rpm/xen
SHA256
1
0
Fork 0
forked from pool/xen
Code Pull Requests Activity

OBS-URL: https://build.opensuse.org/package/show/openSUSE:Factory/xen?expand=0&rev=69

Browse Source
This commit is contained in:
OBS User unknown 2009-02-09 01:49:08 +00:00 committed by Git OBS Bridge
parent 8a8c811c02
commit 5545adae8a
19 changed files with 2092 additions and 188 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
19072-vmx-pat.patch
View File

@ -1,7 +1,15 @@
Index: xen-3.3.1-testing/xen/arch/x86/hvm/vmx/vmcs.c # HG changeset patch
=================================================================== # User Keir Fraser <keir.fraser@citrix.com>
--- xen-3.3.1-testing.orig/xen/arch/x86/hvm/vmx/vmcs.c # Date 1232623303 0
+++ xen-3.3.1-testing/xen/arch/x86/hvm/vmx/vmcs.c # Node ID d52921c18c3d0171bccb4651cca8412f2fff2dd9
# Parent 9f9ba1a7cc924fbc547e05ea21071becafe5e2c2
vmx: utilise the GUEST_PAT and HOST_PAT vmcs area
Signed-off-by: Xin Li <Xin.Li@intel.com>
Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
--- a/xen/arch/x86/hvm/vmx/vmcs.c
+++ b/xen/arch/x86/hvm/vmx/vmcs.c
@@ -166,14 +166,15 @@ static void vmx_init_vmcs_config(void) @@ -166,14 +166,15 @@ static void vmx_init_vmcs_config(void)
#endif #endif
@ -94,10 +102,8 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/vmx/vmcs.c
printk("*** Control State ***\n"); printk("*** Control State ***\n");
printk("PinBased=%08x CPUBased=%08x SecondaryExec=%08x\n", printk("PinBased=%08x CPUBased=%08x SecondaryExec=%08x\n",
Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/vmx/vmcs.h --- a/xen/include/asm-x86/hvm/vmx/vmcs.h
=================================================================== +++ b/xen/include/asm-x86/hvm/vmx/vmcs.h
--- xen-3.3.1-testing.orig/xen/include/asm-x86/hvm/vmx/vmcs.h
+++ xen-3.3.1-testing/xen/include/asm-x86/hvm/vmx/vmcs.h
@@ -150,11 +150,14 @@ extern u32 vmx_pin_based_exec_control; @@ -150,11 +150,14 @@ extern u32 vmx_pin_based_exec_control;
#define VM_EXIT_IA32E_MODE 0x00000200 #define VM_EXIT_IA32E_MODE 0x00000200
@ -113,17 +119,16 @@ Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/vmx/vmcs.h
extern u32 vmx_vmentry_control; extern u32 vmx_vmentry_control;
#define SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES 0x00000001 #define SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES 0x00000001
@@ -181,7 +184,8 @@ extern bool_t cpu_has_vmx_ins_outs_instr @@ -181,6 +184,8 @@ extern bool_t cpu_has_vmx_ins_outs_instr
(vmx_secondary_exec_control & SECONDARY_EXEC_ENABLE_EPT) (vmx_secondary_exec_control & SECONDARY_EXEC_ENABLE_EPT)
#define cpu_has_vmx_vpid \ #define cpu_has_vmx_vpid \
(vmx_secondary_exec_control & SECONDARY_EXEC_ENABLE_VPID) (vmx_secondary_exec_control & SECONDARY_EXEC_ENABLE_VPID)
-
+#define cpu_has_vmx_pat \ +#define cpu_has_vmx_pat \
+ (vmx_vmentry_control & VM_ENTRY_LOAD_GUEST_PAT) + (vmx_vmentry_control & VM_ENTRY_LOAD_GUEST_PAT)
/* GUEST_INTERRUPTIBILITY_INFO flags. */ /* GUEST_INTERRUPTIBILITY_INFO flags. */
#define VMX_INTR_SHADOW_STI 0x00000001 #define VMX_INTR_SHADOW_STI 0x00000001
#define VMX_INTR_SHADOW_MOV_SS 0x00000002 @@ -232,6 +237,8 @@ enum vmcs_field {
@@ -232,6 +236,8 @@ enum vmcs_field {
VMCS_LINK_POINTER_HIGH = 0x00002801, VMCS_LINK_POINTER_HIGH = 0x00002801,
GUEST_IA32_DEBUGCTL = 0x00002802, GUEST_IA32_DEBUGCTL = 0x00002802,
GUEST_IA32_DEBUGCTL_HIGH = 0x00002803, GUEST_IA32_DEBUGCTL_HIGH = 0x00002803,
@ -132,7 +137,7 @@ Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/vmx/vmcs.h
GUEST_PDPTR0 = 0x0000280a, GUEST_PDPTR0 = 0x0000280a,
GUEST_PDPTR0_HIGH = 0x0000280b, GUEST_PDPTR0_HIGH = 0x0000280b,
GUEST_PDPTR1 = 0x0000280c, GUEST_PDPTR1 = 0x0000280c,
@@ -240,6 +246,8 @@ enum vmcs_field { @@ -240,6 +247,8 @@ enum vmcs_field {
GUEST_PDPTR2_HIGH = 0x0000280f, GUEST_PDPTR2_HIGH = 0x0000280f,
GUEST_PDPTR3 = 0x00002810, GUEST_PDPTR3 = 0x00002810,
GUEST_PDPTR3_HIGH = 0x00002811, GUEST_PDPTR3_HIGH = 0x00002811,

109
19079-snp_ctl-1.patch
View File

@ -1,7 +1,18 @@
Index: xen-3.3.1-testing/xen/arch/x86/hvm/mtrr.c # HG changeset patch
=================================================================== # User Keir Fraser <keir.fraser@citrix.com>
--- xen-3.3.1-testing.orig/xen/arch/x86/hvm/mtrr.c # Date 1232721749 0
+++ xen-3.3.1-testing/xen/arch/x86/hvm/mtrr.c # Node ID 175a425e9b55e63c240b0a2ad61f5ed251e85ead
# Parent f3240cd3cd2b9d48acf3d82caa2ca1cab1f66325
EPT/VT-d: Enhance MTRR/PAT virtualization when EPT/VT-d both enabled
Set effective memory type for EPT according to the VT-d snoop control
capability, and also includes some cleanups for EPT & VT-d both enabled.
Signed-off-by: Edwin Zhai <Edwin.Zhai@intel.com>
Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
--- a/xen/arch/x86/hvm/mtrr.c
+++ b/xen/arch/x86/hvm/mtrr.c
@@ -702,12 +702,15 @@ HVM_REGISTER_SAVE_RESTORE(MTRR, hvm_save @@ -702,12 +702,15 @@ HVM_REGISTER_SAVE_RESTORE(MTRR, hvm_save
1, HVMSR_PER_VCPU); 1, HVMSR_PER_VCPU);
@ -41,10 +52,8 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/mtrr.c
gmtrr_mtype = get_mtrr_type(&v->arch.hvm_vcpu.mtrr, (gfn << PAGE_SHIFT)); gmtrr_mtype = get_mtrr_type(&v->arch.hvm_vcpu.mtrr, (gfn << PAGE_SHIFT));
hmtrr_mtype = get_mtrr_type(&mtrr_state, (mfn << PAGE_SHIFT)); hmtrr_mtype = get_mtrr_type(&mtrr_state, (mfn << PAGE_SHIFT));
return ((gmtrr_mtype <= hmtrr_mtype) ? gmtrr_mtype : hmtrr_mtype); return ((gmtrr_mtype <= hmtrr_mtype) ? gmtrr_mtype : hmtrr_mtype);
Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c --- a/xen/arch/x86/mm/hap/p2m-ept.c
=================================================================== +++ b/xen/arch/x86/mm/hap/p2m-ept.c
--- xen-3.3.1-testing.orig/xen/arch/x86/mm/hap/p2m-ept.c
+++ xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
@@ -66,6 +66,7 @@ static int ept_set_middle_entry(struct d @@ -66,6 +66,7 @@ static int ept_set_middle_entry(struct d
list_add_tail(&pg->list, &d->arch.p2m->pages); list_add_tail(&pg->list, &d->arch.p2m->pages);
@ -53,20 +62,14 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
ept_entry->sp_avail = 0; ept_entry->sp_avail = 0;
ept_entry->avail1 = 0; ept_entry->avail1 = 0;
ept_entry->mfn = page_to_mfn(pg); ept_entry->mfn = page_to_mfn(pg);
@@ -114,9 +115,19 @@ static int ept_next_level(struct domain @@ -114,9 +115,13 @@ static int ept_next_level(struct domain
} }
} }
+/* The parameter need_modify_vtd_table in _ept_set_entry(): +/*
+ * + * TODO: ept_set_entry() computes 'need_modify_vtd_table' for itself,
+ * Usually VT-d page table is created during the p2m table creation time, + * by observing whether any gfn->mfn translations are modified.
+ * and it only cares about the gfn to mfn translations,
+ * EPT table takes the same effect as the p2m table, so we create
+ * VT-d page table in ept_set_entry() function.
+ * But when EPT entires are modified not related to the gfn and mfn translations * We don't need to recreat the same VT-d page tables.
+ * The need_modify_vtd_table parameter is used for this.
+ */ + */
+
static int static int
-ept_set_entry(struct domain *d, unsigned long gfn, mfn_t mfn, -ept_set_entry(struct domain *d, unsigned long gfn, mfn_t mfn,
- unsigned int order, p2m_type_t p2mt) - unsigned int order, p2m_type_t p2mt)
@ -75,7 +78,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
{ {
ept_entry_t *table = NULL; ept_entry_t *table = NULL;
unsigned long gfn_remainder = gfn, offset = 0; unsigned long gfn_remainder = gfn, offset = 0;
@@ -124,6 +135,8 @@ ept_set_entry(struct domain *d, unsigned @@ -124,6 +129,8 @@ ept_set_entry(struct domain *d, unsigned
u32 index; u32 index;
int i, rv = 0, ret = 0; int i, rv = 0, ret = 0;
int walk_level = order / EPT_TABLE_ORDER; int walk_level = order / EPT_TABLE_ORDER;
@ -84,7 +87,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
/* we only support 4k and 2m pages now */ /* we only support 4k and 2m pages now */
@@ -157,7 +170,9 @@ ept_set_entry(struct domain *d, unsigned @@ -157,7 +164,9 @@ ept_set_entry(struct domain *d, unsigned
{ {
if ( mfn_valid(mfn_x(mfn)) || (p2mt == p2m_mmio_direct) ) if ( mfn_valid(mfn_x(mfn)) || (p2mt == p2m_mmio_direct) )
{ {
@ -95,7 +98,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
ept_entry->sp_avail = walk_level ? 1 : 0; ept_entry->sp_avail = walk_level ? 1 : 0;
if ( ret == GUEST_TABLE_SUPER_PAGE ) if ( ret == GUEST_TABLE_SUPER_PAGE )
@@ -208,7 +223,10 @@ ept_set_entry(struct domain *d, unsigned @@ -208,7 +217,10 @@ ept_set_entry(struct domain *d, unsigned
{ {
split_ept_entry = split_table + i; split_ept_entry = split_table + i;
split_ept_entry->emt = epte_get_entry_emt(d, split_ept_entry->emt = epte_get_entry_emt(d,
@ -107,7 +110,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
split_ept_entry->sp_avail = 0; split_ept_entry->sp_avail = 0;
split_ept_entry->mfn = split_mfn+i; split_ept_entry->mfn = split_mfn+i;
@@ -223,7 +241,10 @@ ept_set_entry(struct domain *d, unsigned @@ -223,7 +235,10 @@ ept_set_entry(struct domain *d, unsigned
/* Set the destinated 4k page as normal */ /* Set the destinated 4k page as normal */
split_ept_entry = split_table + offset; split_ept_entry = split_table + offset;
@ -119,7 +122,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
split_ept_entry->mfn = mfn_x(mfn); split_ept_entry->mfn = mfn_x(mfn);
split_ept_entry->avail1 = p2mt; split_ept_entry->avail1 = p2mt;
ept_p2m_type_to_flags(split_ept_entry, p2mt); ept_p2m_type_to_flags(split_ept_entry, p2mt);
@@ -246,7 +267,8 @@ out: @@ -246,7 +261,8 @@ out:
/* Now the p2m table is not shared with vt-d page table */ /* Now the p2m table is not shared with vt-d page table */
@ -129,7 +132,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
{ {
if ( p2mt == p2m_ram_rw ) if ( p2mt == p2m_ram_rw )
{ {
@@ -273,6 +295,17 @@ out: @@ -273,6 +289,17 @@ out:
return rv; return rv;
} }
@ -147,7 +150,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
/* Read ept p2m entries */ /* Read ept p2m entries */
static mfn_t ept_get_entry(struct domain *d, unsigned long gfn, p2m_type_t *t) static mfn_t ept_get_entry(struct domain *d, unsigned long gfn, p2m_type_t *t)
{ {
@@ -393,18 +426,30 @@ void ept_change_entry_emt_with_range(str @@ -393,18 +420,30 @@ void ept_change_entry_emt_with_range(str
* Set emt for super page. * Set emt for super page.
*/ */
order = EPT_TABLE_ORDER; order = EPT_TABLE_ORDER;
@ -183,10 +186,8 @@ Index: xen-3.3.1-testing/xen/arch/x86/mm/hap/p2m-ept.c
} }
} }
Index: xen-3.3.1-testing/xen/drivers/passthrough/iommu.c --- a/xen/drivers/passthrough/iommu.c
=================================================================== +++ b/xen/drivers/passthrough/iommu.c
--- xen-3.3.1-testing.orig/xen/drivers/passthrough/iommu.c
+++ xen-3.3.1-testing/xen/drivers/passthrough/iommu.c
@@ -40,6 +40,7 @@ int iommu_enabled = 0; @@ -40,6 +40,7 @@ int iommu_enabled = 0;
int iommu_pv_enabled = 0; int iommu_pv_enabled = 0;
int force_iommu = 0; int force_iommu = 0;
@ -195,10 +196,8 @@ Index: xen-3.3.1-testing/xen/drivers/passthrough/iommu.c
static void __init parse_iommu_param(char *s) static void __init parse_iommu_param(char *s)
{ {
Index: xen-3.3.1-testing/xen/drivers/passthrough/vtd/dmar.c --- a/xen/drivers/passthrough/vtd/dmar.c
=================================================================== +++ b/xen/drivers/passthrough/vtd/dmar.c
--- xen-3.3.1-testing.orig/xen/drivers/passthrough/vtd/dmar.c
+++ xen-3.3.1-testing/xen/drivers/passthrough/vtd/dmar.c
@@ -29,6 +29,7 @@ @@ -29,6 +29,7 @@
#include <xen/pci_regs.h> #include <xen/pci_regs.h>
#include <asm/string.h> #include <asm/string.h>
@ -238,10 +237,8 @@ Index: xen-3.3.1-testing/xen/drivers/passthrough/vtd/dmar.c
return 0; return 0;
Index: xen-3.3.1-testing/xen/drivers/passthrough/vtd/iommu.c --- a/xen/drivers/passthrough/vtd/iommu.c
=================================================================== +++ b/xen/drivers/passthrough/vtd/iommu.c
--- xen-3.3.1-testing.orig/xen/drivers/passthrough/vtd/iommu.c
+++ xen-3.3.1-testing/xen/drivers/passthrough/vtd/iommu.c
@@ -1495,6 +1495,11 @@ int intel_iommu_map_page( @@ -1495,6 +1495,11 @@ int intel_iommu_map_page(
pte_present = dma_pte_present(*pte); pte_present = dma_pte_present(*pte);
dma_set_pte_addr(*pte, (paddr_t)mfn << PAGE_SHIFT_4K); dma_set_pte_addr(*pte, (paddr_t)mfn << PAGE_SHIFT_4K);
@ -254,10 +251,8 @@ Index: xen-3.3.1-testing/xen/drivers/passthrough/vtd/iommu.c
iommu_flush_cache_entry(pte); iommu_flush_cache_entry(pte);
unmap_vtd_domain_page(page); unmap_vtd_domain_page(page);
Index: xen-3.3.1-testing/xen/drivers/passthrough/vtd/iommu.h --- a/xen/drivers/passthrough/vtd/iommu.h
=================================================================== +++ b/xen/drivers/passthrough/vtd/iommu.h
--- xen-3.3.1-testing.orig/xen/drivers/passthrough/vtd/iommu.h
+++ xen-3.3.1-testing/xen/drivers/passthrough/vtd/iommu.h
@@ -104,6 +104,7 @@ @@ -104,6 +104,7 @@
#define ecap_ext_intr(e) ((e >> 4) & 0x1) #define ecap_ext_intr(e) ((e >> 4) & 0x1)
#define ecap_cache_hints(e) ((e >> 5) & 0x1) #define ecap_cache_hints(e) ((e >> 5) & 0x1)
@ -279,10 +274,8 @@ Index: xen-3.3.1-testing/xen/drivers/passthrough/vtd/iommu.h
#define dma_set_pte_prot(p, prot) \ #define dma_set_pte_prot(p, prot) \
do {(p).val = ((p).val & ~3) | ((prot) & 3); } while (0) do {(p).val = ((p).val & ~3) | ((prot) & 3); } while (0)
#define dma_pte_addr(p) ((p).val & PAGE_MASK_4K) #define dma_pte_addr(p) ((p).val & PAGE_MASK_4K)
Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/vmx/vmx.h --- a/xen/include/asm-x86/hvm/vmx/vmx.h
=================================================================== +++ b/xen/include/asm-x86/hvm/vmx/vmx.h
--- xen-3.3.1-testing.orig/xen/include/asm-x86/hvm/vmx/vmx.h
+++ xen-3.3.1-testing/xen/include/asm-x86/hvm/vmx/vmx.h
@@ -33,7 +33,8 @@ typedef union { @@ -33,7 +33,8 @@ typedef union {
u64 r : 1, u64 r : 1,
w : 1, w : 1,
@ -293,23 +286,25 @@ Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/vmx/vmx.h
sp_avail : 1, sp_avail : 1,
avail1 : 4, avail1 : 4,
mfn : 45, mfn : 45,
Index: xen-3.3.1-testing/xen/include/asm-x86/mtrr.h --- a/xen/include/asm-x86/mtrr.h
=================================================================== +++ b/xen/include/asm-x86/mtrr.h
--- xen-3.3.1-testing.orig/xen/include/asm-x86/mtrr.h @@ -64,9 +64,11 @@ extern int mtrr_del_page(int reg, unsign
+++ xen-3.3.1-testing/xen/include/asm-x86/mtrr.h
@@ -64,7 +64,7 @@ extern int mtrr_del_page(int reg, unsign
extern void mtrr_centaur_report_mcr(int mcr, u32 lo, u32 hi); extern void mtrr_centaur_report_mcr(int mcr, u32 lo, u32 hi);
extern u32 get_pat_flags(struct vcpu *v, u32 gl1e_flags, paddr_t gpaddr, extern u32 get_pat_flags(struct vcpu *v, u32 gl1e_flags, paddr_t gpaddr,
paddr_t spaddr); paddr_t spaddr);
-extern uint8_t epte_get_entry_emt(struct domain *d, unsigned long gfn, unsigned long mfn); -extern uint8_t epte_get_entry_emt(struct domain *d, unsigned long gfn, unsigned long mfn);
+extern uint8_t epte_get_entry_emt(struct domain *d, unsigned long gfn, unsigned long mfn, uint8_t *igmt, int direct_mmio); -extern void ept_change_entry_emt_with_range(struct domain *d, unsigned long start_gfn,
extern void ept_change_entry_emt_with_range(struct domain *d, unsigned long start_gfn, - unsigned long end_gfn);
unsigned long end_gfn); +extern uint8_t epte_get_entry_emt(
+ struct domain *d, unsigned long gfn, unsigned long mfn,
+ uint8_t *igmt, int direct_mmio);
+extern void ept_change_entry_emt_with_range(
+ struct domain *d, unsigned long start_gfn, unsigned long end_gfn);
extern unsigned char pat_type_2_pte_flags(unsigned char pat_type); extern unsigned char pat_type_2_pte_flags(unsigned char pat_type);
Index: xen-3.3.1-testing/xen/include/xen/iommu.h
=================================================================== #endif /* __ASM_X86_MTRR_H__ */
--- xen-3.3.1-testing.orig/xen/include/xen/iommu.h --- a/xen/include/xen/iommu.h
+++ xen-3.3.1-testing/xen/include/xen/iommu.h +++ b/xen/include/xen/iommu.h
@@ -32,6 +32,7 @@ extern int iommu_enabled; @@ -32,6 +32,7 @@ extern int iommu_enabled;
extern int iommu_pv_enabled; extern int iommu_pv_enabled;
extern int force_iommu; extern int force_iommu;

