aboutsummaryrefslogtreecommitdiffstats
path: root/arch
diff options
context:
space:
mode:
Diffstat (limited to 'arch')
-rw-r--r--arch/arm/kernel/sys_arm.c17
-rw-r--r--arch/arm/mach-orion5x/addr-map.c66
-rw-r--r--arch/arm/mach-orion5x/common.c11
-rw-r--r--arch/arm/mach-orion5x/common.h1
-rw-r--r--arch/avr32/kernel/sys_avr32.c13
-rw-r--r--arch/blackfin/kernel/sys_bfin.c17
-rw-r--r--arch/cris/kernel/sys_cris.c5
-rw-r--r--arch/frv/kernel/sys_frv.c17
-rw-r--r--arch/h8300/kernel/sys_h8300.c17
-rw-r--r--arch/m32r/kernel/sys_m32r.c5
-rw-r--r--arch/m68k/kernel/sys_m68k.c17
-rw-r--r--arch/m68k/kernel/traps.c17
-rw-r--r--arch/m68k/mac/config.c24
-rw-r--r--arch/m68knommu/kernel/sys_m68k.c17
-rw-r--r--arch/mn10300/kernel/sys_mn10300.c17
-rw-r--r--arch/parisc/kernel/sys_parisc.c13
-rw-r--r--arch/powerpc/kernel/syscalls.c17
-rw-r--r--arch/powerpc/kvm/booke_guest.c6
-rw-r--r--arch/powerpc/kvm/powerpc.c20
-rw-r--r--arch/powerpc/lib/Makefile1
-rw-r--r--arch/powerpc/lib/devres.c42
-rw-r--r--arch/powerpc/platforms/chrp/pegasos_eth.c4
-rw-r--r--arch/powerpc/platforms/pseries/scanlog.c19
-rw-r--r--arch/powerpc/sysdev/mv64x60_dev.c2
-rw-r--r--arch/ppc/syslib/mv64x60.c3
-rw-r--r--arch/s390/Kconfig7
-rw-r--r--arch/s390/kernel/compat_wrapper.S2
-rw-r--r--arch/s390/kernel/entry.S29
-rw-r--r--arch/s390/kernel/entry64.S57
-rw-r--r--arch/s390/kernel/ptrace.c100
-rw-r--r--arch/s390/kernel/sys_s390.c17
-rw-r--r--arch/s390/kvm/Kconfig1
-rw-r--r--arch/s390/kvm/intercept.c3
-rw-r--r--arch/s390/kvm/kvm-s390.c5
-rw-r--r--arch/s390/mm/Makefile1
-rw-r--r--arch/s390/mm/init.c3
-rw-r--r--arch/s390/mm/page-states.c79
-rw-r--r--arch/sh/kernel/sys_sh64.c17
-rw-r--r--arch/sparc/kernel/process.c20
-rw-r--r--arch/sparc/kernel/signal.c20
-rw-r--r--arch/sparc/kernel/sys_sparc.c3
-rw-r--r--arch/sparc64/kernel/process.c18
-rw-r--r--arch/sparc64/kernel/signal.c21
-rw-r--r--arch/sparc64/kernel/signal32.c18
-rw-r--r--arch/sparc64/kernel/smp.c27
-rw-r--r--arch/sparc64/kernel/sys_sparc.c4
-rw-r--r--arch/sparc64/kernel/sys_sparc32.c31
-rw-r--r--arch/sparc64/kernel/systbls.S2
-rw-r--r--arch/sparc64/mm/init.c16
-rw-r--r--arch/um/drivers/line.c4
-rw-r--r--arch/um/include/line.h2
-rw-r--r--arch/um/kernel/syscall.c17
-rw-r--r--arch/v850/kernel/syscalls.c17
-rw-r--r--arch/x86/Kconfig2
-rw-r--r--arch/x86/boot/compressed/relocs.c2
-rw-r--r--arch/x86/kernel/Makefile4
-rw-r--r--arch/x86/kernel/acpi/Makefile2
-rw-r--r--arch/x86/kernel/acpi/realmode/Makefile5
-rw-r--r--arch/x86/kernel/kvmclock.c4
-rw-r--r--arch/x86/kernel/mpparse.c7
-rw-r--r--arch/x86/kernel/reboot.c1
-rw-r--r--arch/x86/kernel/setup.c2
-rw-r--r--arch/x86/kernel/smpboot.c4
-rw-r--r--arch/x86/kernel/sys_i386_32.c17
-rw-r--r--arch/x86/kernel/sys_x86_64.c17
-rw-r--r--arch/x86/kvm/i8254.c2
-rw-r--r--arch/x86/kvm/mmu.c89
-rw-r--r--arch/x86/kvm/mmu.h37
-rw-r--r--arch/x86/kvm/svm.c10
-rw-r--r--arch/x86/kvm/vmx.c375
-rw-r--r--arch/x86/kvm/vmx.h38
-rw-r--r--arch/x86/kvm/x86.c22
-rw-r--r--arch/x86/kvm/x86_emulate.c1
-rw-r--r--arch/x86/mm/discontig_32.c26
-rw-r--r--arch/x86/mm/pgtable_32.c7
-rw-r--r--arch/x86/pci/Makefile_3212
-rw-r--r--arch/x86/pci/acpi.c41
-rw-r--r--arch/x86/pci/common.c69
-rw-r--r--arch/x86/pci/fixup.c2
-rw-r--r--arch/x86/pci/init.c4
-rw-r--r--arch/x86/pci/pci.h3
-rw-r--r--arch/x86/vdso/vdso32-setup.c2
-rw-r--r--arch/x86/video/fbdev.c2
83 files changed, 956 insertions, 782 deletions
diff --git a/arch/arm/kernel/sys_arm.c b/arch/arm/kernel/sys_arm.c
index 9bd1870d980e..0128687ba0f7 100644
--- a/arch/arm/kernel/sys_arm.c
+++ b/arch/arm/kernel/sys_arm.c
@@ -34,23 +34,6 @@ extern unsigned long do_mremap(unsigned long addr, unsigned long old_len,
34 unsigned long new_len, unsigned long flags, 34 unsigned long new_len, unsigned long flags,
35 unsigned long new_addr); 35 unsigned long new_addr);
36 36
37/*
38 * sys_pipe() is the normal C calling standard for creating
39 * a pipe. It's not the way unix traditionally does this, though.
40 */
41asmlinkage int sys_pipe(unsigned long __user *fildes)
42{
43 int fd[2];
44 int error;
45
46 error = do_pipe(fd);
47 if (!error) {
48 if (copy_to_user(fildes, fd, 2*sizeof(int)))
49 error = -EFAULT;
50 }
51 return error;
52}
53
54/* common code for old and new mmaps */ 37/* common code for old and new mmaps */
55inline long do_mmap2( 38inline long do_mmap2(
56 unsigned long addr, unsigned long len, 39 unsigned long addr, unsigned long len,
diff --git a/arch/arm/mach-orion5x/addr-map.c b/arch/arm/mach-orion5x/addr-map.c
index 9608503d67f5..e63fb05dc893 100644
--- a/arch/arm/mach-orion5x/addr-map.c
+++ b/arch/arm/mach-orion5x/addr-map.c
@@ -34,11 +34,7 @@
34 * Non-CPU Masters address decoding -- 34 * Non-CPU Masters address decoding --
35 * Unlike the CPU, we setup the access from Orion's master interfaces to DDR 35 * Unlike the CPU, we setup the access from Orion's master interfaces to DDR
36 * banks only (the typical use case). 36 * banks only (the typical use case).
37 * Setup access for each master to DDR is issued by common.c. 37 * Setup access for each master to DDR is issued by platform device setup.
38 *
39 * Note: although orion_setbits() and orion_clrbits() are not atomic
40 * no locking is necessary here since code in this file is only called
41 * at boot time when there is no concurrency issues.
42 */ 38 */
43 39
44/* 40/*
@@ -48,10 +44,6 @@
48#define TARGET_DEV_BUS 1 44#define TARGET_DEV_BUS 1
49#define TARGET_PCI 3 45#define TARGET_PCI 3
50#define TARGET_PCIE 4 46#define TARGET_PCIE 4
51#define ATTR_DDR_CS(n) (((n) ==0) ? 0xe : \
52 ((n) == 1) ? 0xd : \
53 ((n) == 2) ? 0xb : \
54 ((n) == 3) ? 0x7 : 0xf)
55#define ATTR_PCIE_MEM 0x59 47#define ATTR_PCIE_MEM 0x59
56#define ATTR_PCIE_IO 0x51 48#define ATTR_PCIE_IO 0x51
57#define ATTR_PCIE_WA 0x79 49#define ATTR_PCIE_WA 0x79
@@ -61,17 +53,12 @@
61#define ATTR_DEV_CS1 0x1d 53#define ATTR_DEV_CS1 0x1d
62#define ATTR_DEV_CS2 0x1b 54#define ATTR_DEV_CS2 0x1b
63#define ATTR_DEV_BOOT 0xf 55#define ATTR_DEV_BOOT 0xf
64#define WIN_EN 1
65 56
66/* 57/*
67 * Helpers to get DDR bank info 58 * Helpers to get DDR bank info
68 */ 59 */
69#define DDR_BASE_CS(n) ORION5X_DDR_REG(0x1500 + ((n) * 8)) 60#define DDR_BASE_CS(n) ORION5X_DDR_REG(0x1500 + ((n) << 3))
70#define DDR_SIZE_CS(n) ORION5X_DDR_REG(0x1504 + ((n) * 8)) 61#define DDR_SIZE_CS(n) ORION5X_DDR_REG(0x1504 + ((n) << 3))
71#define DDR_MAX_CS 4
72#define DDR_REG_TO_SIZE(reg) (((reg) | 0xffffff) + 1)
73#define DDR_REG_TO_BASE(reg) ((reg) & 0xff000000)
74#define DDR_BANK_EN 1
75 62
76/* 63/*
77 * CPU Address Decode Windows registers 64 * CPU Address Decode Windows registers
@@ -81,17 +68,6 @@
81#define CPU_WIN_REMAP_LO(n) ORION5X_BRIDGE_REG(0x008 | ((n) << 4)) 68#define CPU_WIN_REMAP_LO(n) ORION5X_BRIDGE_REG(0x008 | ((n) << 4))
82#define CPU_WIN_REMAP_HI(n) ORION5X_BRIDGE_REG(0x00c | ((n) << 4)) 69#define CPU_WIN_REMAP_HI(n) ORION5X_BRIDGE_REG(0x00c | ((n) << 4))
83 70
84/*
85 * Gigabit Ethernet Address Decode Windows registers
86 */
87#define ETH_WIN_BASE(win) ORION5X_ETH_REG(0x200 + ((win) * 8))
88#define ETH_WIN_SIZE(win) ORION5X_ETH_REG(0x204 + ((win) * 8))
89#define ETH_WIN_REMAP(win) ORION5X_ETH_REG(0x280 + ((win) * 4))
90#define ETH_WIN_EN ORION5X_ETH_REG(0x290)
91#define ETH_WIN_PROT ORION5X_ETH_REG(0x294)
92#define ETH_MAX_WIN 6
93#define ETH_MAX_REMAP_WIN 4
94
95 71
96struct mbus_dram_target_info orion5x_mbus_dram_info; 72struct mbus_dram_target_info orion5x_mbus_dram_info;
97 73
@@ -202,39 +178,3 @@ void __init orion5x_setup_pcie_wa_win(u32 base, u32 size)
202{ 178{
203 setup_cpu_win(7, base, size, TARGET_PCIE, ATTR_PCIE_WA, -1); 179 setup_cpu_win(7, base, size, TARGET_PCIE, ATTR_PCIE_WA, -1);
204} 180}
205
206void __init orion5x_setup_eth_wins(void)
207{
208 int i;
209
210 /*
211 * First, disable and clear windows
212 */
213 for (i = 0; i < ETH_MAX_WIN; i++) {
214 orion5x_write(ETH_WIN_BASE(i), 0);
215 orion5x_write(ETH_WIN_SIZE(i), 0);
216 orion5x_setbits(ETH_WIN_EN, 1 << i);
217 orion5x_clrbits(ETH_WIN_PROT, 0x3 << (i * 2));
218 if (i < ETH_MAX_REMAP_WIN)
219 orion5x_write(ETH_WIN_REMAP(i), 0);
220 }
221
222 /*
223 * Setup windows for DDR banks.
224 */
225 for (i = 0; i < DDR_MAX_CS; i++) {
226 u32 base, size;
227 size = orion5x_read(DDR_SIZE_CS(i));
228 base = orion5x_read(DDR_BASE_CS(i));
229 if (size & DDR_BANK_EN) {
230 base = DDR_REG_TO_BASE(base);
231 size = DDR_REG_TO_SIZE(size);
232 orion5x_write(ETH_WIN_SIZE(i), (size-1) & 0xffff0000);
233 orion5x_write(ETH_WIN_BASE(i), (base & 0xffff0000) |
234 (ATTR_DDR_CS(i) << 8) |
235 TARGET_DDR);
236 orion5x_clrbits(ETH_WIN_EN, 1 << i);
237 orion5x_setbits(ETH_WIN_PROT, 0x3 << (i * 2));
238 }
239 }
240}
diff --git a/arch/arm/mach-orion5x/common.c b/arch/arm/mach-orion5x/common.c
index 968deb58be01..4f13fd037f04 100644
--- a/arch/arm/mach-orion5x/common.c
+++ b/arch/arm/mach-orion5x/common.c
@@ -190,6 +190,11 @@ static struct platform_device orion5x_ehci1 = {
190 * (The Orion and Discovery (MV643xx) families use the same Ethernet driver) 190 * (The Orion and Discovery (MV643xx) families use the same Ethernet driver)
191 ****************************************************************************/ 191 ****************************************************************************/
192 192
193struct mv643xx_eth_shared_platform_data orion5x_eth_shared_data = {
194 .dram = &orion5x_mbus_dram_info,
195 .t_clk = ORION5X_TCLK,
196};
197
193static struct resource orion5x_eth_shared_resources[] = { 198static struct resource orion5x_eth_shared_resources[] = {
194 { 199 {
195 .start = ORION5X_ETH_PHYS_BASE + 0x2000, 200 .start = ORION5X_ETH_PHYS_BASE + 0x2000,
@@ -201,6 +206,9 @@ static struct resource orion5x_eth_shared_resources[] = {
201static struct platform_device orion5x_eth_shared = { 206static struct platform_device orion5x_eth_shared = {
202 .name = MV643XX_ETH_SHARED_NAME, 207 .name = MV643XX_ETH_SHARED_NAME,
203 .id = 0, 208 .id = 0,
209 .dev = {
210 .platform_data = &orion5x_eth_shared_data,
211 },
204 .num_resources = 1, 212 .num_resources = 1,
205 .resource = orion5x_eth_shared_resources, 213 .resource = orion5x_eth_shared_resources,
206}; 214};
@@ -223,7 +231,9 @@ static struct platform_device orion5x_eth = {
223 231
224void __init orion5x_eth_init(struct mv643xx_eth_platform_data *eth_data) 232void __init orion5x_eth_init(struct mv643xx_eth_platform_data *eth_data)
225{ 233{
234 eth_data->shared = &orion5x_eth_shared;
226 orion5x_eth.dev.platform_data = eth_data; 235 orion5x_eth.dev.platform_data = eth_data;
236
227 platform_device_register(&orion5x_eth_shared); 237 platform_device_register(&orion5x_eth_shared);
228 platform_device_register(&orion5x_eth); 238 platform_device_register(&orion5x_eth);
229} 239}
@@ -360,7 +370,6 @@ void __init orion5x_init(void)
360 * Setup Orion address map 370 * Setup Orion address map
361 */ 371 */
362 orion5x_setup_cpu_mbus_bridge(); 372 orion5x_setup_cpu_mbus_bridge();
363 orion5x_setup_eth_wins();
364 373
365 /* 374 /*
366 * Register devices. 375 * Register devices.
diff --git a/arch/arm/mach-orion5x/common.h b/arch/arm/mach-orion5x/common.h
index 14adf8d1a54a..bd0f05de6e18 100644
--- a/arch/arm/mach-orion5x/common.h
+++ b/arch/arm/mach-orion5x/common.h
@@ -22,7 +22,6 @@ void orion5x_setup_dev0_win(u32 base, u32 size);
22void orion5x_setup_dev1_win(u32 base, u32 size); 22void orion5x_setup_dev1_win(u32 base, u32 size);
23void orion5x_setup_dev2_win(u32 base, u32 size); 23void orion5x_setup_dev2_win(u32 base, u32 size);
24void orion5x_setup_pcie_wa_win(u32 base, u32 size); 24void orion5x_setup_pcie_wa_win(u32 base, u32 size);
25void orion5x_setup_eth_wins(void);
26 25
27/* 26/*
28 * Shared code used internally by other Orion core functions. 27 * Shared code used internally by other Orion core functions.
diff --git a/arch/avr32/kernel/sys_avr32.c b/arch/avr32/kernel/sys_avr32.c
index 8deb6003ee62..8e8911e55c8f 100644
--- a/arch/avr32/kernel/sys_avr32.c
+++ b/arch/avr32/kernel/sys_avr32.c
@@ -14,19 +14,6 @@
14#include <asm/mman.h> 14#include <asm/mman.h>
15#include <asm/uaccess.h> 15#include <asm/uaccess.h>
16 16
17asmlinkage int sys_pipe(unsigned long __user *filedes)
18{
19 int fd[2];
20 int error;
21
22 error = do_pipe(fd);
23 if (!error) {
24 if (copy_to_user(filedes, fd, sizeof(fd)))
25 error = -EFAULT;
26 }
27 return error;
28}
29
30asmlinkage long sys_mmap2(unsigned long addr, unsigned long len, 17asmlinkage long sys_mmap2(unsigned long addr, unsigned long len,
31 unsigned long prot, unsigned long flags, 18 unsigned long prot, unsigned long flags,
32 unsigned long fd, off_t offset) 19 unsigned long fd, off_t offset)
diff --git a/arch/blackfin/kernel/sys_bfin.c b/arch/blackfin/kernel/sys_bfin.c
index efb7b25a2633..fce49d7cf001 100644
--- a/arch/blackfin/kernel/sys_bfin.c
+++ b/arch/blackfin/kernel/sys_bfin.c
@@ -45,23 +45,6 @@
45#include <asm/cacheflush.h> 45#include <asm/cacheflush.h>
46#include <asm/dma.h> 46#include <asm/dma.h>
47 47
48/*
49 * sys_pipe() is the normal C calling standard for creating
50 * a pipe. It's not the way unix traditionally does this, though.
51 */
52asmlinkage int sys_pipe(unsigned long __user *fildes)
53{
54 int fd[2];
55 int error;
56
57 error = do_pipe(fd);
58 if (!error) {
59 if (copy_to_user(fildes, fd, 2 * sizeof(int)))
60 error = -EFAULT;
61 }
62 return error;
63}
64
65/* common code for old and new mmaps */ 48/* common code for old and new mmaps */
66static inline long 49static inline long
67do_mmap2(unsigned long addr, unsigned long len, 50do_mmap2(unsigned long addr, unsigned long len,
diff --git a/arch/cris/kernel/sys_cris.c b/arch/cris/kernel/sys_cris.c
index 8b9984197edc..d124066e1728 100644
--- a/arch/cris/kernel/sys_cris.c
+++ b/arch/cris/kernel/sys_cris.c
@@ -40,8 +40,11 @@ asmlinkage int sys_pipe(unsigned long __user * fildes)
40 error = do_pipe(fd); 40 error = do_pipe(fd);
41 unlock_kernel(); 41 unlock_kernel();
42 if (!error) { 42 if (!error) {
43 if (copy_to_user(fildes, fd, 2*sizeof(int))) 43 if (copy_to_user(fildes, fd, 2*sizeof(int))) {
44 sys_close(fd[0]);
45 sys_close(fd[1]);
44 error = -EFAULT; 46 error = -EFAULT;
47 }
45 } 48 }
46 return error; 49 return error;
47} 50}
diff --git a/arch/frv/kernel/sys_frv.c b/arch/frv/kernel/sys_frv.c
index 04c6b1677ccf..49b2cf2c38f3 100644
--- a/arch/frv/kernel/sys_frv.c
+++ b/arch/frv/kernel/sys_frv.c
@@ -28,23 +28,6 @@
28#include <asm/setup.h> 28#include <asm/setup.h>
29#include <asm/uaccess.h> 29#include <asm/uaccess.h>
30 30
31/*
32 * sys_pipe() is the normal C calling standard for creating
33 * a pipe. It's not the way unix traditionally does this, though.
34 */
35asmlinkage long sys_pipe(unsigned long __user * fildes)
36{
37 int fd[2];
38 int error;
39
40 error = do_pipe(fd);
41 if (!error) {
42 if (copy_to_user(fildes, fd, 2*sizeof(int)))
43 error = -EFAULT;
44 }
45 return error;
46}
47
48asmlinkage long sys_mmap2(unsigned long addr, unsigned long len, 31asmlinkage long sys_mmap2(unsigned long addr, unsigned long len,
49 unsigned long prot, unsigned long flags, 32 unsigned long prot, unsigned long flags,
50 unsigned long fd, unsigned long pgoff) 33 unsigned long fd, unsigned long pgoff)
diff --git a/arch/h8300/kernel/sys_h8300.c b/arch/h8300/kernel/sys_h8300.c
index 00608be6d567..2745656dcc52 100644
--- a/arch/h8300/kernel/sys_h8300.c
+++ b/arch/h8300/kernel/sys_h8300.c
@@ -27,23 +27,6 @@
27#include <asm/traps.h> 27#include <asm/traps.h>
28#include <asm/unistd.h> 28#include <asm/unistd.h>
29 29
30/*
31 * sys_pipe() is the normal C calling standard for creating
32 * a pipe. It's not the way unix traditionally does this, though.
33 */
34asmlinkage int sys_pipe(unsigned long * fildes)
35{
36 int fd[2];
37 int error;
38
39 error = do_pipe(fd);
40 if (!error) {
41 if (copy_to_user(fildes, fd, 2*sizeof(int)))
42 error = -EFAULT;
43 }
44 return error;
45}
46
47/* common code for old and new mmaps */ 30/* common code for old and new mmaps */
48static inline long do_mmap2( 31static inline long do_mmap2(
49 unsigned long addr, unsigned long len, 32 unsigned long addr, unsigned long len,
diff --git a/arch/m32r/kernel/sys_m32r.c b/arch/m32r/kernel/sys_m32r.c
index 6d7a80fdad48..319c79720b8a 100644
--- a/arch/m32r/kernel/sys_m32r.c
+++ b/arch/m32r/kernel/sys_m32r.c
@@ -90,8 +90,11 @@ sys_pipe(unsigned long r0, unsigned long r1, unsigned long r2,
90 90
91 error = do_pipe(fd); 91 error = do_pipe(fd);
92 if (!error) { 92 if (!error) {
93 if (copy_to_user((void __user *)r0, fd, 2*sizeof(int))) 93 if (copy_to_user((void __user *)r0, fd, 2*sizeof(int))) {
94 sys_close(fd[0]);
95 sys_close(fd[1]);
94 error = -EFAULT; 96 error = -EFAULT;
97 }
95 } 98 }
96 return error; 99 return error;
97} 100}
diff --git a/arch/m68k/kernel/sys_m68k.c b/arch/m68k/kernel/sys_m68k.c
index e892f17ba3fa..7f54efaf60bb 100644
--- a/arch/m68k/kernel/sys_m68k.c
+++ b/arch/m68k/kernel/sys_m68k.c
@@ -30,23 +30,6 @@
30#include <asm/page.h> 30#include <asm/page.h>
31#include <asm/unistd.h> 31#include <asm/unistd.h>
32 32
33/*
34 * sys_pipe() is the normal C calling standard for creating
35 * a pipe. It's not the way unix traditionally does this, though.
