aboutsummaryrefslogtreecommitdiffstats
path: root/arch/powerpc
diff options
context:
space:
mode:
Diffstat (limited to 'arch/powerpc')
-rw-r--r--arch/powerpc/Kconfig1
-rw-r--r--arch/powerpc/Kconfig.debug21
-rw-r--r--arch/powerpc/include/asm/kvm.h184
-rw-r--r--arch/powerpc/include/asm/kvm_44x.h1
-rw-r--r--arch/powerpc/include/asm/kvm_e500.h2
-rw-r--r--arch/powerpc/include/asm/kvm_host.h5
-rw-r--r--arch/powerpc/include/asm/kvm_ppc.h9
-rw-r--r--arch/powerpc/include/asm/pgalloc.h21
-rw-r--r--arch/powerpc/include/asm/pte-hash64-64k.h2
-rw-r--r--arch/powerpc/include/asm/thread_info.h2
-rw-r--r--arch/powerpc/kernel/asm-offsets.c1
-rw-r--r--arch/powerpc/kernel/kgdb.c2
-rw-r--r--arch/powerpc/kernel/process.c23
-rw-r--r--arch/powerpc/kernel/vmlinux.lds.S2
-rw-r--r--arch/powerpc/kvm/44x.c10
-rw-r--r--arch/powerpc/kvm/44x_emulate.c2
-rw-r--r--arch/powerpc/kvm/booke.c154
-rw-r--r--arch/powerpc/kvm/booke_interrupts.S1
-rw-r--r--arch/powerpc/kvm/e500.c76
-rw-r--r--arch/powerpc/kvm/e500_emulate.c7
-rw-r--r--arch/powerpc/kvm/e500_tlb.c13
-rw-r--r--arch/powerpc/kvm/emulate.c15
-rw-r--r--arch/powerpc/kvm/powerpc.c21
-rw-r--r--arch/powerpc/kvm/timing.c31
-rw-r--r--arch/powerpc/mm/pgtable.c104
-rw-r--r--arch/powerpc/mm/tlb_hash32.c3
-rw-r--r--arch/powerpc/mm/tlb_hash64.c5
-rw-r--r--arch/powerpc/mm/tlb_nohash.c3
28 files changed, 558 insertions, 163 deletions
diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index a3128ca0fe11..423145a6f7ba 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -140,6 +140,7 @@ config PPC
140 select IRQ_PER_CPU 140 select IRQ_PER_CPU
141 select GENERIC_IRQ_SHOW 141 select GENERIC_IRQ_SHOW
142 select GENERIC_IRQ_SHOW_LEVEL 142 select GENERIC_IRQ_SHOW_LEVEL
143 select HAVE_RCU_TABLE_FREE if SMP
143 144
144config EARLY_PRINTK 145config EARLY_PRINTK
145 bool 146 bool
diff --git a/arch/powerpc/Kconfig.debug b/arch/powerpc/Kconfig.debug
index a597dd77b903..e72dcf6a421d 100644
--- a/arch/powerpc/Kconfig.debug
+++ b/arch/powerpc/Kconfig.debug
@@ -35,27 +35,6 @@ config DEBUG_STACKOVERFLOW
35 This option will cause messages to be printed if free stack space 35 This option will cause messages to be printed if free stack space
36 drops below a certain limit. 36 drops below a certain limit.
37 37
38config DEBUG_STACK_USAGE
39 bool "Stack utilization instrumentation"
40 depends on DEBUG_KERNEL
41 help
42 Enables the display of the minimum amount of free stack which each
43 task has ever had available in the sysrq-T and sysrq-P debug output.
44
45 This option will slow down process creation somewhat.
46
47config DEBUG_PER_CPU_MAPS
48 bool "Debug access to per_cpu maps"
49 depends on DEBUG_KERNEL
50 depends on SMP
51 default n
52 ---help---
53 Say Y to verify that the per_cpu map being accessed has
54 been setup. Adds a fair amount of code to kernel memory
55 and decreases performance.
