif_io_lock);
+	info->connected = suspend ?
+		BLKIF_STATE_SUSPENDED : BLKIF_STATE_DISCONNECTED;
+	/* No more blkif_request(). */
+	if (info->rq)
+		blk_stop_queue(info->rq);
+	/* No more gnttab callback work. */
+	gnttab_cancel_free_callback(&info->callback);
+	spin_unlock_irq(&blkif_io_lock);
+
+	/* Flush gnttab callback work. Must be done with no locks held. */
+	flush_scheduled_work();
+
+	/* Free resources associated with old device channel. */
+	if (info->ring_ref != GRANT_INVALID_REF) {
+		gnttab_end_foreign_access(info->ring_ref, 0,
+					  (unsigned long)info->ring.sring);
+		info->ring_ref = GRANT_INVALID_REF;
+		info->ring.sring = NULL;
+	}
+	if (info->irq)
+		unbind_from_irqhandler(info->irq, info);
+	info->evtchn = info->irq = 0;
+
+}
+
+static void blkif_completion(struct blk_shadow *s)
+{
+	int i;
+	for (i = 0; i < s->req.nr_segments; i++)
+		gnttab_end_foreign_access(s->req.seg[i].gref, 0, 0UL);
+}
+
+static void blkif_recover(struct blkfront_info *info)
+{
+	int i;
+	struct blkif_request *req;
+	struct blk_shadow *copy;
+	int j;
+
+	/* Stage 1: Make a safe copy of the shadow state. */
+	copy = kmalloc(sizeof(info->shadow), GFP_KERNEL | __GFP_NOFAIL);
+	memcpy(copy, info->shadow, sizeof(info->shadow));
+
+	/* Stage 2: Set up free list. */
+	memset(&info->shadow, 0, sizeof(info->shadow));
+	for (i = 0; i < BLK_RING_SIZE; i++)
+		info->shadow[i].req.id = i+1;
+	info->shadow_free = info->ring.req_prod_pvt;
+	info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
+
+	/* Stage 3: Find pending requests and requeue them. */
+	for (i = 0; i < BLK_RING_SIZE; i++) {
+		/* Not in use? */
+		if (copy[i].request == 0)
+			continue;
+
+		/* Grab a request slot and copy shadow state into it. */
+		req = RING_GET_REQUEST(
+			&info->ring, info->ring.req_prod_pvt);
+		*req = copy[i].req;
+
+		/* We get a new request id, and must reset the shadow state. */
+		req->id = GET_ID_FROM_FREELIST(info);
+		memcpy(&info->shadow[req->id], &copy[i], sizeof(copy[i]));
+
+		/* Rewrite any grant references invalidated by susp/resume. */
+		for (j = 0; j < req->nr_segments; j++)
+			gnttab_grant_foreign_access_ref(
+				req->seg[j].gref,
+				info->xbdev->otherend_id,
+				pfn_to_mfn(info->shadow[req->id].frame[j]),
+				rq_data_dir(
+					(struct request *)
+					info->shadow[req->id].request));
+		info->shadow[req->id].req = *req;
+
+		info->ring.req_prod_pvt++;
+	}
+
+	kfree(copy);
+
+	(void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
+
+	spin_lock_irq(&blkif_io_lock);
+
+	/* Now safe for us to use the shared ring */
+	info->connected = BLKIF_STATE_CONNECTED;
+
+	/* Send off requeued requests */
+	flush_requests(info);
+
+	/* Kick any other new requests queued since we resumed */
+	kick_pending_request_queues(info);
+
+	spin_unlock_irq(&blkif_io_lock);
+}
+
+
+/* ** Driver Registration ** */
+
+
+static struct xenbus_device_id blkfront_ids[] = {
+	{ "vbd" },
+	{ "" }
+};
+
+
+static struct xenbus_driver blkfront = {
+	.name = "vbd",
+	.owner = THIS_MODULE,
+	.ids = blkfront_ids,
+	.probe = blkfront_probe,