36 */
37asmlinkage int sys_pipe(unsigned long __user * fildes)
38{
39 int fd[2];
40 int error;
41
42 error = do_pipe(fd);
43 if (!error) {
44 if (copy_to_user(fildes, fd, 2*sizeof(int)))
45 error = -EFAULT;
46 }
47 return error;
48}
49
50/* common code for old and new mmaps */ 33/* common code for old and new mmaps */
51static inline long do_mmap2( 34static inline long do_mmap2(
52 unsigned long addr, unsigned long len, 35 unsigned long addr, unsigned long len,
diff --git a/arch/m68k/kernel/traps.c b/arch/m68k/kernel/traps.c
index fd4858e2dd63..75b8340b254b 100644
--- a/arch/m68k/kernel/traps.c
+++ b/arch/m68k/kernel/traps.c
@@ -468,15 +468,26 @@ static inline void access_error040(struct frame *fp)
468 * (if do_page_fault didn't fix the mapping, 468 * (if do_page_fault didn't fix the mapping,
469 * the writeback won't do good) 469 * the writeback won't do good)
470 */ 470 */
471disable_wb:
471#ifdef DEBUG 472#ifdef DEBUG
472 printk(".. disabling wb2\n"); 473 printk(".. disabling wb2\n");
473#endif 474#endif
474 if (fp->un.fmt7.wb2a == fp->un.fmt7.faddr) 475 if (fp->un.fmt7.wb2a == fp->un.fmt7.faddr)
475 fp->un.fmt7.wb2s &= ~WBV_040; 476 fp->un.fmt7.wb2s &= ~WBV_040;
477 if (fp->un.fmt7.wb3a == fp->un.fmt7.faddr)
478 fp->un.fmt7.wb3s &= ~WBV_040;
476 } 479 }
477 } else if (send_fault_sig(&fp->ptregs) > 0) { 480 } else {
478 printk("68040 access error, ssw=%x\n", ssw); 481 /* In case of a bus error we either kill the process or expect
479 trap_c(fp); 482 * the kernel to catch the fault, which then is also responsible
483 * for cleaning up the mess.
484 */
485 current->thread.signo = SIGBUS;
486 current->thread.faddr = fp->un.fmt7.faddr;
487 if (send_fault_sig(&fp->ptregs) >= 0)
488 printk("68040 bus error (ssw=%x, faddr=%lx)\n", ssw,
489 fp->un.fmt7.faddr);
490 goto disable_wb;
480 } 491 }
481 492
482 do_040writebacks(fp); 493 do_040writebacks(fp);
diff --git a/arch/m68k/mac/config.c b/arch/m68k/mac/config.c
index 735a49b4b936..ad3e3bacae39 100644
--- a/arch/m68k/mac/config.c
+++ b/arch/m68k/mac/config.c
@@ -48,9 +48,6 @@
48struct mac_booter_data mac_bi_data; 48struct mac_booter_data mac_bi_data;
49int mac_bisize = sizeof mac_bi_data; 49int mac_bisize = sizeof mac_bi_data;
50 50
51struct mac_hw_present mac_hw_present;
52EXPORT_SYMBOL(mac_hw_present);
53
54/* New m68k bootinfo stuff and videobase */ 51/* New m68k bootinfo stuff and videobase */
55 52
56extern int m68k_num_memory; 53extern int m68k_num_memory;
@@ -817,27 +814,6 @@ void __init mac_identify(void)
817 m68k_ramdisk.addr, m68k_ramdisk.size); 814 m68k_ramdisk.addr, m68k_ramdisk.size);
818#endif 815#endif
819 816
820 /*
821 * TODO: set the various fields in macintosh_config->hw_present here!
822 */
823 switch (macintosh_config->scsi_type) {
824 case MAC_SCSI_OLD:
825 MACHW_SET(MAC_SCSI_80);
826 break;
827 case MAC_SCSI_QUADRA:
828 case MAC_SCSI_QUADRA2:
829 case MAC_SCSI_QUADRA3:
830 MACHW_SET(MAC_SCSI_96);
831 if ((macintosh_config->ident == MAC_MODEL_Q900) ||
832 (macintosh_config->ident == MAC_MODEL_Q950))
833 MACHW_SET(MAC_SCSI_96_2);
834 break;
835 default:
836 printk(KERN_WARNING "config.c: wtf: unknown scsi, using 53c80\n");
837 MACHW_SET(MAC_SCSI_80);
838 break;
839 }
840
841 iop_init(); 817 iop_init();
842 via_init(); 818 via_init();
843 oss_init(); 819 oss_init();
diff --git a/arch/m68knommu/kernel/sys_m68k.c b/arch/m68knommu/kernel/sys_m68k.c
index 65f7a95f056e..700281638629 100644
--- a/arch/m68knommu/kernel/sys_m68k.c
+++ b/arch/m68knommu/kernel/sys_m68k.c
@@ -28,23 +28,6 @@
28#include <asm/cacheflush.h> 28#include <asm/cacheflush.h>
29#include <asm/unistd.h> 29#include <asm/unistd.h>
30 30
31/*
32 * sys_pipe() is the normal C calling standard for creating
33 * a pipe. It's not the way unix traditionally does this, though.
34 */
35asmlinkage int sys_pipe(unsigned long * fildes)
36{
37 int fd[2];
38 int error;
39
40 error = do_pipe(fd);
41 if (!error) {
42 if (copy_to_user(fildes, fd, 2*sizeof(int)))
43 error = -EFAULT;
44 }
45 return error;
46}
47
48/* common code for old and new mmaps */ 31/* common code for old and new mmaps */
49static inline long do_mmap2( 32static inline long do_mmap2(
50 unsigned long addr, unsigned long len, 33 unsigned long addr, unsigned long len,
diff --git a/arch/mn10300/kernel/sys_mn10300.c b/arch/mn10300/kernel/sys_mn10300.c
index 5f17a1ebc825..bca5a84dc72c 100644
--- a/arch/mn10300/kernel/sys_mn10300.c
+++ b/arch/mn10300/kernel/sys_mn10300.c
@@ -29,23 +29,6 @@
29#define MIN_MAP_ADDR PAGE_SIZE /* minimum fixed mmap address */ 29#define MIN_MAP_ADDR PAGE_SIZE /* minimum fixed mmap address */
30 30
31/* 31/*
32 * sys_pipe() is the normal C calling standard for creating
33 * a pipe. It's not the way Unix traditionally does this, though.
34 */
35asmlinkage long sys_pipe(unsigned long __user *fildes)
36{
37 int fd[2];
38 int error;
39
40 error = do_pipe(fd);
41 if (!error) {
42 if (copy_to_user(fildes, fd, 2 * sizeof(int)))
43 error = -EFAULT;
44 }
45 return error;
46}
47
48/*
49 * memory mapping syscall 32 * memory mapping syscall
50 */ 33 */
51asmlinkage long sys_mmap2(unsigned long addr, unsigned long len, 34asmlinkage long sys_mmap2(unsigned long addr, unsigned long len,
diff --git a/arch/parisc/kernel/sys_parisc.c b/arch/parisc/kernel/sys_parisc.c
index 4f589216b39e..71b31957c8f1 100644
--- a/arch/parisc/kernel/sys_parisc.c
+++ b/arch/parisc/kernel/sys_parisc.c
@@ -33,19 +33,6 @@
33#include <linux/utsname.h> 33#include <linux/utsname.h>
34#include <linux/personality.h> 34#include <linux/personality.h>
35 35
36int sys_pipe(int __user *fildes)
37{
38 int fd[2];
39 int error;
40
41 error = do_pipe(fd);
42 if (!error) {
43 if (copy_to_user(fildes, fd, 2*sizeof(int)))
44 error = -EFAULT;
45 }
46 return error;
47}
48
49static unsigned long get_unshared_area(unsigned long addr, unsigned long len) 36static unsigned long get_unshared_area(unsigned long addr, unsigned long len)
50{ 37{
51 struct vm_area_struct *vma; 38 struct vm_area_struct *vma;
diff --git a/arch/powerpc/kernel/syscalls.c b/arch/powerpc/kernel/syscalls.c
index e722a4eeb5d0..4fe69ca24481 100644
--- a/arch/powerpc/kernel/syscalls.c
+++ b/arch/powerpc/kernel/syscalls.c
@@ -136,23 +136,6 @@ int sys_ipc(uint call, int first, unsigned long second, long third,
136 return ret; 136 return ret;
137} 137}
138 138
139/*
140 * sys_pipe() is the normal C calling standard for creating
141 * a pipe. It's not the way unix traditionally does this, though.
142 */
143int sys_pipe(int __user *fildes)
144{
145 int fd[2];
146 int error;
147
148 error = do_pipe(fd);
149 if (!error) {
150 if (copy_to_user(fildes, fd, 2*sizeof(int)))
151 error = -EFAULT;
152 }
153 return error;
154}
155
156static inline unsigned long do_mmap2(unsigned long addr, size_t len, 139static inline unsigned long do_mmap2(unsigned long addr, size_t len,
157 unsigned long prot, unsigned long flags, 140 unsigned long prot, unsigned long flags,
158 unsigned long fd, unsigned long off, int shift) 141 unsigned long fd, unsigned long off, int shift)
diff --git a/arch/powerpc/kvm/booke_guest.c b/arch/powerpc/kvm/booke_guest.c
index 6d9884a6884a..712d89a28c46 100644
--- a/arch/powerpc/kvm/booke_guest.c
+++ b/arch/powerpc/kvm/booke_guest.c
@@ -49,6 +49,7 @@ struct kvm_stats_debugfs_item debugfs_entries[] = {
49 { "inst_emu", VCPU_STAT(emulated_inst_exits) }, 49 { "inst_emu", VCPU_STAT(emulated_inst_exits) },
50 { "dec", VCPU_STAT(dec_exits) }, 50 { "dec", VCPU_STAT(dec_exits) },
51 { "ext_intr", VCPU_STAT(ext_intr_exits) }, 51 { "ext_intr", VCPU_STAT(ext_intr_exits) },
52 { "halt_wakeup", VCPU_STAT(halt_wakeup) },
52 { NULL } 53 { NULL }
53}; 54};
54 55
@@ -338,6 +339,11 @@ int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
338 } 339 }
339 break; 340 break;
340 341
342 case BOOKE_INTERRUPT_FP_UNAVAIL:
343 kvmppc_queue_exception(vcpu, exit_nr);
344 r = RESUME_GUEST;
345 break;
346
341 case BOOKE_INTERRUPT_DATA_STORAGE: 347 case BOOKE_INTERRUPT_DATA_STORAGE:
342 vcpu->arch.dear = vcpu->arch.fault_dear; 348 vcpu->arch.dear = vcpu->arch.fault_dear;
343 vcpu->arch.esr = vcpu->arch.fault_esr; 349 vcpu->arch.esr = vcpu->arch.fault_esr;
diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
index bad40bd2d3ac..777e0f34e0ea 100644
--- a/arch/powerpc/kvm/powerpc.c
+++ b/arch/powerpc/kvm/powerpc.c
@@ -36,13 +36,12 @@ gfn_t unalias_gfn(struct kvm *kvm, gfn_t gfn)
36 36
37int kvm_cpu_has_interrupt(struct kvm_vcpu *v) 37int kvm_cpu_has_interrupt(struct kvm_vcpu *v)
38{ 38{
39 /* XXX implement me */ 39 return !!(v->arch.pending_exceptions);
40 return 0;
41} 40}
42 41
43int kvm_arch_vcpu_runnable(struct kvm_vcpu *v) 42int kvm_arch_vcpu_runnable(struct kvm_vcpu *v)
44{ 43{
45 return 1; 44 return !(v->arch.msr & MSR_WE);
46} 45}
47 46
48 47
@@ -214,6 +213,11 @@ static void kvmppc_decrementer_func(unsigned long data)
214 struct kvm_vcpu *vcpu = (struct kvm_vcpu *)data; 213 struct kvm_vcpu *vcpu = (struct kvm_vcpu *)data;
215 214
216 kvmppc_queue_exception(vcpu, BOOKE_INTERRUPT_DECREMENTER); 215 kvmppc_queue_exception(vcpu, BOOKE_INTERRUPT_DECREMENTER);
216
217 if (waitqueue_active(&vcpu->wq)) {
218 wake_up_interruptible(&vcpu->wq);
219 vcpu->stat.halt_wakeup++;
220 }
217} 221}
218 222
219int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu) 223int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
@@ -339,6 +343,8 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *run)
339 int r; 343 int r;
340 sigset_t sigsaved; 344 sigset_t sigsaved;
341 345
346 vcpu_load(vcpu);
347
342 if (vcpu->sigset_active) 348 if (vcpu->sigset_active)
343 sigprocmask(SIG_SETMASK, &vcpu->sigset, &sigsaved); 349 sigprocmask(SIG_SETMASK, &vcpu->sigset, &sigsaved);
344 350
@@ -363,12 +369,20 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *run)
363 if (vcpu->sigset_active) 369 if (vcpu->sigset_active)
364 sigprocmask(SIG_SETMASK, &sigsaved, NULL); 370 sigprocmask(SIG_SETMASK, &sigsaved, NULL);
365 371
372 vcpu_put(vcpu);
373
366 return r; 374 return r;
367} 375}
368 376
369int kvm_vcpu_ioctl_interrupt(struct kvm_vcpu *vcpu, struct kvm_interrupt *irq) 377int kvm_vcpu_ioctl_interrupt(struct kvm_vcpu *vcpu, struct kvm_interrupt *irq)
370{ 378{
371 kvmppc_queue_exception(vcpu, BOOKE_INTERRUPT_EXTERNAL); 379 kvmppc_queue_exception(vcpu, BOOKE_INTERRUPT_EXTERNAL);
380
381 if (waitqueue_active(&vcpu->wq)) {
382 wake_up_interruptible(&vcpu->wq);
383 vcpu->stat.halt_wakeup++;
384 }
385
372 return 0; 386 return 0;
373} 387}
374 388
diff --git a/arch/powerpc/lib/Makefile b/arch/powerpc/lib/Makefile
index 4bb023f4c869..f1d2cdc5331b 100644
--- a/arch/powerpc/lib/Makefile
+++ b/arch/powerpc/lib/Makefile
@@ -23,3 +23,4 @@ obj-$(CONFIG_SMP) += locks.o
23endif 23endif
24 24
25obj-$(CONFIG_PPC_LIB_RHEAP) += rheap.o 25obj-$(CONFIG_PPC_LIB_RHEAP) += rheap.o
26obj-$(CONFIG_HAS_IOMEM) += devres.o
diff --git a/arch/powerpc/lib/devres.c b/arch/powerpc/lib/devres.c
new file mode 100644
index 000000000000..292115d98ea9
--- /dev/null
+++ b/arch/powerpc/lib/devres.c
@@ -0,0 +1,42 @@
1/*
2 * Copyright (C) 2008 Freescale Semiconductor, Inc.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License
6 * as published by the Free Software Foundation; either version
7 * 2 of the License, or (at your option) any later version.
8 */
9
10#include <linux/device.h> /* devres_*(), devm_ioremap_release() */
11#include <linux/io.h> /* ioremap_flags() */
12#include <linux/module.h> /* EXPORT_SYMBOL() */
13
14/**
15 * devm_ioremap_prot - Managed ioremap_flags()
16 * @dev: Generic device to remap IO address for
17 * @offset: BUS offset to map
18 * @size: Size of map
19 * @flags: Page flags
20 *
21 * Managed ioremap_prot(). Map is automatically unmapped on driver
22 * detach.
23 */
24void __iomem *devm_ioremap_prot(struct device *dev, resource_size_t offset,
25 size_t size, unsigned long flags)
26{
27 void __iomem **ptr, *addr;
28
29 ptr = devres_alloc(devm_ioremap_release, sizeof(*ptr), GFP_KERNEL);
30 if (!ptr)
31 return NULL;
32
33 addr = ioremap_flags(offset, size, flags);
34 if (addr) {
35 *ptr = addr;
36 devres_add(dev, ptr);
37 } else
38 devres_free(ptr);
39
40 return addr;
41}
42EXPORT_SYMBOL(devm_ioremap_prot);
diff --git a/arch/powerpc/platforms/chrp/pegasos_eth.c b/arch/powerpc/platforms/chrp/pegasos_eth.c
index 5bcc58d9a4dd..130ff72d99dd 100644
--- a/arch/powerpc/platforms/chrp/pegasos_eth.c
+++ b/arch/powerpc/platforms/chrp/pegasos_eth.c
@@ -58,7 +58,9 @@ static struct resource mv643xx_eth0_resources[] = {
58 58
59 59
60static struct mv643xx_eth_platform_data eth0_pd = { 60static struct mv643xx_eth_platform_data eth0_pd = {
61 .shared = &mv643xx_eth_shared_device,
61 .port_number = 0, 62 .port_number = 0,
63
62 .tx_sram_addr = PEGASOS2_SRAM_BASE_ETH0, 64 .tx_sram_addr = PEGASOS2_SRAM_BASE_ETH0,
63 .tx_sram_size = PEGASOS2_SRAM_TXRING_SIZE, 65 .tx_sram_size = PEGASOS2_SRAM_TXRING_SIZE,
64 .tx_queue_size = PEGASOS2_SRAM_TXRING_SIZE/16, 66 .tx_queue_size = PEGASOS2_SRAM_TXRING_SIZE/16,
@@ -88,7 +90,9 @@ static struct resource mv643xx_eth1_resources[] = {
88}; 90};
89 91
90static struct mv643xx_eth_platform_data eth1_pd = { 92static struct mv643xx_eth_platform_data eth1_pd = {
93 .shared = &mv643xx_eth_shared_device,
91 .port_number = 1, 94 .port_number = 1,
95
92 .tx_sram_addr = PEGASOS2_SRAM_BASE_ETH1, 96 .tx_sram_addr = PEGASOS2_SRAM_BASE_ETH1,
93 .tx_sram_size = PEGASOS2_SRAM_TXRING_SIZE, 97 .tx_sram_size = PEGASOS2_SRAM_TXRING_SIZE,
94 .tx_queue_size = PEGASOS2_SRAM_TXRING_SIZE/16, 98 .tx_queue_size = PEGASOS2_SRAM_TXRING_SIZE/16,
diff --git a/arch/powerpc/platforms/pseries/scanlog.c b/arch/powerpc/platforms/pseries/scanlog.c
index bec3803f0618..417eca79df69 100644
--- a/arch/powerpc/platforms/pseries/scanlog.c
+++ b/arch/powerpc/platforms/pseries/scanlog.c
@@ -55,11 +55,6 @@ static ssize_t scanlog_read(struct file *file, char __user *buf,
55 dp = PDE(inode); 55 dp = PDE(inode);
56 data = (unsigned int *)dp->data; 56 data = (unsigned int *)dp->data;
57 57
58 if (!data) {
59 printk(KERN_ERR "scanlog: read failed no data\n");
60 return -EIO;
61 }
62
63 if (count > RTAS_DATA_BUF_SIZE) 58 if (count > RTAS_DATA_BUF_SIZE)
64 count = RTAS_DATA_BUF_SIZE; 59 count = RTAS_DATA_BUF_SIZE;
65 60
@@ -146,11 +141,6 @@ static int scanlog_open(struct inode * inode, struct file * file)
146 struct proc_dir_entry *dp = PDE(inode); 141 struct proc_dir_entry *dp = PDE(inode);
147 unsigned int *data = (unsigned int *)dp->data; 142 unsigned int *data = (unsigned int *)dp->data;
148 143
149 if (!data) {
150 printk(KERN_ERR "scanlog: open failed no data\n");
151 return -EIO;
152 }
153
154 if (data[0] != 0) { 144 if (data[0] != 0) {
155 /* This imperfect test stops a second copy of the 145 /* This imperfect test stops a second copy of the
156 * data (or a reset while data is being copied) 146 * data (or a reset while data is being copied)
@@ -168,10 +158,6 @@ static int scanlog_release(struct inode * inode, struct file * file)
168 struct proc_dir_entry *dp = PDE(inode); 158 struct proc_dir_entry *dp = PDE(inode);
169 unsigned int *data = (unsigned int *)dp->data; 159 unsigned int *data = (unsigned int *)dp->data;
170 160
171 if (!data) {
172 printk(KERN_ERR "scanlog: release failed no data\n");
173 return -EIO;
174 }
175 data[0] = 0; 161 data[0] = 0;
176 162
177 return 0; 163 return 0;
@@ -200,12 +186,11 @@ static int __init scanlog_init(void)
200 if (!data) 186 if (!data)
201 goto err; 187 goto err;
202 188
203 ent = proc_create("ppc64/rtas/scan-log-dump", S_IRUSR, NULL, 189 ent = proc_create_data("ppc64/rtas/scan-log-dump", S_IRUSR, NULL,
204 &scanlog_fops); 190 &scanlog_fops, data);
205 if (!ent) 191 if (!ent)
206 goto err; 192 goto err;
207 193
208 ent->data = data;
209 proc_ppc64_scan_log_dump = ent; 194 proc_ppc64_scan_log_dump = ent;
210 195
211 return 0; 196 return 0;
diff --git a/arch/powerpc/sysdev/mv64x60_dev.c b/arch/powerpc/sysdev/mv64x60_dev.c
index 41af1223e2a0..a132e0de8ca5 100644
--- a/arch/powerpc/sysdev/mv64x60_dev.c
+++ b/arch/powerpc/sysdev/mv64x60_dev.c
@@ -239,6 +239,8 @@ static int __init mv64x60_eth_device_setup(struct device_node *np, int id,
239 239
240 memset(&pdata, 0, sizeof(pdata)); 240 memset(&pdata, 0, sizeof(pdata));
241 241
242 pdata.shared = shared_pdev;
243
242 prop = of_get_property(np, "reg", NULL); 244 prop = of_get_property(np, "reg", NULL);
243 if (!prop) 245 if (!prop)
244 return -ENODEV; 246 return -ENODEV;
diff --git a/arch/ppc/syslib/mv64x60.c b/arch/ppc/syslib/mv64x60.c
index 90fe904d3614..418f3053de52 100644
--- a/arch/ppc/syslib/mv64x60.c
+++ b/arch/ppc/syslib/mv64x60.c
@@ -341,6 +341,7 @@ static struct resource mv64x60_eth0_resources[] = {
341}; 341};
342 342
343static struct mv643xx_eth_platform_data eth0_pd = { 343static struct mv643xx_eth_platform_data eth0_pd = {
344 .shared = &mv64x60_eth_shared_device;
344 .port_number = 0, 345 .port_number = 0,
345}; 346};
346 347
@@ -366,6 +367,7 @@ static struct resource mv64x60_eth1_resources[] = {
366}; 367};
367 368
368static struct mv643xx_eth_platform_data eth1_pd = { 369static struct mv643xx_eth_platform_data eth1_pd = {
370 .shared = &mv64x60_eth_shared_device;
369 .port_number = 1, 371 .port_number = 1,
370}; 372};
371 373
@@ -391,6 +393,7 @@ static struct resource mv64x60_eth2_resources[] = {
391}; 393};
392 394
393static struct mv643xx_eth_platform_data eth2_pd = { 395static struct mv643xx_eth_platform_data eth2_pd = {
396 .shared = &mv64x60_eth_shared_device;
394 .port_number = 2, 397 .port_number = 2,
395}; 398};
396 399
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 29a7940f284f..1d035082e78e 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -430,6 +430,13 @@ config CMM_IUCV
430 Select this option to enable the special message interface to 430 Select this option to enable the special message interface to
431 the cooperative memory management. 431 the cooperative memory management.
432 432
433config PAGE_STATES
434 bool "Unused page notification"
435 help
436 This enables the notification of unused pages to the
437 hypervisor. The ESSA instruction is used to do the states
438 changes between a page that has content and the unused state.