56
57 Say N if unsure.
58
59config HCALL_STATS 38config HCALL_STATS
60 bool "Hypervisor call instrumentation" 39 bool "Hypervisor call instrumentation"
61 depends on PPC_PSERIES && DEBUG_FS && TRACEPOINTS 40 depends on PPC_PSERIES && DEBUG_FS && TRACEPOINTS
diff --git a/arch/powerpc/include/asm/kvm.h b/arch/powerpc/include/asm/kvm.h
index 18ea6963ad77..d2ca5ed3877b 100644
--- a/arch/powerpc/include/asm/kvm.h
+++ b/arch/powerpc/include/asm/kvm.h
@@ -45,6 +45,114 @@ struct kvm_regs {
45 __u64 gpr[32]; 45 __u64 gpr[32];
46}; 46};
47 47
48#define KVM_SREGS_E_IMPL_NONE 0
49#define KVM_SREGS_E_IMPL_FSL 1
50
51#define KVM_SREGS_E_FSL_PIDn (1 << 0) /* PID1/PID2 */
52
53/*
54 * Feature bits indicate which sections of the sregs struct are valid,
55 * both in KVM_GET_SREGS and KVM_SET_SREGS. On KVM_SET_SREGS, registers
56 * corresponding to unset feature bits will not be modified. This allows
57 * restoring a checkpoint made without that feature, while keeping the
58 * default values of the new registers.
59 *
60 * KVM_SREGS_E_BASE contains:
61 * CSRR0/1 (refers to SRR2/3 on 40x)
62 * ESR
63 * DEAR
64 * MCSR
65 * TSR
66 * TCR
67 * DEC
68 * TB
69 * VRSAVE (USPRG0)
70 */
71#define KVM_SREGS_E_BASE (1 << 0)
72
73/*
74 * KVM_SREGS_E_ARCH206 contains:
75 *
76 * PIR
77 * MCSRR0/1
78 * DECAR
79 * IVPR
80 */
81#define KVM_SREGS_E_ARCH206 (1 << 1)
82
83/*
84 * Contains EPCR, plus the upper half of 64-bit registers
85 * that are 32-bit on 32-bit implementations.
86 */
87#define KVM_SREGS_E_64 (1 << 2)
88
89#define KVM_SREGS_E_SPRG8 (1 << 3)
90#define KVM_SREGS_E_MCIVPR (1 << 4)
91
92/*
93 * IVORs are used -- contains IVOR0-15, plus additional IVORs
94 * in combination with an appropriate feature bit.
95 */
96#define KVM_SREGS_E_IVOR (1 << 5)
97
98/*
99 * Contains MAS0-4, MAS6-7, TLBnCFG, MMUCFG.
100 * Also TLBnPS if MMUCFG[MAVN] = 1.
101 */
102#define KVM_SREGS_E_ARCH206_MMU (1 << 6)
103
104/* DBSR, DBCR, IAC, DAC, DVC */
105#define KVM_SREGS_E_DEBUG (1 << 7)
106
107/* Enhanced debug -- DSRR0/1, SPRG9 */
108#define KVM_SREGS_E_ED (1 << 8)
109
110/* Embedded Floating Point (SPE) -- IVOR32-34 if KVM_SREGS_E_IVOR */
111#define KVM_SREGS_E_SPE (1 << 9)
112
113/* External Proxy (EXP) -- EPR */
114#define KVM_SREGS_EXP (1 << 10)
115
116/* External PID (E.PD) -- EPSC/EPLC */
117#define KVM_SREGS_E_PD (1 << 11)
118
119/* Processor Control (E.PC) -- IVOR36-37 if KVM_SREGS_E_IVOR */
120#define KVM_SREGS_E_PC (1 << 12)
121
122/* Page table (E.PT) -- EPTCFG */
123#define KVM_SREGS_E_PT (1 << 13)
124
125/* Embedded Performance Monitor (E.PM) -- IVOR35 if KVM_SREGS_E_IVOR */
126#define KVM_SREGS_E_PM (1 << 14)
127
128/*
129 * Special updates:
130 *
131 * Some registers may change even while a vcpu is not running.