+	.remove = blkfront_remove,
+	.resume = blkfront_resume,
+	.otherend_changed = backend_changed,
+};
+
+
+static int __init xlblk_init(void)
+{
+	if (!is_running_on_xen())
+		return -ENODEV;
+
+	if (xlvbd_alloc_major() < 0)
+		return -ENODEV;
+
+	return xenbus_register_frontend(&blkfront);
+}
+module_init(xlblk_init);
+
+
+static void xlblk_exit(void)
+{
+	return xenbus_unregister_driver(&blkfront);
+}
+module_exit(xlblk_exit);
+
+MODULE_LICENSE("Dual BSD/GPL");
===================================================================
--- /dev/null
+++ b/drivers/block/xen/block.h
@@ -0,0 +1,157 @@
+/******************************************************************************
+ * block.h
+ * 
+ * Shared definitions between all levels of XenLinux Virtual block devices.
+ * 
+ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
+ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
+ * Copyright (c) 2004-2005, Christian Limpach
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#ifndef __XEN_DRIVERS_BLOCK_H__
+#define __XEN_DRIVERS_BLOCK_H__
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include <linux/fs.h>
+#include <linux/hdreg.h>
+#include <linux/blkdev.h>
+#include <linux/major.h>
+#include <asm/hypervisor.h>
+#include <xen/xenbus.h>
+#include <xen/grant_table.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/io/blkif.h>
+#include <xen/interface/io/ring.h>
+#include <asm/io.h>
+#include <asm/atomic.h>
+#include <asm/uaccess.h>
+
+#if 1
+#define IPRINTK(fmt, args...)				\
+	printk(KERN_INFO "xen_blk: " fmt, ##args)
+#else
+#define IPRINTK(fmt, args...) ((void)0)
+#endif
+
+#if 1
+#define WPRINTK(fmt, args...)				\
+	printk(KERN_WARNING "xen_blk: " fmt, ##args)
+#else
+#define WPRINTK(fmt, args...) ((void)0)
+#endif
+
+#define DPRINTK(_f, _a...) pr_debug(_f, ## _a)
+
+#if 0
+#define DPRINTK_IOCTL(_f, _a...) printk(KERN_ALERT _f, ## _a)
+#else
+#define DPRINTK_IOCTL(_f, _a...) ((void)0)
+#endif
+
+struct xlbd_type_info
+{
+	int partn_shift;
+	int disks_per_major;
+	char *devname;
+	char *diskname;
+};
+
+struct xlbd_major_info
+{
+	int major;
+	int index;
+	int usage;
+	struct xlbd_type_info *type;
+};
+
+struct blk_shadow {
+	struct blkif_request req;
+	unsigned long request;
+	unsigned long frame[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+};
+
+#define BLK_RING_SIZE __RING_SIZE((struct blkif_sring *)0, PAGE_SIZE)
+
+/*
+ * We have one of these per vbd, whether ide, scsi or 'other'.  They
+ * hang in private_data off the gendisk structure. We may end up
+ * putting all kinds of interesting stuff here :-)
+ */
+struct blkfront_info
+{
+	struct xenbus_device *xbdev;
+	dev_t dev;
+ 	struct gendisk *gd;
+	int vdevice;
+	blkif_vdev_t handle;
+	int connected;
+	int ring_ref;
+	struct blkif_front_ring ring;
+	unsigned int evtchn, irq;
+	struct xlbd_major_info *mi;
+	request_queue_t *rq;
+	struct work_struct work;
+	struct gnttab_free_callback callback;
+	struct blk_shadow shadow[BLK_RING_SIZE];
+	unsigned long shadow_free;
+	int feature_barrier;
+
+	/**