146
19088-x86-page-non-atomic-owner.patch Normal file
View File

@ -0,0 +1,146 @@
# HG changeset patch
# User Keir Fraser <keir.fraser@citrix.com>
# Date 1232981779 0
# Node ID 055c589f4791811797867736857b08fdd0fd6d49
# Parent c9dc7dcacc1d0c064a131da98a4063fa2cedd716
x86: No need for CMPXCHG8B on page_info structure.
References: bnc#470949
Updates and checks on count_info and page owner can safely be
non-atomic.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -1887,36 +1887,29 @@ void put_page(struct page_info *page)
int get_page(struct page_info *page, struct domain *domain)
{
- u32 x, nx, y = page->count_info;
- u32 d, nd = page->u.inuse._domain;
- u32 _domain = pickle_domptr(domain);
+ u32 x, y = page->count_info;
do {
- x = y;
- nx = x + 1;
- d = nd;
+ x = y;
if ( unlikely((x & PGC_count_mask) == 0) || /* Not allocated? */
/* Keep one spare reference to be acquired by get_page_light(). */
- unlikely(((nx + 1) & PGC_count_mask) <= 1) || /* Overflow? */
- unlikely(d != _domain) ) /* Wrong owner? */
- {
- if ( !_shadow_mode_refcounts(domain) && !domain->is_dying )
- gdprintk(XENLOG_INFO,
- "Error pfn %lx: rd=%p, od=%p, caf=%08x, taf=%"
- PRtype_info "\n",
- page_to_mfn(page), domain, unpickle_domptr(d),
- x, page->u.inuse.type_info);
- return 0;
- }
- asm volatile (
- LOCK_PREFIX "cmpxchg8b %2"
- : "=d" (nd), "=a" (y),
- "=m" (*(volatile u64 *)(&page->count_info))
- : "0" (d), "1" (x), "c" (d), "b" (nx) );
+ unlikely(((x + 2) & PGC_count_mask) <= 1) ) /* Overflow? */
+ goto fail;
}
- while ( unlikely(nd != d) || unlikely(y != x) );
+ while ( (y = cmpxchg(&page->count_info, x, x + 1)) != x );
- return 1;
+ if ( likely(page_get_owner(page) == domain) )
+ return 1;
+
+ put_page(page);
+
+ fail:
+ if ( !_shadow_mode_refcounts(domain) && !domain->is_dying )
+ gdprintk(XENLOG_INFO,
+ "Error pfn %lx: rd=%p, od=%p, caf=%08x, taf=%" PRtype_info,
+ page_to_mfn(page), domain, page_get_owner(page),
+ y, page->u.inuse.type_info);
+ return 0;
}
/*
@@ -3438,49 +3431,47 @@ int replace_grant_host_mapping(
int steal_page(
struct domain *d, struct page_info *page, unsigned int memflags)
{
- u32 _d, _nd, x, y;
+ u32 x, y;
spin_lock(&d->page_alloc_lock);
+ if ( is_xen_heap_page(page) || (page_get_owner(page) != d) )
+ goto fail;
+
/*
- * The tricky bit: atomically release ownership while there is just one
- * benign reference to the page (PGC_allocated). If that reference
- * disappears then the deallocation routine will safely spin.
+ * We require there is just one reference (PGC_allocated). We temporarily
+ * drop this reference now so that we can safely swizzle the owner.
*/
- _d = pickle_domptr(d);
- _nd = page->u.inuse._domain;
- y = page->count_info;
+ y = page->count_info;
do {
x = y;
- if ( unlikely((x & (PGC_count_mask|PGC_allocated)) !=
- (1 | PGC_allocated)) || unlikely(_nd != _d) )
- {
- MEM_LOG("gnttab_transfer: Bad page %p: ed=%p(%u), sd=%p,"
- " caf=%08x, taf=%" PRtype_info "\n",
- (void *) page_to_mfn(page),
- d, d->domain_id, unpickle_domptr(_nd), x,
- page->u.inuse.type_info);
- spin_unlock(&d->page_alloc_lock);
- return -1;
- }
- asm volatile (
- LOCK_PREFIX "cmpxchg8b %2"
- : "=d" (_nd), "=a" (y),
- "=m" (*(volatile u64 *)(&page->count_info))
- : "0" (_d), "1" (x), "c" (NULL), "b" (x) );
- } while (unlikely(_nd != _d) || unlikely(y != x));
+ if ( (x & (PGC_count_mask|PGC_allocated)) != (1 | PGC_allocated) )
+ goto fail;
+ y = cmpxchg(&page->count_info, x, x & ~PGC_count_mask);
+ } while ( y != x );
- /*
- * Unlink from 'd'. At least one reference remains (now anonymous), so
- * noone else is spinning to try to delete this page from 'd'.
- */
+ /* Swizzle the owner then reinstate the PGC_allocated reference. */
+ page_set_owner(page, NULL);
+ y = page->count_info;
+ do {
+ x = y;
+ BUG_ON((x & (PGC_count_mask|PGC_allocated)) != PGC_allocated);
+ } while ( (y = cmpxchg(&page->count_info, x, x | 1)) != x );
+
+ /* Unlink from original owner. */
if ( !(memflags & MEMF_no_refcount) )
d->tot_pages--;
list_del(&page->list);
spin_unlock(&d->page_alloc_lock);
-
return 0;
+
+ fail:
+ spin_unlock(&d->page_alloc_lock);
+ MEM_LOG("Bad page %p: ed=%p(%u), sd=%p, caf=%08x, taf=%" PRtype_info,
+ (void *)page_to_mfn(page), d, d->domain_id,
+ page_get_owner(page), page->count_info, page->u.inuse.type_info);
+ return -1;
}
int do_update_va_mapping(unsigned long va, u64 val64,