132 * To avoid losing these changes, by default these registers are
133 * not updated by KVM_SET_SREGS. To force an update, set the bit
134 * in u.e.update_special corresponding to the register to be updated.
135 *
136 * The update_special field is zero on return from KVM_GET_SREGS.
137 *
138 * When restoring a checkpoint, the caller can set update_special
139 * to 0xffffffff to ensure that everything is restored, even new features
140 * that the caller doesn't know about.
141 */
142#define KVM_SREGS_E_UPDATE_MCSR (1 << 0)
143#define KVM_SREGS_E_UPDATE_TSR (1 << 1)
144#define KVM_SREGS_E_UPDATE_DEC (1 << 2)
145#define KVM_SREGS_E_UPDATE_DBSR (1 << 3)
146
147/*
148 * In KVM_SET_SREGS, reserved/pad fields must be left untouched from a
149 * previous KVM_GET_REGS.
150 *
151 * Unless otherwise indicated, setting any register with KVM_SET_SREGS
152 * directly sets its value. It does not trigger any special semantics such
153 * as write-one-to-clear. Calling KVM_SET_SREGS on an unmodified struct
154 * just received from KVM_GET_SREGS is always a no-op.
155 */
48struct kvm_sregs { 156struct kvm_sregs {
49 __u32 pvr; 157 __u32 pvr;
50 union { 158 union {
@@ -62,6 +170,82 @@ struct kvm_sregs {
62 __u64 dbat[8]; 170 __u64 dbat[8];
63 } ppc32; 171 } ppc32;
64 } s; 172 } s;
173 struct {
174 union {
175 struct { /* KVM_SREGS_E_IMPL_FSL */
176 __u32 features; /* KVM_SREGS_E_FSL_ */
177 __u32 svr;
178 __u64 mcar;
179 __u32 hid0;
180
181 /* KVM_SREGS_E_FSL_PIDn */
182 __u32 pid1, pid2;
183 } fsl;
184 __u8 pad[256];
185 } impl;
186
187 __u32 features; /* KVM_SREGS_E_ */
188 __u32 impl_id; /* KVM_SREGS_E_IMPL_ */
189 __u32 update_special; /* KVM_SREGS_E_UPDATE_ */
190 __u32 pir; /* read-only */
191 __u64 sprg8;
192 __u64 sprg9; /* E.ED */
193 __u64 csrr0;
194 __u64 dsrr0; /* E.ED */
195 __u64 mcsrr0;
196 __u32 csrr1;
197 __u32 dsrr1; /* E.ED */
198 __u32 mcsrr1;
199 __u32 esr;
200 __u64 dear;
201 __u64 ivpr;
202 __u64 mcivpr;
203 __u64 mcsr; /* KVM_SREGS_E_UPDATE_MCSR */
204
205 __u32 tsr; /* KVM_SREGS_E_UPDATE_TSR */
206 __u32 tcr;
207 __u32 decar;
208 __u32 dec; /* KVM_SREGS_E_UPDATE_DEC */
209
210 /*
211 * Userspace can read TB directly, but the
212 * value reported here is consistent with "dec".
213 *
214 * Read-only.