+	 * The number of people holding this device open.  We won't allow a
+	 * hot-unplug unless this is 0.
+	 */
+	int users;
+};
+
+extern spinlock_t blkif_io_lock;
+
+extern int blkif_open(struct inode *inode, struct file *filep);
+extern int blkif_release(struct inode *inode, struct file *filep);
+extern int blkif_ioctl(struct inode *inode, struct file *filep,
+		       unsigned command, unsigned long argument);
+extern int blkif_getgeo(struct block_device *, struct hd_geometry *);
+extern int blkif_check(dev_t dev);
+extern int blkif_revalidate(dev_t dev);
+extern void do_blkif_request (request_queue_t *rq);
+
+/* Virtual block device subsystem. */
+int xlvbd_alloc_major(void);
+/* Note that xlvbd_add doesn't call add_disk for you: you're expected
+   to call add_disk on info->gd once the disk is properly connected
+   up. */
+int xlvbd_add(blkif_sector_t capacity, int device,
+	      u16 vdisk_info, u16 sector_size, struct blkfront_info *info);
+void xlvbd_del(struct blkfront_info *info);
+int xlvbd_barrier(struct blkfront_info *info);
+
+#endif /* __XEN_DRIVERS_BLOCK_H__ */
===================================================================
--- /dev/null
+++ b/drivers/block/xen/vbd.c
@@ -0,0 +1,231 @@
+/******************************************************************************
+ * vbd.c
+ * 
+ * XenLinux virtual block device driver (xvd).
+ * 
+ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
+ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
+ * Copyright (c) 2004-2005, Christian Limpach
+ * 
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#include "block.h"
+#include <linux/blkdev.h>
+#include <linux/list.h>
+
+#define BLKIF_MAJOR(dev) ((dev)>>8)
+#define BLKIF_MINOR(dev) ((dev) & 0xff)
+
+static struct xlbd_type_info xvd_type_info = {
+	.partn_shift = 4,
+	.disks_per_major = 16,
+	.devname = "xvd",
+	.diskname = "xvd"
+};
+
+static struct xlbd_major_info xvd_major_info = {
+	.major = XENVBD_MAJOR,
+	.type = &xvd_type_info
+};
+
+/* Information about our VBDs. */
+#define MAX_VBDS 64
+static LIST_HEAD(vbds_list);
+
+static struct block_device_operations xlvbd_block_fops =
+{
+	.owner = THIS_MODULE,
+	.open = blkif_open,
+	.release = blkif_release,
+	.ioctl  = blkif_ioctl,
+	.getgeo = blkif_getgeo
+};
+
+DEFINE_SPINLOCK(blkif_io_lock);
+
+int
+xlvbd_alloc_major(void)
+{
+	printk("Registering block device major %i\n", xvd_major_info.major);
+	if (register_blkdev(xvd_major_info.major,
+			    xvd_major_info.type->devname)) {
+		WPRINTK("can't get major %d with name %s\n",
+			xvd_major_info.major, xvd_major_info.type->devname);
+		return -1;
+	}
+	return 0;
+}
+
+static int
+xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
+{
+	request_queue_t *rq;
+
+	rq = blk_init_queue(do_blkif_request, &blkif_io_lock);
+	if (rq == NULL)
+		return -1;
+
+	elevator_init(rq, "noop");
+
+	/* Hard sector size and max sectors impersonate the equiv. hardware. */
+	blk_queue_hardsect_size(rq, sector_size);
+	blk_queue_max_sectors(rq, 512);
+
+	/* Each segment in a request is up to an aligned page in size. */