445
19089-x86_64-widen-page-refcounts.patch Normal file
View File

@ -0,0 +1,445 @@
# HG changeset patch
# User Keir Fraser <keir.fraser@citrix.com>
# Date 1232986782 0
# Node ID 39517e863cc89a085341e1d53317aaa7ceddd127
# Parent 055c589f4791811797867736857b08fdd0fd6d49
x86_64: Widen page counts to avoid overflow.
References: bnc#470949
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
# HG changeset patch
# User Keir Fraser <keir.fraser@citrix.com>
# Date 1232988758 0
# Node ID 728d1892f0e24c2531df2d61a2d95177400ceb17
# Parent 90909b81b3b9cf9b303e2bc457580603da3ac7fd
x86: Clean up shadow_page_info after page_info changes.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
# HG changeset patch
# User Keir Fraser <keir.fraser@citrix.com>
# Date 1233056759 0
# Node ID 6e623569455c08b57e43e3355f6809b3a4ba0707
# Parent 7b56dbd1b439e0996083810489398cb51dc43aa6
x86: clean up struct page_info
Remove the now unnecessary (and leading to misalignment of cpumask on
x86-64) 'packed' attributes.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -143,7 +143,7 @@ void dump_pageframe_info(struct domain *
{
list_for_each_entry ( page, &d->page_list, list )
{
- printk(" DomPage %p: caf=%08x, taf=%" PRtype_info "\n",
+ printk(" DomPage %p: caf=%08lx, taf=%" PRtype_info "\n",
_p(page_to_mfn(page)),
page->count_info, page->u.inuse.type_info);
}
@@ -151,7 +151,7 @@ void dump_pageframe_info(struct domain *
list_for_each_entry ( page, &d->xenpage_list, list )
{
- printk(" XenPage %p: caf=%08x, taf=%" PRtype_info "\n",
+ printk(" XenPage %p: caf=%08lx, taf=%" PRtype_info "\n",
_p(page_to_mfn(page)),
page->count_info, page->u.inuse.type_info);
}
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -714,8 +714,8 @@ get_page_from_l1e(
else if ( pte_flags_to_cacheattr(l1f) !=
((page->count_info >> PGC_cacheattr_base) & 7) )
{
- uint32_t x, nx, y = page->count_info;
- uint32_t cacheattr = pte_flags_to_cacheattr(l1f);
+ unsigned long x, nx, y = page->count_info;
+ unsigned long cacheattr = pte_flags_to_cacheattr(l1f);
if ( is_xen_heap_page(page) )
{
@@ -1869,7 +1869,7 @@ static int mod_l4_entry(l4_pgentry_t *pl
void put_page(struct page_info *page)
{
- u32 nx, x, y = page->count_info;
+ unsigned long nx, x, y = page->count_info;
do {
x = y;
@@ -1887,7 +1887,7 @@ void put_page(struct page_info *page)
int get_page(struct page_info *page, struct domain *domain)
{
- u32 x, y = page->count_info;
+ unsigned long x, y = page->count_info;
do {
x = y;
@@ -1906,7 +1906,7 @@ int get_page(struct page_info *page, str
fail:
if ( !_shadow_mode_refcounts(domain) && !domain->is_dying )
gdprintk(XENLOG_INFO,
- "Error pfn %lx: rd=%p, od=%p, caf=%08x, taf=%" PRtype_info,
+ "Error pfn %lx: rd=%p, od=%p, caf=%08lx, taf=%" PRtype_info,
page_to_mfn(page), domain, page_get_owner(page),
y, page->u.inuse.type_info);
return 0;
@@ -1922,7 +1922,7 @@ int get_page(struct page_info *page, str
*/
static void get_page_light(struct page_info *page)
{
- u32 x, nx, y = page->count_info;
+ unsigned long x, nx, y = page->count_info;
do {
x = y;
@@ -1963,7 +1963,7 @@ static int alloc_page_type(struct page_i
rc = alloc_segdesc_page(page);
break;
default:
- printk("Bad type in alloc_page_type %lx t=%" PRtype_info " c=%x\n",
+ printk("Bad type in alloc_page_type %lx t=%" PRtype_info " c=%lx\n",
type, page->u.inuse.type_info,
page->count_info);
rc = -EINVAL;
@@ -1987,7 +1987,7 @@ static int alloc_page_type(struct page_i
{
ASSERT(rc < 0);
MEM_LOG("Error while validating mfn %lx (pfn %lx) for type %"
- PRtype_info ": caf=%08x taf=%" PRtype_info,
+ PRtype_info ": caf=%08lx taf=%" PRtype_info,
page_to_mfn(page), get_gpfn_from_mfn(page_to_mfn(page)),
type, page->count_info, page->u.inuse.type_info);
page->u.inuse.type_info = 0;
@@ -3144,7 +3144,7 @@ static int create_grant_pte_mapping(
void *va;
unsigned long gmfn, mfn;
struct page_info *page;
- u32 type;
+ unsigned long type;
l1_pgentry_t ol1e;
struct domain *d = v->domain;
@@ -3205,7 +3205,7 @@ static int destroy_grant_pte_mapping(
void *va;
unsigned long gmfn, mfn;
struct page_info *page;
- u32 type;
+ unsigned long type;
l1_pgentry_t ol1e;
gmfn = addr >> PAGE_SHIFT;
@@ -3431,7 +3431,7 @@ int replace_grant_host_mapping(
int steal_page(
struct domain *d, struct page_info *page, unsigned int memflags)
{
- u32 x, y;
+ unsigned long x, y;
spin_lock(&d->page_alloc_lock);
@@ -3468,7 +3468,7 @@ int steal_page(
fail:
spin_unlock(&d->page_alloc_lock);
- MEM_LOG("Bad page %p: ed=%p(%u), sd=%p, caf=%08x, taf=%" PRtype_info,
+ MEM_LOG("Bad page %p: ed=%p(%u), sd=%p, caf=%08lx, taf=%" PRtype_info,
(void *)page_to_mfn(page), d, d->domain_id,
page_get_owner(page), page->count_info, page->u.inuse.type_info);
return -1;
--- a/xen/arch/x86/mm/hap/hap.c
+++ b/xen/arch/x86/mm/hap/hap.c
@@ -166,7 +166,7 @@ void hap_free_p2m_page(struct domain *d,
ASSERT(page_get_owner(pg) == d);
/* Should have just the one ref we gave it in alloc_p2m_page() */
if ( (pg->count_info & PGC_count_mask) != 1 )
- HAP_ERROR("Odd p2m page count c=%#x t=%"PRtype_info"\n",
+ HAP_ERROR("Odd p2m page count c=%#lx t=%"PRtype_info"\n",
pg->count_info, pg->u.inuse.type_info);
pg->count_info = 0;
/* Free should not decrement domain's total allocation, since
--- a/xen/arch/x86/mm/shadow/common.c
+++ b/xen/arch/x86/mm/shadow/common.c
@@ -1678,7 +1678,7 @@ shadow_free_p2m_page(struct domain *d, s
/* Should have just the one ref we gave it in alloc_p2m_page() */
if ( (pg->count_info & PGC_count_mask) != 1 )
{
- SHADOW_ERROR("Odd p2m page count c=%#x t=%"PRtype_info"\n",
+ SHADOW_ERROR("Odd p2m page count c=%#lx t=%"PRtype_info"\n",
pg->count_info, pg->u.inuse.type_info);
}
pg->count_info = 0;
@@ -1796,14 +1796,21 @@ static unsigned int sh_set_allocation(st
sp = list_entry(d->arch.paging.shadow.freelists[order].next,
struct shadow_page_info, list);
list_del(&sp->list);
-#if defined(__x86_64__)
/*
- * Re-instate lock field which we overwrite with shadow_page_info.
- * This was safe, since the lock is only used on guest pages.
+ * The pages were allocated anonymously, but the owner field
+ * may get overwritten, so need to clear it here.
*/
for ( j = 0; j < 1U << order; j++ )
+ {
+ page_set_owner(&((struct page_info *)sp)[j], NULL);
+#if defined(__x86_64__)
+ /*
+ * Re-instate lock field which we overwrite with shadow_page_info.
+ * This was safe, since the lock is only used on guest pages.
+ */
spin_lock_init(&((struct page_info *)sp)[j].lock);
#endif
+ }
d->arch.paging.shadow.free_pages -= 1 << order;
d->arch.paging.shadow.total_pages -= 1 << order;
free_domheap_pages((struct page_info *)sp, order);
@@ -2516,7 +2523,7 @@ int sh_remove_all_mappings(struct vcpu *
&& (page->u.inuse.type_info & PGT_count_mask) == 0) )
{
SHADOW_ERROR("can't find all mappings of mfn %lx: "
- "c=%08x t=%08lx\n", mfn_x(gmfn),
+ "c=%08lx t=%08lx\n", mfn_x(gmfn),
page->count_info, page->u.inuse.type_info);
}
}
@@ -3591,7 +3598,6 @@ int shadow_track_dirty_vram(struct domai
for ( i = 0; i < nr; i++ ) {
mfn_t mfn = gfn_to_mfn(d, begin_pfn + i, &t);
struct page_info *page;
- u32 count_info;
int dirty = 0;
paddr_t sl1ma = d->dirty_vram->sl1ma[i];
@@ -3602,8 +3608,7 @@ int shadow_track_dirty_vram(struct domai
else
{
page = mfn_to_page(mfn);
- count_info = page->u.inuse.type_info & PGT_count_mask;
- switch (count_info)
+ switch (page->u.inuse.type_info & PGT_count_mask)
{
case 0:
/* No guest reference, nothing to track. */
--- a/xen/arch/x86/mm/shadow/multi.c
+++ b/xen/arch/x86/mm/shadow/multi.c
@@ -1334,9 +1334,8 @@ static inline void shadow_vram_get_l1e(s
if ( (gfn >= d->dirty_vram->begin_pfn) && (gfn < d->dirty_vram->end_pfn) ) {
unsigned long i = gfn - d->dirty_vram->begin_pfn;
struct page_info *page = mfn_to_page(mfn);
- u32 count_info = page->u.inuse.type_info & PGT_count_mask;
- if ( count_info == 1 )
+ if ( (page->u.inuse.type_info & PGT_count_mask) == 1 )
/* Initial guest reference, record it */
d->dirty_vram->sl1ma[i] = pfn_to_paddr(mfn_x(sl1mfn))
| ((unsigned long)sl1e & ~PAGE_MASK);
@@ -1362,12 +1361,11 @@ static inline void shadow_vram_put_l1e(s
if ( (gfn >= d->dirty_vram->begin_pfn) && (gfn < d->dirty_vram->end_pfn) ) {
unsigned long i = gfn - d->dirty_vram->begin_pfn;
struct page_info *page = mfn_to_page(mfn);
- u32 count_info = page->u.inuse.type_info & PGT_count_mask;
int dirty = 0;
paddr_t sl1ma = pfn_to_paddr(mfn_x(sl1mfn))
| ((unsigned long)sl1e & ~PAGE_MASK);
- if ( count_info == 1 ) {
+ if ( (page->u.inuse.type_info & PGT_count_mask) == 1 ) {
/* Last reference */
if ( d->dirty_vram->sl1ma[i] == INVALID_PADDR ) {
/* We didn't know it was that one, let's say it is dirty */
--- a/xen/arch/x86/mm/shadow/private.h
+++ b/xen/arch/x86/mm/shadow/private.h
@@ -201,12 +201,11 @@ struct shadow_page_info
u32 tlbflush_timestamp;
};
struct {
- unsigned int type:5; /* What kind of shadow is this? */
- unsigned int pinned:1; /* Is the shadow pinned? */
- unsigned int count:26; /* Reference count */
- u32 mbz; /* Must be zero: this is where the
- * owner field lives in page_info */
- } __attribute__((packed));
+ unsigned long mbz; /* Must be zero: count_info is here. */
+ unsigned long type:5; /* What kind of shadow is this? */
+ unsigned long pinned:1; /* Is the shadow pinned? */
+ unsigned long count:26; /* Reference count */
+ };
union {
/* For unused shadow pages, a list of pages of this order; for
* pinnable shadows, if pinned, a list of other pinned shadows
@@ -229,7 +228,7 @@ static inline void shadow_check_page_str
BUILD_BUG_ON(sizeof (struct shadow_page_info) !=
sizeof (struct page_info));
BUILD_BUG_ON(offsetof(struct shadow_page_info, mbz) !=
- offsetof(struct page_info, u.inuse._domain));
+ offsetof(struct page_info, count_info));
};
/* Shadow type codes */
--- a/xen/arch/x86/x86_32/mm.c
+++ b/xen/arch/x86/x86_32/mm.c
@@ -159,15 +159,6 @@ void __init subarch_init_memory(void)
unsigned long m2p_start_mfn;
unsigned int i, j;
- /*
- * We are rather picky about the layout of 'struct page_info'. The
- * count_info and domain fields must be adjacent, as we perform atomic
- * 64-bit operations on them. Also, just for sanity, we assert the size
- * of the structure here.
- */
- BUILD_BUG_ON(offsetof(struct page_info, u.inuse._domain) !=
- (offsetof(struct page_info, count_info) + sizeof(u32)));
- BUILD_BUG_ON((offsetof(struct page_info, count_info) & 7) != 0);
BUILD_BUG_ON(sizeof(struct page_info) != 24);
/* M2P table is mappable read-only by privileged domains. */
--- a/xen/arch/x86/x86_64/mm.c
+++ b/xen/arch/x86/x86_64/mm.c
@@ -225,17 +225,6 @@ void __init subarch_init_memory(void)
l3_pgentry_t l3e;
l2_pgentry_t l2e;
- /*
- * We are rather picky about the layout of 'struct page_info'. The
- * count_info and domain fields must be adjacent, as we perform atomic
- * 64-bit operations on them.
- */
- BUILD_BUG_ON(offsetof(struct page_info, u.inuse._domain) !=
- (offsetof(struct page_info, count_info) + sizeof(u32)));
- BUILD_BUG_ON((offsetof(struct page_info, count_info) & 7) != 0);
- BUILD_BUG_ON(sizeof(struct page_info) !=
- (32 + BITS_TO_LONGS(NR_CPUS)*sizeof(long)));
-
/* M2P table is mappable read-only by privileged domains. */
for ( v = RDWR_MPT_VIRT_START;
v != RDWR_MPT_VIRT_END;
--- a/xen/common/xenoprof.c
+++ b/xen/common/xenoprof.c
@@ -142,8 +142,8 @@ share_xenoprof_page_with_guest(struct do
struct page_info *page = mfn_to_page(mfn + i);
if ( (page->count_info & (PGC_allocated|PGC_count_mask)) != 0 )
{
- gdprintk(XENLOG_INFO, "mfn 0x%lx page->count_info 0x%x\n",
- mfn + i, page->count_info);
+ gdprintk(XENLOG_INFO, "mfn 0x%lx page->count_info 0x%lx\n",
+ mfn + i, (unsigned long)page->count_info);
return -EBUSY;
}
page_set_owner(page, NULL);
--- a/xen/include/asm-x86/mm.h
+++ b/xen/include/asm-x86/mm.h
@@ -23,7 +23,7 @@ struct page_info
struct list_head list;
/* Reference count and various PGC_xxx flags and fields. */
- u32 count_info;
+ unsigned long count_info;
/* Context-dependent fields follow... */
union {
@@ -34,7 +34,7 @@ struct page_info
u32 _domain; /* pickled format */
/* Type reference count and various PGT_xxx flags and fields. */
unsigned long type_info;
- } __attribute__ ((packed)) inuse;
+ } inuse;
/* Page is on a free list: ((count_info & PGC_count_mask) == 0). */
struct {
@@ -42,7 +42,7 @@ struct page_info
u32 order;
/* Mask of possibly-tainted TLBs. */
cpumask_t cpumask;
- } __attribute__ ((packed)) free;
+ } free;
} u;
@@ -102,48 +102,53 @@ struct page_info
};
};
+#define PG_shift(idx) (BITS_PER_LONG - (idx))
+#define PG_mask(x, idx) (x ## UL << PG_shift(idx))
+
/* The following page types are MUTUALLY EXCLUSIVE. */
-#define PGT_none (0U<<29) /* no special uses of this page */
-#define PGT_l1_page_table (1U<<29) /* using this page as an L1 page table? */
-#define PGT_l2_page_table (2U<<29) /* using this page as an L2 page table? */
-#define PGT_l3_page_table (3U<<29) /* using this page as an L3 page table? */
-#define PGT_l4_page_table (4U<<29) /* using this page as an L4 page table? */
-#define PGT_seg_desc_page (5U<<29) /* using this page in a GDT/LDT? */
-#define PGT_writable_page (7U<<29) /* has writable mappings of this page? */
-#define PGT_type_mask (7U<<29) /* Bits 29-31. */
+#define PGT_none PG_mask(0, 3) /* no special uses of this page */
+#define PGT_l1_page_table PG_mask(1, 3) /* using as an L1 page table? */
+#define PGT_l2_page_table PG_mask(2, 3) /* using as an L2 page table? */
+#define PGT_l3_page_table PG_mask(3, 3) /* using as an L3 page table? */
+#define PGT_l4_page_table PG_mask(4, 3) /* using as an L4 page table? */
+#define PGT_seg_desc_page PG_mask(5, 3) /* using this page in a GDT/LDT? */
+#define PGT_writable_page PG_mask(7, 3) /* has writable mappings? */
+#define PGT_type_mask PG_mask(7, 3) /* Bits 29-31. */
/* Owning guest has pinned this page to its current type? */
-#define _PGT_pinned 28
-#define PGT_pinned (1U<<_PGT_pinned)
+#define _PGT_pinned PG_shift(4)
+#define PGT_pinned PG_mask(1, 4)
/* Has this page been validated for use as its current type? */
-#define _PGT_validated 27
-#define PGT_validated (1U<<_PGT_validated)
+#define _PGT_validated PG_shift(5)
+#define PGT_validated PG_mask(1, 5)
/* PAE only: is this an L2 page directory containing Xen-private mappings? */
-#define _PGT_pae_xen_l2 26
-#define PGT_pae_xen_l2 (1U<<_PGT_pae_xen_l2)
+#define _PGT_pae_xen_l2 PG_shift(6)
+#define PGT_pae_xen_l2 PG_mask(1, 6)
/* Has this page been *partially* validated for use as its current type? */
-#define _PGT_partial 25
-#define PGT_partial (1U<<_PGT_partial)
+#define _PGT_partial PG_shift(7)
+#define PGT_partial PG_mask(1, 7)
- /* 25-bit count of uses of this frame as its current type. */
-#define PGT_count_mask ((1U<<25)-1)
+ /* Count of uses of this frame as its current type. */
+#define PGT_count_width PG_shift(7)
+#define PGT_count_mask ((1UL<<PGT_count_width)-1)
/* Cleared when the owning guest 'frees' this page. */
-#define _PGC_allocated 31
-#define PGC_allocated (1U<<_PGC_allocated)
+#define _PGC_allocated PG_shift(1)
+#define PGC_allocated PG_mask(1, 1)
#if defined(__i386__)
/* Page is locked? */
-# define _PGC_locked 30
-# define PGC_locked (1U<<_PGC_out_of_sync)
+# define _PGC_locked PG_shift(2)
+# define PGC_locked PG_mask(1, 2)
#endif
/* Set when is using a page as a page table */
-#define _PGC_page_table 29
-#define PGC_page_table (1U<<_PGC_page_table)
+#define _PGC_page_table PG_shift(3)
+#define PGC_page_table PG_mask(1, 3)
/* 3-bit PAT/PCD/PWT cache-attribute hint. */
-#define PGC_cacheattr_base 26
-#define PGC_cacheattr_mask (7U<<PGC_cacheattr_base)
- /* 26-bit count of references to this frame. */
-#define PGC_count_mask ((1U<<26)-1)
+#define PGC_cacheattr_base PG_shift(6)
+#define PGC_cacheattr_mask PG_mask(7, 6)
+ /* Count of references to this frame. */
+#define PGC_count_width PG_shift(6)
+#define PGC_count_mask ((1UL<<PGC_count_width)-1)
#define is_xen_heap_page(page) is_xen_heap_mfn(page_to_mfn(page))
#define is_xen_heap_mfn(mfn) ({ \