439
433config VIRT_TIMER 440config VIRT_TIMER
434 bool "Virtual CPU timer support" 441 bool "Virtual CPU timer support"
435 help 442 help
diff --git a/arch/s390/kernel/compat_wrapper.S b/arch/s390/kernel/compat_wrapper.S
index 743d54f0b8db..d003a6e16afb 100644
--- a/arch/s390/kernel/compat_wrapper.S
+++ b/arch/s390/kernel/compat_wrapper.S
@@ -121,7 +121,7 @@ sys32_ptrace_wrapper:
121 lgfr %r3,%r3 # long 121 lgfr %r3,%r3 # long
122 llgtr %r4,%r4 # long 122 llgtr %r4,%r4 # long
123 llgfr %r5,%r5 # long 123 llgfr %r5,%r5 # long
124 jg sys_ptrace # branch to system call 124 jg compat_sys_ptrace # branch to system call
125 125
126 .globl sys32_alarm_wrapper 126 .globl sys32_alarm_wrapper
127sys32_alarm_wrapper: 127sys32_alarm_wrapper:
diff --git a/arch/s390/kernel/entry.S b/arch/s390/kernel/entry.S
index bdbb3bcd78a5..708cf9cf9a35 100644
--- a/arch/s390/kernel/entry.S
+++ b/arch/s390/kernel/entry.S
@@ -279,8 +279,6 @@ sysc_do_restart:
279 st %r2,SP_R2(%r15) # store return value (change R2 on stack) 279 st %r2,SP_R2(%r15) # store return value (change R2 on stack)
280 280
281sysc_return: 281sysc_return:
282 tm SP_PSW+1(%r15),0x01 # returning to user ?
283 bno BASED(sysc_restore)
284 tm __TI_flags+3(%r9),_TIF_WORK_SVC 282 tm __TI_flags+3(%r9),_TIF_WORK_SVC
285 bnz BASED(sysc_work) # there is work to do (signals etc.) 283 bnz BASED(sysc_work) # there is work to do (signals etc.)
286sysc_restore: 284sysc_restore:
@@ -312,6 +310,8 @@ sysc_work_loop:
312# One of the work bits is on. Find out which one. 310# One of the work bits is on. Find out which one.
313# 311#
314sysc_work: 312sysc_work:
313 tm SP_PSW+1(%r15),0x01 # returning to user ?
314 bno BASED(sysc_restore)
315 tm __TI_flags+3(%r9),_TIF_MCCK_PENDING 315 tm __TI_flags+3(%r9),_TIF_MCCK_PENDING
316 bo BASED(sysc_mcck_pending) 316 bo BASED(sysc_mcck_pending)
317 tm __TI_flags+3(%r9),_TIF_NEED_RESCHED 317 tm __TI_flags+3(%r9),_TIF_NEED_RESCHED
@@ -602,12 +602,6 @@ io_no_vtime:
602 la %r2,SP_PTREGS(%r15) # address of register-save area 602 la %r2,SP_PTREGS(%r15) # address of register-save area
603 basr %r14,%r1 # branch to standard irq handler 603 basr %r14,%r1 # branch to standard irq handler
604io_return: 604io_return:
605 tm SP_PSW+1(%r15),0x01 # returning to user ?
606#ifdef CONFIG_PREEMPT
607 bno BASED(io_preempt) # no -> check for preemptive scheduling
608#else
609 bno BASED(io_restore) # no-> skip resched & signal
610#endif
611 tm __TI_flags+3(%r9),_TIF_WORK_INT 605 tm __TI_flags+3(%r9),_TIF_WORK_INT
612 bnz BASED(io_work) # there is work to do (signals etc.) 606 bnz BASED(io_work) # there is work to do (signals etc.)
613io_restore: 607io_restore:
@@ -629,10 +623,18 @@ io_restore_trace_psw:
629 .long 0, io_restore_trace + 0x80000000 623 .long 0, io_restore_trace + 0x80000000
630#endif 624#endif
631 625
632#ifdef CONFIG_PREEMPT 626#
633io_preempt: 627# switch to kernel stack, then check the TIF bits
628#
629io_work:
630 tm SP_PSW+1(%r15),0x01 # returning to user ?
631#ifndef CONFIG_PREEMPT
632 bno BASED(io_restore) # no-> skip resched & signal
633#else
634 bnz BASED(io_work_user) # no -> check for preemptive scheduling
635 # check for preemptive scheduling
634 icm %r0,15,__TI_precount(%r9) 636 icm %r0,15,__TI_precount(%r9)
635 bnz BASED(io_restore) 637 bnz BASED(io_restore) # preemption disabled
636 l %r1,SP_R15(%r15) 638 l %r1,SP_R15(%r15)
637 s %r1,BASED(.Lc_spsize) 639 s %r1,BASED(.Lc_spsize)
638 mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15) 640 mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15)
@@ -646,10 +648,7 @@ io_resume_loop:
646 br %r1 # call schedule 648 br %r1 # call schedule
647#endif 649#endif
648 650
649# 651io_work_user:
650# switch to kernel stack, then check the TIF bits
651#
652io_work:
653 l %r1,__LC_KERNEL_STACK 652 l %r1,__LC_KERNEL_STACK
654 s %r1,BASED(.Lc_spsize) 653 s %r1,BASED(.Lc_spsize)
655 mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15) 654 mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15)
diff --git a/arch/s390/kernel/entry64.S b/arch/s390/kernel/entry64.S
index 5a4a7bcd2bba..fee10177dbfc 100644
--- a/arch/s390/kernel/entry64.S
+++ b/arch/s390/kernel/entry64.S
@@ -271,8 +271,6 @@ sysc_noemu:
271 stg %r2,SP_R2(%r15) # store return value (change R2 on stack) 271 stg %r2,SP_R2(%r15) # store return value (change R2 on stack)
272 272
273sysc_return: 273sysc_return:
274 tm SP_PSW+1(%r15),0x01 # returning to user ?
275 jno sysc_restore
276 tm __TI_flags+7(%r9),_TIF_WORK_SVC 274 tm __TI_flags+7(%r9),_TIF_WORK_SVC
277 jnz sysc_work # there is work to do (signals etc.) 275 jnz sysc_work # there is work to do (signals etc.)
278sysc_restore: 276sysc_restore:
@@ -304,6 +302,8 @@ sysc_work_loop:
304# One of the work bits is on. Find out which one. 302# One of the work bits is on. Find out which one.
305# 303#
306sysc_work: 304sysc_work:
305 tm SP_PSW+1(%r15),0x01 # returning to user ?
306 jno sysc_restore
307 tm __TI_flags+7(%r9),_TIF_MCCK_PENDING 307 tm __TI_flags+7(%r9),_TIF_MCCK_PENDING
308 jo sysc_mcck_pending 308 jo sysc_mcck_pending
309 tm __TI_flags+7(%r9),_TIF_NEED_RESCHED 309 tm __TI_flags+7(%r9),_TIF_NEED_RESCHED
@@ -585,12 +585,6 @@ io_no_vtime:
585 la %r2,SP_PTREGS(%r15) # address of register-save area 585 la %r2,SP_PTREGS(%r15) # address of register-save area
586 brasl %r14,do_IRQ # call standard irq handler 586 brasl %r14,do_IRQ # call standard irq handler
587io_return: 587io_return:
588 tm SP_PSW+1(%r15),0x01 # returning to user ?
589#ifdef CONFIG_PREEMPT
590 jno io_preempt # no -> check for preemptive scheduling
591#else
592 jno io_restore # no-> skip resched & signal
593#endif
594 tm __TI_flags+7(%r9),_TIF_WORK_INT 588 tm __TI_flags+7(%r9),_TIF_WORK_INT
595 jnz io_work # there is work to do (signals etc.) 589 jnz io_work # there is work to do (signals etc.)
596io_restore: 590io_restore:
@@ -612,10 +606,41 @@ io_restore_trace_psw:
612 .quad 0, io_restore_trace 606 .quad 0, io_restore_trace
613#endif 607#endif
614 608
615#ifdef CONFIG_PREEMPT 609#
616io_preempt: 610# There is work todo, we need to check if we return to userspace, then
611# check, if we are in SIE, if yes leave it
612#
613io_work:
614 tm SP_PSW+1(%r15),0x01 # returning to user ?
615#ifndef CONFIG_PREEMPT
616#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE)
617 jnz io_work_user # yes -> no need to check for SIE
618 la %r1, BASED(sie_opcode) # we return to kernel here
619 lg %r2, SP_PSW+8(%r15)
620 clc 0(2,%r1), 0(%r2) # is current instruction = SIE?
621 jne io_restore # no-> return to kernel
622 lg %r1, SP_PSW+8(%r15) # yes-> add 4 bytes to leave SIE
623 aghi %r1, 4
624 stg %r1, SP_PSW+8(%r15)
625 j io_restore # return to kernel
626#else
627 jno io_restore # no-> skip resched & signal
628#endif
629#else
630 jnz io_work_user # yes -> do resched & signal
631#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE)
632 la %r1, BASED(sie_opcode)
633 lg %r2, SP_PSW+8(%r15)
634 clc 0(2,%r1), 0(%r2) # is current instruction = SIE?
635 jne 0f # no -> leave PSW alone
636 lg %r1, SP_PSW+8(%r15) # yes-> add 4 bytes to leave SIE
637 aghi %r1, 4
638 stg %r1, SP_PSW+8(%r15)
6390:
640#endif
641 # check for preemptive scheduling
617 icm %r0,15,__TI_precount(%r9) 642 icm %r0,15,__TI_precount(%r9)
618 jnz io_restore 643 jnz io_restore # preemption is disabled
619 # switch to kernel stack 644 # switch to kernel stack
620 lg %r1,SP_R15(%r15) 645 lg %r1,SP_R15(%r15)
621 aghi %r1,-SP_SIZE 646 aghi %r1,-SP_SIZE
@@ -629,10 +654,7 @@ io_resume_loop:
629 jg preempt_schedule_irq 654 jg preempt_schedule_irq
630#endif 655#endif
631 656
632# 657io_work_user:
633# switch to kernel stack, then check TIF bits
634#
635io_work:
636 lg %r1,__LC_KERNEL_STACK 658 lg %r1,__LC_KERNEL_STACK
637 aghi %r1,-SP_SIZE 659 aghi %r1,-SP_SIZE
638 mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15) 660 mvc SP_PTREGS(__PT_SIZE,%r1),SP_PTREGS(%r15)
@@ -653,6 +675,11 @@ io_work_loop:
653 j io_restore 675 j io_restore
654io_work_done: 676io_work_done:
655 677
678#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE)
679sie_opcode:
680 .long 0xb2140000
681#endif
682
656# 683#
657# _TIF_MCCK_PENDING is set, call handler 684# _TIF_MCCK_PENDING is set, call handler
658# 685#
diff --git a/arch/s390/kernel/ptrace.c b/arch/s390/kernel/ptrace.c
index 7f4270163744..35827b9bd4d1 100644
--- a/arch/s390/kernel/ptrace.c
+++ b/arch/s390/kernel/ptrace.c
@@ -292,8 +292,7 @@ poke_user(struct task_struct *child, addr_t addr, addr_t data)
292 return 0; 292 return 0;
293} 293}
294 294
295static int 295long arch_ptrace(struct task_struct *child, long request, long addr, long data)
296do_ptrace_normal(struct task_struct *child, long request, long addr, long data)
297{ 296{
298 ptrace_area parea; 297 ptrace_area parea;
299 int copied, ret; 298 int copied, ret;
@@ -529,35 +528,19 @@ poke_user_emu31(struct task_struct *child, addr_t addr, addr_t data)
529 return 0; 528 return 0;
530} 529}
531 530
532static int 531long compat_arch_ptrace(struct task_struct *child, compat_long_t request,
533do_ptrace_emu31(struct task_struct *child, long request, long addr, long data) 532 compat_ulong_t caddr, compat_ulong_t cdata)
534{ 533{
535 unsigned int tmp; /* 4 bytes !! */ 534 unsigned long addr = caddr;
535 unsigned long data = cdata;
536 ptrace_area_emu31 parea; 536 ptrace_area_emu31 parea;
537 int copied, ret; 537 int copied, ret;
538 538
539 switch (request) { 539 switch (request) {
540 case PTRACE_PEEKTEXT:
541 case PTRACE_PEEKDATA:
542 /* read word at location addr. */
543 copied = access_process_vm(child, addr, &tmp, sizeof(tmp), 0);
544 if (copied != sizeof(tmp))
545 return -EIO;
546 return put_user(tmp, (unsigned int __force __user *) data);
547
548 case PTRACE_PEEKUSR: 540 case PTRACE_PEEKUSR:
549 /* read the word at location addr in the USER area. */ 541 /* read the word at location addr in the USER area. */
550 return peek_user_emu31(child, addr, data); 542 return peek_user_emu31(child, addr, data);
551 543
552 case PTRACE_POKETEXT:
553 case PTRACE_POKEDATA:
554 /* write the word at location addr. */
555 tmp = data;
556 copied = access_process_vm(child, addr, &tmp, sizeof(tmp), 1);
557 if (copied != sizeof(tmp))
558 return -EIO;
559 return 0;
560
561 case PTRACE_POKEUSR: 544 case PTRACE_POKEUSR:
562 /* write the word at location addr in the USER area */ 545 /* write the word at location addr in the USER area */
563 return poke_user_emu31(child, addr, data); 546 return poke_user_emu31(child, addr, data);
@@ -587,82 +570,11 @@ do_ptrace_emu31(struct task_struct *child, long request, long addr, long data)
587 copied += sizeof(unsigned int); 570 copied += sizeof(unsigned int);
588 } 571 }
589 return 0; 572 return 0;
590 case PTRACE_GETEVENTMSG:
591 return put_user((__u32) child->ptrace_message,
592 (unsigned int __force __user *) data);
593 case PTRACE_GETSIGINFO:
594 if (child->last_siginfo == NULL)
595 return -EINVAL;
596 return copy_siginfo_to_user32((compat_siginfo_t
597 __force __user *) data,
598 child->last_siginfo);
599 case PTRACE_SETSIGINFO:
600 if (child->last_siginfo == NULL)
601 return -EINVAL;
602 return copy_siginfo_from_user32(child->last_siginfo,
603 (compat_siginfo_t
604 __force __user *) data);
605 } 573 }
606 return ptrace_request(child, request, addr, data); 574 return compat_ptrace_request(child, request, addr, data);
607} 575}
608#endif 576#endif
609 577
610long arch_ptrace(struct task_struct *child, long request, long addr, long data)
611{
612 switch (request) {
613 case PTRACE_SYSCALL:
614 /* continue and stop at next (return from) syscall */
615 case PTRACE_CONT:
616 /* restart after signal. */
617 if (!valid_signal(data))
618 return -EIO;
619 if (request == PTRACE_SYSCALL)
620 set_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
621 else
622 clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
623 child->exit_code = data;
624 /* make sure the single step bit is not set. */
625 user_disable_single_step(child);
626 wake_up_process(child);
627 return 0;
628
629 case PTRACE_KILL:
630 /*
631 * make the child exit. Best I can do is send it a sigkill.
632 * perhaps it should be put in the status that it wants to
633 * exit.
634 */
635 if (child->exit_state == EXIT_ZOMBIE) /* already dead */
636 return 0;
637 child->exit_code = SIGKILL;
638 /* make sure the single step bit is not set. */
639 user_disable_single_step(child);
640 wake_up_process(child);
641 return 0;
642
643 case PTRACE_SINGLESTEP:
644 /* set the trap flag. */
645 if (!valid_signal(data))
646 return -EIO;
647 clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
648 child->exit_code = data;
649 user_enable_single_step(child);
650 /* give it a chance to run. */
651 wake_up_process(child);
652 return 0;
653
654 /* Do requests that differ for 31/64 bit */
655 default:
656#ifdef CONFIG_COMPAT
657 if (test_thread_flag(TIF_31BIT))
658 return do_ptrace_emu31(child, request, addr, data);
659#endif
660 return do_ptrace_normal(child, request, addr, data);
661 }
662 /* Not reached. */
663 return -EIO;
664}
665
666asmlinkage void 578asmlinkage void
667syscall_trace(struct pt_regs *regs, int entryexit) 579syscall_trace(struct pt_regs *regs, int entryexit)
668{ 580{
diff --git a/arch/s390/kernel/sys_s390.c b/arch/s390/kernel/sys_s390.c
index 988d0d64c2c8..5fdb799062b7 100644
--- a/arch/s390/kernel/sys_s390.c
+++ b/arch/s390/kernel/sys_s390.c
@@ -32,23 +32,6 @@
32#include <asm/uaccess.h> 32#include <asm/uaccess.h>
33#include "entry.h" 33#include "entry.h"
34 34
35/*
36 * sys_pipe() is the normal C calling standard for creating
37 * a pipe. It's not the way Unix traditionally does this, though.
38 */
39asmlinkage long sys_pipe(unsigned long __user *fildes)
40{
41 int fd[2];
42 int error;
43
44 error = do_pipe(fd);
45 if (!error) {
46 if (copy_to_user(fildes, fd, 2*sizeof(int)))
47 error = -EFAULT;
48 }
49 return error;
50}
51
52/* common code for old and new mmaps */ 35/* common code for old and new mmaps */
53static inline long do_mmap2( 36static inline long do_mmap2(
54 unsigned long addr, unsigned long len, 37 unsigned long addr, unsigned long len,
diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig
index 1761b74d639b..e051cad1f1e0 100644
--- a/arch/s390/kvm/Kconfig
+++ b/arch/s390/kvm/Kconfig
@@ -22,7 +22,6 @@ config KVM
22 select PREEMPT_NOTIFIERS 22 select PREEMPT_NOTIFIERS
23 select ANON_INODES 23 select ANON_INODES
24 select S390_SWITCH_AMODE 24 select S390_SWITCH_AMODE
25 select PREEMPT
26 ---help--- 25 ---help---
27 Support hosting paravirtualized guest machines using the SIE 26 Support hosting paravirtualized guest machines using the SIE
28 virtualization capability on the mainframe. This should work 27 virtualization capability on the mainframe. This should work
diff --git a/arch/s390/kvm/intercept.c b/arch/s390/kvm/intercept.c
index 349581a26103..47a0b642174c 100644
--- a/arch/s390/kvm/intercept.c
+++ b/arch/s390/kvm/intercept.c
@@ -105,6 +105,9 @@ static intercept_handler_t instruction_handlers[256] = {
105static int handle_noop(struct kvm_vcpu *vcpu) 105static int handle_noop(struct kvm_vcpu *vcpu)
106{ 106{
107 switch (vcpu->arch.sie_block->icptcode) { 107 switch (vcpu->arch.sie_block->icptcode) {
108 case 0x0:
109 vcpu->stat.exit_null++;
110 break;
108 case 0x10: 111 case 0x10:
109 vcpu->stat.exit_external_request++; 112 vcpu->stat.exit_external_request++;
110 break; 113 break;
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 98d1e73e01f1..0ac36a649eba 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -31,6 +31,7 @@
31 31
32struct kvm_stats_debugfs_item debugfs_entries[] = { 32struct kvm_stats_debugfs_item debugfs_entries[] = {
33 { "userspace_handled", VCPU_STAT(exit_userspace) }, 33 { "userspace_handled", VCPU_STAT(exit_userspace) },
34 { "exit_null", VCPU_STAT(exit_null) },
34 { "exit_validity", VCPU_STAT(exit_validity) }, 35 { "exit_validity", VCPU_STAT(exit_validity) },
35 { "exit_stop_request", VCPU_STAT(exit_stop_request) }, 36 { "exit_stop_request", VCPU_STAT(exit_stop_request) },
36 { "exit_external_request", VCPU_STAT(exit_external_request) }, 37 { "exit_external_request", VCPU_STAT(exit_external_request) },
@@ -221,10 +222,6 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
221 vcpu->arch.guest_fpregs.fpc &= FPC_VALID_MASK; 222 vcpu->arch.guest_fpregs.fpc &= FPC_VALID_MASK;
222 restore_fp_regs(&vcpu->arch.guest_fpregs); 223 restore_fp_regs(&vcpu->arch.guest_fpregs);
223 restore_access_regs(vcpu->arch.guest_acrs); 224 restore_access_regs(vcpu->arch.guest_acrs);
224
225 if (signal_pending(current))
226 atomic_set_mask(CPUSTAT_STOP_INT,
227 &vcpu->arch.sie_block->cpuflags);
228} 225}
229 226
230void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu) 227void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
diff --git a/arch/s390/mm/Makefile b/arch/s390/mm/Makefile
index fb988a48a754..2a7458134544 100644
--- a/arch/s390/mm/Makefile
+++ b/arch/s390/mm/Makefile
@@ -5,3 +5,4 @@
5obj-y := init.o fault.o extmem.o mmap.o vmem.o pgtable.o 5obj-y := init.o fault.o extmem.o mmap.o vmem.o pgtable.o
6obj-$(CONFIG_CMM) += cmm.o 6obj-$(CONFIG_CMM) += cmm.o
7obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o 7obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o
8obj-$(CONFIG_PAGE_STATES) += page-states.o
diff --git a/arch/s390/mm/init.c b/arch/s390/mm/init.c
index fa31de6ae97a..29f3a63806b9 100644
--- a/arch/s390/mm/init.c
+++ b/arch/s390/mm/init.c
@@ -126,6 +126,9 @@ void __init mem_init(void)
126 /* clear the zero-page */ 126 /* clear the zero-page */
127 memset(empty_zero_page, 0, PAGE_SIZE); 127 memset(empty_zero_page, 0, PAGE_SIZE);
128 128
129 /* Setup guest page hinting */
130 cmma_init();
131
129 /* this will put all low memory onto the freelists */ 132 /* this will put all low memory onto the freelists */
130 totalram_pages += free_all_bootmem(); 133 totalram_pages += free_all_bootmem();
131 134
diff --git a/arch/s390/mm/page-states.c b/arch/s390/mm/page-states.c
new file mode 100644
index 000000000000..fc0ad73ffd90
--- /dev/null
+++ b/arch/s390/mm/page-states.c
@@ -0,0 +1,79 @@
1/*
2 * arch/s390/mm/page-states.c
3 *
4 * Copyright IBM Corp. 2008
5 *
6 * Guest page hinting for unused pages.
7 *
8 * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>
9 */
10
11#include <linux/kernel.h>
12#include <linux/errno.h>
13#include <linux/types.h>
14#include <linux/mm.h>
15#include <linux/init.h>
16
17#define ESSA_SET_STABLE 1
18#define ESSA_SET_UNUSED 2
19
20static int cmma_flag;
21
22static int __init cmma(char *str)
23{
24 char *parm;
25 parm = strstrip(str);
26 if (strcmp(parm, "yes") == 0 || strcmp(parm, "on") == 0) {
27 cmma_flag = 1;
28 return 1;
29 }
30 cmma_flag = 0;
31 if (strcmp(parm, "no") == 0 || strcmp(parm, "off") == 0)
32 return 1;
33 return 0;
34}
35
36__setup("cmma=", cmma);
37
38void __init cmma_init(void)
39{
40 register unsigned long tmp asm("0") = 0;
41 register int rc asm("1") = -EOPNOTSUPP;
42
43 if (!cmma_flag)
44 return;
45 asm volatile(
46 " .insn rrf,0xb9ab0000,%1,%1,0,0\n"
47 "0: la %0,0\n"
48 "1:\n"
49 EX_TABLE(0b,1b)
50 : "+&d" (rc), "+&d" (tmp));
51 if (rc)
52 cmma_flag = 0;
53}
54
55void arch_free_page(struct page *page, int order)
56{
57 int i, rc;
58
59 if (!cmma_flag)
60 return;
61 for (i = 0; i < (1 << order); i++)
62 asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0"
63 : "=&d" (rc)
64 : "a" ((page_to_pfn(page) + i) << PAGE_SHIFT),
65 "i" (ESSA_SET_UNUSED));
66}
67
68void arch_alloc_page(struct page *page, int order)
69{
70 int i, rc;
71
72 if (!cmma_flag)
73 return;
74 for (i = 0; i < (1 << order); i++)
75 asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0"
76 : "=&d" (rc)
77 : "a" ((page_to_pfn(page) + i) << PAGE_SHIFT),
78 "i" (ESSA_SET_STABLE));
79}
diff --git a/arch/sh/kernel/sys_sh64.c b/arch/sh/kernel/sys_sh64.c
index 578004d71e02..91fb8445a5a0 100644
--- a/arch/sh/kernel/sys_sh64.c
+++ b/arch/sh/kernel/sys_sh64.c
@@ -31,23 +31,6 @@
31#include <asm/unistd.h> 31#include <asm/unistd.h>
32 32
33/* 33/*
34 * sys_pipe() is the normal C calling standard for creating
35 * a pipe. It's not the way Unix traditionally does this, though.