+	blk_queue_segment_boundary(rq, PAGE_SIZE - 1);
+	blk_queue_max_segment_size(rq, PAGE_SIZE);
+
+	/* Ensure a merged request will fit in a single I/O ring slot. */
+	blk_queue_max_phys_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
+	blk_queue_max_hw_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
+
+	/* Make sure buffer addresses are sector-aligned. */
+	blk_queue_dma_alignment(rq, 511);
+
+	gd->queue = rq;
+
+	return 0;
+}
+
+static int
+xlvbd_alloc_gendisk(int minor, blkif_sector_t capacity, int vdevice,
+		    u16 vdisk_info, u16 sector_size,
+		    struct blkfront_info *info)
+{
+	struct gendisk *gd;
+	struct xlbd_major_info *mi;
+	int nr_minors = 1;
+	int err = -ENODEV;
+
+	BUG_ON(info->gd != NULL);
+	BUG_ON(info->mi != NULL);
+	BUG_ON(info->rq != NULL);
+
+	mi = &xvd_major_info;
+	info->mi = mi;
+
+	if ((minor & ((1 << mi->type->partn_shift) - 1)) == 0)
+		nr_minors = 1 << mi->type->partn_shift;
+
+	gd = alloc_disk(nr_minors);
+	if (gd == NULL)
+		goto out;
+
+	if (nr_minors > 1)
+		sprintf(gd->disk_name, "%s%c", mi->type->diskname,
+			'a' + mi->index * mi->type->disks_per_major +
+			(minor >> mi->type->partn_shift));
+	else
+		sprintf(gd->disk_name, "%s%c%d", mi->type->diskname,
+			'a' + mi->index * mi->type->disks_per_major +
+			(minor >> mi->type->partn_shift),
+			minor & ((1 << mi->type->partn_shift) - 1));
+
+	gd->major = mi->major;
+	gd->first_minor = minor;
+	gd->fops = &xlvbd_block_fops;
+	gd->private_data = info;
+	gd->driverfs_dev = &(info->xbdev->dev);
+	set_capacity(gd, capacity);
+
+	if (xlvbd_init_blk_queue(gd, sector_size)) {
+		del_gendisk(gd);
+		goto out;
+	}
+
+	info->rq = gd->queue;
+	info->gd = gd;
+
+	if (info->feature_barrier)
+		xlvbd_barrier(info);
+
+	if (vdisk_info & VDISK_READONLY)
+		set_disk_ro(gd, 1);
+
+	if (vdisk_info & VDISK_REMOVABLE)
+		gd->flags |= GENHD_FL_REMOVABLE;
+
+	if (vdisk_info & VDISK_CDROM)
+		gd->flags |= GENHD_FL_CD;
+
+	return 0;
+
+ out:
+	info->mi = NULL;
+	return err;
+}
+
+int
+xlvbd_add(blkif_sector_t capacity, int vdevice, u16 vdisk_info,
+	  u16 sector_size, struct blkfront_info *info)
+{
+	struct block_device *bd;
+	int err = 0;
+
+	info->dev = MKDEV(BLKIF_MAJOR(vdevice), BLKIF_MINOR(vdevice));
+
+	bd = bdget(info->dev);
+	if (bd == NULL)
+		return -ENODEV;
+
+	err = xlvbd_alloc_gendisk(BLKIF_MINOR(vdevice), capacity, vdevice,
+				  vdisk_info, sector_size, info);
+
+	bdput(bd);
+	return err;
+}
+
+void
+xlvbd_del(struct blkfront_info *info)
+{
+	if (info->mi == NULL)
+		return;
+
+	BUG_ON(info->gd == NULL);
+	del_gendisk(info->gd);
+	put_disk(info->gd);
+	info->gd = NULL;
+
+	info->mi = NULL;
+
+	BUG_ON(info->rq == NULL);
+	blk_cleanup_queue(info->rq);
+	info->rq = NULL;
+}
+
+int
+xlvbd_barrier(struct blkfront_info *info)
+{
+	int err;
+
+	err = blk_queue_ordered(info->rq,
+		info->feature_barrier ? QUEUE_ORDERED_DRAIN : QUEUE_ORDERED_NONE, NULL);
+	if (err)
+		return err;
+	printk("blkfront: %s: barriers %s\n",
+	       info->gd->disk_name, info->feature_barrier ? "enabled" : "disabled");
+	return 0;
+}