834
19103-x86_64-fold-page-lock.patch Normal file
View File

@ -0,0 +1,834 @@
# HG changeset patch
# User Keir Fraser <keir.fraser@citrix.com>
# Date 1233072141 0
# Node ID bcf77bfd1161d1e2693d6762bcd436ad98ec0779
# Parent dbf53b739af0434adff50172fc071f718b57b450
x86: Fold page_info lock into type_info.
References: bnc#470949
Fix some racey looking code at the same time.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -202,11 +202,6 @@ void __init init_frametable(void)
}
memset(frame_table, 0, nr_pages << PAGE_SHIFT);
-
-#if defined(__x86_64__)
- for ( i = 0; i < max_page; i ++ )
- spin_lock_init(&frame_table[i].lock);
-#endif
}
void __init arch_init_memory(void)
@@ -1499,24 +1494,31 @@ static int free_l4_table(struct page_inf
#define free_l4_table(page, preemptible) (-EINVAL)
#endif
-static void page_lock(struct page_info *page)
+static int page_lock(struct page_info *page)
{
-#if defined(__i386__)
- while ( unlikely(test_and_set_bit(_PGC_locked, &page->count_info)) )
- while ( test_bit(_PGC_locked, &page->count_info) )
+ unsigned long x, nx;
+
+ do {
+ while ( (x = page->u.inuse.type_info) & PGT_locked )
cpu_relax();
-#else
- spin_lock(&page->lock);
-#endif
+ nx = x + (1 | PGT_locked);
+ if ( !(x & PGT_validated) ||
+ !(x & PGT_count_mask) ||
+ !(nx & PGT_count_mask) )
+ return 0;
+ } while ( cmpxchg(&page->u.inuse.type_info, x, nx) != x );
+
+ return 1;
}
static void page_unlock(struct page_info *page)
{
-#if defined(__i386__)
- clear_bit(_PGC_locked, &page->count_info);
-#else
- spin_unlock(&page->lock);
-#endif
+ unsigned long x, nx, y = page->u.inuse.type_info;
+
+ do {
+ x = y;
+ nx = x - (1 | PGT_locked);
+ } while ( (y = cmpxchg(&page->u.inuse.type_info, x, nx)) != x );
}
/* How to write an entry to the guest pagetables.
@@ -1579,19 +1581,15 @@ static int mod_l1_entry(l1_pgentry_t *pl
struct vcpu *curr = current;
struct domain *d = curr->domain;
unsigned long mfn;
- struct page_info *l1pg = mfn_to_page(gl1mfn);
p2m_type_t p2mt;
int rc = 1;
- page_lock(l1pg);
-
if ( unlikely(__copy_from_user(&ol1e, pl1e, sizeof(ol1e)) != 0) )
- return page_unlock(l1pg), 0;
+ return 0;
if ( unlikely(paging_mode_refcounts(d)) )
{
rc = UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, curr, preserve_ad);
- page_unlock(l1pg);
return rc;
}
@@ -1600,13 +1598,12 @@ static int mod_l1_entry(l1_pgentry_t *pl
/* Translate foreign guest addresses. */
mfn = mfn_x(gfn_to_mfn(FOREIGNDOM, l1e_get_pfn(nl1e), &p2mt));
if ( !p2m_is_ram(p2mt) || unlikely(mfn == INVALID_MFN) )
- return page_unlock(l1pg), 0;
+ return 0;
ASSERT((mfn & ~(PADDR_MASK >> PAGE_SHIFT)) == 0);
nl1e = l1e_from_pfn(mfn, l1e_get_flags(nl1e));
if ( unlikely(l1e_get_flags(nl1e) & l1_disallow_mask(d)) )
{
- page_unlock(l1pg);
MEM_LOG("Bad L1 flags %x",
l1e_get_flags(nl1e) & l1_disallow_mask(d));
return 0;
@@ -1618,12 +1615,11 @@ static int mod_l1_entry(l1_pgentry_t *pl
adjust_guest_l1e(nl1e, d);
rc = UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, curr,
preserve_ad);
- page_unlock(l1pg);
return rc;
}
if ( unlikely(!get_page_from_l1e(nl1e, FOREIGNDOM)) )
- return page_unlock(l1pg), 0;
+ return 0;
adjust_guest_l1e(nl1e, d);
if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, curr,
@@ -1636,11 +1632,9 @@ static int mod_l1_entry(l1_pgentry_t *pl
else if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, curr,
preserve_ad)) )
{
- page_unlock(l1pg);
return 0;
}
- page_unlock(l1pg);
put_page_from_l1e(ol1e, d);
return rc;
}
@@ -1650,13 +1644,13 @@ static int mod_l1_entry(l1_pgentry_t *pl
static int mod_l2_entry(l2_pgentry_t *pl2e,
l2_pgentry_t nl2e,
unsigned long pfn,
- unsigned long type,
int preserve_ad)
{
l2_pgentry_t ol2e;
struct vcpu *curr = current;
struct domain *d = curr->domain;
struct page_info *l2pg = mfn_to_page(pfn);
+ unsigned long type = l2pg->u.inuse.type_info;
int rc = 1;
if ( unlikely(!is_guest_l2_slot(d, type, pgentry_ptr_to_slot(pl2e))) )
@@ -1665,16 +1659,13 @@ static int mod_l2_entry(l2_pgentry_t *pl
return 0;
}
- page_lock(l2pg);
-
if ( unlikely(__copy_from_user(&ol2e, pl2e, sizeof(ol2e)) != 0) )
- return page_unlock(l2pg), 0;
+ return 0;
if ( l2e_get_flags(nl2e) & _PAGE_PRESENT )
{
if ( unlikely(l2e_get_flags(nl2e) & L2_DISALLOW_MASK) )
{
- page_unlock(l2pg);
MEM_LOG("Bad L2 flags %x",
l2e_get_flags(nl2e) & L2_DISALLOW_MASK);
return 0;
@@ -1685,12 +1676,11 @@ static int mod_l2_entry(l2_pgentry_t *pl
{
adjust_guest_l2e(nl2e, d);
rc = UPDATE_ENTRY(l2, pl2e, ol2e, nl2e, pfn, curr, preserve_ad);
- page_unlock(l2pg);
return rc;
}
if ( unlikely(get_page_from_l2e(nl2e, pfn, d) < 0) )
- return page_unlock(l2pg), 0;
+ return 0;
adjust_guest_l2e(nl2e, d);
if ( unlikely(!UPDATE_ENTRY(l2, pl2e, ol2e, nl2e, pfn, curr,
@@ -1703,11 +1693,9 @@ static int mod_l2_entry(l2_pgentry_t *pl
else if ( unlikely(!UPDATE_ENTRY(l2, pl2e, ol2e, nl2e, pfn, curr,
preserve_ad)) )
{
- page_unlock(l2pg);
return 0;
}
- page_unlock(l2pg);
put_page_from_l2e(ol2e, pfn);
return rc;
}
@@ -1722,7 +1710,6 @@ static int mod_l3_entry(l3_pgentry_t *pl
l3_pgentry_t ol3e;
struct vcpu *curr = current;
struct domain *d = curr->domain;
- struct page_info *l3pg = mfn_to_page(pfn);
int rc = 0;
if ( unlikely(!is_guest_l3_slot(pgentry_ptr_to_slot(pl3e))) )
@@ -1738,16 +1725,13 @@ static int mod_l3_entry(l3_pgentry_t *pl
if ( is_pv_32bit_domain(d) && (pgentry_ptr_to_slot(pl3e) >= 3) )
return -EINVAL;
- page_lock(l3pg);
-
if ( unlikely(__copy_from_user(&ol3e, pl3e, sizeof(ol3e)) != 0) )
- return page_unlock(l3pg), -EFAULT;
+ return -EFAULT;
if ( l3e_get_flags(nl3e) & _PAGE_PRESENT )
{
if ( unlikely(l3e_get_flags(nl3e) & l3_disallow_mask(d)) )
{
- page_unlock(l3pg);
MEM_LOG("Bad L3 flags %x",
l3e_get_flags(nl3e) & l3_disallow_mask(d));
return -EINVAL;
@@ -1758,13 +1742,12 @@ static int mod_l3_entry(l3_pgentry_t *pl
{
adjust_guest_l3e(nl3e, d);
rc = UPDATE_ENTRY(l3, pl3e, ol3e, nl3e, pfn, curr, preserve_ad);
- page_unlock(l3pg);
return rc ? 0 : -EFAULT;
}
rc = get_page_from_l3e(nl3e, pfn, d, 0, preemptible);
if ( unlikely(rc < 0) )
- return page_unlock(l3pg), rc;
+ return rc;
rc = 0;
adjust_guest_l3e(nl3e, d);
@@ -1778,7 +1761,6 @@ static int mod_l3_entry(l3_pgentry_t *pl
else if ( unlikely(!UPDATE_ENTRY(l3, pl3e, ol3e, nl3e, pfn, curr,
preserve_ad)) )
{
- page_unlock(l3pg);
return -EFAULT;
}
@@ -1790,7 +1772,6 @@ static int mod_l3_entry(l3_pgentry_t *pl
pae_flush_pgd(pfn, pgentry_ptr_to_slot(pl3e), nl3e);
}
- page_unlock(l3pg);
put_page_from_l3e(ol3e, pfn, 0, 0);
return rc;
}
@@ -1807,7 +1788,6 @@ static int mod_l4_entry(l4_pgentry_t *pl
struct vcpu *curr = current;
struct domain *d = curr->domain;
l4_pgentry_t ol4e;
- struct page_info *l4pg = mfn_to_page(pfn);
int rc = 0;
if ( unlikely(!is_guest_l4_slot(d, pgentry_ptr_to_slot(pl4e))) )
@@ -1816,16 +1796,13 @@ static int mod_l4_entry(l4_pgentry_t *pl
return -EINVAL;
}
- page_lock(l4pg);
-
if ( unlikely(__copy_from_user(&ol4e, pl4e, sizeof(ol4e)) != 0) )
- return page_unlock(l4pg), -EFAULT;
+ return -EFAULT;
if ( l4e_get_flags(nl4e) & _PAGE_PRESENT )
{
if ( unlikely(l4e_get_flags(nl4e) & L4_DISALLOW_MASK) )
{
- page_unlock(l4pg);
MEM_LOG("Bad L4 flags %x",
l4e_get_flags(nl4e) & L4_DISALLOW_MASK);
return -EINVAL;
@@ -1836,13 +1813,12 @@ static int mod_l4_entry(l4_pgentry_t *pl
{
adjust_guest_l4e(nl4e, d);
rc = UPDATE_ENTRY(l4, pl4e, ol4e, nl4e, pfn, curr, preserve_ad);
- page_unlock(l4pg);
return rc ? 0 : -EFAULT;
}
rc = get_page_from_l4e(nl4e, pfn, d, 0, preemptible);
if ( unlikely(rc < 0) )
- return page_unlock(l4pg), rc;
+ return rc;
rc = 0;
adjust_guest_l4e(nl4e, d);
@@ -1856,11 +1832,9 @@ static int mod_l4_entry(l4_pgentry_t *pl
else if ( unlikely(!UPDATE_ENTRY(l4, pl4e, ol4e, nl4e, pfn, curr,
preserve_ad)) )
{
- page_unlock(l4pg);
return -EFAULT;
}
- page_unlock(l4pg);
put_page_from_l4e(ol4e, pfn, 0, 0);
return rc;
}
@@ -2918,7 +2892,6 @@ int do_mmu_update(
unsigned int cmd, done = 0;
struct vcpu *v = current;
struct domain *d = v->domain;
- unsigned long type_info;
struct domain_mmap_cache mapcache;
if ( unlikely(count & MMU_UPDATE_PREEMPTED) )
@@ -2990,24 +2963,9 @@ int do_mmu_update(
(unsigned long)(req.ptr & ~PAGE_MASK));
page = mfn_to_page(mfn);
- switch ( (type_info = page->u.inuse.type_info) & PGT_type_mask )
+ if ( page_lock(page) )
{
- case PGT_l1_page_table:
- case PGT_l2_page_table:
- case PGT_l3_page_table:
- case PGT_l4_page_table:
- {
- if ( paging_mode_refcounts(d) )
- {
- MEM_LOG("mmu update on auto-refcounted domain!");
- break;
- }
-
- if ( unlikely(!get_page_type(
- page, type_info & (PGT_type_mask|PGT_pae_xen_l2))) )
- goto not_a_pt;
-
- switch ( type_info & PGT_type_mask )
+ switch ( page->u.inuse.type_info & PGT_type_mask )
{
case PGT_l1_page_table:
{
@@ -3019,7 +2977,7 @@ int do_mmu_update(
case PGT_l2_page_table:
{
l2_pgentry_t l2e = l2e_from_intpte(req.val);
- okay = mod_l2_entry(va, l2e, mfn, type_info,
+ okay = mod_l2_entry(va, l2e, mfn,
cmd == MMU_PT_UPDATE_PRESERVE_AD);
}
break;
@@ -3041,31 +2999,23 @@ int do_mmu_update(
}
break;
#endif
+ case PGT_writable_page:
+ perfc_incr(writable_mmu_updates);
+ okay = paging_write_guest_entry(v, va, req.val, _mfn(mfn));
+ break;
}
-
- put_page_type(page);
+ page_unlock(page);
if ( rc == -EINTR )
rc = -EAGAIN;
}
- break;
-
- default:
- not_a_pt:
+ else if ( get_page_type(page, PGT_writable_page) )
{
- if ( unlikely(!get_page_type(page, PGT_writable_page)) )
- break;
-
perfc_incr(writable_mmu_updates);
-
okay = paging_write_guest_entry(v, va, req.val, _mfn(mfn));
-
put_page_type(page);
}
- break;
- }
unmap_domain_page_with_cache(va, &mapcache);
-
put_page(page);
break;
@@ -3144,7 +3094,6 @@ static int create_grant_pte_mapping(
void *va;
unsigned long gmfn, mfn;
struct page_info *page;
- unsigned long type;
l1_pgentry_t ol1e;
struct domain *d = v->domain;
@@ -3165,21 +3114,23 @@ static int create_grant_pte_mapping(
va = (void *)((unsigned long)va + ((unsigned long)pte_addr & ~PAGE_MASK));
page = mfn_to_page(mfn);
- type = page->u.inuse.type_info & PGT_type_mask;
- if ( (type != PGT_l1_page_table) || !get_page_type(page, type) )
+ if ( !page_lock(page) )
{
- MEM_LOG("Grant map attempted to update a non-L1 page");
rc = GNTST_general_error;
goto failed;
}
- page_lock(page);
+ if ( (page->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table )
+ {
+ page_unlock(page);
+ rc = GNTST_general_error;
+ goto failed;
+ }
ol1e = *(l1_pgentry_t *)va;
if ( !UPDATE_ENTRY(l1, (l1_pgentry_t *)va, ol1e, nl1e, mfn, v, 0) )
{
page_unlock(page);
- put_page_type(page);
rc = GNTST_general_error;
goto failed;
}
@@ -3189,8 +3140,6 @@ static int create_grant_pte_mapping(
if ( !paging_mode_refcounts(d) )
put_page_from_l1e(ol1e, d);
- put_page_type(page);
-
failed:
unmap_domain_page(va);
put_page(page);
@@ -3205,7 +3154,6 @@ static int destroy_grant_pte_mapping(
void *va;
unsigned long gmfn, mfn;
struct page_info *page;
- unsigned long type;
l1_pgentry_t ol1e;
gmfn = addr >> PAGE_SHIFT;
@@ -3221,15 +3169,18 @@ static int destroy_grant_pte_mapping(
va = (void *)((unsigned long)va + ((unsigned long)addr & ~PAGE_MASK));
page = mfn_to_page(mfn);
- type = page->u.inuse.type_info & PGT_type_mask;
- if ( (type != PGT_l1_page_table) || !get_page_type(page, type) )
+ if ( !page_lock(page) )
{
- MEM_LOG("Grant map attempted to update a non-L1 page");
rc = GNTST_general_error;
goto failed;
}
- page_lock(page);
+ if ( (page->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table )
+ {
+ page_unlock(page);
+ rc = GNTST_general_error;
+ goto failed;
+ }
ol1e = *(l1_pgentry_t *)va;
@@ -3239,7 +3190,6 @@ static int destroy_grant_pte_mapping(
page_unlock(page);
MEM_LOG("PTE entry %lx for address %"PRIx64" doesn't match frame %lx",
(unsigned long)l1e_get_intpte(ol1e), addr, frame);
- put_page_type(page);
rc = GNTST_general_error;
goto failed;
}
@@ -3253,13 +3203,11 @@ static int destroy_grant_pte_mapping(
{
page_unlock(page);
MEM_LOG("Cannot delete PTE entry at %p", va);
- put_page_type(page);
rc = GNTST_general_error;
goto failed;
}
page_unlock(page);
- put_page_type(page);
failed:
unmap_domain_page(va);
@@ -3287,21 +3235,40 @@ static int create_grant_va_mapping(
MEM_LOG("Could not find L1 PTE for address %lx", va);
return GNTST_general_error;
}
+
+ if ( !get_page_from_pagenr(gl1mfn, current->domain) )
+ {
+ guest_unmap_l1e(v, pl1e);
+ return GNTST_general_error;
+ }
+
l1pg = mfn_to_page(gl1mfn);
- page_lock(l1pg);
+ if ( !page_lock(l1pg) )
+ {
+ put_page(l1pg);
+ guest_unmap_l1e(v, pl1e);
+ return GNTST_general_error;
+ }
+
+ if ( (l1pg->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table )
+ {
+ page_unlock(l1pg);
+ put_page(l1pg);
+ guest_unmap_l1e(v, pl1e);
+ return GNTST_general_error;
+ }
+
ol1e = *pl1e;
okay = UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, v, 0);
+
page_unlock(l1pg);
+ put_page(l1pg);
guest_unmap_l1e(v, pl1e);
- pl1e = NULL;
- if ( !okay )
- return GNTST_general_error;
-
- if ( !paging_mode_refcounts(d) )
+ if ( okay && !paging_mode_refcounts(d) )
put_page_from_l1e(ol1e, d);
- return GNTST_okay;
+ return okay ? GNTST_okay : GNTST_general_error;
}
static int replace_grant_va_mapping(
@@ -3319,31 +3286,48 @@ static int replace_grant_va_mapping(
return GNTST_general_error;
}
+ if ( !get_page_from_pagenr(gl1mfn, current->domain) )
+ {
+ rc = GNTST_general_error;
+ goto out;
+ }
+
l1pg = mfn_to_page(gl1mfn);
- page_lock(l1pg);
+ if ( !page_lock(l1pg) )
+ {
+ rc = GNTST_general_error;
+ put_page(l1pg);
+ goto out;
+ }
+
+ if ( (l1pg->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table )
+ {
+ rc = GNTST_general_error;
+ goto unlock_and_out;
+ }
+
ol1e = *pl1e;
/* Check that the virtual address supplied is actually mapped to frame. */
if ( unlikely(l1e_get_pfn(ol1e) != frame) )
{
- page_unlock(l1pg);
MEM_LOG("PTE entry %lx for address %lx doesn't match frame %lx",
l1e_get_pfn(ol1e), addr, frame);
rc = GNTST_general_error;
- goto out;
+ goto unlock_and_out;
}
/* Delete pagetable entry. */
if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, v, 0)) )
{
- page_unlock(l1pg);
MEM_LOG("Cannot delete PTE entry at %p", (unsigned long *)pl1e);
rc = GNTST_general_error;
- goto out;
+ goto unlock_and_out;
}
+ unlock_and_out:
page_unlock(l1pg);
-
+ put_page(l1pg);
out:
guest_unmap_l1e(v, pl1e);
return rc;
@@ -3405,20 +3389,42 @@ int replace_grant_host_mapping(
return GNTST_general_error;
}
+ if ( !get_page_from_pagenr(gl1mfn, current->domain) )
+ {
+ guest_unmap_l1e(curr, pl1e);
+ return GNTST_general_error;
+ }
+
l1pg = mfn_to_page(gl1mfn);
- page_lock(l1pg);
+ if ( !page_lock(l1pg) )
+ {
+ put_page(l1pg);
+ guest_unmap_l1e(curr, pl1e);
+ return GNTST_general_error;
+ }
+
+ if ( (l1pg->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table )
+ {
+ page_unlock(l1pg);
+ put_page(l1pg);
+ guest_unmap_l1e(curr, pl1e);
+ return GNTST_general_error;
+ }
+
ol1e = *pl1e;
if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, l1e_empty(),
gl1mfn, curr, 0)) )
{
page_unlock(l1pg);
+ put_page(l1pg);
MEM_LOG("Cannot delete PTE entry at %p", (unsigned long *)pl1e);
guest_unmap_l1e(curr, pl1e);
return GNTST_general_error;
}
page_unlock(l1pg);
+ put_page(l1pg);
guest_unmap_l1e(curr, pl1e);
rc = replace_grant_va_mapping(addr, frame, ol1e, curr);
@@ -3480,28 +3486,45 @@ int do_update_va_mapping(unsigned long v
l1_pgentry_t val = l1e_from_intpte(val64);
struct vcpu *v = current;
struct domain *d = v->domain;
+ struct page_info *gl1pg;
l1_pgentry_t *pl1e;
unsigned long vmask, bmap_ptr, gl1mfn;
cpumask_t pmask;
- int rc = 0;
+ int rc;
perfc_incr(calls_to_update_va);
- if ( unlikely(!access_ok(va, 1) && !paging_mode_external(d)) )
- return -EINVAL;
-
rc = xsm_update_va_mapping(d, val);
if ( rc )
return rc;
+ rc = -EINVAL;
pl1e = guest_map_l1e(v, va, &gl1mfn);
+ if ( unlikely(!pl1e || !get_page_from_pagenr(gl1mfn, d)) )
+ goto out;
- if ( unlikely(!pl1e || !mod_l1_entry(pl1e, val, gl1mfn, 0)) )
- rc = -EINVAL;
+ gl1pg = mfn_to_page(gl1mfn);
+ if ( !page_lock(gl1pg) )
+ {
+ put_page(gl1pg);
+ goto out;
+ }
+
+ if ( (gl1pg->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table )
+ {
+ page_unlock(gl1pg);
+ put_page(gl1pg);
+ goto out;
+ }
+
+ rc = mod_l1_entry(pl1e, val, gl1mfn, 0) ? 0 : -EINVAL;
+ page_unlock(gl1pg);
+ put_page(gl1pg);
+
+ out:
if ( pl1e )
guest_unmap_l1e(v, pl1e);
- pl1e = NULL;
process_deferred_ops();
@@ -4122,15 +4145,25 @@ int ptwr_do_page_fault(struct vcpu *v, u
/* Attempt to read the PTE that maps the VA being accessed. */
guest_get_eff_l1e(v, addr, &pte);
- page = l1e_get_page(pte);
/* We are looking only for read-only mappings of p.t. pages. */
if ( ((l1e_get_flags(pte) & (_PAGE_PRESENT|_PAGE_RW)) != _PAGE_PRESENT) ||
- !mfn_valid(l1e_get_pfn(pte)) ||
- ((page->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table) ||
- ((page->u.inuse.type_info & PGT_count_mask) == 0) ||
- (page_get_owner(page) != d) )
+ !get_page_from_pagenr(l1e_get_pfn(pte), d) )
+ goto bail;
+
+ page = l1e_get_page(pte);
+ if ( !page_lock(page) )
+ {
+ put_page(page);
+ goto bail;
+ }
+
+ if ( (page->u.inuse.type_info & PGT_type_mask) != PGT_l1_page_table )
+ {
+ page_unlock(page);
+ put_page(page);
goto bail;
+ }
ptwr_ctxt.ctxt.regs = regs;
ptwr_ctxt.ctxt.force_writeback = 0;
@@ -4139,9 +4172,11 @@ int ptwr_do_page_fault(struct vcpu *v, u
ptwr_ctxt.cr2 = addr;
ptwr_ctxt.pte = pte;
- page_lock(page);
rc = x86_emulate(&ptwr_ctxt.ctxt, &ptwr_emulate_ops);
+
page_unlock(page);
+ put_page(page);
+
if ( rc == X86EMUL_UNHANDLEABLE )
goto bail;
--- a/xen/arch/x86/mm/shadow/common.c
+++ b/xen/arch/x86/mm/shadow/common.c
@@ -1685,9 +1685,6 @@ shadow_free_p2m_page(struct domain *d, s
/* Free should not decrement domain's total allocation, since
* these pages were allocated without an owner. */
page_set_owner(pg, NULL);
-#if defined(__x86_64__)
- spin_lock_init(&pg->lock);
-#endif
free_domheap_pages(pg, 0);
d->arch.paging.shadow.p2m_pages--;
perfc_decr(shadow_alloc_count);
@@ -1801,16 +1798,7 @@ static unsigned int sh_set_allocation(st
* may get overwritten, so need to clear it here.
*/
for ( j = 0; j < 1U << order; j++ )
- {
page_set_owner(&((struct page_info *)sp)[j], NULL);
-#if defined(__x86_64__)
- /*
- * Re-instate lock field which we overwrite with shadow_page_info.
- * This was safe, since the lock is only used on guest pages.
- */
- spin_lock_init(&((struct page_info *)sp)[j].lock);
-#endif
- }
d->arch.paging.shadow.free_pages -= 1 << order;
d->arch.paging.shadow.total_pages -= 1 << order;
free_domheap_pages((struct page_info *)sp, order);
--- a/xen/include/asm-x86/mm.h
+++ b/xen/include/asm-x86/mm.h
@@ -46,10 +46,6 @@ struct page_info
} u;
-#if defined(__x86_64__)
- spinlock_t lock;
-#endif
-
union {
/*
* Timestamp from 'TLB clock', used to avoid extra safety flushes.
@@ -127,27 +123,25 @@ struct page_info
/* Has this page been *partially* validated for use as its current type? */
#define _PGT_partial PG_shift(7)
#define PGT_partial PG_mask(1, 7)
+ /* Page is locked? */
+#define _PGT_locked PG_shift(8)
+#define PGT_locked PG_mask(1, 8)
/* Count of uses of this frame as its current type. */
-#define PGT_count_width PG_shift(7)
+#define PGT_count_width PG_shift(8)
#define PGT_count_mask ((1UL<<PGT_count_width)-1)
/* Cleared when the owning guest 'frees' this page. */
#define _PGC_allocated PG_shift(1)
#define PGC_allocated PG_mask(1, 1)
-#if defined(__i386__)
- /* Page is locked? */
-# define _PGC_locked PG_shift(2)
-# define PGC_locked PG_mask(1, 2)
-#endif
/* Set when is using a page as a page table */
-#define _PGC_page_table PG_shift(3)
-#define PGC_page_table PG_mask(1, 3)
+#define _PGC_page_table PG_shift(2)
+#define PGC_page_table PG_mask(1, 2)
/* 3-bit PAT/PCD/PWT cache-attribute hint. */
-#define PGC_cacheattr_base PG_shift(6)
-#define PGC_cacheattr_mask PG_mask(7, 6)
+#define PGC_cacheattr_base PG_shift(5)
+#define PGC_cacheattr_mask PG_mask(7, 5)
/* Count of references to this frame. */
-#define PGC_count_width PG_shift(6)
+#define PGC_count_width PG_shift(5)
#define PGC_count_mask ((1UL<<PGC_count_width)-1)
#define is_xen_heap_page(page) is_xen_heap_mfn(page_to_mfn(page))
--- a/xen/include/asm-x86/paging.h
+++ b/xen/include/asm-x86/paging.h
@@ -336,7 +336,7 @@ void paging_dump_vcpu_info(struct vcpu *
* Access to the guest pagetables */
/* Get a mapping of a PV guest's l1e for this virtual address. */
-static inline void *
+static inline l1_pgentry_t *
guest_map_l1e(struct vcpu *v, unsigned long addr, unsigned long *gl1mfn)
{
l2_pgentry_t l2e;
@@ -354,15 +354,14 @@ guest_map_l1e(struct vcpu *v, unsigned l
!= _PAGE_PRESENT )
return NULL;
*gl1mfn = l2e_get_pfn(l2e);
- return &__linear_l1_table[l1_linear_offset(addr)];
+ return (l1_pgentry_t *)map_domain_page(*gl1mfn) + l1_table_offset(addr);
}
/* Pull down the mapping we got from guest_map_l1e() */
static inline void
guest_unmap_l1e(struct vcpu *v, void *p)
{
- if ( unlikely(paging_mode_translate(v->domain)) )
- unmap_domain_page(p);
+ unmap_domain_page(p);
}
/* Read the guest's l1e that maps this address. */