36 */
37asmlinkage int sys_pipe(unsigned long * fildes)
38{
39 int fd[2];
40 int error;
41
42 error = do_pipe(fd);
43 if (!error) {
44 if (copy_to_user(fildes, fd, 2*sizeof(int)))
45 error = -EFAULT;
46 }
47 return error;
48}
49
50/*
51 * Do a system call from kernel instead of calling sys_execve so we 34 * Do a system call from kernel instead of calling sys_execve so we
52 * end up with proper pt_regs. 35 * end up with proper pt_regs.
53 */ 36 */
diff --git a/arch/sparc/kernel/process.c b/arch/sparc/kernel/process.c
index e7f35198ae34..36431f377dee 100644
--- a/arch/sparc/kernel/process.c
+++ b/arch/sparc/kernel/process.c
@@ -419,14 +419,26 @@ asmlinkage int sparc_do_fork(unsigned long clone_flags,
419 unsigned long stack_size) 419 unsigned long stack_size)
420{ 420{
421 unsigned long parent_tid_ptr, child_tid_ptr; 421 unsigned long parent_tid_ptr, child_tid_ptr;
422 unsigned long orig_i1 = regs->u_regs[UREG_I1];
423 long ret;
422 424
423 parent_tid_ptr = regs->u_regs[UREG_I2]; 425 parent_tid_ptr = regs->u_regs[UREG_I2];
424 child_tid_ptr = regs->u_regs[UREG_I4]; 426 child_tid_ptr = regs->u_regs[UREG_I4];
425 427
426 return do_fork(clone_flags, stack_start, 428 ret = do_fork(clone_flags, stack_start,
427 regs, stack_size, 429 regs, stack_size,
428 (int __user *) parent_tid_ptr, 430 (int __user *) parent_tid_ptr,
429 (int __user *) child_tid_ptr); 431 (int __user *) child_tid_ptr);
432
433 /* If we get an error and potentially restart the system
434 * call, we're screwed because copy_thread() clobbered
435 * the parent's %o1. So detect that case and restore it
436 * here.
437 */
438 if ((unsigned long)ret >= -ERESTART_RESTARTBLOCK)
439 regs->u_regs[UREG_I1] = orig_i1;
440
441 return ret;
430} 442}
431 443
432/* Copy a Sparc thread. The fork() return value conventions 444/* Copy a Sparc thread. The fork() return value conventions
diff --git a/arch/sparc/kernel/signal.c b/arch/sparc/kernel/signal.c
index 3c312290c3c2..368157926d24 100644
--- a/arch/sparc/kernel/signal.c
+++ b/arch/sparc/kernel/signal.c
@@ -245,15 +245,29 @@ static inline int invalid_frame_pointer(void __user *fp, int fplen)
245 245
246static inline void __user *get_sigframe(struct sigaction *sa, struct pt_regs *regs, unsigned long framesize) 246static inline void __user *get_sigframe(struct sigaction *sa, struct pt_regs *regs, unsigned long framesize)
247{ 247{
248 unsigned long sp; 248 unsigned long sp = regs->u_regs[UREG_FP];
249 249
250 sp = regs->u_regs[UREG_FP]; 250 /*
251 * If we are on the alternate signal stack and would overflow it, don't.
252 * Return an always-bogus address instead so we will die with SIGSEGV.
253 */
254 if (on_sig_stack(sp) && !likely(on_sig_stack(sp - framesize)))
255 return (void __user *) -1L;
251 256
252 /* This is the X/Open sanctioned signal stack switching. */ 257 /* This is the X/Open sanctioned signal stack switching. */
253 if (sa->sa_flags & SA_ONSTACK) { 258 if (sa->sa_flags & SA_ONSTACK) {
254 if (!on_sig_stack(sp) && !((current->sas_ss_sp + current->sas_ss_size) & 7)) 259 if (sas_ss_flags(sp) == 0)
255 sp = current->sas_ss_sp + current->sas_ss_size; 260 sp = current->sas_ss_sp + current->sas_ss_size;
256 } 261 }
262
263 /* Always align the stack frame. This handles two cases. First,
264 * sigaltstack need not be mindful of platform specific stack
265 * alignment. Second, if we took this signal because the stack
266 * is not aligned properly, we'd like to take the signal cleanly
267 * and report that.
268 */
269 sp &= ~7UL;
270
257 return (void __user *)(sp - framesize); 271 return (void __user *)(sp - framesize);
258} 272}
259 273
diff --git a/arch/sparc/kernel/sys_sparc.c b/arch/sparc/kernel/sys_sparc.c
index f188b5dc9fd0..e995491c4436 100644
--- a/arch/sparc/kernel/sys_sparc.c
+++ b/arch/sparc/kernel/sys_sparc.c
@@ -223,8 +223,7 @@ int sparc_mmap_check(unsigned long addr, unsigned long len, unsigned long flags)
223{ 223{
224 if (ARCH_SUN4C_SUN4 && 224 if (ARCH_SUN4C_SUN4 &&
225 (len > 0x20000000 || 225 (len > 0x20000000 ||
226 ((flags & MAP_FIXED) && 226 (addr < 0xe0000000 && addr + len > 0x20000000)))
227 addr < 0xe0000000 && addr + len > 0x20000000)))
228 return -EINVAL; 227 return -EINVAL;
229 228
230 /* See asm-sparc/uaccess.h */ 229 /* See asm-sparc/uaccess.h */
diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c
index 500ac6d483a0..4129c0449856 100644
--- a/arch/sparc64/kernel/process.c
+++ b/arch/sparc64/kernel/process.c
@@ -503,6 +503,8 @@ asmlinkage long sparc_do_fork(unsigned long clone_flags,
503 unsigned long stack_size) 503 unsigned long stack_size)
504{ 504{
505 int __user *parent_tid_ptr, *child_tid_ptr; 505 int __user *parent_tid_ptr, *child_tid_ptr;
506 unsigned long orig_i1 = regs->u_regs[UREG_I1];
507 long ret;
506 508
507#ifdef CONFIG_COMPAT 509#ifdef CONFIG_COMPAT
508 if (test_thread_flag(TIF_32BIT)) { 510 if (test_thread_flag(TIF_32BIT)) {
@@ -515,9 +517,19 @@ asmlinkage long sparc_do_fork(unsigned long clone_flags,
515 child_tid_ptr = (int __user *) regs->u_regs[UREG_I4]; 517 child_tid_ptr = (int __user *) regs->u_regs[UREG_I4];
516 } 518 }
517 519
518 return do_fork(clone_flags, stack_start, 520 ret = do_fork(clone_flags, stack_start,
519 regs, stack_size, 521 regs, stack_size,
520 parent_tid_ptr, child_tid_ptr); 522 parent_tid_ptr, child_tid_ptr);
523
524 /* If we get an error and potentially restart the system
525 * call, we're screwed because copy_thread() clobbered
526 * the parent's %o1. So detect that case and restore it
527 * here.
528 */
529 if ((unsigned long)ret >= -ERESTART_RESTARTBLOCK)
530 regs->u_regs[UREG_I1] = orig_i1;
531
532 return ret;
521} 533}
522 534
523/* Copy a Sparc thread. The fork() return value conventions 535/* Copy a Sparc thread. The fork() return value conventions
diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c
index 45d6bf632daa..07c0443ea3f5 100644
--- a/arch/sparc64/kernel/signal.c
+++ b/arch/sparc64/kernel/signal.c
@@ -376,16 +376,29 @@ save_fpu_state(struct pt_regs *regs, __siginfo_fpu_t __user *fpu)
376 376
377static inline void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, unsigned long framesize) 377static inline void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, unsigned long framesize)
378{ 378{
379 unsigned long sp; 379 unsigned long sp = regs->u_regs[UREG_FP] + STACK_BIAS;
380 380
381 sp = regs->u_regs[UREG_FP] + STACK_BIAS; 381 /*
382 * If we are on the alternate signal stack and would overflow it, don't.
383 * Return an always-bogus address instead so we will die with SIGSEGV.
384 */
385 if (on_sig_stack(sp) && !likely(on_sig_stack(sp - framesize)))
386 return (void __user *) -1L;
382 387
383 /* This is the X/Open sanctioned signal stack switching. */ 388 /* This is the X/Open sanctioned signal stack switching. */
384 if (ka->sa.sa_flags & SA_ONSTACK) { 389 if (ka->sa.sa_flags & SA_ONSTACK) {
385 if (!on_sig_stack(sp) && 390 if (sas_ss_flags(sp) == 0)
386 !((current->sas_ss_sp + current->sas_ss_size) & 7))
387 sp = current->sas_ss_sp + current->sas_ss_size; 391 sp = current->sas_ss_sp + current->sas_ss_size;
388 } 392 }
393
394 /* Always align the stack frame. This handles two cases. First,
395 * sigaltstack need not be mindful of platform specific stack
396 * alignment. Second, if we took this signal because the stack
397 * is not aligned properly, we'd like to take the signal cleanly
398 * and report that.
399 */
400 sp &= ~7UL;
401
389 return (void __user *)(sp - framesize); 402 return (void __user *)(sp - framesize);
390} 403}
391 404
diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c
index 9415d2c918c5..0f6b7b156efd 100644
--- a/arch/sparc64/kernel/signal32.c
+++ b/arch/sparc64/kernel/signal32.c
@@ -406,11 +406,27 @@ static void __user *get_sigframe(struct sigaction *sa, struct pt_regs *regs, uns
406 regs->u_regs[UREG_FP] &= 0x00000000ffffffffUL; 406 regs->u_regs[UREG_FP] &= 0x00000000ffffffffUL;
407 sp = regs->u_regs[UREG_FP]; 407 sp = regs->u_regs[UREG_FP];
408 408
409 /*
410 * If we are on the alternate signal stack and would overflow it, don't.
411 * Return an always-bogus address instead so we will die with SIGSEGV.
412 */
413 if (on_sig_stack(sp) && !likely(on_sig_stack(sp - framesize)))
414 return (void __user *) -1L;
415
409 /* This is the X/Open sanctioned signal stack switching. */ 416 /* This is the X/Open sanctioned signal stack switching. */
410 if (sa->sa_flags & SA_ONSTACK) { 417 if (sa->sa_flags & SA_ONSTACK) {
411 if (!on_sig_stack(sp) && !((current->sas_ss_sp + current->sas_ss_size) & 7)) 418 if (sas_ss_flags(sp) == 0)
412 sp = current->sas_ss_sp + current->sas_ss_size; 419 sp = current->sas_ss_sp + current->sas_ss_size;
413 } 420 }
421
422 /* Always align the stack frame. This handles two cases. First,
423 * sigaltstack need not be mindful of platform specific stack
424 * alignment. Second, if we took this signal because the stack
425 * is not aligned properly, we'd like to take the signal cleanly
426 * and report that.
427 */
428 sp &= ~7UL;
429
414 return (void __user *)(sp - framesize); 430 return (void __user *)(sp - framesize);
415} 431}
416 432
diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c
index 3aba47624df4..0d6403a630ac 100644
--- a/arch/sparc64/kernel/smp.c
+++ b/arch/sparc64/kernel/smp.c
@@ -865,21 +865,14 @@ void smp_call_function_client(int irq, struct pt_regs *regs)
865 void *info = call_data->info; 865 void *info = call_data->info;
866 866
867 clear_softint(1 << irq); 867 clear_softint(1 << irq);
868
869 irq_enter();
870
871 if (!call_data->wait) {
872 /* let initiator proceed after getting data */
873 atomic_inc(&call_data->finished);
874 }
875
876 func(info);
877
878 irq_exit();
879
880 if (call_data->wait) { 868 if (call_data->wait) {
881 /* let initiator proceed only after completion */ 869 /* let initiator proceed only after completion */
870 func(info);
882 atomic_inc(&call_data->finished); 871 atomic_inc(&call_data->finished);
872 } else {
873 /* let initiator proceed after getting data */
874 atomic_inc(&call_data->finished);
875 func(info);
883 } 876 }
884} 877}
885 878
@@ -1041,9 +1034,7 @@ void smp_receive_signal(int cpu)
1041 1034
1042void smp_receive_signal_client(int irq, struct pt_regs *regs) 1035void smp_receive_signal_client(int irq, struct pt_regs *regs)
1043{ 1036{
1044 irq_enter();
1045 clear_softint(1 << irq); 1037 clear_softint(1 << irq);
1046 irq_exit();
1047} 1038}
1048 1039
1049void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs) 1040void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs)
@@ -1051,8 +1042,6 @@ void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs)
1051 struct mm_struct *mm; 1042 struct mm_struct *mm;
1052 unsigned long flags; 1043 unsigned long flags;
1053 1044
1054 irq_enter();
1055
1056 clear_softint(1 << irq); 1045 clear_softint(1 << irq);
1057 1046
1058 /* See if we need to allocate a new TLB context because 1047 /* See if we need to allocate a new TLB context because
@@ -1072,8 +1061,6 @@ void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs)
1072 load_secondary_context(mm); 1061 load_secondary_context(mm);
1073 __flush_tlb_mm(CTX_HWBITS(mm->context), 1062 __flush_tlb_mm(CTX_HWBITS(mm->context),
1074 SECONDARY_CONTEXT); 1063 SECONDARY_CONTEXT);
1075
1076 irq_exit();
1077} 1064}
1078 1065
1079void smp_new_mmu_context_version(void) 1066void smp_new_mmu_context_version(void)
@@ -1239,8 +1226,6 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs)
1239{ 1226{
1240 clear_softint(1 << irq); 1227 clear_softint(1 << irq);
1241 1228
1242 irq_enter();
1243
1244 preempt_disable(); 1229 preempt_disable();
1245 1230
1246 __asm__ __volatile__("flushw"); 1231 __asm__ __volatile__("flushw");
@@ -1253,8 +1238,6 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs)
1253 prom_world(0); 1238 prom_world(0);
1254 1239
1255 preempt_enable(); 1240 preempt_enable();
1256
1257 irq_exit();
1258} 1241}
1259 1242
1260/* /proc/profile writes can call this, don't __init it please. */ 1243/* /proc/profile writes can call this, don't __init it please. */
diff --git a/arch/sparc64/kernel/sys_sparc.c b/arch/sparc64/kernel/sys_sparc.c
index 8d4761f15fa9..0dbc941f130e 100644
--- a/arch/sparc64/kernel/sys_sparc.c
+++ b/arch/sparc64/kernel/sys_sparc.c
@@ -549,13 +549,13 @@ int sparc64_mmap_check(unsigned long addr, unsigned long len,
549 if (len >= STACK_TOP32) 549 if (len >= STACK_TOP32)
550 return -EINVAL; 550 return -EINVAL;
551 551
552 if ((flags & MAP_FIXED) && addr > STACK_TOP32 - len) 552 if (addr > STACK_TOP32 - len)
553 return -EINVAL; 553 return -EINVAL;
554 } else { 554 } else {
555 if (len >= VA_EXCLUDE_START) 555 if (len >= VA_EXCLUDE_START)
556 return -EINVAL; 556 return -EINVAL;
557 557
558 if ((flags & MAP_FIXED) && invalid_64bit_range(addr, len)) 558 if (invalid_64bit_range(addr, len))
559 return -EINVAL; 559 return -EINVAL;
560 } 560 }
561 561
diff --git a/arch/sparc64/kernel/sys_sparc32.c b/arch/sparc64/kernel/sys_sparc32.c
index 161ce4710fe7..1aa4288125f2 100644
--- a/arch/sparc64/kernel/sys_sparc32.c
+++ b/arch/sparc64/kernel/sys_sparc32.c
@@ -236,13 +236,6 @@ asmlinkage long sys32_getegid16(void)
236 236
237/* 32-bit timeval and related flotsam. */ 237/* 32-bit timeval and related flotsam. */
238 238
239static long get_tv32(struct timeval *o, struct compat_timeval __user *i)
240{
241 return (!access_ok(VERIFY_READ, i, sizeof(*i)) ||
242 (__get_user(o->tv_sec, &i->tv_sec) |
243 __get_user(o->tv_usec, &i->tv_usec)));
244}
245
246static inline long put_tv32(struct compat_timeval __user *o, struct timeval *i) 239static inline long put_tv32(struct compat_timeval __user *o, struct timeval *i)
247{ 240{
248 return (!access_ok(VERIFY_WRITE, o, sizeof(*o)) || 241 return (!access_ok(VERIFY_WRITE, o, sizeof(*o)) ||
@@ -757,30 +750,6 @@ asmlinkage long sys32_settimeofday(struct compat_timeval __user *tv,
757 return do_sys_settimeofday(tv ? &kts : NULL, tz ? &ktz : NULL); 750 return do_sys_settimeofday(tv ? &kts : NULL, tz ? &ktz : NULL);
758} 751}
759 752
760asmlinkage long sys32_utimes(char __user *filename,
761 struct compat_timeval __user *tvs)
762{
763 struct timespec tv[2];
764
765 if (tvs) {
766 struct timeval ktvs[2];
767 if (get_tv32(&ktvs[0], tvs) ||
768 get_tv32(&ktvs[1], 1+tvs))
769 return -EFAULT;
770
771 if (ktvs[0].tv_usec < 0 || ktvs[0].tv_usec >= 1000000 ||
772 ktvs[1].tv_usec < 0 || ktvs[1].tv_usec >= 1000000)
773 return -EINVAL;
774
775 tv[0].tv_sec = ktvs[0].tv_sec;
776 tv[0].tv_nsec = 1000 * ktvs[0].tv_usec;
777 tv[1].tv_sec = ktvs[1].tv_sec;
778 tv[1].tv_nsec = 1000 * ktvs[1].tv_usec;
779 }
780
781 return do_utimes(AT_FDCWD, filename, tvs ? tv : NULL, 0);
782}
783
784/* These are here just in case some old sparc32 binary calls it. */ 753/* These are here just in case some old sparc32 binary calls it. */
785asmlinkage long sys32_pause(void) 754asmlinkage long sys32_pause(void)
786{ 755{
diff --git a/arch/sparc64/kernel/systbls.S b/arch/sparc64/kernel/systbls.S
index a4fef2ba1ae1..8b5282d433c4 100644
--- a/arch/sparc64/kernel/systbls.S
+++ b/arch/sparc64/kernel/systbls.S
@@ -45,7 +45,7 @@ sys_call_table32:
45/*120*/ .word compat_sys_readv, compat_sys_writev, sys32_settimeofday, sys32_fchown16, sys_fchmod 45/*120*/ .word compat_sys_readv, compat_sys_writev, sys32_settimeofday, sys32_fchown16, sys_fchmod
46 .word sys_nis_syscall, sys32_setreuid16, sys32_setregid16, sys_rename, sys_truncate 46 .word sys_nis_syscall, sys32_setreuid16, sys32_setregid16, sys_rename, sys_truncate
47/*130*/ .word sys_ftruncate, sys_flock, compat_sys_lstat64, sys_nis_syscall, sys_nis_syscall 47/*130*/ .word sys_ftruncate, sys_flock, compat_sys_lstat64, sys_nis_syscall, sys_nis_syscall
48 .word sys_nis_syscall, sys32_mkdir, sys_rmdir, sys32_utimes, compat_sys_stat64 48 .word sys_nis_syscall, sys32_mkdir, sys_rmdir, compat_sys_utimes, compat_sys_stat64
49/*140*/ .word sys32_sendfile64, sys_nis_syscall, sys32_futex, sys_gettid, compat_sys_getrlimit 49/*140*/ .word sys32_sendfile64, sys_nis_syscall, sys32_futex, sys_gettid, compat_sys_getrlimit
50 .word compat_sys_setrlimit, sys_pivot_root, sys32_prctl, sys_pciconfig_read, sys_pciconfig_write 50 .word compat_sys_setrlimit, sys_pivot_root, sys32_prctl, sys_pciconfig_read, sys_pciconfig_write
51/*150*/ .word sys_nis_syscall, sys_inotify_init, sys_inotify_add_watch, sys_poll, sys_getdents64 51/*150*/ .word sys_nis_syscall, sys_inotify_init, sys_inotify_add_watch, sys_poll, sys_getdents64
diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c
index 4cad0b32b0af..ec3e2c72302a 100644
--- a/arch/sparc64/mm/init.c
+++ b/arch/sparc64/mm/init.c
@@ -771,6 +771,9 @@ static void __init find_ramdisk(unsigned long phys_base)
771 initrd_end = ramdisk_image + sparc_ramdisk_size; 771 initrd_end = ramdisk_image + sparc_ramdisk_size;
772 772
773 lmb_reserve(initrd_start, initrd_end); 773 lmb_reserve(initrd_start, initrd_end);
774
775 initrd_start += PAGE_OFFSET;
776 initrd_end += PAGE_OFFSET;
774 } 777 }
775#endif 778#endif
776} 779}
@@ -2362,16 +2365,3 @@ void __flush_tlb_all(void)
2362 __asm__ __volatile__("wrpr %0, 0, %%pstate" 2365 __asm__ __volatile__("wrpr %0, 0, %%pstate"
2363 : : "r" (pstate)); 2366 : : "r" (pstate));
2364} 2367}
2365
2366#ifdef CONFIG_MEMORY_HOTPLUG
2367
2368void online_page(struct page *page)
2369{
2370 ClearPageReserved(page);
2371 init_page_count(page);
2372 __free_page(page);
2373 totalram_pages++;
2374 num_physpages++;
2375}
2376
2377#endif /* CONFIG_MEMORY_HOTPLUG */
diff --git a/arch/um/drivers/line.c b/arch/um/drivers/line.c
index 10b86e1cc659..5047490fc299 100644
--- a/arch/um/drivers/line.c
+++ b/arch/um/drivers/line.c
@@ -191,9 +191,9 @@ void line_flush_chars(struct tty_struct *tty)
191 line_flush_buffer(tty); 191 line_flush_buffer(tty);
192} 192}
193 193
194void line_put_char(struct tty_struct *tty, unsigned char ch) 194int line_put_char(struct tty_struct *tty, unsigned char ch)
195{ 195{
196 line_write(tty, &ch, sizeof(ch)); 196 return line_write(tty, &ch, sizeof(ch));
197} 197}
198 198
199int line_write(struct tty_struct *tty, const unsigned char *buf, int len) 199int line_write(struct tty_struct *tty, const unsigned char *buf, int len)
diff --git a/arch/um/include/line.h b/arch/um/include/line.h
index 1223f2c844b4..979b73e6352d 100644
--- a/arch/um/include/line.h
+++ b/arch/um/include/line.h
@@ -71,7 +71,7 @@ extern int line_setup(struct line *lines, unsigned int sizeof_lines,
71 char *init, char **error_out); 71 char *init, char **error_out);
72extern int line_write(struct tty_struct *tty, const unsigned char *buf, 72extern int line_write(struct tty_struct *tty, const unsigned char *buf,
73 int len); 73 int len);
74extern void line_put_char(struct tty_struct *tty, unsigned char ch); 74extern int line_put_char(struct tty_struct *tty, unsigned char ch);
75extern void line_set_termios(struct tty_struct *tty, struct ktermios * old); 75extern void line_set_termios(struct tty_struct *tty, struct ktermios * old);
76extern int line_chars_in_buffer(struct tty_struct *tty); 76extern int line_chars_in_buffer(struct tty_struct *tty);
77extern void line_flush_buffer(struct tty_struct *tty); 77extern void line_flush_buffer(struct tty_struct *tty);
diff --git a/arch/um/kernel/syscall.c b/arch/um/kernel/syscall.c
index 9cffc628a37e..128ee85bc8d9 100644
--- a/arch/um/kernel/syscall.c
+++ b/arch/um/kernel/syscall.c
@@ -73,23 +73,6 @@ long old_mmap(unsigned long addr, unsigned long len,
73 out: 73 out:
74 return err; 74 return err;
75} 75}
76/*
77 * sys_pipe() is the normal C calling standard for creating
78 * a pipe. It's not the way unix traditionally does this, though.