-- 

Re: [patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Arjan van de Ven]

> +#include "../../../arch/i386/paravirt-xen/events.h"
> +#include "../../../arch/i386/paravirt-xen/xen-page.h"

this shows the headers are clearly in the wrong place...
> +
> +	err = xenbus_printf(xbt, dev->nodename,
> +			    "ring-ref","%u", info->ring_ref);

why do you need your own printf?

> +static inline int GET_ID_FROM_FREELIST(

does this really need screaming?


> +
> +int blkif_ioctl(struct inode *inode, struct file *filep,
> +		unsigned command, unsigned long argument)
> +{
> +	int i;
> +
> +	DPRINTK_IOCTL("command: 0x%x, argument: 0x%lx, dev: 0x%04x\n",
> +		      command, (long)argument, inode->i_rdev);
> +
> +	switch (command) {
> +	case CDROMMULTISESSION:
> +		DPRINTK("FIXME: support multisession CDs later\n");
> +		for (i = 0; i < sizeof(struct cdrom_multisession); i++)
> +			if (put_user(0, (char __user *)(argument + i)))
> +				return -EFAULT;
> +		return 0;
> +
> +	default:
> +		/*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
> +		  command);*/
> +		return -EINVAL; /* same return as native Linux */
> +	}

eh so you implement no ioctls.. why then implement the ioctl method at
all?


> +static struct xenbus_driver blkfront = {
> +	.name = "vbd",
> +	.owner = THIS_MODULE,
> +	.ids = blkfront_ids,
> +	.probe = blkfront_probe,
> +	.remove = blkfront_remove,
> +	.resume = blkfront_resume,
> +	.otherend_changed = backend_changed,
> +};

this can be const

> +
> +#define DPRINTK(_f, _a...) pr_debug(_f, ## _a)

why this silly abstraction? Just use pr_debug in the code directly

Re: [patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Greg KH]

On Sat, Jan 13, 2007 at 05:07:28PM -0800, Arjan van de Ven wrote:
> > +
> > +#define DPRINTK(_f, _a...) pr_debug(_f, ## _a)
> 
> why this silly abstraction? Just use pr_debug in the code directly

Actually, for drivers, like this one, you should use the dev_printk()
and friends (dev_dbg, dev_err, etc.) instead so that userspace knows
exactly which device and driver the message comes from.

thanks,

greg k-h

Re: [Xen-devel] Re: [patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Mark Williamson]

> > +
> > +	err = xenbus_printf(xbt, dev->nodename,
> > +			    "ring-ref","%u", info->ring_ref);
>
> why do you need your own printf?

xenbus_printf isn't a printf replacement - it is used for writing a formatted 
string into XenStore (which contains  VM configuration data in a 
human-readable form).

Internally it does a vsnprintf into a buffer and writes the resulting string 
to the XenStore.

Cheers,
Mark

> > +static inline int GET_ID_FROM_FREELIST(
>
> does this really need screaming?
>
> > +
> > +int blkif_ioctl(struct inode *inode, struct file *filep,
> > +		unsigned command, unsigned long argument)
> > +{
> > +	int i;
> > +
> > +	DPRINTK_IOCTL("command: 0x%x, argument: 0x%lx, dev: 0x%04x\n",
> > +		      command, (long)argument, inode->i_rdev);
> > +
> > +	switch (command) {
> > +	case CDROMMULTISESSION:
> > +		DPRINTK("FIXME: support multisession CDs later\n");
> > +		for (i = 0; i < sizeof(struct cdrom_multisession); i++)
> > +			if (put_user(0, (char __user *)(argument + i)))
> > +				return -EFAULT;
> > +		return 0;
> > +
> > +	default:
> > +		/*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
> > +		  command);*/
> > +		return -EINVAL; /* same return as native Linux */
> > +	}
>
> eh so you implement no ioctls.. why then implement the ioctl method at
> all?