14
19151-xend-class-dereg.patch Normal file
View File

@ -0,0 +1,14 @@
Index: xen-3.3.1-testing/tools/python/xen/xend/XendAPIStore.py
===================================================================
--- xen-3.3.1-testing.orig/tools/python/xen/xend/XendAPIStore.py
+++ xen-3.3.1-testing/tools/python/xen/xend/XendAPIStore.py
@@ -33,7 +33,8 @@ def register(uuid, type, inst):
def deregister(uuid, type):
old = get(uuid, type)
- del __classes[(uuid, type)]
+ if old is not None:
+ del __classes[(uuid, type)]
return old
def get(uuid, type):

13
19152-xm-man-page.patch Normal file
View File

@ -0,0 +1,13 @@
Index: xen-3.3.1-testing/docs/man/xm.pod.1
===================================================================
--- xen-3.3.1-testing.orig/docs/man/xm.pod.1
+++ xen-3.3.1-testing/docs/man/xm.pod.1
@@ -67,6 +67,8 @@ The attached console will perform much l
so running curses based interfaces over the console B<is not
advised>. Vi tends to get very odd when using it over this interface.
+Use the key combination Ctrl+] to detach the domain console.
+
=item B<create> I<configfile> [I<OPTIONS>] [I<vars>]..
The create subcommand requires a config file and can optionally take a

17
19153-xm-noxen-error.patch Normal file
View File

@ -0,0 +1,17 @@
Index: xen-3.3.1-testing/tools/python/xen/xm/main.py
===================================================================
--- xen-3.3.1-testing.orig/tools/python/xen/xm/main.py
+++ xen-3.3.1-testing/tools/python/xen/xm/main.py
@@ -58,7 +58,11 @@ from xen.util.acmpolicy import ACM_LABEL
import XenAPI
import xen.lowlevel.xc
-xc = xen.lowlevel.xc.xc()
+try:
+ xc = xen.lowlevel.xc.xc()
+except Exception, ex:
+ print >>sys.stderr, ("Is xen kernel running?")
+ sys.exit(1)
import inspect
from xen.xend import XendOptions

104
blktap-error-handling.patch
View File

@ -55,11 +55,95 @@ Index: xen-3.3.1-testing/tools/blktap/lib/xenbus.c
#include "blktaplib.h" #include "blktaplib.h"
#include "list.h" #include "list.h"
#include "xs_api.h" #include "xs_api.h"
@@ -149,6 +150,37 @@ static int backend_remove(struct xs_hand @@ -149,6 +150,125 @@ static int backend_remove(struct xs_hand
return 0; return 0;
} }
+static int check_image(struct backend_info *be, const char** errmsg) +static int check_sharing(struct xs_handle *h, struct backend_info *be)
+{
+ char *dom_uuid;
+ char *cur_dom_uuid;
+ char *path;
+ char *mode;
+ char *params;
+ char **domains;
+ char **devices;
+ int i, j;
+ unsigned int num_dom, num_dev;
+ blkif_info_t *info;
+ int ret = 0;
+
+ /* If the mode contains '!' or doesn't contain 'w' don't check anything */
+ xs_gather(h, be->backpath, "mode", NULL, &mode, NULL);
+ if (strchr(mode, '!'))
+ goto out;
+ if (strchr(mode, 'w') == NULL)
+ goto out;
+
+ /* Get the UUID of the domain we want to attach to */
+ if (asprintf(&path, "/local/domain/%ld", be->frontend_id) == -1)
+ goto fail;
+ xs_gather(h, path, "vm", NULL, &dom_uuid, NULL);
+ free(path);
+
+ /* Iterate through the devices of all VMs */
+ domains = xs_directory(h, XBT_NULL, "backend/tap", &num_dom);
+ for (i = 0; !ret && (i < num_dom); i++) {
+
+ /* If it's the same VM, no action needed */
+ if (asprintf(&path, "/local/domain/%s", domains[i]) == -1) {
+ ret = -1;
+ break;
+ }
+ xs_gather(h, path, "vm", NULL, &cur_dom_uuid, NULL);
+ free(path);
+
+ if (!strcmp(cur_dom_uuid, dom_uuid)) {
+ free(cur_dom_uuid);
+ continue;
+ }
+
+ /* Check the devices */
+ if (asprintf(&path, "backend/tap/%s", domains[i]) == -1) {
+ ret = -1;
+ free(cur_dom_uuid);
+ break;
+ }
+ devices = xs_directory(h, XBT_NULL, path, &num_dev);
+ free(path);
+
+ for (j = 0; !ret && (j < num_dev); j++) {
+ if (asprintf(&path, "backend/tap/%s/%s", domains[i], devices[j]) == -1) {
+ ret = -1;
+ break;
+ }
+ xs_gather(h, path, "params", NULL, &params, NULL);
+ free(path);
+
+ info = be->blkif->info;
+ if (strcmp(params, info->params)) {
+ ret = -1;
+ }
+
+ free(params);
+ }
+
+ free(cur_dom_uuid);
+ free(devices);
+ }
+ free(domains);
+ free(dom_uuid);
+ goto out;
+
+fail:
+ ret = -1;
+out:
+ free(mode);
+ return ret;
+}
+
+static int check_image(struct xs_handle *h, struct backend_info *be,
+ const char** errmsg)
+{ +{
+ const char *path; + const char *path;
+ int mode; + int mode;
@ -85,7 +169,11 @@ Index: xen-3.3.1-testing/tools/blktap/lib/xenbus.c
+ return -1; + return -1;
+ } + }
+ +
+ /* TODO Check that the image is not attached to a different VM */ + /* Check that the image is not attached to a different VM */
+ if (check_sharing(h, be)) {
+ *errmsg = "File already in use by other domain";
+ return -1;
+ }
+ +
+ return 0; + return 0;
+} +}
@ -93,7 +181,7 @@ Index: xen-3.3.1-testing/tools/blktap/lib/xenbus.c
static void ueblktap_setup(struct xs_handle *h, char *bepath) static void ueblktap_setup(struct xs_handle *h, char *bepath)
{ {
struct backend_info *be; struct backend_info *be;
@@ -156,6 +188,7 @@ static void ueblktap_setup(struct xs_han @@ -156,6 +276,7 @@ static void ueblktap_setup(struct xs_han
int len, er, deverr; int len, er, deverr;
long int pdev = 0, handle; long int pdev = 0, handle;
blkif_info_t *blk; blkif_info_t *blk;
@ -101,17 +189,17 @@ Index: xen-3.3.1-testing/tools/blktap/lib/xenbus.c
be = be_lookup_be(bepath); be = be_lookup_be(bepath);
if (be == NULL) if (be == NULL)
@@ -211,6 +244,9 @@ static void ueblktap_setup(struct xs_han @@ -211,6 +332,9 @@ static void ueblktap_setup(struct xs_han
be->pdev = pdev; be->pdev = pdev;
} }
+ if (check_image(be, &errmsg)) + if (check_image(h, be, &errmsg))
+ goto fail; + goto fail;
+ +
er = blkif_init(be->blkif, handle, be->pdev, be->readonly); er = blkif_init(be->blkif, handle, be->pdev, be->readonly);
if (er != 0) { if (er != 0) {
DPRINTF("Unable to open device %s\n",blk->params); DPRINTF("Unable to open device %s\n",blk->params);
@@ -246,12 +282,21 @@ static void ueblktap_setup(struct xs_han @@ -246,12 +370,21 @@ static void ueblktap_setup(struct xs_han
} }
be->blkif->state = CONNECTED; be->blkif->state = CONNECTED;
@ -134,7 +222,7 @@ Index: xen-3.3.1-testing/tools/blktap/lib/xenbus.c
close: close:
if (path) if (path)
free(path); free(path);
@@ -286,7 +331,8 @@ static void ueblktap_probe(struct xs_han @@ -286,7 +419,8 @@ static void ueblktap_probe(struct xs_han
len = strsep_len(bepath, '/', 7); len = strsep_len(bepath, '/', 7);
if (len < 0) if (len < 0)
goto free_be; goto free_be;