79 */
80long sys_pipe(unsigned long __user * fildes)
81{
82 int fd[2];
83 long error;
84
85 error = do_pipe(fd);
86 if (!error) {
87 if (copy_to_user(fildes, fd, sizeof(fd)))
88 error = -EFAULT;
89 }
90 return error;
91}
92
93 76
94long sys_uname(struct old_utsname __user * name) 77long sys_uname(struct old_utsname __user * name)
95{ 78{
diff --git a/arch/v850/kernel/syscalls.c b/arch/v850/kernel/syscalls.c
index 003db9c8c44a..1a83daf8e24f 100644
--- a/arch/v850/kernel/syscalls.c
+++ b/arch/v850/kernel/syscalls.c
@@ -132,23 +132,6 @@ sys_ipc (uint call, int first, int second, int third, void *ptr, long fifth)
132 return ret; 132 return ret;
133} 133}
134 134
135/*
136 * sys_pipe() is the normal C calling standard for creating
137 * a pipe. It's not the way unix traditionally does this, though.
138 */
139int sys_pipe (int *fildes)
140{
141 int fd[2];
142 int error;
143
144 error = do_pipe (fd);
145 if (!error) {
146 if (copy_to_user (fildes, fd, 2*sizeof (int)))
147 error = -EFAULT;
148 }
149 return error;
150}
151
152static inline unsigned long 135static inline unsigned long
153do_mmap2 (unsigned long addr, size_t len, 136do_mmap2 (unsigned long addr, size_t len,
154 unsigned long prot, unsigned long flags, 137 unsigned long prot, unsigned long flags,
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index c3f880902d66..bbcafaa160c0 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -18,6 +18,7 @@ config X86_64
18### Arch settings 18### Arch settings
19config X86 19config X86
20 def_bool y 20 def_bool y
21 select HAVE_UNSTABLE_SCHED_CLOCK
21 select HAVE_IDE 22 select HAVE_IDE
22 select HAVE_OPROFILE 23 select HAVE_OPROFILE
23 select HAVE_KPROBES 24 select HAVE_KPROBES
@@ -1661,6 +1662,7 @@ config GEODE_MFGPT_TIMER
1661 1662
1662config OLPC 1663config OLPC
1663 bool "One Laptop Per Child support" 1664 bool "One Laptop Per Child support"
1665 depends on MGEODE_LX
1664 default n 1666 default n
1665 help 1667 help
1666 Add support for detecting the unique features of the OLPC 1668 Add support for detecting the unique features of the OLPC
diff --git a/arch/x86/boot/compressed/relocs.c b/arch/x86/boot/compressed/relocs.c
index d01ea42187e6..edaadea90aaf 100644
--- a/arch/x86/boot/compressed/relocs.c
+++ b/arch/x86/boot/compressed/relocs.c
@@ -191,7 +191,7 @@ static void read_ehdr(FILE *fp)
191 die("Cannot read ELF header: %s\n", 191 die("Cannot read ELF header: %s\n",
192 strerror(errno)); 192 strerror(errno));
193 } 193 }
194 if (memcmp(ehdr.e_ident, ELFMAG, 4) != 0) { 194 if (memcmp(ehdr.e_ident, ELFMAG, SELFMAG) != 0) {
195 die("No ELF magic\n"); 195 die("No ELF magic\n");
196 } 196 }
197 if (ehdr.e_ident[EI_CLASS] != ELFCLASS32) { 197 if (ehdr.e_ident[EI_CLASS] != ELFCLASS32) {
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index bbdacb398d48..5e618c3b4720 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -83,9 +83,7 @@ obj-$(CONFIG_KVM_GUEST) += kvm.o
83obj-$(CONFIG_KVM_CLOCK) += kvmclock.o 83obj-$(CONFIG_KVM_CLOCK) += kvmclock.o
84obj-$(CONFIG_PARAVIRT) += paravirt.o paravirt_patch_$(BITS).o 84obj-$(CONFIG_PARAVIRT) += paravirt.o paravirt_patch_$(BITS).o
85 85
86ifdef CONFIG_INPUT_PCSPKR 86obj-$(CONFIG_PCSPKR_PLATFORM) += pcspeaker.o
87obj-y += pcspeaker.o
88endif
89 87
90obj-$(CONFIG_SCx200) += scx200.o 88obj-$(CONFIG_SCx200) += scx200.o
91scx200-y += scx200_32.o 89scx200-y += scx200_32.o
diff --git a/arch/x86/kernel/acpi/Makefile b/arch/x86/kernel/acpi/Makefile
index 7335959b6aff..fd5ca97a2ad5 100644
--- a/arch/x86/kernel/acpi/Makefile
+++ b/arch/x86/kernel/acpi/Makefile
@@ -10,5 +10,5 @@ endif
10$(obj)/wakeup_rm.o: $(obj)/realmode/wakeup.bin 10$(obj)/wakeup_rm.o: $(obj)/realmode/wakeup.bin
11 11
12$(obj)/realmode/wakeup.bin: FORCE 12$(obj)/realmode/wakeup.bin: FORCE
13 $(Q)$(MAKE) $(build)=$(obj)/realmode $@ 13 $(Q)$(MAKE) $(build)=$(obj)/realmode
14 14
diff --git a/arch/x86/kernel/acpi/realmode/Makefile b/arch/x86/kernel/acpi/realmode/Makefile
index 092900854acc..1c31cc0e9def 100644
--- a/arch/x86/kernel/acpi/realmode/Makefile
+++ b/arch/x86/kernel/acpi/realmode/Makefile
@@ -6,7 +6,8 @@
6# for more details. 6# for more details.
7# 7#
8 8
9targets := wakeup.bin wakeup.elf 9always := wakeup.bin
10targets := wakeup.elf wakeup.lds
10 11
11wakeup-y += wakeup.o wakemain.o video-mode.o copy.o 12wakeup-y += wakeup.o wakemain.o video-mode.o copy.o
12 13
@@ -48,7 +49,7 @@ LDFLAGS_wakeup.elf := -T
48 49
49CPPFLAGS_wakeup.lds += -P -C 50CPPFLAGS_wakeup.lds += -P -C
50 51
51$(obj)/wakeup.elf: $(src)/wakeup.lds $(WAKEUP_OBJS) FORCE 52$(obj)/wakeup.elf: $(obj)/wakeup.lds $(WAKEUP_OBJS) FORCE
52 $(call if_changed,ld) 53 $(call if_changed,ld)
53 54
54OBJCOPYFLAGS_wakeup.bin := -O binary 55OBJCOPYFLAGS_wakeup.bin := -O binary
diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c
index ddee04043aeb..4bc1be5d5472 100644
--- a/arch/x86/kernel/kvmclock.c
+++ b/arch/x86/kernel/kvmclock.c
@@ -133,6 +133,7 @@ static int kvm_register_clock(void)
133 return native_write_msr_safe(MSR_KVM_SYSTEM_TIME, low, high); 133 return native_write_msr_safe(MSR_KVM_SYSTEM_TIME, low, high);
134} 134}
135 135
136#ifdef CONFIG_X86_LOCAL_APIC
136static void kvm_setup_secondary_clock(void) 137static void kvm_setup_secondary_clock(void)
137{ 138{
138 /* 139 /*
@@ -143,6 +144,7 @@ static void kvm_setup_secondary_clock(void)
143 /* ok, done with our trickery, call native */ 144 /* ok, done with our trickery, call native */
144 setup_secondary_APIC_clock(); 145 setup_secondary_APIC_clock();
145} 146}
147#endif
146 148
147/* 149/*
148 * After the clock is registered, the host will keep writing to the 150 * After the clock is registered, the host will keep writing to the
@@ -177,7 +179,9 @@ void __init kvmclock_init(void)
177 pv_time_ops.get_wallclock = kvm_get_wallclock; 179 pv_time_ops.get_wallclock = kvm_get_wallclock;
178 pv_time_ops.set_wallclock = kvm_set_wallclock; 180 pv_time_ops.set_wallclock = kvm_set_wallclock;
179 pv_time_ops.sched_clock = kvm_clock_read; 181 pv_time_ops.sched_clock = kvm_clock_read;
182#ifdef CONFIG_X86_LOCAL_APIC
180 pv_apic_ops.setup_secondary_clock = kvm_setup_secondary_clock; 183 pv_apic_ops.setup_secondary_clock = kvm_setup_secondary_clock;
184#endif
181 machine_ops.shutdown = kvm_shutdown; 185 machine_ops.shutdown = kvm_shutdown;
182#ifdef CONFIG_KEXEC 186#ifdef CONFIG_KEXEC
183 machine_ops.crash_shutdown = kvm_crash_shutdown; 187 machine_ops.crash_shutdown = kvm_crash_shutdown;
diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index 3e2c54dc8b29..404683b94e79 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -794,6 +794,11 @@ void __init find_smp_config(void)
794 ACPI-based MP Configuration 794 ACPI-based MP Configuration
795 -------------------------------------------------------------------------- */ 795 -------------------------------------------------------------------------- */
796 796
797/*
798 * Keep this outside and initialized to 0, for !CONFIG_ACPI builds:
799 */
800int es7000_plat;
801
797#ifdef CONFIG_ACPI 802#ifdef CONFIG_ACPI
798 803
799#ifdef CONFIG_X86_IO_APIC 804#ifdef CONFIG_X86_IO_APIC
@@ -909,8 +914,6 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi)
909 MP_intsrc_info(&intsrc); 914 MP_intsrc_info(&intsrc);
910} 915}
911 916
912int es7000_plat;
913
914void __init mp_config_acpi_legacy_irqs(void) 917void __init mp_config_acpi_legacy_irqs(void)
915{ 918{
916 struct mpc_config_intsrc intsrc; 919 struct mpc_config_intsrc intsrc;
diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c
index 07c6d42ab5ff..f6be7d5f82f8 100644
--- a/arch/x86/kernel/reboot.c
+++ b/arch/x86/kernel/reboot.c
@@ -149,7 +149,6 @@ static struct dmi_system_id __initdata reboot_dmi_table[] = {
149 .matches = { 149 .matches = {
150 DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), 150 DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
151 DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex 745"), 151 DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex 745"),
152 DMI_MATCH(DMI_BOARD_NAME, "0WF810"),
153 }, 152 },
154 }, 153 },
155 { /* Handle problems with rebooting on Dell Optiplex 745's DFF*/ 154 { /* Handle problems with rebooting on Dell Optiplex 745's DFF*/
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index c0c68c18a788..cc6f5eb20b24 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -95,7 +95,7 @@ void __init setup_per_cpu_areas(void)
95 95
96 /* Copy section for each CPU (we discard the original) */ 96 /* Copy section for each CPU (we discard the original) */
97 size = PERCPU_ENOUGH_ROOM; 97 size = PERCPU_ENOUGH_ROOM;
98 printk(KERN_INFO "PERCPU: Allocating %lu bytes of per cpu data\n", 98 printk(KERN_INFO "PERCPU: Allocating %zd bytes of per cpu data\n",
99 size); 99 size);
100 100
101 for_each_possible_cpu(i) { 101 for_each_possible_cpu(i) {
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 84241a256dc8..6b087ab6cd8f 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -299,7 +299,7 @@ static void __cpuinit smp_callin(void)
299/* 299/*
300 * Activate a secondary processor. 300 * Activate a secondary processor.
301 */ 301 */
302void __cpuinit start_secondary(void *unused) 302static void __cpuinit start_secondary(void *unused)
303{ 303{
304 /* 304 /*
305 * Don't put *anything* before cpu_init(), SMP booting is too 305 * Don't put *anything* before cpu_init(), SMP booting is too
@@ -1306,7 +1306,7 @@ static void remove_siblinginfo(int cpu)
1306 cpu_clear(cpu, cpu_sibling_setup_map); 1306 cpu_clear(cpu, cpu_sibling_setup_map);
1307} 1307}
1308 1308
1309int additional_cpus __initdata = -1; 1309static int additional_cpus __initdata = -1;
1310 1310
1311static __init int setup_additional_cpus(char *s) 1311static __init int setup_additional_cpus(char *s)
1312{ 1312{
diff --git a/arch/x86/kernel/sys_i386_32.c b/arch/x86/kernel/sys_i386_32.c
index a86d26f036e1..d2ab52cc1d6b 100644
--- a/arch/x86/kernel/sys_i386_32.c
+++ b/arch/x86/kernel/sys_i386_32.c
@@ -22,23 +22,6 @@
22#include <asm/uaccess.h> 22#include <asm/uaccess.h>
23#include <asm/unistd.h> 23#include <asm/unistd.h>
24 24
25/*
26 * sys_pipe() is the normal C calling standard for creating
27 * a pipe. It's not the way Unix traditionally does this, though.
28 */
29asmlinkage int sys_pipe(unsigned long __user * fildes)
30{
31 int fd[2];
32 int error;
33
34 error = do_pipe(fd);
35 if (!error) {
36 if (copy_to_user(fildes, fd, 2*sizeof(int)))
37 error = -EFAULT;
38 }
39 return error;
40}
41
42asmlinkage long sys_mmap2(unsigned long addr, unsigned long len, 25asmlinkage long sys_mmap2(unsigned long addr, unsigned long len,
43 unsigned long prot, unsigned long flags, 26 unsigned long prot, unsigned long flags,
44 unsigned long fd, unsigned long pgoff) 27 unsigned long fd, unsigned long pgoff)
diff --git a/arch/x86/kernel/sys_x86_64.c b/arch/x86/kernel/sys_x86_64.c
index bd802a5e1aa3..3b360ef33817 100644
--- a/arch/x86/kernel/sys_x86_64.c
+++ b/arch/x86/kernel/sys_x86_64.c
@@ -17,23 +17,6 @@
17#include <asm/uaccess.h> 17#include <asm/uaccess.h>
18#include <asm/ia32.h> 18#include <asm/ia32.h>
19 19
20/*
21 * sys_pipe() is the normal C calling standard for creating
22 * a pipe. It's not the way Unix traditionally does this, though.
23 */
24asmlinkage long sys_pipe(int __user *fildes)
25{
26 int fd[2];
27 int error;
28
29 error = do_pipe(fd);
30 if (!error) {
31 if (copy_to_user(fildes, fd, 2*sizeof(int)))
32 error = -EFAULT;
33 }
34 return error;
35}
36
37asmlinkage long sys_mmap(unsigned long addr, unsigned long len, unsigned long prot, unsigned long flags, 20asmlinkage long sys_mmap(unsigned long addr, unsigned long len, unsigned long prot, unsigned long flags,
38 unsigned long fd, unsigned long off) 21 unsigned long fd, unsigned long off)
39{ 22{
diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index 4c943eabacc3..3324d90038e4 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -288,6 +288,8 @@ static void pit_load_count(struct kvm *kvm, int channel, u32 val)
288 * mode 1 is one shot, mode 2 is period, otherwise del timer */ 288 * mode 1 is one shot, mode 2 is period, otherwise del timer */
289 switch (ps->channels[0].mode) { 289 switch (ps->channels[0].mode) {
290 case 1: 290 case 1:
291 /* FIXME: enhance mode 4 precision */
292 case 4:
291 create_pit_timer(&ps->pit_timer, val, 0); 293 create_pit_timer(&ps->pit_timer, val, 0);
292 break; 294 break;
293 case 2: 295 case 2:
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 2ad6f5481671..36c5406b1813 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -79,36 +79,6 @@ static int dbg = 1;
79 } 79 }
80#endif 80#endif
81 81
82#define PT64_PT_BITS 9
83#define PT64_ENT_PER_PAGE (1 << PT64_PT_BITS)
84#define PT32_PT_BITS 10
85#define PT32_ENT_PER_PAGE (1 << PT32_PT_BITS)
86
87#define PT_WRITABLE_SHIFT 1
88
89#define PT_PRESENT_MASK (1ULL << 0)
90#define PT_WRITABLE_MASK (1ULL << PT_WRITABLE_SHIFT)
91#define PT_USER_MASK (1ULL << 2)
92#define PT_PWT_MASK (1ULL << 3)
93#define PT_PCD_MASK (1ULL << 4)
94#define PT_ACCESSED_MASK (1ULL << 5)
95#define PT_DIRTY_MASK (1ULL << 6)
96#define PT_PAGE_SIZE_MASK (1ULL << 7)
97#define PT_PAT_MASK (1ULL << 7)
98#define PT_GLOBAL_MASK (1ULL << 8)
99#define PT64_NX_SHIFT 63
100#define PT64_NX_MASK (1ULL << PT64_NX_SHIFT)
101
102#define PT_PAT_SHIFT 7
103#define PT_DIR_PAT_SHIFT 12
104#define PT_DIR_PAT_MASK (1ULL << PT_DIR_PAT_SHIFT)
105
106#define PT32_DIR_PSE36_SIZE 4
107#define PT32_DIR_PSE36_SHIFT 13
108#define PT32_DIR_PSE36_MASK \
109 (((1ULL << PT32_DIR_PSE36_SIZE) - 1) << PT32_DIR_PSE36_SHIFT)
110
111
112#define PT_FIRST_AVAIL_BITS_SHIFT 9 82#define PT_FIRST_AVAIL_BITS_SHIFT 9
113#define PT64_SECOND_AVAIL_BITS_SHIFT 52 83#define PT64_SECOND_AVAIL_BITS_SHIFT 52
114 84
@@ -154,10 +124,6 @@ static int dbg = 1;
154#define PFERR_USER_MASK (1U << 2) 124#define PFERR_USER_MASK (1U << 2)
155#define PFERR_FETCH_MASK (1U << 4) 125#define PFERR_FETCH_MASK (1U << 4)
156 126
157#define PT64_ROOT_LEVEL 4
158#define PT32_ROOT_LEVEL 2
159#define PT32E_ROOT_LEVEL 3
160
161#define PT_DIRECTORY_LEVEL 2 127#define PT_DIRECTORY_LEVEL 2
162#define PT_PAGE_TABLE_LEVEL 1 128#define PT_PAGE_TABLE_LEVEL 1
163 129
@@ -186,6 +152,12 @@ static struct kmem_cache *mmu_page_header_cache;
186 152
187static u64 __read_mostly shadow_trap_nonpresent_pte; 153static u64 __read_mostly shadow_trap_nonpresent_pte;
188static u64 __read_mostly shadow_notrap_nonpresent_pte; 154static u64 __read_mostly shadow_notrap_nonpresent_pte;
155static u64 __read_mostly shadow_base_present_pte;
156static u64 __read_mostly shadow_nx_mask;
157static u64 __read_mostly shadow_x_mask; /* mutual exclusive with nx_mask */
158static u64 __read_mostly shadow_user_mask;
159static u64 __read_mostly shadow_accessed_mask;
160static u64 __read_mostly shadow_dirty_mask;
189 161
190void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte) 162void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte)
191{ 163{
@@ -194,6 +166,23 @@ void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte)
194} 166}
195EXPORT_SYMBOL_GPL(kvm_mmu_set_nonpresent_ptes); 167EXPORT_SYMBOL_GPL(kvm_mmu_set_nonpresent_ptes);
196 168
169void kvm_mmu_set_base_ptes(u64 base_pte)
170{
171 shadow_base_present_pte = base_pte;
172}
173EXPORT_SYMBOL_GPL(kvm_mmu_set_base_ptes);
174
175void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
176 u64 dirty_mask, u64 nx_mask, u64 x_mask)
177{
178 shadow_user_mask = user_mask;
179 shadow_accessed_mask = accessed_mask;
180 shadow_dirty_mask = dirty_mask;
181 shadow_nx_mask = nx_mask;
182 shadow_x_mask = x_mask;
183}
184EXPORT_SYMBOL_GPL(kvm_mmu_set_mask_ptes);
185
197static int is_write_protection(struct kvm_vcpu *vcpu) 186static int is_write_protection(struct kvm_vcpu *vcpu)
198{ 187{
199 return vcpu->arch.cr0 & X86_CR0_WP; 188 return vcpu->arch.cr0 & X86_CR0_WP;
@@ -232,7 +221,7 @@ static int is_writeble_pte(unsigned long pte)
232 221
233static int is_dirty_pte(unsigned long pte) 222static int is_dirty_pte(unsigned long pte)
234{ 223{
235 return pte & PT_DIRTY_MASK; 224 return pte & shadow_dirty_mask;
236} 225}
237 226
238static int is_rmap_pte(u64 pte) 227static int is_rmap_pte(u64 pte)
@@ -387,7 +376,6 @@ static void account_shadowed(struct kvm *kvm, gfn_t gfn)
387 376
388 write_count = slot_largepage_idx(gfn, gfn_to_memslot(kvm, gfn)); 377 write_count = slot_largepage_idx(gfn, gfn_to_memslot(kvm, gfn));
389 *write_count += 1; 378 *write_count += 1;
390 WARN_ON(*write_count > KVM_PAGES_PER_HPAGE);
391} 379}
392 380
393static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) 381static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn)
@@ -547,7 +535,7 @@ static void rmap_remove(struct kvm *kvm, u64 *spte)
547 return; 535 return;
548 sp = page_header(__pa(spte)); 536 sp = page_header(__pa(spte));
549 pfn = spte_to_pfn(*spte); 537 pfn = spte_to_pfn(*spte);
550 if (*spte & PT_ACCESSED_MASK) 538 if (*spte & shadow_accessed_mask)
551 kvm_set_pfn_accessed(pfn); 539 kvm_set_pfn_accessed(pfn);
552 if (is_writeble_pte(*spte)) 540 if (is_writeble_pte(*spte))
553 kvm_release_pfn_dirty(pfn); 541 kvm_release_pfn_dirty(pfn);
@@ -1073,17 +1061,17 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte,
1073 * whether the guest actually used the pte (in order to detect 1061 * whether the guest actually used the pte (in order to detect
1074 * demand paging). 1062 * demand paging).