I'm not familiar with this code...  but perhaps the (fake) multisession 
handling is to keep userspace that queries this happy?  I can't really think 
of anywhere this would apply off the top of my head, though.

Cheers,
Mark

> > +static struct xenbus_driver blkfront = {
> > +	.name = "vbd",
> > +	.owner = THIS_MODULE,
> > +	.ids = blkfront_ids,
> > +	.probe = blkfront_probe,
> > +	.remove = blkfront_remove,
> > +	.resume = blkfront_resume,
> > +	.otherend_changed = backend_changed,
> > +};
>
> this can be const
>
> > +
> > +#define DPRINTK(_f, _a...) pr_debug(_f, ## _a)
>
> why this silly abstraction? Just use pr_debug in the code directly
>
>
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel

-- 
Dave: Just a question. What use is a unicyle with no seat?  And no pedals!
Mark: To answer a question with a question: What use is a skateboard?
Dave: Skateboards have wheels.
Mark: My wheel has a wheel!

Re: [patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Jan Engelhardt]

On Jan 12 2007 17:45, Jeremy Fitzhardinge wrote:
>
>The block device frontend driver allows the kernel to access block
>devices exported exported by a virtual machine containing a physical
>block device driver.

Is this significantly different from ubd/hostfs that it actually warrants a
reinvention?

>+	(void)xenbus_switch_state(info->xbdev, XenbusStateConnected);

Cast remove, if xenbus_switch_state does not have __must_check.
Also elsewhere.


	-`J'
-- 

Re: [patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Muli Ben-Yehuda]

On Sun, Jan 14, 2007 at 12:05:13PM +0100, Jan Engelhardt wrote:

> >+	(void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
> 
> Cast remove, if xenbus_switch_state does not have __must_check.
> Also elsewhere.

Hmm, why? this way you know that the programmer isn't checking the
return value *on purpose*. More information is good.

Cheers,
Muli

Re: [patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Jan Engelhardt]

On Jan 14 2007 13:24, Muli Ben-Yehuda wrote:
>On Sun, Jan 14, 2007 at 12:05:13PM +0100, Jan Engelhardt wrote:
>
>> >+	(void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
>> 
>> Cast remove, if xenbus_switch_state does not have __must_check.
>> Also elsewhere.
>
>Hmm, why? this way you know that the programmer isn't checking the
>return value *on purpose*. More information is good.

Then I'd prefer a comment. Plus, you don't do (void) on every
xenbus_switch_state, so I have to assume (1) you either forgot to check the
return value or (2) you don't follow your own "on purpose" policy. If you
really want to enforce to use the return value and explicitly not in some
places, then please tag it __must_check. http://lkml.org/lkml/2006/12/21/282
might be something to read.


	-`J'
-- 

Re: [patch 20/20] XEN-paravirt: Add Xen virtual block device driver.

[Keir Fraser]

On 14/1/07 11:05 am, "Jan Engelhardt" <jengelh@linux01.gwdg.de> wrote:

>> The block device frontend driver allows the kernel to access block
>> devices exported exported by a virtual machine containing a physical
>> block device driver.
> 
> Is this significantly different from ubd/hostfs that it actually warrants a
> reinvention?

It is certainly unlike hostfs because hostfs provides file-level access to
host storage, not block-level. It's unlike both ubd and hostfs in that both
of those (I believe) make significant use of the syscall interface (and so
assume they run in a process on a Linux host). Also our driver appears to be
lower level, pushing more responsibility for features like CoW into the VMM.
Arguably that's a more generically reusable and flexible strategy although
it requires more VMM run-time support (which a Xen system provides).

>> + (void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
> 
> Cast remove, if xenbus_switch_state does not have __must_check.
> Also elsewhere.

Okay, we should certainly follow the general rule here.

 Thanks,
 Keir