81
blktap-pv-cdrom.patch
View File

@ -1,8 +1,6 @@
Index: xen-3.3.1-testing/tools/blktap/drivers/block-cdrom.c --- /dev/null
=================================================================== +++ b/tools/blktap/drivers/block-cdrom.c
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 @@ -0,0 +1,536 @@
+++ xen-3.3.1-testing/tools/blktap/drivers/block-cdrom.c 2008-09-28 13:14:49.000000000 -0600
@@ -0,0 +1,535 @@
+/* block-cdrom.c +/* block-cdrom.c
+ * + *
+ * simple slow synchronous cdrom disk implementation. Based off + * simple slow synchronous cdrom disk implementation. Based off
@ -113,16 +111,17 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/block-cdrom.c
+ ioctl(prv->fd, BLKSSZGET, &s->sector_size); + ioctl(prv->fd, BLKSSZGET, &s->sector_size);
+ +
+ if (s->sector_size != CDROM_DEFAULT_SECTOR_SIZE) + if (s->sector_size != CDROM_DEFAULT_SECTOR_SIZE)
+ DPRINTF("Note: sector size is %ld (not %d)\n", + DPRINTF("Note: sector size is %llu (not %d)\n",
+ s->sector_size, CDROM_DEFAULT_SECTOR_SIZE); + (long long unsigned)s->sector_size,
+ CDROM_DEFAULT_SECTOR_SIZE);
+ } + }
+#else +#else
+ s->sector_size = CDROM_DEFAULT_SECTOR_SIZE; + s->sector_size = CDROM_DEFAULT_SECTOR_SIZE;
+#endif +#endif
+ DPRINTF("Block Device: Image size: %llu", + DPRINTF("Block Device: Image size: %llu"
+ (long long unsigned)s->size); + " media_present: %d sector_size: %llu\n",
+ DPRINTF("\t media_present: %d sector_size: %lu\n", + (long long unsigned)s->size, prv->media_present,
+ prv->media_present, s->sector_size); + (long long unsigned)s->sector_size);
+ } else { + } else {
+ /*Local file? try fstat instead*/ + /*Local file? try fstat instead*/
+ prv->dev_type = FILE_DEVICE; + prv->dev_type = FILE_DEVICE;
@ -538,10 +537,8 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/block-cdrom.c
+ .td_get_parent_id = tdcdrom_get_parent_id, + .td_get_parent_id = tdcdrom_get_parent_id,
+ .td_validate_parent = tdcdrom_validate_parent + .td_validate_parent = tdcdrom_validate_parent
+}; +};
Index: xen-3.3.1-testing/xen/include/public/io/cdromif.h --- /dev/null
=================================================================== +++ b/xen/include/public/io/cdromif.h
--- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/include/public/io/cdromif.h 2008-09-28 13:11:57.000000000 -0600
@@ -0,0 +1,120 @@ @@ -0,0 +1,120 @@
+/****************************************************************************** +/******************************************************************************
+ * cdromif.h + * cdromif.h
@ -663,11 +660,9 @@ Index: xen-3.3.1-testing/xen/include/public/io/cdromif.h
+ sizeof(struct vcd_generic_command) - sizeof(struct request_sense)) + sizeof(struct vcd_generic_command) - sizeof(struct request_sense))
+ +
+#endif +#endif
Index: xen-3.3.1-testing/tools/blktap/drivers/Makefile --- a/tools/blktap/drivers/Makefile
=================================================================== +++ b/tools/blktap/drivers/Makefile
--- xen-3.3.1-testing.orig/tools/blktap/drivers/Makefile 2008-09-28 13:11:54.000000000 -0600 @@ -24,8 +24,9 @@ CRYPT_LIB := -lcrypto
+++ xen-3.3.1-testing/tools/blktap/drivers/Makefile 2008-09-28 13:11:57.000000000 -0600
@@ -24,8 +24,9 @@
$(warning *** libgcrypt not installed: falling back to libcrypto ***) $(warning *** libgcrypt not installed: falling back to libcrypto ***)
endif endif
@ -679,7 +674,7 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/Makefile
BLK-OBJS-y := block-aio.o BLK-OBJS-y := block-aio.o
BLK-OBJS-y += block-sync.o BLK-OBJS-y += block-sync.o
@@ -33,6 +34,7 @@ @@ -33,6 +34,7 @@ BLK-OBJS-y += block-vmdk.o
BLK-OBJS-y += block-ram.o BLK-OBJS-y += block-ram.o
BLK-OBJS-y += block-qcow.o BLK-OBJS-y += block-qcow.o
BLK-OBJS-y += block-qcow2.o BLK-OBJS-y += block-qcow2.o
@ -687,11 +682,9 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/Makefile
BLK-OBJS-y += aes.o BLK-OBJS-y += aes.o
BLK-OBJS-y += tapaio.o BLK-OBJS-y += tapaio.o
BLK-OBJS-$(CONFIG_Linux) += blk_linux.o BLK-OBJS-$(CONFIG_Linux) += blk_linux.o
Index: xen-3.3.1-testing/tools/blktap/drivers/tapdisk.h --- a/tools/blktap/drivers/tapdisk.h
=================================================================== +++ b/tools/blktap/drivers/tapdisk.h
--- xen-3.3.1-testing.orig/tools/blktap/drivers/tapdisk.h 2008-09-28 13:11:56.000000000 -0600 @@ -137,6 +137,9 @@ struct tap_disk {
+++ xen-3.3.1-testing/tools/blktap/drivers/tapdisk.h 2008-09-28 13:11:57.000000000 -0600
@@ -137,6 +137,9 @@
int (*td_get_parent_id) (struct disk_driver *dd, struct disk_id *id); int (*td_get_parent_id) (struct disk_driver *dd, struct disk_id *id);
int (*td_validate_parent)(struct disk_driver *dd, int (*td_validate_parent)(struct disk_driver *dd,
struct disk_driver *p, td_flag_t flags); struct disk_driver *p, td_flag_t flags);
@ -701,7 +694,7 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/tapdisk.h
}; };
typedef struct disk_info { typedef struct disk_info {
@@ -160,6 +163,7 @@ @@ -160,6 +163,7 @@ extern struct tap_disk tapdisk_vmdk;
extern struct tap_disk tapdisk_ram; extern struct tap_disk tapdisk_ram;
extern struct tap_disk tapdisk_qcow; extern struct tap_disk tapdisk_qcow;
extern struct tap_disk tapdisk_qcow2; extern struct tap_disk tapdisk_qcow2;
@ -709,7 +702,7 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/tapdisk.h
/*Define Individual Disk Parameters here */ /*Define Individual Disk Parameters here */
@@ -240,6 +244,17 @@ @@ -240,6 +244,17 @@ static disk_info_t ioemu_disk = {
#endif #endif
}; };
@ -727,7 +720,7 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/tapdisk.h
/*Main disk info array */ /*Main disk info array */
static disk_info_t *dtypes[] = { static disk_info_t *dtypes[] = {
&aio_disk, &aio_disk,
@@ -249,6 +264,7 @@ @@ -249,6 +264,7 @@ static disk_info_t *dtypes[] = {
&qcow_disk, &qcow_disk,
&qcow2_disk, &qcow2_disk,
&ioemu_disk, &ioemu_disk,
@ -735,11 +728,9 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/tapdisk.h
}; };
typedef struct driver_list_entry { typedef struct driver_list_entry {
Index: xen-3.3.1-testing/tools/blktap/lib/blktaplib.h --- a/tools/blktap/lib/blktaplib.h
=================================================================== +++ b/tools/blktap/lib/blktaplib.h
--- xen-3.3.1-testing.orig/tools/blktap/lib/blktaplib.h 2008-09-28 13:11:56.000000000 -0600 @@ -221,6 +221,7 @@ typedef struct msg_pid {
+++ xen-3.3.1-testing/tools/blktap/lib/blktaplib.h 2008-09-28 13:11:57.000000000 -0600
@@ -221,6 +221,7 @@
#define DISK_TYPE_QCOW 4 #define DISK_TYPE_QCOW 4
#define DISK_TYPE_QCOW2 5 #define DISK_TYPE_QCOW2 5
#define DISK_TYPE_IOEMU 6 #define DISK_TYPE_IOEMU 6
@ -747,10 +738,8 @@ Index: xen-3.3.1-testing/tools/blktap/lib/blktaplib.h
/* xenstore/xenbus: */ /* xenstore/xenbus: */
#define DOMNAME "Domain-0" #define DOMNAME "Domain-0"
Index: xen-3.3.1-testing/xen/include/public/io/blkif.h --- a/xen/include/public/io/blkif.h
=================================================================== +++ b/xen/include/public/io/blkif.h
--- xen-3.3.1-testing.orig/xen/include/public/io/blkif.h 2008-09-16 10:31:07.000000000 -0600
+++ xen-3.3.1-testing/xen/include/public/io/blkif.h 2008-09-28 13:11:57.000000000 -0600
@@ -76,6 +76,10 @@ @@ -76,6 +76,10 @@
* "feature-flush-cache" node! * "feature-flush-cache" node!
*/ */
@ -762,11 +751,9 @@ Index: xen-3.3.1-testing/xen/include/public/io/blkif.h
/* /*
* Maximum scatter/gather segments per request. * Maximum scatter/gather segments per request.
Index: xen-3.3.1-testing/tools/blktap/drivers/tapdisk.c --- a/tools/blktap/drivers/tapdisk.c
=================================================================== +++ b/tools/blktap/drivers/tapdisk.c
--- xen-3.3.1-testing.orig/tools/blktap/drivers/tapdisk.c 2008-09-16 10:31:02.000000000 -0600 @@ -735,6 +735,22 @@ static void get_io_request(struct td_sta
+++ xen-3.3.1-testing/tools/blktap/drivers/tapdisk.c 2008-09-28 13:11:57.000000000 -0600
@@ -735,6 +735,22 @@
goto out; goto out;
} }
break; break;
@ -789,11 +776,9 @@ Index: xen-3.3.1-testing/tools/blktap/drivers/tapdisk.c
default: default:
DPRINTF("Unknown block operation\n"); DPRINTF("Unknown block operation\n");
break; break;
Index: xen-3.3.1-testing/tools/python/xen/xend/server/BlktapController.py --- a/tools/python/xen/xend/server/BlktapController.py
=================================================================== +++ b/tools/python/xen/xend/server/BlktapController.py
--- xen-3.3.1-testing.orig/tools/python/xen/xend/server/BlktapController.py 2008-09-16 10:31:03.000000000 -0600 @@ -14,8 +14,8 @@ blktap_disk_types = [
+++ xen-3.3.1-testing/tools/python/xen/xend/server/BlktapController.py 2008-09-28 13:11:57.000000000 -0600
@@ -14,8 +14,8 @@
'ram', 'ram',
'qcow', 'qcow',
'qcow2', 'qcow2',

2
dom-print.patch
View File

@ -22,7 +22,7 @@
+ continue; + continue;
+ } + }
+ } + }
printk(" DomPage %p: caf=%08x, taf=%" PRtype_info "\n", printk(" DomPage %p: caf=%08lx, taf=%" PRtype_info "\n",
_p(page_to_mfn(page)), _p(page_to_mfn(page)),
page->count_info, page->u.inuse.type_info); page->count_info, page->u.inuse.type_info);
--- a/xen/common/keyhandler.c --- a/xen/common/keyhandler.c