1075 */ 1063 */
1076 spte = PT_PRESENT_MASK | PT_DIRTY_MASK; 1064 spte = shadow_base_present_pte | shadow_dirty_mask;
1077 if (!speculative) 1065 if (!speculative)
1078 pte_access |= PT_ACCESSED_MASK; 1066 pte_access |= PT_ACCESSED_MASK;
1079 if (!dirty) 1067 if (!dirty)
1080 pte_access &= ~ACC_WRITE_MASK; 1068 pte_access &= ~ACC_WRITE_MASK;
1081 if (!(pte_access & ACC_EXEC_MASK)) 1069 if (pte_access & ACC_EXEC_MASK)
1082 spte |= PT64_NX_MASK; 1070 spte |= shadow_x_mask;
1083 1071 else
1084 spte |= PT_PRESENT_MASK; 1072 spte |= shadow_nx_mask;
1085 if (pte_access & ACC_USER_MASK) 1073 if (pte_access & ACC_USER_MASK)
1086 spte |= PT_USER_MASK; 1074 spte |= shadow_user_mask;
1087 if (largepage) 1075 if (largepage)
1088 spte |= PT_PAGE_SIZE_MASK; 1076 spte |= PT_PAGE_SIZE_MASK;
1089 1077
@@ -1188,8 +1176,9 @@ static int __direct_map(struct kvm_vcpu *vcpu, gpa_t v, int write,
1188 return -ENOMEM; 1176 return -ENOMEM;
1189 } 1177 }
1190 1178
1191 table[index] = __pa(new_table->spt) | PT_PRESENT_MASK 1179 table[index] = __pa(new_table->spt)
1192 | PT_WRITABLE_MASK | PT_USER_MASK; 1180 | PT_PRESENT_MASK | PT_WRITABLE_MASK
1181 | shadow_user_mask | shadow_x_mask;
1193 } 1182 }
1194 table_addr = table[index] & PT64_BASE_ADDR_MASK; 1183 table_addr = table[index] & PT64_BASE_ADDR_MASK;
1195 } 1184 }
@@ -1244,7 +1233,6 @@ static void mmu_free_roots(struct kvm_vcpu *vcpu)
1244 if (!VALID_PAGE(vcpu->arch.mmu.root_hpa)) 1233 if (!VALID_PAGE(vcpu->arch.mmu.root_hpa))
1245 return; 1234 return;
1246 spin_lock(&vcpu->kvm->mmu_lock); 1235 spin_lock(&vcpu->kvm->mmu_lock);
1247#ifdef CONFIG_X86_64
1248 if (vcpu->arch.mmu.shadow_root_level == PT64_ROOT_LEVEL) { 1236 if (vcpu->arch.mmu.shadow_root_level == PT64_ROOT_LEVEL) {
1249 hpa_t root = vcpu->arch.mmu.root_hpa; 1237 hpa_t root = vcpu->arch.mmu.root_hpa;
1250 1238
@@ -1256,7 +1244,6 @@ static void mmu_free_roots(struct kvm_vcpu *vcpu)
1256 spin_unlock(&vcpu->kvm->mmu_lock); 1244 spin_unlock(&vcpu->kvm->mmu_lock);
1257 return; 1245 return;
1258 } 1246 }
1259#endif
1260 for (i = 0; i < 4; ++i) { 1247 for (i = 0; i < 4; ++i) {
1261 hpa_t root = vcpu->arch.mmu.pae_root[i]; 1248 hpa_t root = vcpu->arch.mmu.pae_root[i];
1262 1249
@@ -1282,7 +1269,6 @@ static void mmu_alloc_roots(struct kvm_vcpu *vcpu)
1282 1269
1283 root_gfn = vcpu->arch.cr3 >> PAGE_SHIFT; 1270 root_gfn = vcpu->arch.cr3 >> PAGE_SHIFT;
1284 1271
1285#ifdef CONFIG_X86_64
1286 if (vcpu->arch.mmu.shadow_root_level == PT64_ROOT_LEVEL) { 1272 if (vcpu->arch.mmu.shadow_root_level == PT64_ROOT_LEVEL) {
1287 hpa_t root = vcpu->arch.mmu.root_hpa; 1273 hpa_t root = vcpu->arch.mmu.root_hpa;
1288 1274
@@ -1297,7 +1283,6 @@ static void mmu_alloc_roots(struct kvm_vcpu *vcpu)
1297 vcpu->arch.mmu.root_hpa = root; 1283 vcpu->arch.mmu.root_hpa = root;
1298 return; 1284 return;
1299 } 1285 }
1300#endif
1301 metaphysical = !is_paging(vcpu); 1286 metaphysical = !is_paging(vcpu);
1302 if (tdp_enabled) 1287 if (tdp_enabled)
1303 metaphysical = 1; 1288 metaphysical = 1;
@@ -1377,7 +1362,7 @@ static int tdp_page_fault(struct kvm_vcpu *vcpu, gva_t gpa,
1377 spin_lock(&vcpu->kvm->mmu_lock); 1362 spin_lock(&vcpu->kvm->mmu_lock);
1378 kvm_mmu_free_some_pages(vcpu); 1363 kvm_mmu_free_some_pages(vcpu);
1379 r = __direct_map(vcpu, gpa, error_code & PFERR_WRITE_MASK, 1364 r = __direct_map(vcpu, gpa, error_code & PFERR_WRITE_MASK,
1380 largepage, gfn, pfn, TDP_ROOT_LEVEL); 1365 largepage, gfn, pfn, kvm_x86_ops->get_tdp_level());
1381 spin_unlock(&vcpu->kvm->mmu_lock); 1366 spin_unlock(&vcpu->kvm->mmu_lock);
1382 1367
1383 return r; 1368 return r;
@@ -1484,7 +1469,7 @@ static int init_kvm_tdp_mmu(struct kvm_vcpu *vcpu)
1484 context->page_fault = tdp_page_fault; 1469 context->page_fault = tdp_page_fault;
1485 context->free = nonpaging_free; 1470 context->free = nonpaging_free;
1486 context->prefetch_page = nonpaging_prefetch_page; 1471 context->prefetch_page = nonpaging_prefetch_page;
1487 context->shadow_root_level = TDP_ROOT_LEVEL; 1472 context->shadow_root_level = kvm_x86_ops->get_tdp_level();
1488 context->root_hpa = INVALID_PAGE; 1473 context->root_hpa = INVALID_PAGE;
1489 1474
1490 if (!is_paging(vcpu)) { 1475 if (!is_paging(vcpu)) {
@@ -1633,7 +1618,7 @@ static bool last_updated_pte_accessed(struct kvm_vcpu *vcpu)
1633{ 1618{
1634 u64 *spte = vcpu->arch.last_pte_updated; 1619 u64 *spte = vcpu->arch.last_pte_updated;
1635 1620
1636 return !!(spte && (*spte & PT_ACCESSED_MASK)); 1621 return !!(spte && (*spte & shadow_accessed_mask));
1637} 1622}
1638 1623
1639static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, 1624static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
index e64e9f56a65e..1730757bbc7a 100644
--- a/arch/x86/kvm/mmu.h
+++ b/arch/x86/kvm/mmu.h
@@ -3,11 +3,38 @@
3 3
4#include <linux/kvm_host.h> 4#include <linux/kvm_host.h>
5 5
6#ifdef CONFIG_X86_64 6#define PT64_PT_BITS 9
7#define TDP_ROOT_LEVEL PT64_ROOT_LEVEL 7#define PT64_ENT_PER_PAGE (1 << PT64_PT_BITS)
8#else 8#define PT32_PT_BITS 10
9#define TDP_ROOT_LEVEL PT32E_ROOT_LEVEL 9#define PT32_ENT_PER_PAGE (1 << PT32_PT_BITS)
10#endif 10
11#define PT_WRITABLE_SHIFT 1
12
13#define PT_PRESENT_MASK (1ULL << 0)
14#define PT_WRITABLE_MASK (1ULL << PT_WRITABLE_SHIFT)
15#define PT_USER_MASK (1ULL << 2)
16#define PT_PWT_MASK (1ULL << 3)
17#define PT_PCD_MASK (1ULL << 4)
18#define PT_ACCESSED_MASK (1ULL << 5)
19#define PT_DIRTY_MASK (1ULL << 6)
20#define PT_PAGE_SIZE_MASK (1ULL << 7)
21#define PT_PAT_MASK (1ULL << 7)
22#define PT_GLOBAL_MASK (1ULL << 8)
23#define PT64_NX_SHIFT 63
24#define PT64_NX_MASK (1ULL << PT64_NX_SHIFT)
25
26#define PT_PAT_SHIFT 7
27#define PT_DIR_PAT_SHIFT 12
28#define PT_DIR_PAT_MASK (1ULL << PT_DIR_PAT_SHIFT)
29
30#define PT32_DIR_PSE36_SIZE 4
31#define PT32_DIR_PSE36_SHIFT 13
32#define PT32_DIR_PSE36_MASK \
33 (((1ULL << PT32_DIR_PSE36_SIZE) - 1) << PT32_DIR_PSE36_SHIFT)
34
35#define PT64_ROOT_LEVEL 4
36#define PT32_ROOT_LEVEL 2
37#define PT32E_ROOT_LEVEL 3
11 38
12static inline void kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu) 39static inline void kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu)
13{ 40{
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 89e0be2c10d0..ab22615eee89 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -1863,6 +1863,15 @@ static bool svm_cpu_has_accelerated_tpr(void)
1863 return false; 1863 return false;
1864} 1864}
1865 1865
1866static int get_npt_level(void)
1867{
1868#ifdef CONFIG_X86_64
1869 return PT64_ROOT_LEVEL;
1870#else
1871 return PT32E_ROOT_LEVEL;
1872#endif
1873}
1874
1866static struct kvm_x86_ops svm_x86_ops = { 1875static struct kvm_x86_ops svm_x86_ops = {
1867 .cpu_has_kvm_support = has_svm, 1876 .cpu_has_kvm_support = has_svm,
1868 .disabled_by_bios = is_disabled, 1877 .disabled_by_bios = is_disabled,
@@ -1920,6 +1929,7 @@ static struct kvm_x86_ops svm_x86_ops = {
1920 .inject_pending_vectors = do_interrupt_requests, 1929 .inject_pending_vectors = do_interrupt_requests,
1921 1930
1922 .set_tss_addr = svm_set_tss_addr, 1931 .set_tss_addr = svm_set_tss_addr,
1932 .get_tdp_level = get_npt_level,
1923}; 1933};
1924 1934
1925static int __init svm_init(void) 1935static int __init svm_init(void)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 8e5d6645b90d..bfe4db11989c 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -42,6 +42,9 @@ module_param(enable_vpid, bool, 0);
42static int flexpriority_enabled = 1; 42static int flexpriority_enabled = 1;
43module_param(flexpriority_enabled, bool, 0); 43module_param(flexpriority_enabled, bool, 0);
44 44
45static int enable_ept = 1;
46module_param(enable_ept, bool, 0);
47
45struct vmcs { 48struct vmcs {
46 u32 revision_id; 49 u32 revision_id;
47 u32 abort; 50 u32 abort;
@@ -84,7 +87,7 @@ static inline struct vcpu_vmx *to_vmx(struct kvm_vcpu *vcpu)
84 return container_of(vcpu, struct vcpu_vmx, vcpu); 87 return container_of(vcpu, struct vcpu_vmx, vcpu);
85} 88}
86 89
87static int init_rmode_tss(struct kvm *kvm); 90static int init_rmode(struct kvm *kvm);
88 91
89static DEFINE_PER_CPU(struct vmcs *, vmxarea); 92static DEFINE_PER_CPU(struct vmcs *, vmxarea);
90static DEFINE_PER_CPU(struct vmcs *, current_vmcs); 93static DEFINE_PER_CPU(struct vmcs *, current_vmcs);
@@ -107,6 +110,11 @@ static struct vmcs_config {
107 u32 vmentry_ctrl; 110 u32 vmentry_ctrl;
108} vmcs_config; 111} vmcs_config;
109 112
113struct vmx_capability {
114 u32 ept;
115 u32 vpid;
116} vmx_capability;
117
110#define VMX_SEGMENT_FIELD(seg) \ 118#define VMX_SEGMENT_FIELD(seg) \
111 [VCPU_SREG_##seg] = { \ 119 [VCPU_SREG_##seg] = { \
112 .selector = GUEST_##seg##_SELECTOR, \ 120 .selector = GUEST_##seg##_SELECTOR, \
@@ -214,6 +222,32 @@ static inline bool cpu_has_vmx_virtualize_apic_accesses(void)
214 SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES); 222 SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES);
215} 223}
216 224
225static inline int cpu_has_vmx_invept_individual_addr(void)
226{
227 return (!!(vmx_capability.ept & VMX_EPT_EXTENT_INDIVIDUAL_BIT));
228}
229
230static inline int cpu_has_vmx_invept_context(void)
231{
232 return (!!(vmx_capability.ept & VMX_EPT_EXTENT_CONTEXT_BIT));
233}
234
235static inline int cpu_has_vmx_invept_global(void)
236{
237 return (!!(vmx_capability.ept & VMX_EPT_EXTENT_GLOBAL_BIT));
238}
239
240static inline int cpu_has_vmx_ept(void)
241{
242 return (vmcs_config.cpu_based_2nd_exec_ctrl &
243 SECONDARY_EXEC_ENABLE_EPT);
244}
245
246static inline int vm_need_ept(void)
247{
248 return (cpu_has_vmx_ept() && enable_ept);
249}
250
217static inline int vm_need_virtualize_apic_accesses(struct kvm *kvm) 251static inline int vm_need_virtualize_apic_accesses(struct kvm *kvm)
218{ 252{
219 return ((cpu_has_vmx_virtualize_apic_accesses()) && 253 return ((cpu_has_vmx_virtualize_apic_accesses()) &&
@@ -250,6 +284,18 @@ static inline void __invvpid(int ext, u16 vpid, gva_t gva)
250 : : "a"(&operand), "c"(ext) : "cc", "memory"); 284 : : "a"(&operand), "c"(ext) : "cc", "memory");
251} 285}
252 286
287static inline void __invept(int ext, u64 eptp, gpa_t gpa)
288{
289 struct {
290 u64 eptp, gpa;
291 } operand = {eptp, gpa};
292
293 asm volatile (ASM_VMX_INVEPT
294 /* CF==1 or ZF==1 --> rc = -1 */
295 "; ja 1f ; ud2 ; 1:\n"
296 : : "a" (&operand), "c" (ext) : "cc", "memory");
297}
298
253static struct kvm_msr_entry *find_msr_entry(struct vcpu_vmx *vmx, u32 msr) 299static struct kvm_msr_entry *find_msr_entry(struct vcpu_vmx *vmx, u32 msr)
254{ 300{
255 int i; 301 int i;
@@ -301,6 +347,33 @@ static inline void vpid_sync_vcpu_all(struct vcpu_vmx *vmx)
301 __invvpid(VMX_VPID_EXTENT_SINGLE_CONTEXT, vmx->vpid, 0); 347 __invvpid(VMX_VPID_EXTENT_SINGLE_CONTEXT, vmx->vpid, 0);
302} 348}
303 349
350static inline void ept_sync_global(void)
351{
352 if (cpu_has_vmx_invept_global())
353 __invept(VMX_EPT_EXTENT_GLOBAL, 0, 0);
354}
355
356static inline void ept_sync_context(u64 eptp)
357{
358 if (vm_need_ept()) {
359 if (cpu_has_vmx_invept_context())
360 __invept(VMX_EPT_EXTENT_CONTEXT, eptp, 0);
361 else
362 ept_sync_global();
363 }
364}
365
366static inline void ept_sync_individual_addr(u64 eptp, gpa_t gpa)
367{
368 if (vm_need_ept()) {
369 if (cpu_has_vmx_invept_individual_addr())
370 __invept(VMX_EPT_EXTENT_INDIVIDUAL_ADDR,
371 eptp, gpa);
372 else
373 ept_sync_context(eptp);
374 }
375}
376
304static unsigned long vmcs_readl(unsigned long field) 377static unsigned long vmcs_readl(unsigned long field)
305{ 378{
306 unsigned long value; 379 unsigned long value;
@@ -388,6 +461,8 @@ static void update_exception_bitmap(struct kvm_vcpu *vcpu)
388 eb |= 1u << 1; 461 eb |= 1u << 1;
389 if (vcpu->arch.rmode.active) 462 if (vcpu->arch.rmode.active)
390 eb = ~0; 463 eb = ~0;
464 if (vm_need_ept())
465 eb &= ~(1u << PF_VECTOR); /* bypass_guest_pf = 0 */
391 vmcs_write32(EXCEPTION_BITMAP, eb); 466 vmcs_write32(EXCEPTION_BITMAP, eb);
392} 467}
393 468
@@ -985,7 +1060,7 @@ static __init int adjust_vmx_controls(u32 ctl_min, u32 ctl_opt,
985static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf) 1060static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
986{ 1061{
987 u32 vmx_msr_low, vmx_msr_high; 1062 u32 vmx_msr_low, vmx_msr_high;
988 u32 min, opt; 1063 u32 min, opt, min2, opt2;
989 u32 _pin_based_exec_control = 0; 1064 u32 _pin_based_exec_control = 0;
990 u32 _cpu_based_exec_control = 0; 1065 u32 _cpu_based_exec_control = 0;
991 u32 _cpu_based_2nd_exec_control = 0; 1066 u32 _cpu_based_2nd_exec_control = 0;
@@ -1003,6 +1078,8 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
1003 CPU_BASED_CR8_LOAD_EXITING | 1078 CPU_BASED_CR8_LOAD_EXITING |
1004 CPU_BASED_CR8_STORE_EXITING | 1079 CPU_BASED_CR8_STORE_EXITING |
1005#endif 1080#endif
1081 CPU_BASED_CR3_LOAD_EXITING |
1082 CPU_BASED_CR3_STORE_EXITING |
1006 CPU_BASED_USE_IO_BITMAPS | 1083 CPU_BASED_USE_IO_BITMAPS |
1007 CPU_BASED_MOV_DR_EXITING | 1084 CPU_BASED_MOV_DR_EXITING |
1008 CPU_BASED_USE_TSC_OFFSETING; 1085 CPU_BASED_USE_TSC_OFFSETING;
@@ -1018,11 +1095,13 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
1018 ~CPU_BASED_CR8_STORE_EXITING; 1095 ~CPU_BASED_CR8_STORE_EXITING;
1019#endif 1096#endif
1020 if (_cpu_based_exec_control & CPU_BASED_ACTIVATE_SECONDARY_CONTROLS) { 1097 if (_cpu_based_exec_control & CPU_BASED_ACTIVATE_SECONDARY_CONTROLS) {
1021 min = 0; 1098 min2 = 0;
1022 opt = SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES | 1099 opt2 = SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES |
1023 SECONDARY_EXEC_WBINVD_EXITING | 1100 SECONDARY_EXEC_WBINVD_EXITING |
1024 SECONDARY_EXEC_ENABLE_VPID; 1101 SECONDARY_EXEC_ENABLE_VPID |
1025 if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PROCBASED_CTLS2, 1102 SECONDARY_EXEC_ENABLE_EPT;
1103 if (adjust_vmx_controls(min2, opt2,
1104 MSR_IA32_VMX_PROCBASED_CTLS2,
1026 &_cpu_based_2nd_exec_control) < 0) 1105 &_cpu_based_2nd_exec_control) < 0)
1027 return -EIO; 1106 return -EIO;
1028 } 1107 }
@@ -1031,6 +1110,16 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
1031 SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES)) 1110 SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES))
1032 _cpu_based_exec_control &= ~CPU_BASED_TPR_SHADOW; 1111 _cpu_based_exec_control &= ~CPU_BASED_TPR_SHADOW;
1033#endif 1112#endif
1113 if (_cpu_based_2nd_exec_control & SECONDARY_EXEC_ENABLE_EPT) {
1114 /* CR3 accesses don't need to cause VM Exits when EPT enabled */
1115 min &= ~(CPU_BASED_CR3_LOAD_EXITING |
1116 CPU_BASED_CR3_STORE_EXITING);
1117 if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PROCBASED_CTLS,
1118 &_cpu_based_exec_control) < 0)
1119 return -EIO;
1120 rdmsr(MSR_IA32_VMX_EPT_VPID_CAP,
1121 vmx_capability.ept, vmx_capability.vpid);
1122 }
1034 1123
1035 min = 0; 1124 min = 0;
1036#ifdef CONFIG_X86_64 1125#ifdef CONFIG_X86_64
@@ -1256,7 +1345,7 @@ static void enter_rmode(struct kvm_vcpu *vcpu)
1256 fix_rmode_seg(VCPU_SREG_FS, &vcpu->arch.rmode.fs); 1345 fix_rmode_seg(VCPU_SREG_FS, &vcpu->arch.rmode.fs);
1257 1346
1258 kvm_mmu_reset_context(vcpu); 1347 kvm_mmu_reset_context(vcpu);
1259 init_rmode_tss(vcpu->kvm); 1348 init_rmode(vcpu->kvm);
1260} 1349}
1261 1350
1262#ifdef CONFIG_X86_64 1351#ifdef CONFIG_X86_64
@@ -1304,8 +1393,64 @@ static void vmx_decache_cr4_guest_bits(struct kvm_vcpu *vcpu)
1304 vcpu->arch.cr4 |= vmcs_readl(GUEST_CR4) & ~KVM_GUEST_CR4_MASK; 1393 vcpu->arch.cr4 |= vmcs_readl(GUEST_CR4) & ~KVM_GUEST_CR4_MASK;
1305} 1394}
1306 1395
1396static void ept_load_pdptrs(struct kvm_vcpu *vcpu)
1397{
1398 if (is_paging(vcpu) && is_pae(vcpu) && !is_long_mode(vcpu)) {
1399 if (!load_pdptrs(vcpu, vcpu->arch.cr3)) {
1400 printk(KERN_ERR "EPT: Fail to load pdptrs!\n");
1401 return;
1402 }
1403 vmcs_write64(GUEST_PDPTR0, vcpu->arch.pdptrs[0]);
1404 vmcs_write64(GUEST_PDPTR1, vcpu->arch.pdptrs[1]);
1405 vmcs_write64(GUEST_PDPTR2, vcpu->arch.pdptrs[2]);
1406 vmcs_write64(GUEST_PDPTR3, vcpu->arch.pdptrs[3]);
1407 }
1408}
1409
1410static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
1411
1412static void ept_update_paging_mode_cr0(unsigned long *hw_cr0,
1413 unsigned long cr0,
1414 struct kvm_vcpu *vcpu)
1415{
1416 if (!(cr0 & X86_CR0_PG)) {
1417 /* From paging/starting to nonpaging */
1418 vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
1419 vmcs_config.cpu_based_exec_ctrl |
1420 (CPU_BASED_CR3_LOAD_EXITING |
1421 CPU_BASED_CR3_STORE_EXITING));
1422 vcpu->arch.cr0 = cr0;
1423 vmx_set_cr4(vcpu, vcpu->arch.cr4);
1424 *hw_cr0 |= X86_CR0_PE | X86_CR0_PG;
1425 *hw_cr0 &= ~X86_CR0_WP;
1426 } else if (!is_paging(vcpu)) {
1427 /* From nonpaging to paging */
1428 vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
1429 vmcs_config.cpu_based_exec_ctrl &
1430 ~(CPU_BASED_CR3_LOAD_EXITING |
1431 CPU_BASED_CR3_STORE_EXITING));
1432 vcpu->arch.cr0 = cr0;
1433 vmx_set_cr4(vcpu, vcpu->arch.cr4);
1434 if (!(vcpu->arch.cr0 & X86_CR0_WP))
1435 *hw_cr0 &= ~X86_CR0_WP;
1436 }
1437}
1438
1439static void ept_update_paging_mode_cr4(unsigned long *hw_cr4,
1440 struct kvm_vcpu *vcpu)
1441{
1442 if (!is_paging(vcpu)) {
1443 *hw_cr4 &= ~X86_CR4_PAE;
1444 *hw_cr4 |= X86_CR4_PSE;
1445 } else if (!(vcpu->arch.cr4 & X86_CR4_PAE))
1446 *hw_cr4 &= ~X86_CR4_PAE;
1447}
1448
1307static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0) 1449static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
1308{ 1450{
1451 unsigned long hw_cr0 = (cr0 & ~KVM_GUEST_CR0_MASK) |
1452 KVM_VM_CR0_ALWAYS_ON;
1453
1309 vmx_fpu_deactivate(vcpu); 1454 vmx_fpu_deactivate(vcpu);
1310 1455
1311 if (vcpu->arch.rmode.active && (cr0 & X86_CR0_PE)) 1456 if (vcpu->arch.rmode.active && (cr0 & X86_CR0_PE))
@@ -1323,29 +1468,61 @@ static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
1323 } 1468 }
1324#endif 1469#endif
1325 1470
1471 if (vm_need_ept())
1472 ept_update_paging_mode_cr0(&hw_cr0, cr0, vcpu);
1473
1326 vmcs_writel(CR0_READ_SHADOW, cr0); 1474 vmcs_writel(CR0_READ_SHADOW, cr0);
1327 vmcs_writel(GUEST_CR0, 1475 vmcs_writel(GUEST_CR0, hw_cr0);
1328 (cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
1329 vcpu->arch.cr0 = cr0; 1476 vcpu->arch.cr0 = cr0;
1330 1477
1331 if (!(cr0 & X86_CR0_TS) || !(cr0 & X86_CR0_PE)) 1478 if (!(cr0 & X86_CR0_TS) || !(cr0 & X86_CR0_PE))
1332 vmx_fpu_activate(vcpu); 1479 vmx_fpu_activate(vcpu);
1333} 1480}
1334 1481
1482static u64 construct_eptp(unsigned long root_hpa)
1483{
1484 u64 eptp;
1485
1486 /* TODO write the value reading from MSR */
1487 eptp = VMX_EPT_DEFAULT_MT |
1488 VMX_EPT_DEFAULT_GAW << VMX_EPT_GAW_EPTP_SHIFT;
1489 eptp |= (root_hpa & PAGE_MASK);
1490
1491 return eptp;
1492}
1493
1335static void vmx_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3) 1494static void vmx_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
1336{ 1495{
1496 unsigned long guest_cr3;
1497 u64 eptp;
1498
1499 guest_cr3 = cr3;
1500 if (vm_need_ept()) {
1501 eptp = construct_eptp(cr3);
1502 vmcs_write64(EPT_POINTER, eptp);
1503 ept_sync_context(eptp);
1504 ept_load_pdptrs(vcpu);
1505 guest_cr3 = is_paging(vcpu) ? vcpu->arch.cr3 :
1506 VMX_EPT_IDENTITY_PAGETABLE_ADDR;
1507 }
1508
1337 vmx_flush_tlb(vcpu); 1509 vmx_flush_tlb(vcpu);
1338 vmcs_writel(GUEST_CR3, cr3); 1510 vmcs_writel(GUEST_CR3, guest_cr3);
1339 if (vcpu->arch.cr0 & X86_CR0_PE) 1511 if (vcpu->arch.cr0 & X86_CR0_PE)
1340 vmx_fpu_deactivate(vcpu); 1512 vmx_fpu_deactivate(vcpu);
1341} 1513}
1342 1514
1343static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4) 1515static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
1344{ 1516{
1345 vmcs_writel(CR4_READ_SHADOW, cr4); 1517 unsigned long hw_cr4 = cr4 | (vcpu->arch.rmode.active ?
1346 vmcs_writel(GUEST_CR4, cr4 | (vcpu->arch.rmode.active ? 1518 KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON);
1347 KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON)); 1519
1348 vcpu->arch.cr4 = cr4; 1520 vcpu->arch.cr4 = cr4;
1521 if (vm_need_ept())
1522 ept_update_paging_mode_cr4(&hw_cr4, vcpu);
1523
1524 vmcs_writel(CR4_READ_SHADOW, cr4);
1525 vmcs_writel(GUEST_CR4, hw_cr4);
1349} 1526}
1350 1527
1351static void vmx_set_efer(struct kvm_vcpu *vcpu, u64 efer) 1528static void vmx_set_efer(struct kvm_vcpu *vcpu, u64 efer)
@@ -1530,6 +1707,41 @@ out:
1530 return ret; 1707 return ret;
1531} 1708}
1532 1709
1710static int init_rmode_identity_map(struct kvm *kvm)
1711{
1712 int i, r, ret;
1713 pfn_t identity_map_pfn;
1714 u32 tmp;
1715
1716 if (!vm_need_ept())
1717 return 1;
1718 if (unlikely(!kvm->arch.ept_identity_pagetable)) {
1719 printk(KERN_ERR "EPT: identity-mapping pagetable "
1720 "haven't been allocated!\n");
1721 return 0;
1722 }
1723 if (likely(kvm->arch.ept_identity_pagetable_done))
1724 return 1;
1725 ret = 0;
1726 identity_map_pfn = VMX_EPT_IDENTITY_PAGETABLE_ADDR >> PAGE_SHIFT;
1727 r = kvm_clear_guest_page(kvm, identity_map_pfn, 0, PAGE_SIZE);
1728 if (r < 0)
1729 goto out;
1730 /* Set up identity-mapping pagetable for EPT in real mode */
1731 for (i = 0; i < PT32_ENT_PER_PAGE; i++) {
1732 tmp = (i << 22) + (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER |
1733 _PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_PSE);
1734 r = kvm_write_guest_page(kvm, identity_map_pfn,
1735 &tmp, i * sizeof(tmp), sizeof(tmp));
1736 if (r < 0)
1737 goto out;
1738 }
1739 kvm->arch.ept_identity_pagetable_done = true;
1740 ret = 1;
1741out:
1742 return ret;
1743}
1744
1533static void seg_setup(int seg) 1745static void seg_setup(int seg)
1534{ 1746{
1535 struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg]; 1747 struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
@@ -1564,6 +1776,31 @@ out:
1564 return r; 1776 return r;
1565} 1777}
1566 1778
1779static int alloc_identity_pagetable(struct kvm *kvm)
1780{
1781 struct kvm_userspace_memory_region kvm_userspace_mem;
1782 int r = 0;
1783
1784 down_write(&kvm->slots_lock);
1785 if (kvm->arch.ept_identity_pagetable)
1786 goto out;
1787 kvm_userspace_mem.slot = IDENTITY_PAGETABLE_PRIVATE_MEMSLOT;
1788 kvm_userspace_mem.flags = 0;
1789 kvm_userspace_mem.guest_phys_addr = VMX_EPT_IDENTITY_PAGETABLE_ADDR;
1790 kvm_userspace_mem.memory_size = PAGE_SIZE;
1791 r = __kvm_set_memory_region(kvm, &kvm_userspace_mem, 0);
1792 if (r)
1793 goto out;
1794
1795 down_read(&current->mm->mmap_sem);
1796 kvm->arch.ept_identity_pagetable = gfn_to_page(kvm,
1797 VMX_EPT_IDENTITY_PAGETABLE_ADDR >> PAGE_SHIFT);
1798 up_read(&current->mm->mmap_sem);
1799out:
1800 up_write(&kvm->slots_lock);
1801 return r;
1802}
1803
1567static void allocate_vpid(struct vcpu_vmx *vmx) 1804static void allocate_vpid(struct vcpu_vmx *vmx)
1568{ 1805{
1569 int vpid; 1806 int vpid;
@@ -1638,6 +1875,9 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
1638 CPU_BASED_CR8_LOAD_EXITING; 1875 CPU_BASED_CR8_LOAD_EXITING;
1639#endif 1876#endif
1640 } 1877 }
1878 if (!vm_need_ept())
1879 exec_control |= CPU_BASED_CR3_STORE_EXITING |
1880 CPU_BASED_CR3_LOAD_EXITING;
1641 vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, exec_control); 1881 vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, exec_control);
1642 1882
1643 if (cpu_has_secondary_exec_ctrls()) { 1883 if (cpu_has_secondary_exec_ctrls()) {
@@ -1647,6 +1887,8 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
1647 ~SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES; 1887 ~SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES;
1648 if (vmx->vpid == 0) 1888 if (vmx->vpid == 0)
1649 exec_control &= ~SECONDARY_EXEC_ENABLE_VPID; 1889 exec_control &= ~SECONDARY_EXEC_ENABLE_VPID;
1890 if (!vm_need_ept())
1891 exec_control &= ~SECONDARY_EXEC_ENABLE_EPT;
1650 vmcs_write32(SECONDARY_VM_EXEC_CONTROL, exec_control); 1892 vmcs_write32(SECONDARY_VM_EXEC_CONTROL, exec_control);
1651 } 1893 }
1652 1894
@@ -1722,6 +1964,15 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
1722 return 0; 1964 return 0;
1723} 1965}
1724 1966
1967static int init_rmode(struct kvm *kvm)
1968{
1969 if (!init_rmode_tss(kvm))
1970 return 0;
1971 if (!init_rmode_identity_map(kvm))
1972 return 0;
1973 return 1;
1974}
1975
1725static int vmx_vcpu_reset(struct kvm_vcpu *vcpu) 1976static int vmx_vcpu_reset(struct kvm_vcpu *vcpu)
1726{ 1977{
1727 struct vcpu_vmx *vmx = to_vmx(vcpu); 1978 struct vcpu_vmx *vmx = to_vmx(vcpu);
@@ -1729,7 +1980,7 @@ static int vmx_vcpu_reset(struct kvm_vcpu *vcpu)
1729 int ret; 1980 int ret;
1730 1981
1731 down_read(&vcpu->kvm->slots_lock); 1982 down_read(&vcpu->kvm->slots_lock);
1732 if (!init_rmode_tss(vmx->vcpu.kvm)) { 1983 if (!init_rmode(vmx->vcpu.kvm)) {
1733 ret = -ENOMEM; 1984 ret = -ENOMEM;
1734 goto out; 1985 goto out;
1735 } 1986 }
@@ -1994,6 +2245,9 @@ static int handle_exception(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
1994 if (intr_info & INTR_INFO_DELIVER_CODE_MASK) 2245 if (intr_info & INTR_INFO_DELIVER_CODE_MASK)
1995 error_code = vmcs_read32(VM_EXIT_INTR_ERROR_CODE); 2246 error_code = vmcs_read32(VM_EXIT_INTR_ERROR_CODE);
1996 if (is_page_fault(intr_info)) { 2247 if (is_page_fault(intr_info)) {
2248 /* EPT won't cause page fault directly */
2249 if (vm_need_ept())
2250 BUG();
1997 cr2 = vmcs_readl(EXIT_QUALIFICATION); 2251 cr2 = vmcs_readl(EXIT_QUALIFICATION);
1998 KVMTRACE_3D(PAGE_FAULT, vcpu, error_code, (u32)cr2, 2252 KVMTRACE_3D(PAGE_FAULT, vcpu, error_code, (u32)cr2,
1999 (u32)((u64)cr2 >> 32), handler); 2253 (u32)((u64)cr2 >> 32), handler);
@@ -2323,6 +2577,64 @@ static int handle_task_switch(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
2323 return kvm_task_switch(vcpu, tss_selector, reason); 2577 return kvm_task_switch(vcpu, tss_selector, reason);
2324} 2578}
2325 2579
2580static int handle_ept_violation(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
2581{
2582 u64 exit_qualification;
2583 enum emulation_result er;
2584 gpa_t gpa;
2585 unsigned long hva;
2586 int gla_validity;
2587 int r;
2588
2589 exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
2590
2591 if (exit_qualification & (1 << 6)) {
2592 printk(KERN_ERR "EPT: GPA exceeds GAW!\n");
2593 return -ENOTSUPP;
2594 }
2595
2596 gla_validity = (exit_qualification >> 7) & 0x3;
2597 if (gla_validity != 0x3 && gla_validity != 0x1 && gla_validity != 0) {
2598 printk(KERN_ERR "EPT: Handling EPT violation failed!\n");
2599 printk(KERN_ERR "EPT: GPA: 0x%lx, GVA: 0x%lx\n",
2600 (long unsigned int)vmcs_read64(GUEST_PHYSICAL_ADDRESS),
2601 (long unsigned int)vmcs_read64(GUEST_LINEAR_ADDRESS));
2602 printk(KERN_ERR "EPT: Exit qualification is 0x%lx\n",
2603 (long unsigned int)exit_qualification);
2604 kvm_run->exit_reason = KVM_EXIT_UNKNOWN;
2605 kvm_run->hw.hardware_exit_reason = 0;
2606 return -ENOTSUPP;
2607 }
2608
2609 gpa = vmcs_read64(GUEST_PHYSICAL_ADDRESS);
2610 hva = gfn_to_hva(vcpu->kvm, gpa >> PAGE_SHIFT);
2611 if (!kvm_is_error_hva(hva)) {
2612 r = kvm_mmu_page_fault(vcpu, gpa & PAGE_MASK, 0);
2613 if (r < 0) {
2614 printk(KERN_ERR "EPT: Not enough memory!\n");
2615 return -ENOMEM;
2616 }
2617 return 1;
2618 } else {
2619 /* must be MMIO */
2620 er = emulate_instruction(vcpu, kvm_run, 0, 0, 0);
2621
2622 if (er == EMULATE_FAIL) {
2623 printk(KERN_ERR
2624 "EPT: Fail to handle EPT violation vmexit!er is %d\n",
2625 er);
2626 printk(KERN_ERR "EPT: GPA: 0x%lx, GVA: 0x%lx\n",
2627 (long unsigned int)vmcs_read64(GUEST_PHYSICAL_ADDRESS),
2628 (long unsigned int)vmcs_read64(GUEST_LINEAR_ADDRESS));
2629 printk(KERN_ERR "EPT: Exit qualification is 0x%lx\n",
2630 (long unsigned int)exit_qualification);
2631 return -ENOTSUPP;
2632 } else if (er == EMULATE_DO_MMIO)
2633 return 0;
2634 }
2635 return 1;
2636}
2637
2326/* 2638/*
2327 * The exit handlers return 1 if the exit was handled fully and guest execution 2639 * The exit handlers return 1 if the exit was handled fully and guest execution
2328 * may resume. Otherwise they set the kvm_run parameter to indicate what needs 2640 * may resume. Otherwise they set the kvm_run parameter to indicate what needs
@@ -2346,6 +2658,7 @@ static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu,
2346 [EXIT_REASON_APIC_ACCESS] = handle_apic_access, 2658 [EXIT_REASON_APIC_ACCESS] = handle_apic_access,
2347 [EXIT_REASON_WBINVD] = handle_wbinvd, 2659 [EXIT_REASON_WBINVD] = handle_wbinvd,
2348 [EXIT_REASON_TASK_SWITCH] = handle_task_switch, 2660 [EXIT_REASON_TASK_SWITCH] = handle_task_switch,
2661 [EXIT_REASON_EPT_VIOLATION] = handle_ept_violation,
2349}; 2662};
2350 2663
2351static const int kvm_vmx_max_exit_handlers = 2664static const int kvm_vmx_max_exit_handlers =
@@ -2364,6 +2677,13 @@ static int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
2364 KVMTRACE_3D(VMEXIT, vcpu, exit_reason, (u32)vmcs_readl(GUEST_RIP), 2677 KVMTRACE_3D(VMEXIT, vcpu, exit_reason, (u32)vmcs_readl(GUEST_RIP),
2365 (u32)((u64)vmcs_readl(GUEST_RIP) >> 32), entryexit); 2678 (u32)((u64)vmcs_readl(GUEST_RIP) >> 32), entryexit);
2366 2679
2680 /* Access CR3 don't cause VMExit in paging mode, so we need
2681 * to sync with guest real CR3. */
2682 if (vm_need_ept() && is_paging(vcpu)) {
2683 vcpu->arch.cr3 = vmcs_readl(GUEST_CR3);
2684 ept_load_pdptrs(vcpu);
2685 }
2686
2367 if (unlikely(vmx->fail)) { 2687 if (unlikely(vmx->fail)) {
2368 kvm_run->exit_reason = KVM_EXIT_FAIL_ENTRY; 2688 kvm_run->exit_reason = KVM_EXIT_FAIL_ENTRY;
2369 kvm_run->fail_entry.hardware_entry_failure_reason 2689 kvm_run->fail_entry.hardware_entry_failure_reason
@@ -2372,7 +2692,8 @@ static int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
2372 } 2692 }
2373 2693
2374 if ((vectoring_info & VECTORING_INFO_VALID_MASK) && 2694 if ((vectoring_info & VECTORING_INFO_VALID_MASK) &&
2375 exit_reason != EXIT_REASON_EXCEPTION_NMI) 2695 (exit_reason != EXIT_REASON_EXCEPTION_NMI &&
2696 exit_reason != EXIT_REASON_EPT_VIOLATION))
2376 printk(KERN_WARNING "%s: unexpected, valid vectoring info and " 2697 printk(KERN_WARNING "%s: unexpected, valid vectoring info and "
2377 "exit reason is 0x%x\n", __func__, exit_reason); 2698 "exit reason is 0x%x\n", __func__, exit_reason);
2378 if (exit_reason < kvm_vmx_max_exit_handlers 2699 if (exit_reason < kvm_vmx_max_exit_handlers
@@ -2674,6 +2995,15 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
2674 return ERR_PTR(-ENOMEM); 2995 return ERR_PTR(-ENOMEM);
2675 2996
2676 allocate_vpid(vmx); 2997 allocate_vpid(vmx);
2998 if (id == 0 && vm_need_ept()) {
2999 kvm_mmu_set_base_ptes(VMX_EPT_READABLE_MASK |
3000 VMX_EPT_WRITABLE_MASK |
3001 VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT);
3002 kvm_mmu_set_mask_ptes(0ull, VMX_EPT_FAKE_ACCESSED_MASK,
3003 VMX_EPT_FAKE_DIRTY_MASK, 0ull,
3004 VMX_EPT_EXECUTABLE_MASK);
3005 kvm_enable_tdp();
3006 }
2677 3007
2678 err = kvm_vcpu_init(&vmx->vcpu, kvm, id); 3008 err = kvm_vcpu_init(&vmx->vcpu, kvm, id);
2679 if (err) 3009 if (err)
@@ -2706,6 +3036,10 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
2706 if (alloc_apic_access_page(kvm) != 0) 3036 if (alloc_apic_access_page(kvm) != 0)
2707 goto free_vmcs; 3037 goto free_vmcs;
2708 3038
3039 if (vm_need_ept())
3040 if (alloc_identity_pagetable(kvm) != 0)
3041 goto free_vmcs;
3042
2709 return &vmx->vcpu; 3043 return &vmx->vcpu;
2710 3044
2711free_vmcs: 3045free_vmcs:
@@ -2735,6 +3069,11 @@ static void __init vmx_check_processor_compat(void *rtn)
2735 } 3069 }
2736} 3070}
2737 3071
3072static int get_ept_level(void)
3073{
3074 return VMX_EPT_DEFAULT_GAW + 1;
3075}
3076
2738static struct kvm_x86_ops vmx_x86_ops = { 3077static struct kvm_x86_ops vmx_x86_ops = {
2739 .cpu_has_kvm_support = cpu_has_kvm_support, 3078 .cpu_has_kvm_support = cpu_has_kvm_support,
2740 .disabled_by_bios = vmx_disabled_by_bios, 3079 .disabled_by_bios = vmx_disabled_by_bios,
@@ -2791,6 +3130,7 @@ static struct kvm_x86_ops vmx_x86_ops = {
2791 .inject_pending_vectors = do_interrupt_requests, 3130 .inject_pending_vectors = do_interrupt_requests,
2792 3131
2793 .set_tss_addr = vmx_set_tss_addr, 3132 .set_tss_addr = vmx_set_tss_addr,
3133 .get_tdp_level = get_ept_level,
2794}; 3134};
2795 3135
2796static int __init vmx_init(void) 3136static int __init vmx_init(void)
@@ -2843,9 +3183,14 @@ static int __init vmx_init(void)
2843 vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP); 3183 vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP);
2844 vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP); 3184 vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP);
2845 3185
3186 if (cpu_has_vmx_ept())
3187 bypass_guest_pf = 0;
3188
2846 if (bypass_guest_pf) 3189 if (bypass_guest_pf)
2847 kvm_mmu_set_nonpresent_ptes(~0xffeull, 0ull); 3190 kvm_mmu_set_nonpresent_ptes(~0xffeull, 0ull);
2848 3191
3192 ept_sync_global();
3193
2849 return 0; 3194 return 0;
2850 3195
2851out2: 3196out2:
diff --git a/arch/x86/kvm/vmx.h b/arch/x86/kvm/vmx.h
index 5dff4606b988..79d94c610dfe 100644
--- a/arch/x86/kvm/vmx.h
+++ b/arch/x86/kvm/vmx.h
@@ -35,6 +35,8 @@
35#define CPU_BASED_MWAIT_EXITING 0x00000400 35#define CPU_BASED_MWAIT_EXITING 0x00000400
36#define CPU_BASED_RDPMC_EXITING 0x00000800 36#define CPU_BASED_RDPMC_EXITING 0x00000800
37#define CPU_BASED_RDTSC_EXITING 0x00001000 37#define CPU_BASED_RDTSC_EXITING 0x00001000
38#define CPU_BASED_CR3_LOAD_EXITING 0x00008000
39#define CPU_BASED_CR3_STORE_EXITING 0x00010000
38#define CPU_BASED_CR8_LOAD_EXITING 0x00080000 40#define CPU_BASED_CR8_LOAD_EXITING 0x00080000
39#define CPU_BASED_CR8_STORE_EXITING 0x00100000 41#define CPU_BASED_CR8_STORE_EXITING 0x00100000
40#define CPU_BASED_TPR_SHADOW 0x00200000 42#define CPU_BASED_TPR_SHADOW 0x00200000
@@ -49,6 +51,7 @@
49 * Definitions of Secondary Processor-Based VM-Execution Controls. 51 * Definitions of Secondary Processor-Based VM-Execution Controls.