51
hv_xen_base.patch
View File

@ -1,9 +1,9 @@
%patch %patch
Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/domain.h Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/domain.h
=================================================================== ===================================================================
--- xen-3.3.1-testing.orig/xen/include/asm-x86/hvm/domain.h --- xen-3.3.1-testing.orig/xen/include/asm-x86/hvm/domain.h 2009-01-05 13:27:58.000000000 -0700
+++ xen-3.3.1-testing/xen/include/asm-x86/hvm/domain.h +++ xen-3.3.1-testing/xen/include/asm-x86/hvm/domain.h 2009-01-22 13:23:44.000000000 -0700
@@ -82,6 +82,7 @@ struct hvm_domain { @@ -82,6 +82,7 @@
struct vmx_domain vmx; struct vmx_domain vmx;
struct svm_domain svm; struct svm_domain svm;
}; };
@ -13,8 +13,8 @@ Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/domain.h
#endif /* __ASM_X86_HVM_DOMAIN_H__ */ #endif /* __ASM_X86_HVM_DOMAIN_H__ */
Index: xen-3.3.1-testing/xen/arch/x86/hvm/Makefile Index: xen-3.3.1-testing/xen/arch/x86/hvm/Makefile
=================================================================== ===================================================================
--- xen-3.3.1-testing.orig/xen/arch/x86/hvm/Makefile --- xen-3.3.1-testing.orig/xen/arch/x86/hvm/Makefile 2009-01-05 13:27:57.000000000 -0700
+++ xen-3.3.1-testing/xen/arch/x86/hvm/Makefile +++ xen-3.3.1-testing/xen/arch/x86/hvm/Makefile 2009-01-22 13:23:44.000000000 -0700
@@ -1,5 +1,6 @@ @@ -1,5 +1,6 @@
subdir-y += svm subdir-y += svm
subdir-y += vmx subdir-y += vmx
@ -24,8 +24,8 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/Makefile
obj-y += hvm.o obj-y += hvm.o
Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
=================================================================== ===================================================================
--- xen-3.3.1-testing.orig/xen/arch/x86/hvm/hvm.c --- xen-3.3.1-testing.orig/xen/arch/x86/hvm/hvm.c 2009-01-05 13:27:57.000000000 -0700
+++ xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c +++ xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c 2009-01-22 13:23:44.000000000 -0700
@@ -44,6 +44,7 @@ @@ -44,6 +44,7 @@
#include <asm/mc146818rtc.h> #include <asm/mc146818rtc.h>
#include <asm/spinlock.h> #include <asm/spinlock.h>
@ -34,7 +34,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
#include <asm/hvm/vpt.h> #include <asm/hvm/vpt.h>
#include <asm/hvm/support.h> #include <asm/hvm/support.h>
#include <asm/hvm/cacheattr.h> #include <asm/hvm/cacheattr.h>
@@ -361,6 +362,7 @@ void hvm_domain_relinquish_resources(str @@ -361,6 +362,7 @@
void hvm_domain_destroy(struct domain *d) void hvm_domain_destroy(struct domain *d)
{ {
@ -42,7 +42,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
hvm_funcs.domain_destroy(d); hvm_funcs.domain_destroy(d);
rtc_deinit(d); rtc_deinit(d);
stdvga_deinit(d); stdvga_deinit(d);
@@ -644,8 +646,14 @@ int hvm_vcpu_initialise(struct vcpu *v) @@ -644,8 +646,14 @@
{ {
int rc; int rc;
@ -57,7 +57,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
if ( (rc = hvm_funcs.vcpu_initialise(v)) != 0 ) if ( (rc = hvm_funcs.vcpu_initialise(v)) != 0 )
goto fail2; goto fail2;
@@ -692,12 +700,14 @@ int hvm_vcpu_initialise(struct vcpu *v) @@ -692,12 +700,14 @@
hvm_funcs.vcpu_destroy(v); hvm_funcs.vcpu_destroy(v);
fail2: fail2:
vlapic_destroy(v); vlapic_destroy(v);
@ -72,7 +72,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
hvm_vcpu_cacheattr_destroy(v); hvm_vcpu_cacheattr_destroy(v);
vlapic_destroy(v); vlapic_destroy(v);
hvm_funcs.vcpu_destroy(v); hvm_funcs.vcpu_destroy(v);
@@ -1645,7 +1655,7 @@ void hvm_cpuid(unsigned int input, unsig @@ -1645,7 +1655,7 @@
struct vcpu *v = current; struct vcpu *v = current;
if ( cpuid_hypervisor_leaves(input, eax, ebx, ecx, edx) ) if ( cpuid_hypervisor_leaves(input, eax, ebx, ecx, edx) )
@ -81,7 +81,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
domain_cpuid(v->domain, input, *ecx, eax, ebx, ecx, edx); domain_cpuid(v->domain, input, *ecx, eax, ebx, ecx, edx);
@@ -1657,6 +1667,8 @@ void hvm_cpuid(unsigned int input, unsig @@ -1657,6 +1667,8 @@
if ( vlapic_hw_disabled(vcpu_vlapic(v)) ) if ( vlapic_hw_disabled(vcpu_vlapic(v)) )
__clear_bit(X86_FEATURE_APIC & 31, edx); __clear_bit(X86_FEATURE_APIC & 31, edx);
} }
@ -90,7 +90,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
} }
void hvm_rdtsc_intercept(struct cpu_user_regs *regs) void hvm_rdtsc_intercept(struct cpu_user_regs *regs)
@@ -1747,6 +1759,8 @@ int hvm_msr_read_intercept(struct cpu_us @@ -1747,6 +1759,8 @@
break; break;
default: default:
@ -99,7 +99,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
return hvm_funcs.msr_read_intercept(regs); return hvm_funcs.msr_read_intercept(regs);
} }
@@ -1835,6 +1849,8 @@ int hvm_msr_write_intercept(struct cpu_u @@ -1835,6 +1849,8 @@
break; break;
default: default:
@ -108,7 +108,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
return hvm_funcs.msr_write_intercept(regs); return hvm_funcs.msr_write_intercept(regs);
} }
@@ -2002,6 +2018,10 @@ int hvm_do_hypercall(struct cpu_user_reg @@ -2002,6 +2018,10 @@
case 0: case 0:
break; break;
} }
@ -119,7 +119,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
if ( (eax >= NR_hypercalls) || !hvm_hypercall32_table[eax] ) if ( (eax >= NR_hypercalls) || !hvm_hypercall32_table[eax] )
{ {
@@ -2503,6 +2523,15 @@ long do_hvm_op(unsigned long op, XEN_GUE @@ -2503,6 +2523,15 @@
rc = -EINVAL; rc = -EINVAL;
break; break;
@ -137,9 +137,9 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hvm.c
if ( rc == 0 ) if ( rc == 0 )
Index: xen-3.3.1-testing/xen/include/public/arch-x86/hvm/save.h Index: xen-3.3.1-testing/xen/include/public/arch-x86/hvm/save.h
=================================================================== ===================================================================
--- xen-3.3.1-testing.orig/xen/include/public/arch-x86/hvm/save.h --- xen-3.3.1-testing.orig/xen/include/public/arch-x86/hvm/save.h 2009-01-05 13:27:58.000000000 -0700
+++ xen-3.3.1-testing/xen/include/public/arch-x86/hvm/save.h +++ xen-3.3.1-testing/xen/include/public/arch-x86/hvm/save.h 2009-01-23 08:35:50.000000000 -0700
@@ -38,7 +38,7 @@ struct hvm_save_header { @@ -38,7 +38,7 @@
uint32_t version; /* File format version */ uint32_t version; /* File format version */
uint64_t changeset; /* Version of Xen that saved this file */ uint64_t changeset; /* Version of Xen that saved this file */
uint32_t cpuid; /* CPUID[0x01][%eax] on the saving machine */ uint32_t cpuid; /* CPUID[0x01][%eax] on the saving machine */
@ -148,7 +148,7 @@ Index: xen-3.3.1-testing/xen/include/public/arch-x86/hvm/save.h
}; };
DECLARE_HVM_SAVE_TYPE(HEADER, 1, struct hvm_save_header); DECLARE_HVM_SAVE_TYPE(HEADER, 1, struct hvm_save_header);
@@ -421,9 +421,22 @@ struct hvm_hw_mtrr { @@ -421,9 +421,23 @@
DECLARE_HVM_SAVE_TYPE(MTRR, 14, struct hvm_hw_mtrr); DECLARE_HVM_SAVE_TYPE(MTRR, 14, struct hvm_hw_mtrr);
@ -163,6 +163,7 @@ Index: xen-3.3.1-testing/xen/include/public/arch-x86/hvm/save.h
+struct hvm_hyperv_cpu { +struct hvm_hyperv_cpu {
+ uint64_t control_msr; + uint64_t control_msr;
+ uint64_t version_msr; + uint64_t version_msr;
+ uint64_t pad[27]; //KYS: sles10 sp2 compatibility
+}; +};
+DECLARE_HVM_SAVE_TYPE(HYPERV_CPU, 16, struct hvm_hyperv_cpu); +DECLARE_HVM_SAVE_TYPE(HYPERV_CPU, 16, struct hvm_hyperv_cpu);
/* /*
@ -174,8 +175,8 @@ Index: xen-3.3.1-testing/xen/include/public/arch-x86/hvm/save.h
#endif /* __XEN_PUBLIC_HVM_SAVE_X86_H__ */ #endif /* __XEN_PUBLIC_HVM_SAVE_X86_H__ */
Index: xen-3.3.1-testing/xen/arch/x86/hvm/vlapic.c Index: xen-3.3.1-testing/xen/arch/x86/hvm/vlapic.c
=================================================================== ===================================================================
--- xen-3.3.1-testing.orig/xen/arch/x86/hvm/vlapic.c --- xen-3.3.1-testing.orig/xen/arch/x86/hvm/vlapic.c 2009-01-22 13:23:43.000000000 -0700
+++ xen-3.3.1-testing/xen/arch/x86/hvm/vlapic.c +++ xen-3.3.1-testing/xen/arch/x86/hvm/vlapic.c 2009-01-22 13:23:44.000000000 -0700
@@ -34,6 +34,7 @@ @@ -34,6 +34,7 @@
#include <asm/hvm/hvm.h> #include <asm/hvm/hvm.h>
#include <asm/hvm/io.h> #include <asm/hvm/io.h>
@ -184,7 +185,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/vlapic.c
#include <asm/hvm/vmx/vmx.h> #include <asm/hvm/vmx/vmx.h>
#include <public/hvm/ioreq.h> #include <public/hvm/ioreq.h>
#include <public/hvm/params.h> #include <public/hvm/params.h>
@@ -307,6 +308,7 @@ static int vlapic_accept_sipi(struct vcp @@ -307,6 +308,7 @@
hvm_vcpu_reset_state(v, trampoline_vector << 8, 0); hvm_vcpu_reset_state(v, trampoline_vector << 8, 0);
vcpu_unpause(v); vcpu_unpause(v);
@ -194,8 +195,8 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/vlapic.c
} }
Index: xen-3.3.1-testing/xen/include/public/hvm/params.h Index: xen-3.3.1-testing/xen/include/public/hvm/params.h
=================================================================== ===================================================================
--- xen-3.3.1-testing.orig/xen/include/public/hvm/params.h --- xen-3.3.1-testing.orig/xen/include/public/hvm/params.h 2009-01-05 13:27:58.000000000 -0700
+++ xen-3.3.1-testing/xen/include/public/hvm/params.h +++ xen-3.3.1-testing/xen/include/public/hvm/params.h 2009-01-22 13:23:44.000000000 -0700
@@ -93,6 +93,8 @@ @@ -93,6 +93,8 @@
/* ACPI S state: currently support S0 and S3 on x86. */ /* ACPI S state: currently support S0 and S3 on x86. */
#define HVM_PARAM_ACPI_S_STATE 14 #define HVM_PARAM_ACPI_S_STATE 14

46
hv_xen_extension.patch
View File

@ -2,7 +2,7 @@
Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/hvm_extensions.h Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/hvm_extensions.h
=================================================================== ===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/include/asm-x86/hvm/hvm_extensions.h 2008-12-08 13:58:31.000000000 -0700 +++ xen-3.3.1-testing/xen/include/asm-x86/hvm/hvm_extensions.h 2009-01-22 13:23:44.000000000 -0700
@@ -0,0 +1,165 @@ @@ -0,0 +1,165 @@
+/**************************************************************************** +/****************************************************************************
+ | + |
@ -172,14 +172,14 @@ Index: xen-3.3.1-testing/xen/include/asm-x86/hvm/hvm_extensions.h
Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/Makefile Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/Makefile
=================================================================== ===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/Makefile 2008-12-08 13:58:31.000000000 -0700 +++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/Makefile 2009-01-22 13:23:44.000000000 -0700
@@ -0,0 +1,2 @@ @@ -0,0 +1,2 @@
+obj-y += hv_intercept.o +obj-y += hv_intercept.o
+obj-y += hv_hypercall.o +obj-y += hv_hypercall.o
Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_errno.h Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_errno.h
=================================================================== ===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_errno.h 2008-12-08 13:58:31.000000000 -0700 +++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_errno.h 2009-01-22 13:23:44.000000000 -0700
@@ -0,0 +1,62 @@ @@ -0,0 +1,62 @@
+/**************************************************************************** +/****************************************************************************
+ | + |
@ -246,7 +246,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_errno.h
Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.c Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.c
=================================================================== ===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.c 2008-12-11 18:17:25.000000000 -0700 +++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.c 2009-01-22 13:23:44.000000000 -0700
@@ -0,0 +1,153 @@ @@ -0,0 +1,153 @@
+/**************************************************************************** +/****************************************************************************
+ | + |
@ -404,7 +404,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.c
Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.h Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.h
=================================================================== ===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.h 2008-12-08 13:58:31.000000000 -0700 +++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.h 2009-01-22 13:23:44.000000000 -0700
@@ -0,0 +1,46 @@ @@ -0,0 +1,46 @@
+/**************************************************************************** +/****************************************************************************
+ | + |
@ -455,8 +455,8 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_hypercall.h
Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c
=================================================================== ===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c 2008-12-09 12:42:50.000000000 -0700 +++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c 2009-01-26 12:33:27.000000000 -0700
@@ -0,0 +1,1002 @@ @@ -0,0 +1,1008 @@
+/**************************************************************************** +/****************************************************************************
+ | + |
+ | Copyright (c) [2007, 2008] Novell, Inc. + | Copyright (c) [2007, 2008] Novell, Inc.
@ -523,7 +523,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c
+ +
+ +
+static inline void +static inline void
+hv_hypercall_page_initialize(void *hypercall_page, hv_partition_t *curp); +hv_hypercall_page_initialize(void *hypercall_page);
+ +
+static inline void * +static inline void *
+get_virt_from_gmfn(struct domain *d, unsigned long gmfn) +get_virt_from_gmfn(struct domain *d, unsigned long gmfn)
@ -607,8 +607,12 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c
+ spin_unlock(&curp->lock); + spin_unlock(&curp->lock);
+ return; + return;
+ } + }
+ hv_hypercall_page_initialize(hypercall_page, curp); + hv_hypercall_page_initialize(hypercall_page);
+ unmap_domain_page(hypercall_page); + unmap_domain_page(hypercall_page);
+ if (hvm_funcs.guest_x86_mode(current) == 8)
+ curp->long_mode_guest = 1;
+ else
+ curp->long_mode_guest = 0;
+ curp->hypercall_msr = msr_content; + curp->hypercall_msr = msr_content;
+ spin_unlock(&curp->lock); + spin_unlock(&curp->lock);
+ cur_vcpu->flags |= HV_VCPU_UP; + cur_vcpu->flags |= HV_VCPU_UP;
@ -771,16 +775,10 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c
+} +}
+ +
+static inline void +static inline void
+hv_hypercall_page_initialize(void *hypercall_page, hv_partition_t *curp) +hv_hypercall_page_initialize(void *hypercall_page)
+{ +{
+ char *p; + char *p;
+ +
+ if (hvm_funcs.guest_x86_mode(current) == 8)
+ curp->long_mode_guest = 1;
+ else
+ curp->long_mode_guest = 0;
+
+
+ memset(hypercall_page, 0, PAGE_SIZE); + memset(hypercall_page, 0, PAGE_SIZE);
+ p = (char *)(hypercall_page) ; + p = (char *)(hypercall_page) ;
+ /* + /*
@ -1032,18 +1030,26 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c
+{ +{
+ struct hvm_hyperv_dom ctxt; + struct hvm_hyperv_dom ctxt;
+ hv_partition_t *curp; + hv_partition_t *curp;
+ void *hypercall_page;
+ +
+ if ( hvm_load_entry(HYPERV_DOM, h, &ctxt) != 0 ) + if ( hvm_load_entry(HYPERV_DOM, h, &ctxt) != 0 )
+ return -EINVAL; + return -EINVAL;
+ d->arch.hvm_domain.params[HVM_PARAM_EXTEND_HYPERVISOR] = ctxt.ext_id; + d->arch.hvm_domain.params[HVM_PARAM_EXTEND_HYPERVISOR] = 1;
+ if (ctxt.ext_id != 1)
+ return 0;
+ if (hyperv_initialize(d)) + if (hyperv_initialize(d))
+ return -EINVAL; + return -EINVAL;
+ curp = d->arch.hvm_domain.hyperv_handle; + curp = d->arch.hvm_domain.hyperv_handle;
+ +
+ curp->guest_id_msr = ctxt.guestid_msr; + curp->guest_id_msr = ctxt.guestid_msr;
+ curp->hypercall_msr = ctxt.hypercall_msr; + curp->hypercall_msr = ctxt.hypercall_msr;
+ /*
+ * We may have migrated from a sles10 host; re-initialize the
+ * hypercall page.
+ */
+ hypercall_page = get_virt_from_gmfn(d, (curp->hypercall_msr >>12));
+ if (hypercall_page == NULL)
+ return -EINVAL;
+ hv_hypercall_page_initialize(hypercall_page);
+ unmap_domain_page(hypercall_page);
+ curp->long_mode_guest = ctxt.long_mode; + curp->long_mode_guest = ctxt.long_mode;
+ return 0; + return 0;
+} +}
@ -1462,7 +1468,7 @@ Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_intercept.c
Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_shim.h Index: xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_shim.h
=================================================================== ===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_shim.h 2008-12-09 14:50:19.000000000 -0700 +++ xen-3.3.1-testing/xen/arch/x86/hvm/hyperv/hv_shim.h 2009-01-22 13:23:44.000000000 -0700
@@ -0,0 +1,285 @@ @@ -0,0 +1,285 @@
+/**************************************************************************** +/****************************************************************************
+ | + |

24
ioemu-blktap-fv-init.patch Normal file
View File

@ -0,0 +1,24 @@
Index: xen-3.3.1-testing/tools/ioemu-remote/hw/xen_machine_fv.c
===================================================================
--- xen-3.3.1-testing.orig/tools/ioemu-remote/hw/xen_machine_fv.c
+++ xen-3.3.1-testing/tools/ioemu-remote/hw/xen_machine_fv.c
@@ -185,6 +185,7 @@ void qemu_invalidate_map_cache(void)
#endif /* defined(MAPCACHE) */
+extern void init_blktap(void);
static void xen_init_fv(ram_addr_t ram_size, int vga_ram_size,
const char *boot_device, DisplayState *ds,
@@ -210,6 +211,11 @@ static void xen_init_fv(ram_addr_t ram_s
}
#endif
+#ifndef CONFIG_STUBDOM
+ /* Initialize tapdisk client */
+ init_blktap();
+#endif
+
#ifdef CONFIG_STUBDOM /* the hvmop is not supported on older hypervisors */
xc_set_hvm_param(xc_handle, domid, HVM_PARAM_DM_DOMAIN, DOMID_SELF);
#endif

2
x86-show-page-walk-early.patch
View File

@ -122,7 +122,7 @@
} }
--- a/xen/include/asm-x86/mm.h --- a/xen/include/asm-x86/mm.h
+++ b/xen/include/asm-x86/mm.h +++ b/xen/include/asm-x86/mm.h
@@ -331,6 +331,7 @@ TYPE_SAFE(unsigned long,mfn); @@ -330,6 +330,7 @@ TYPE_SAFE(unsigned long,mfn);
#define machine_to_phys_mapping ((unsigned long *)RDWR_MPT_VIRT_START) #define machine_to_phys_mapping ((unsigned long *)RDWR_MPT_VIRT_START)
#define INVALID_M2P_ENTRY (~0UL) #define INVALID_M2P_ENTRY (~0UL)
#define VALID_M2P(_e) (!((_e) & (1UL<<(BITS_PER_LONG-1)))) #define VALID_M2P(_e) (!((_e) & (1UL<<(BITS_PER_LONG-1))))