50 */ 52 */
51#define SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES 0x00000001 53#define SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES 0x00000001
54#define SECONDARY_EXEC_ENABLE_EPT 0x00000002
52#define SECONDARY_EXEC_ENABLE_VPID 0x00000020 55#define SECONDARY_EXEC_ENABLE_VPID 0x00000020
53#define SECONDARY_EXEC_WBINVD_EXITING 0x00000040 56#define SECONDARY_EXEC_WBINVD_EXITING 0x00000040
54 57
@@ -100,10 +103,22 @@ enum vmcs_field {
100 VIRTUAL_APIC_PAGE_ADDR_HIGH = 0x00002013, 103 VIRTUAL_APIC_PAGE_ADDR_HIGH = 0x00002013,
101 APIC_ACCESS_ADDR = 0x00002014, 104 APIC_ACCESS_ADDR = 0x00002014,
102 APIC_ACCESS_ADDR_HIGH = 0x00002015, 105 APIC_ACCESS_ADDR_HIGH = 0x00002015,
106 EPT_POINTER = 0x0000201a,
107 EPT_POINTER_HIGH = 0x0000201b,
108 GUEST_PHYSICAL_ADDRESS = 0x00002400,
109 GUEST_PHYSICAL_ADDRESS_HIGH = 0x00002401,
103 VMCS_LINK_POINTER = 0x00002800, 110 VMCS_LINK_POINTER = 0x00002800,
104 VMCS_LINK_POINTER_HIGH = 0x00002801, 111 VMCS_LINK_POINTER_HIGH = 0x00002801,
105 GUEST_IA32_DEBUGCTL = 0x00002802, 112 GUEST_IA32_DEBUGCTL = 0x00002802,
106 GUEST_IA32_DEBUGCTL_HIGH = 0x00002803, 113 GUEST_IA32_DEBUGCTL_HIGH = 0x00002803,
114 GUEST_PDPTR0 = 0x0000280a,
115 GUEST_PDPTR0_HIGH = 0x0000280b,
116 GUEST_PDPTR1 = 0x0000280c,
117 GUEST_PDPTR1_HIGH = 0x0000280d,
118 GUEST_PDPTR2 = 0x0000280e,
119 GUEST_PDPTR2_HIGH = 0x0000280f,
120 GUEST_PDPTR3 = 0x00002810,
121 GUEST_PDPTR3_HIGH = 0x00002811,
107 PIN_BASED_VM_EXEC_CONTROL = 0x00004000, 122 PIN_BASED_VM_EXEC_CONTROL = 0x00004000,
108 CPU_BASED_VM_EXEC_CONTROL = 0x00004002, 123 CPU_BASED_VM_EXEC_CONTROL = 0x00004002,
109 EXCEPTION_BITMAP = 0x00004004, 124 EXCEPTION_BITMAP = 0x00004004,
@@ -226,6 +241,8 @@ enum vmcs_field {
226#define EXIT_REASON_MWAIT_INSTRUCTION 36 241#define EXIT_REASON_MWAIT_INSTRUCTION 36
227#define EXIT_REASON_TPR_BELOW_THRESHOLD 43 242#define EXIT_REASON_TPR_BELOW_THRESHOLD 43
228#define EXIT_REASON_APIC_ACCESS 44 243#define EXIT_REASON_APIC_ACCESS 44
244#define EXIT_REASON_EPT_VIOLATION 48
245#define EXIT_REASON_EPT_MISCONFIG 49
229#define EXIT_REASON_WBINVD 54 246#define EXIT_REASON_WBINVD 54
230 247
231/* 248/*
@@ -316,15 +333,36 @@ enum vmcs_field {
316#define MSR_IA32_VMX_CR4_FIXED1 0x489 333#define MSR_IA32_VMX_CR4_FIXED1 0x489
317#define MSR_IA32_VMX_VMCS_ENUM 0x48a 334#define MSR_IA32_VMX_VMCS_ENUM 0x48a
318#define MSR_IA32_VMX_PROCBASED_CTLS2 0x48b 335#define MSR_IA32_VMX_PROCBASED_CTLS2 0x48b
336#define MSR_IA32_VMX_EPT_VPID_CAP 0x48c
319 337
320#define MSR_IA32_FEATURE_CONTROL 0x3a 338#define MSR_IA32_FEATURE_CONTROL 0x3a
321#define MSR_IA32_FEATURE_CONTROL_LOCKED 0x1 339#define MSR_IA32_FEATURE_CONTROL_LOCKED 0x1
322#define MSR_IA32_FEATURE_CONTROL_VMXON_ENABLED 0x4 340#define MSR_IA32_FEATURE_CONTROL_VMXON_ENABLED 0x4
323 341
324#define APIC_ACCESS_PAGE_PRIVATE_MEMSLOT 9 342#define APIC_ACCESS_PAGE_PRIVATE_MEMSLOT 9
343#define IDENTITY_PAGETABLE_PRIVATE_MEMSLOT 10
325 344
326#define VMX_NR_VPIDS (1 << 16) 345#define VMX_NR_VPIDS (1 << 16)
327#define VMX_VPID_EXTENT_SINGLE_CONTEXT 1 346#define VMX_VPID_EXTENT_SINGLE_CONTEXT 1
328#define VMX_VPID_EXTENT_ALL_CONTEXT 2 347#define VMX_VPID_EXTENT_ALL_CONTEXT 2
329 348
349#define VMX_EPT_EXTENT_INDIVIDUAL_ADDR 0
350#define VMX_EPT_EXTENT_CONTEXT 1
351#define VMX_EPT_EXTENT_GLOBAL 2
352#define VMX_EPT_EXTENT_INDIVIDUAL_BIT (1ull << 24)
353#define VMX_EPT_EXTENT_CONTEXT_BIT (1ull << 25)
354#define VMX_EPT_EXTENT_GLOBAL_BIT (1ull << 26)
355#define VMX_EPT_DEFAULT_GAW 3
356#define VMX_EPT_MAX_GAW 0x4
357#define VMX_EPT_MT_EPTE_SHIFT 3
358#define VMX_EPT_GAW_EPTP_SHIFT 3
359#define VMX_EPT_DEFAULT_MT 0x6ull
360#define VMX_EPT_READABLE_MASK 0x1ull
361#define VMX_EPT_WRITABLE_MASK 0x2ull
362#define VMX_EPT_EXECUTABLE_MASK 0x4ull
363#define VMX_EPT_FAKE_ACCESSED_MASK (1ull << 62)
364#define VMX_EPT_FAKE_DIRTY_MASK (1ull << 63)
365
366#define VMX_EPT_IDENTITY_PAGETABLE_ADDR 0xfffbc000ul
367
330#endif 368#endif
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 0ce556372a4d..21338bdb28ff 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2417,6 +2417,9 @@ int kvm_arch_init(void *opaque)
2417 2417
2418 kvm_x86_ops = ops; 2418 kvm_x86_ops = ops;
2419 kvm_mmu_set_nonpresent_ptes(0ull, 0ull); 2419 kvm_mmu_set_nonpresent_ptes(0ull, 0ull);
2420 kvm_mmu_set_base_ptes(PT_PRESENT_MASK);
2421 kvm_mmu_set_mask_ptes(PT_USER_MASK, PT_ACCESSED_MASK,
2422 PT_DIRTY_MASK, PT64_NX_MASK, 0);
2420 return 0; 2423 return 0;
2421 2424
2422out: 2425out:
@@ -3019,6 +3022,8 @@ int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs)
3019 3022
3020 kvm_x86_ops->decache_regs(vcpu); 3023 kvm_x86_ops->decache_regs(vcpu);
3021 3024
3025 vcpu->arch.exception.pending = false;
3026
3022 vcpu_put(vcpu); 3027 vcpu_put(vcpu);
3023 3028
3024 return 0; 3029 return 0;
@@ -3481,7 +3486,7 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason)
3481 } 3486 }
3482 3487
3483 if (reason == TASK_SWITCH_IRET || reason == TASK_SWITCH_JMP) { 3488 if (reason == TASK_SWITCH_IRET || reason == TASK_SWITCH_JMP) {
3484 cseg_desc.type &= ~(1 << 8); //clear the B flag 3489 cseg_desc.type &= ~(1 << 1); //clear the B flag
3485 save_guest_segment_descriptor(vcpu, tr_seg.selector, 3490 save_guest_segment_descriptor(vcpu, tr_seg.selector,
3486 &cseg_desc); 3491 &cseg_desc);
3487 } 3492 }
@@ -3507,7 +3512,7 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason)
3507 } 3512 }
3508 3513
3509 if (reason != TASK_SWITCH_IRET) { 3514 if (reason != TASK_SWITCH_IRET) {
3510 nseg_desc.type |= (1 << 8); 3515 nseg_desc.type |= (1 << 1);
3511 save_guest_segment_descriptor(vcpu, tss_selector, 3516 save_guest_segment_descriptor(vcpu, tss_selector,
3512 &nseg_desc); 3517 &nseg_desc);
3513 } 3518 }
@@ -3698,10 +3703,19 @@ void fx_init(struct kvm_vcpu *vcpu)
3698{ 3703{
3699 unsigned after_mxcsr_mask; 3704 unsigned after_mxcsr_mask;
3700 3705
3706 /*
3707 * Touch the fpu the first time in non atomic context as if
3708 * this is the first fpu instruction the exception handler
3709 * will fire before the instruction returns and it'll have to
3710 * allocate ram with GFP_KERNEL.
3711 */
3712 if (!used_math())
3713 fx_save(&vcpu->arch.host_fx_image);
3714
3701 /* Initialize guest FPU by resetting ours and saving into guest's */ 3715 /* Initialize guest FPU by resetting ours and saving into guest's */
3702 preempt_disable(); 3716 preempt_disable();
3703 fx_save(&vcpu->arch.host_fx_image); 3717 fx_save(&vcpu->arch.host_fx_image);
3704 fpu_init(); 3718 fx_finit();
3705 fx_save(&vcpu->arch.guest_fx_image); 3719 fx_save(&vcpu->arch.guest_fx_image);
3706 fx_restore(&vcpu->arch.host_fx_image); 3720 fx_restore(&vcpu->arch.host_fx_image);
3707 preempt_enable(); 3721 preempt_enable();
@@ -3906,6 +3920,8 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
3906 kvm_free_physmem(kvm); 3920 kvm_free_physmem(kvm);
3907 if (kvm->arch.apic_access_page) 3921 if (kvm->arch.apic_access_page)
3908 put_page(kvm->arch.apic_access_page); 3922 put_page(kvm->arch.apic_access_page);
3923 if (kvm->arch.ept_identity_pagetable)
3924 put_page(kvm->arch.ept_identity_pagetable);
3909 kfree(kvm); 3925 kfree(kvm);
3910} 3926}
3911 3927
diff --git a/arch/x86/kvm/x86_emulate.c b/arch/x86/kvm/x86_emulate.c
index 2ca08386f993..f2a696d6a243 100644
--- a/arch/x86/kvm/x86_emulate.c
+++ b/arch/x86/kvm/x86_emulate.c
@@ -1761,6 +1761,7 @@ twobyte_insn:
1761 case 6: /* lmsw */ 1761 case 6: /* lmsw */
1762 realmode_lmsw(ctxt->vcpu, (u16)c->src.val, 1762 realmode_lmsw(ctxt->vcpu, (u16)c->src.val,
1763 &ctxt->eflags); 1763 &ctxt->eflags);
1764 c->dst.type = OP_NONE;
1764 break; 1765 break;
1765 case 7: /* invlpg*/ 1766 case 7: /* invlpg*/
1766 emulate_invlpg(ctxt->vcpu, memop); 1767 emulate_invlpg(ctxt->vcpu, memop);
diff --git a/arch/x86/mm/discontig_32.c b/arch/x86/mm/discontig_32.c
index 18378850e25a..914ccf983687 100644
--- a/arch/x86/mm/discontig_32.c
+++ b/arch/x86/mm/discontig_32.c
@@ -476,29 +476,3 @@ int memory_add_physaddr_to_nid(u64 addr)
476 476
477EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid); 477EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid);
478#endif 478#endif
479
480#ifndef CONFIG_HAVE_ARCH_PARSE_SRAT
481/*
482 * XXX FIXME: Make SLIT table parsing available to 32-bit NUMA
483 *
484 * These stub functions are needed to compile 32-bit NUMA when SRAT is
485 * not set. There are functions in srat_64.c for parsing this table
486 * and it may be possible to make them common functions.
487 */
488void acpi_numa_slit_init (struct acpi_table_slit *slit)
489{
490 printk(KERN_INFO "ACPI: No support for parsing SLIT table\n");
491}
492
493void acpi_numa_processor_affinity_init (struct acpi_srat_cpu_affinity *pa)
494{
495}
496
497void acpi_numa_memory_affinity_init (struct acpi_srat_mem_affinity *ma)
498{
499}
500
501void acpi_numa_arch_fixup(void)
502{
503}
504#endif /* CONFIG_HAVE_ARCH_PARSE_SRAT */
diff --git a/arch/x86/mm/pgtable_32.c b/arch/x86/mm/pgtable_32.c
index 9ee007be9142..369cf065b6a4 100644
--- a/arch/x86/mm/pgtable_32.c
+++ b/arch/x86/mm/pgtable_32.c
@@ -172,10 +172,3 @@ void reserve_top_address(unsigned long reserve)
172 __FIXADDR_TOP = -reserve - PAGE_SIZE; 172 __FIXADDR_TOP = -reserve - PAGE_SIZE;
173 __VMALLOC_RESERVE += reserve; 173 __VMALLOC_RESERVE += reserve;
174} 174}
175
176int pmd_bad(pmd_t pmd)
177{
178 WARN_ON_ONCE(pmd_bad_v1(pmd) != pmd_bad_v2(pmd));
179
180 return pmd_bad_v1(pmd);
181}
diff --git a/arch/x86/pci/Makefile_32 b/arch/x86/pci/Makefile_32
index 7fa519868d70..89ec35d00efd 100644
--- a/arch/x86/pci/Makefile_32
+++ b/arch/x86/pci/Makefile_32
@@ -6,11 +6,19 @@ obj-$(CONFIG_PCI_DIRECT) += direct.o
6obj-$(CONFIG_PCI_OLPC) += olpc.o 6obj-$(CONFIG_PCI_OLPC) += olpc.o
7 7
8pci-y := fixup.o 8pci-y := fixup.o
9
10# Do not change the ordering here. There is a nasty init function
11# ordering dependency which breaks when you move acpi.o below
12# legacy/irq.o
9pci-$(CONFIG_ACPI) += acpi.o 13pci-$(CONFIG_ACPI) += acpi.o
10pci-y += legacy.o irq.o 14pci-y += legacy.o irq.o
11 15
12pci-$(CONFIG_X86_VISWS) += visws.o fixup.o 16# Careful: VISWS and NUMAQ overrule the pci-y above. The colons are
13pci-$(CONFIG_X86_NUMAQ) += numa.o irq.o 17# therefor correct. This needs a proper fix by distangling the code.
18pci-$(CONFIG_X86_VISWS) := visws.o fixup.o
19pci-$(CONFIG_X86_NUMAQ) := numa.o irq.o
20
21# Necessary for NUMAQ as well
14pci-$(CONFIG_NUMA) += mp_bus_to_node.o 22pci-$(CONFIG_NUMA) += mp_bus_to_node.o
15 23
16obj-y += $(pci-y) common.o early.o 24obj-y += $(pci-y) common.o early.o
diff --git a/arch/x86/pci/acpi.c b/arch/x86/pci/acpi.c
index 1a9c0c6a1a18..d95de2f199cd 100644
--- a/arch/x86/pci/acpi.c
+++ b/arch/x86/pci/acpi.c
@@ -6,45 +6,6 @@
6#include <asm/numa.h> 6#include <asm/numa.h>
7#include "pci.h" 7#include "pci.h"
8 8
9static int __devinit can_skip_ioresource_align(const struct dmi_system_id *d)
10{
11 pci_probe |= PCI_CAN_SKIP_ISA_ALIGN;
12 printk(KERN_INFO "PCI: %s detected, can skip ISA alignment\n", d->ident);
13 return 0;
14}
15
16static struct dmi_system_id acpi_pciprobe_dmi_table[] __devinitdata = {
17/*
18 * Systems where PCI IO resource ISA alignment can be skipped
19 * when the ISA enable bit in the bridge control is not set
20 */
21 {
22 .callback = can_skip_ioresource_align,
23 .ident = "IBM System x3800",
24 .matches = {
25 DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
26 DMI_MATCH(DMI_PRODUCT_NAME, "x3800"),
27 },
28 },
29 {
30 .callback = can_skip_ioresource_align,
31 .ident = "IBM System x3850",
32 .matches = {
33 DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
34 DMI_MATCH(DMI_PRODUCT_NAME, "x3850"),
35 },
36 },
37 {
38 .callback = can_skip_ioresource_align,
39 .ident = "IBM System x3950",
40 .matches = {
41 DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
42 DMI_MATCH(DMI_PRODUCT_NAME, "x3950"),
43 },
44 },
45 {}
46};
47
48struct pci_root_info { 9struct pci_root_info {
49 char *name; 10 char *name;
50 unsigned int res_num; 11 unsigned int res_num;
@@ -196,8 +157,6 @@ struct pci_bus * __devinit pci_acpi_scan_root(struct acpi_device *device, int do
196 int pxm; 157 int pxm;
197#endif 158#endif
198 159
199 dmi_check_system(acpi_pciprobe_dmi_table);
200
201 if (domain && !pci_domains_supported) { 160 if (domain && !pci_domains_supported) {
202 printk(KERN_WARNING "PCI: Multiple domains not supported " 161 printk(KERN_WARNING "PCI: Multiple domains not supported "
203 "(dom %d, bus %d)\n", domain, busnum); 162 "(dom %d, bus %d)\n", domain, busnum);
diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index 2a4d751818b7..8545c8a9d107 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -77,17 +77,48 @@ int pcibios_scanned;
77 */ 77 */
78DEFINE_SPINLOCK(pci_config_lock); 78DEFINE_SPINLOCK(pci_config_lock);
79 79
80static void __devinit pcibios_fixup_device_resources(struct pci_dev *dev) 80static int __devinit can_skip_ioresource_align(const struct dmi_system_id *d)
81{ 81{
82 struct resource *rom_r = &dev->resource[PCI_ROM_RESOURCE]; 82 pci_probe |= PCI_CAN_SKIP_ISA_ALIGN;
83 83 printk(KERN_INFO "PCI: %s detected, can skip ISA alignment\n", d->ident);
84 if (rom_r->parent) 84 return 0;
85 return; 85}
86 if (rom_r->start) 86
87 /* we deal with BIOS assigned ROM later */ 87static struct dmi_system_id can_skip_pciprobe_dmi_table[] __devinitdata = {
88 return; 88/*
89 if (!(pci_probe & PCI_ASSIGN_ROMS)) 89 * Systems where PCI IO resource ISA alignment can be skipped
90 rom_r->start = rom_r->end = rom_r->flags = 0; 90 * when the ISA enable bit in the bridge control is not set
91 */
92 {
93 .callback = can_skip_ioresource_align,
94 .ident = "IBM System x3800",
95 .matches = {
96 DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
97 DMI_MATCH(DMI_PRODUCT_NAME, "x3800"),
98 },
99 },
100 {
101 .callback = can_skip_ioresource_align,
102 .ident = "IBM System x3850",
103 .matches = {
104 DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
105 DMI_MATCH(DMI_PRODUCT_NAME, "x3850"),
106 },
107 },
108 {
109 .callback = can_skip_ioresource_align,
110 .ident = "IBM System x3950",
111 .matches = {
112 DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
113 DMI_MATCH(DMI_PRODUCT_NAME, "x3950"),
114 },
115 },
116 {}
117};
118
119void __init dmi_check_skip_isa_align(void)
120{
121 dmi_check_system(can_skip_pciprobe_dmi_table);
91} 122}
92 123
93/* 124/*
@@ -97,11 +128,7 @@ static void __devinit pcibios_fixup_device_resources(struct pci_dev *dev)
97 128
98void __devinit pcibios_fixup_bus(struct pci_bus *b) 129void __devinit pcibios_fixup_bus(struct pci_bus *b)
99{ 130{
100 struct pci_dev *dev;
101
102 pci_read_bridge_bases(b); 131 pci_read_bridge_bases(b);
103 list_for_each_entry(dev, &b->devices, bus_list)
104 pcibios_fixup_device_resources(dev);
105} 132}
106 133
107/* 134/*
@@ -318,13 +345,16 @@ static struct dmi_system_id __devinitdata pciprobe_dmi_table[] = {
318 {} 345 {}
319}; 346};
320 347
348void __init dmi_check_pciprobe(void)
349{
350 dmi_check_system(pciprobe_dmi_table);
351}
352
321struct pci_bus * __devinit pcibios_scan_root(int busnum) 353struct pci_bus * __devinit pcibios_scan_root(int busnum)
322{ 354{
323 struct pci_bus *bus = NULL; 355 struct pci_bus *bus = NULL;
324 struct pci_sysdata *sd; 356 struct pci_sysdata *sd;
325 357
326 dmi_check_system(pciprobe_dmi_table);
327
328 while ((bus = pci_find_next_bus(bus)) != NULL) { 358 while ((bus = pci_find_next_bus(bus)) != NULL) {
329 if (bus->number == busnum) { 359 if (bus->number == busnum) {
330 /* Already scanned */ 360 /* Already scanned */
@@ -462,6 +492,9 @@ char * __devinit pcibios_setup(char *str)
462 } else if (!strcmp(str, "routeirq")) { 492 } else if (!strcmp(str, "routeirq")) {
463 pci_routeirq = 1; 493 pci_routeirq = 1;
464 return NULL; 494 return NULL;
495 } else if (!strcmp(str, "skip_isa_align")) {
496 pci_probe |= PCI_CAN_SKIP_ISA_ALIGN;
497 return NULL;
465 } 498 }
466 return str; 499 return str;
467} 500}
@@ -489,7 +522,7 @@ void pcibios_disable_device (struct pci_dev *dev)
489 pcibios_disable_irq(dev); 522 pcibios_disable_irq(dev);
490} 523}
491 524
492struct pci_bus *pci_scan_bus_on_node(int busno, struct pci_ops *ops, int node) 525struct pci_bus * __devinit pci_scan_bus_on_node(int busno, struct pci_ops *ops, int node)
493{ 526{
494 struct pci_bus *bus = NULL; 527 struct pci_bus *bus = NULL;
495 struct pci_sysdata *sd; 528 struct pci_sysdata *sd;
@@ -512,7 +545,7 @@ struct pci_bus *pci_scan_bus_on_node(int busno, struct pci_ops *ops, int node)
512 return bus; 545 return bus;
513} 546}
514 547
515struct pci_bus *pci_scan_bus_with_sysdata(int busno) 548struct pci_bus * __devinit pci_scan_bus_with_sysdata(int busno)
516{ 549{
517 return pci_scan_bus_on_node(busno, &pci_root_ops, -1); 550 return pci_scan_bus_on_node(busno, &pci_root_ops, -1);
518} 551}
diff --git a/arch/x86/pci/fixup.c b/arch/x86/pci/fixup.c
index b60b2abd480c..ff3a6a336342 100644
--- a/arch/x86/pci/fixup.c
+++ b/arch/x86/pci/fixup.c
@@ -502,7 +502,7 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_SIEMENS, 0x0015,
502 */ 502 */
503static void fam10h_pci_cfg_space_size(struct pci_dev *dev) 503static void fam10h_pci_cfg_space_size(struct pci_dev *dev)
504{ 504{
505 dev->cfg_size = pci_cfg_space_size_ext(dev, 0); 505 dev->cfg_size = pci_cfg_space_size_ext(dev);
506} 506}
507 507
508DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1200, fam10h_pci_cfg_space_size); 508DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_AMD, 0x1200, fam10h_pci_cfg_space_size);
diff --git a/arch/x86/pci/init.c b/arch/x86/pci/init.c
index dd30c6076b5d..e70b9c57b88e 100644
--- a/arch/x86/pci/init.c
+++ b/arch/x86/pci/init.c
@@ -33,6 +33,10 @@ static __init int pci_access_init(void)
33 printk(KERN_ERR 33 printk(KERN_ERR
34 "PCI: Fatal: No config space access function found\n"); 34 "PCI: Fatal: No config space access function found\n");
35 35
36 dmi_check_pciprobe();
37
38 dmi_check_skip_isa_align();
39
36 return 0; 40 return 0;
37} 41}
38arch_initcall(pci_access_init); 42arch_initcall(pci_access_init);
diff --git a/arch/x86/pci/pci.h b/arch/x86/pci/pci.h
index c58805a92db5..f3972b12c60a 100644
--- a/arch/x86/pci/pci.h
+++ b/arch/x86/pci/pci.h
@@ -38,6 +38,9 @@ enum pci_bf_sort_state {
38 pci_dmi_bf, 38 pci_dmi_bf,
39}; 39};
40 40
41extern void __init dmi_check_pciprobe(void);
42extern void __init dmi_check_skip_isa_align(void);
43
41/* pci-i386.c */ 44/* pci-i386.c */
42 45
43extern unsigned int pcibios_max_latency; 46extern unsigned int pcibios_max_latency;
diff --git a/arch/x86/vdso/vdso32-setup.c b/arch/x86/vdso/vdso32-setup.c
index 4dceeb1fc5e0..cf058fecfcee 100644
--- a/arch/x86/vdso/vdso32-setup.c
+++ b/arch/x86/vdso/vdso32-setup.c
@@ -162,7 +162,7 @@ static __init void relocate_vdso(Elf32_Ehdr *ehdr)
162 Elf32_Shdr *shdr; 162 Elf32_Shdr *shdr;
163 int i; 163 int i;
164 164
165 BUG_ON(memcmp(ehdr->e_ident, ELFMAG, 4) != 0 || 165 BUG_ON(memcmp(ehdr->e_ident, ELFMAG, SELFMAG) != 0 ||
166 !elf_check_arch_ia32(ehdr) || 166 !elf_check_arch_ia32(ehdr) ||
167 ehdr->e_type != ET_DYN); 167 ehdr->e_type != ET_DYN);
168 168
diff --git a/arch/x86/video/fbdev.c b/arch/x86/video/fbdev.c
index 4db42bff8c60..69527688f794 100644
--- a/arch/x86/video/fbdev.c
+++ b/arch/x86/video/fbdev.c
@@ -1,5 +1,4 @@
1/* 1/*
2 *
3 * Copyright (C) 2007 Antonino Daplas <adaplas@gmail.com> 2 * Copyright (C) 2007 Antonino Daplas <adaplas@gmail.com>
4 * 3 *
5 * This file is subject to the terms and conditions of the GNU General Public 4 * This file is subject to the terms and conditions of the GNU General Public
@@ -29,3 +28,4 @@ int fb_is_primary_device(struct fb_info *info)
29 return retval; 28 return retval;
30} 29}
31EXPORT_SYMBOL(fb_is_primary_device); 30EXPORT_SYMBOL(fb_is_primary_device);
31MODULE_LICENSE("GPL");
generated by cgit v1.2.2 (git 2.25.0) at 2024-11-19 02:29:42 -0500