123
x86_64-page-info-pack.patch Normal file
View File

@ -0,0 +1,123 @@
References: bnc#470949
--- a/xen/arch/x86/mm/shadow/common.c
+++ b/xen/arch/x86/mm/shadow/common.c
@@ -1902,7 +1902,7 @@ static void sh_hash_audit_bucket(struct
{
if ( !page_is_out_of_sync(gpg) )
{
- SHADOW_ERROR("MFN %#lx shadowed (by %#"PRI_mfn")"
+ SHADOW_ERROR("MFN %#x shadowed (by %#"PRI_mfn")"
" and not OOS but has typecount %#lx\n",
sp->backpointer,
mfn_x(shadow_page_to_mfn(sp)),
@@ -1916,7 +1916,7 @@ static void sh_hash_audit_bucket(struct
if ( (gpg->u.inuse.type_info & PGT_type_mask) == PGT_writable_page
&& (gpg->u.inuse.type_info & PGT_count_mask) != 0 )
{
- SHADOW_ERROR("MFN %#lx shadowed (by %#"PRI_mfn")"
+ SHADOW_ERROR("MFN %#x shadowed (by %#"PRI_mfn")"
" but has typecount %#lx\n",
sp->backpointer, mfn_x(shadow_page_to_mfn(sp)),
gpg->u.inuse.type_info);
--- a/xen/arch/x86/mm/shadow/private.h
+++ b/xen/arch/x86/mm/shadow/private.h
@@ -190,7 +190,7 @@ struct shadow_page_info
struct {
union {
/* When in use, guest page we're a shadow of */
- unsigned long backpointer;
+ unsigned int backpointer;
/* When free, order of the freelist we're on */
unsigned int order;
};
@@ -204,12 +204,17 @@ struct shadow_page_info
/* When free, TLB flush time when freed */
u32 tlbflush_timestamp;
};
+#ifdef __i386__
+ unsigned long mbz; /* Must be zero: count_info is here. */
+#endif
struct {
- unsigned long mbz; /* Must be zero: count_info is here. */
unsigned long type:5; /* What kind of shadow is this? */
unsigned long pinned:1; /* Is the shadow pinned? */
unsigned long count:26; /* Reference count */
};
+#ifndef __i386__
+ unsigned long mbz; /* Must be zero: count_info is here. */
+#endif
union {
/* For unused shadow pages, a list of pages of this order; for
* pinnable shadows, if pinned, a list of other pinned shadows
@@ -645,7 +650,7 @@ static inline int sh_get_ref(struct vcpu
if ( unlikely(nx >= 1U<<26) )
{
- SHADOW_PRINTK("shadow ref overflow, gmfn=%" PRtype_info " smfn=%lx\n",
+ SHADOW_PRINTK("shadow ref overflow, gmfn=%x smfn=%lx\n",
sp->backpointer, mfn_x(smfn));
return 0;
}
--- a/xen/include/asm-x86/mm.h
+++ b/xen/include/asm-x86/mm.h
@@ -15,7 +15,7 @@
* 1. 'struct page_info' contains a 'struct list_head list'.
* 2. Provide a PFN_ORDER() macro for accessing the order of a free page.
*/
-#define PFN_ORDER(_pfn) ((_pfn)->u.free.order)
+#define PFN_ORDER(_pfn) ((_pfn)->v.free.order)
struct page_info
{
@@ -30,16 +30,12 @@ struct page_info
/* Page is in use: ((count_info & PGC_count_mask) != 0). */
struct {
- /* Owner of this page (NULL if page is anonymous). */
- u32 _domain; /* pickled format */
/* Type reference count and various PGT_xxx flags and fields. */
unsigned long type_info;
} inuse;
/* Page is on a free list: ((count_info & PGC_count_mask) == 0). */
struct {
- /* Order-size of the free chunk this page is the head of. */
- u32 order;
/* Mask of possibly-tainted TLBs. */
cpumask_t cpumask;
} free;
@@ -47,6 +43,22 @@ struct page_info
} u;
union {
+
+ /* Page is in use. */
+ struct {
+ /* Owner of this page (NULL if page is anonymous). */
+ u32 _domain; /* pickled format */
+ } inuse;
+
+ /* Page is on a free list. */
+ struct {
+ /* Order-size of the free chunk this page is the head of. */
+ u32 order;
+ } free;
+
+ } v;
+
+ union {
/*
* Timestamp from 'TLB clock', used to avoid extra safety flushes.
* Only valid for: a) free pages, and b) pages with zero type count
@@ -173,8 +185,8 @@ static inline u32 pickle_domptr(struct d
/* OOS fixup entries */
#define SHADOW_OOS_FIXUPS 2
-#define page_get_owner(_p) (unpickle_domptr((_p)->u.inuse._domain))
-#define page_set_owner(_p,_d) ((_p)->u.inuse._domain = pickle_domptr(_d))
+#define page_get_owner(_p) (unpickle_domptr((_p)->v.inuse._domain))
+#define page_set_owner(_p,_d) ((_p)->v.inuse._domain = pickle_domptr(_d))
#define maddr_get_owner(ma) (page_get_owner(maddr_to_page((ma))))
#define vaddr_get_owner(va) (page_get_owner(virt_to_page((va))))

132
x86_64-sh-next-shadow.patch Normal file
View File

@ -0,0 +1,132 @@
References: bnc#470949
--- a/xen/arch/x86/mm/shadow/common.c
+++ b/xen/arch/x86/mm/shadow/common.c
@@ -1456,6 +1456,22 @@ static __init int shadow_blow_tables_key
__initcall(shadow_blow_tables_keyhandler_init);
#endif /* !NDEBUG */
+#ifdef __i386__
+# define next_shadow(pg) ((pg)->next_shadow)
+# define set_next_shadow(pg, n) ((void)((pg)->next_shadow = (n)))
+#else
+static inline struct shadow_page_info *
+next_shadow(const struct shadow_page_info *sp)
+{
+ return sp->next_shadow ? mfn_to_shadow_page(_mfn(sp->next_shadow)) : NULL;
+}
+static inline void
+set_next_shadow(struct shadow_page_info *sp, struct shadow_page_info *next)
+{
+ sp->next_shadow = next ? mfn_x(shadow_page_to_mfn(next)) : 0;
+}
+#endif
+
/* Allocate another shadow's worth of (contiguous, aligned) pages,
* and fill in the type and backpointer fields of their page_infos.
* Never fails to allocate. */
@@ -1525,7 +1541,7 @@ mfn_t shadow_alloc(struct domain *d,
sp[i].pinned = 0;
sp[i].count = 0;
sp[i].backpointer = backpointer;
- sp[i].next_shadow = NULL;
+ set_next_shadow(&sp[i], NULL);
perfc_incr(shadow_alloc_count);
}
return shadow_page_to_mfn(sp);
@@ -1865,7 +1881,7 @@ static void sh_hash_audit_bucket(struct
/* Wrong bucket? */
BUG_ON( sh_hash(sp->backpointer, sp->type) != bucket );
/* Duplicate entry? */
- for ( x = sp->next_shadow; x; x = x->next_shadow )
+ for ( x = next_shadow(sp); x; x = next_shadow(x) )
BUG_ON( x->backpointer == sp->backpointer && x->type == sp->type );
/* Follow the backpointer to the guest pagetable */
if ( sp->type != SH_type_fl1_32_shadow
@@ -1908,7 +1924,7 @@ static void sh_hash_audit_bucket(struct
}
}
/* That entry was OK; on we go */
- sp = sp->next_shadow;
+ sp = next_shadow(sp);
}
}
@@ -2002,7 +2018,7 @@ mfn_t shadow_hash_lookup(struct vcpu *v,
/* Delete sp from the list */
prev->next_shadow = sp->next_shadow;
/* Re-insert it at the head of the list */
- sp->next_shadow = d->arch.paging.shadow.hash_table[key];
+ set_next_shadow(sp, d->arch.paging.shadow.hash_table[key]);
d->arch.paging.shadow.hash_table[key] = sp;
}
}
@@ -2013,7 +2029,7 @@ mfn_t shadow_hash_lookup(struct vcpu *v,
return shadow_page_to_mfn(sp);
}
prev = sp;
- sp = sp->next_shadow;
+ sp = next_shadow(sp);
}
perfc_incr(shadow_hash_lookup_miss);
@@ -2040,7 +2056,7 @@ void shadow_hash_insert(struct vcpu *v,
/* Insert this shadow at the top of the bucket */
sp = mfn_to_shadow_page(smfn);
- sp->next_shadow = d->arch.paging.shadow.hash_table[key];
+ set_next_shadow(sp, d->arch.paging.shadow.hash_table[key]);
d->arch.paging.shadow.hash_table[key] = sp;
sh_hash_audit_bucket(d, key);
@@ -2067,7 +2083,7 @@ void shadow_hash_delete(struct vcpu *v,
sp = mfn_to_shadow_page(smfn);
if ( d->arch.paging.shadow.hash_table[key] == sp )
/* Easy case: we're deleting the head item. */
- d->arch.paging.shadow.hash_table[key] = sp->next_shadow;
+ d->arch.paging.shadow.hash_table[key] = next_shadow(sp);
else
{
/* Need to search for the one we want */
@@ -2076,15 +2092,15 @@ void shadow_hash_delete(struct vcpu *v,
{
ASSERT(x); /* We can't have hit the end, since our target is
* still in the chain somehwere... */
- if ( x->next_shadow == sp )
+ if ( next_shadow(x) == sp )
{
x->next_shadow = sp->next_shadow;
break;
}
- x = x->next_shadow;
+ x = next_shadow(x);
}
}
- sp->next_shadow = NULL;
+ set_next_shadow(sp, NULL);
sh_hash_audit_bucket(d, key);
}
@@ -2118,7 +2134,7 @@ static void hash_foreach(struct vcpu *v,
/* WARNING: This is not safe against changes to the hash table.
* The callback *must* return non-zero if it has inserted or
* deleted anything from the hash (lookups are OK, though). */
- for ( x = d->arch.paging.shadow.hash_table[i]; x; x = x->next_shadow )
+ for ( x = d->arch.paging.shadow.hash_table[i]; x; x = next_shadow(x) )
{
if ( callback_mask & (1 << x->type) )
{
--- a/xen/arch/x86/mm/shadow/private.h
+++ b/xen/arch/x86/mm/shadow/private.h
@@ -196,7 +196,11 @@ struct shadow_page_info
};
union {
/* When in use, next shadow in this hash chain */
+#ifdef __i386__
struct shadow_page_info *next_shadow;
+#else
+ unsigned int next_shadow;
+#endif
/* When free, TLB flush time when freed */
u32 tlbflush_timestamp;
};

37
xen.changes
View File

@ -1,3 +1,40 @@
-------------------------------------------------------------------
Thu Feb 5 12:03:44 MST 2009 - jfehlig@novell.com
- bnc#470133 - Better error handling in xm when not booted Xen
19153-xm-noxen-error.patch
-------------------------------------------------------------------
Wed Feb 4 20:35:41 CET 2009 - kwolf@suse.de
- bnc#472075 - Fix ioemu to initialize its blktap backend also for
fully virtualized guests
ioemu-blktap-fv-init.patch
-------------------------------------------------------------------
Tue Feb 3 13:35:28 MST 2009 - jfehlig@novell.com
- bnc#470855 - Add note to xm man page on how to detach domain
console
19152-xm-man-page.patch
-------------------------------------------------------------------
Mon Feb 2 14:15:55 MST 2009 - jfehlig@novell.com
- bnc#471090 - XendAPIStore: Do not remove non-existent item
class list
19151-xend-class-dereg.patch
-------------------------------------------------------------------
Mon Feb 2 10:41:05 MST 2009 - carnold@novell.com
- bnc#470949 - user mode application may crash kernel
19088-x86-page-non-atomic-owner.patch (Jan Beulich)
19089-x86_64-widen-page-refcounts.patch
19103-x86_64-fold-page-lock.patch
x86_64-page-info-pack.patch
x86_64-sh-next-shadow.patch
------------------------------------------------------------------- -------------------------------------------------------------------
Fri Jan 23 11:47:31 MST 2009 - carnold@novell.com Fri Jan 23 11:47:31 MST 2009 - carnold@novell.com

49
xen.spec
View File

@ -1,5 +1,5 @@
# #
# spec file for package xen (Version 3.3.1_18546_04) # spec file for package xen (Version 3.3.1_18546_06)
# #
# Copyright (c) 2009 SUSE LINUX Products GmbH, Nuernberg, Germany. # Copyright (c) 2009 SUSE LINUX Products GmbH, Nuernberg, Germany.
# #
@ -37,7 +37,7 @@ BuildRequires: glibc-32bit glibc-devel-32bit
%if %{?with_kmp}0 %if %{?with_kmp}0
BuildRequires: kernel-source kernel-syms module-init-tools xorg-x11 BuildRequires: kernel-source kernel-syms module-init-tools xorg-x11
%endif %endif
Version: 3.3.1_18546_04 Version: 3.3.1_18546_06
Release: 1 Release: 1
License: GPL v2 only License: GPL v2 only
Group: System/Kernel Group: System/Kernel
@ -146,6 +146,12 @@ Patch76: 19048-cross-bit-coredumping.patch
Patch77: 19051-cross-bit-coredumping.patch Patch77: 19051-cross-bit-coredumping.patch
Patch78: 19072-vmx-pat.patch Patch78: 19072-vmx-pat.patch
Patch79: 19079-snp_ctl-1.patch Patch79: 19079-snp_ctl-1.patch
Patch80: 19088-x86-page-non-atomic-owner.patch
Patch81: 19089-x86_64-widen-page-refcounts.patch
Patch82: 19103-x86_64-fold-page-lock.patch
Patch83: 19151-xend-class-dereg.patch
Patch84: 19152-xm-man-page.patch
Patch85: 19153-xm-noxen-error.patch
# Our patches # Our patches
Patch100: xen-config.diff Patch100: xen-config.diff
Patch101: xend-config.diff Patch101: xend-config.diff
@ -212,8 +218,9 @@ Patch185: tapdisk-ioemu-logfile.patch
Patch186: blktap-ioemu-close-fix.patch Patch186: blktap-ioemu-close-fix.patch
Patch187: ioemu-blktap-zero-size.patch Patch187: ioemu-blktap-zero-size.patch
Patch188: blktap-error-handling.patch Patch188: blktap-error-handling.patch
Patch189: ioemu-blktap-fv-init.patch
# Jim's domain lock patch # Jim's domain lock patch
Patch190: xend-domain-lock.patch Patch200: xend-domain-lock.patch
# Patches from Jan # Patches from Jan
Patch240: dump-exec-state.patch Patch240: dump-exec-state.patch
Patch241: x86-show-page-walk-early.patch Patch241: x86-show-page-walk-early.patch
@ -228,6 +235,8 @@ Patch352: pvdrv_emulation_control.patch
Patch353: blktap-pv-cdrom.patch Patch353: blktap-pv-cdrom.patch
Patch354: x86-cpufreq-report.patch Patch354: x86-cpufreq-report.patch
Patch355: dom-print.patch Patch355: dom-print.patch
Patch356: x86_64-sh-next-shadow.patch
Patch357: x86_64-page-info-pack.patch
# novell_shim patches # novell_shim patches
Patch400: hv_tools.patch Patch400: hv_tools.patch
Patch401: hv_xen_base.patch Patch401: hv_xen_base.patch
@ -644,6 +653,12 @@ Authors:
%patch77 -p1 %patch77 -p1
%patch78 -p1 %patch78 -p1
%patch79 -p1 %patch79 -p1
%patch80 -p1
%patch81 -p1
%patch82 -p1
%patch83 -p1
%patch84 -p1
%patch85 -p1
%patch100 -p1 %patch100 -p1
%patch101 -p1 %patch101 -p1
%patch102 -p1 %patch102 -p1
@ -706,7 +721,8 @@ Authors:
%patch186 -p1 %patch186 -p1
%patch187 -p1 %patch187 -p1
%patch188 -p1 %patch188 -p1
%patch190 -p1 %patch189 -p1
%patch200 -p1
%patch240 -p1 %patch240 -p1
%patch241 -p1 %patch241 -p1
%patch242 -p1 %patch242 -p1
@ -719,7 +735,8 @@ Authors:
%patch353 -p1 %patch353 -p1
%patch354 -p1 %patch354 -p1
%patch355 -p1 %patch355 -p1
# Don't use shim for now %patch356 -p1
%patch357 -p1
%ifarch x86_64 %ifarch x86_64
%patch400 -p1 %patch400 -p1
%patch401 -p1 %patch401 -p1
@ -1069,6 +1086,28 @@ rm -f $RPM_BUILD_ROOT/%{_libdir}/xen/bin/qemu-dm.debug
/sbin/ldconfig /sbin/ldconfig
%changelog %changelog
* Thu Feb 05 2009 jfehlig@novell.com
- bnc#470133 - Better error handling in xm when not booted Xen
19153-xm-noxen-error.patch
* Wed Feb 04 2009 kwolf@suse.de
- bnc#472075 - Fix ioemu to initialize its blktap backend also for
fully virtualized guests
ioemu-blktap-fv-init.patch
* Tue Feb 03 2009 jfehlig@novell.com
- bnc#470855 - Add note to xm man page on how to detach domain
console
19152-xm-man-page.patch
* Mon Feb 02 2009 jfehlig@novell.com
- bnc#471090 - XendAPIStore: Do not remove non-existent item
class list
19151-xend-class-dereg.patch
* Mon Feb 02 2009 carnold@novell.com
- bnc#470949 - user mode application may crash kernel
19088-x86-page-non-atomic-owner.patch (Jan Beulich)
19089-x86_64-widen-page-refcounts.patch
19103-x86_64-fold-page-lock.patch
x86_64-page-info-pack.patch
x86_64-sh-next-shadow.patch
* Fri Jan 23 2009 carnold@novell.com * Fri Jan 23 2009 carnold@novell.com
- Intel - Remove improper operating condition that results in a - Intel - Remove improper operating condition that results in a
machine check. machine check.