215 */
216 __u64 tb;
217
218 __u32 dbsr; /* KVM_SREGS_E_UPDATE_DBSR */
219 __u32 dbcr[3];
220 __u32 iac[4];
221 __u32 dac[2];
222 __u32 dvc[2];
223 __u8 num_iac; /* read-only */
224 __u8 num_dac; /* read-only */
225 __u8 num_dvc; /* read-only */
226 __u8 pad;
227
228 __u32 epr; /* EXP */
229 __u32 vrsave; /* a.k.a. USPRG0 */
230 __u32 epcr; /* KVM_SREGS_E_64 */
231
232 __u32 mas0;
233 __u32 mas1;
234 __u64 mas2;
235 __u64 mas7_3;
236 __u32 mas4;
237 __u32 mas6;
238
239 __u32 ivor_low[16]; /* IVOR0-15 */
240 __u32 ivor_high[18]; /* IVOR32+, plus room to expand */
241
242 __u32 mmucfg; /* read-only */
243 __u32 eptcfg; /* E.PT, read-only */
244 __u32 tlbcfg[4];/* read-only */
245 __u32 tlbps[4]; /* read-only */
246
247 __u32 eplc, epsc; /* E.PD */
248 } e;
65 __u8 pad[1020]; 249 __u8 pad[1020];
66 } u; 250 } u;
67}; 251};
diff --git a/arch/powerpc/include/asm/kvm_44x.h b/arch/powerpc/include/asm/kvm_44x.h
index d22d39942a92..a0e57618ff33 100644
--- a/arch/powerpc/include/asm/kvm_44x.h
+++ b/arch/powerpc/include/asm/kvm_44x.h
@@ -61,7 +61,6 @@ static inline struct kvmppc_vcpu_44x *to_44x(struct kvm_vcpu *vcpu)
61 return container_of(vcpu, struct kvmppc_vcpu_44x, vcpu); 61 return container_of(vcpu, struct kvmppc_vcpu_44x, vcpu);
62} 62}
63 63
64void kvmppc_set_pid(struct kvm_vcpu *vcpu, u32 new_pid);
65void kvmppc_44x_tlb_put(struct kvm_vcpu *vcpu); 64void kvmppc_44x_tlb_put(struct kvm_vcpu *vcpu);
66void kvmppc_44x_tlb_load(struct kvm_vcpu *vcpu); 65void kvmppc_44x_tlb_load(struct kvm_vcpu *vcpu);
67 66
diff --git a/arch/powerpc/include/asm/kvm_e500.h b/arch/powerpc/include/asm/kvm_e500.h
index 7fea26fffb25..7a2a565f88c4 100644
--- a/arch/powerpc/include/asm/kvm_e500.h
+++ b/arch/powerpc/include/asm/kvm_e500.h
@@ -43,6 +43,7 @@ struct kvmppc_vcpu_e500 {
43 43
44 u32 host_pid[E500_PID_NUM]; 44 u32 host_pid[E500_PID_NUM];
45 u32 pid[E500_PID_NUM]; 45 u32 pid[E500_PID_NUM];
46 u32 svr;
46 47
47 u32 mas0; 48 u32 mas0;
48 u32 mas1; 49 u32 mas1;
@@ -58,6 +59,7 @@ struct kvmppc_vcpu_e500 {
58 u32 hid1; 59 u32 hid1;
59 u32 tlb0cfg; 60 u32 tlb0cfg;
60 u32 tlb1cfg; 61 u32 tlb1cfg;
62 u64 mcar;
61 63
62 struct kvm_vcpu vcpu; 64 struct kvm_vcpu vcpu;
63}; 65};
diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index bba3b9b72a39..186f150b9b89 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h
@@ -223,6 +223,7 @@ struct kvm_vcpu_arch {
223 ulong hflags; 223 ulong hflags;
224 ulong guest_owned_ext; 224 ulong guest_owned_ext;
225#endif 225#endif
226 u32 vrsave; /* also USPRG0 */
226 u32 mmucr; 227 u32 mmucr;
227 ulong sprg4; 228 ulong sprg4;
228 ulong sprg5; 229 ulong sprg5;
@@ -232,6 +233,9 @@ struct kvm_vcpu_arch {
232 ulong csrr1; 233 ulong csrr1;
233 ulong dsrr0; 234 ulong dsrr0;
234 ulong dsrr1; 235 ulong dsrr1;
236 ulong mcsrr0;
237 ulong mcsrr1;
238 ulong mcsr;
235 ulong esr; 239 ulong esr;
236 u32 dec; 240 u32 dec;
237 u32 decar; 241 u32 decar;
@@ -255,6 +259,7 @@ struct kvm_vcpu_arch {
255 u32 dbsr; 259 u32 dbsr;
256 260
257#ifdef CONFIG_KVM_EXIT_TIMING 261#ifdef CONFIG_KVM_EXIT_TIMING
262 struct mutex exit_timing_lock;
258 struct kvmppc_exit_timing timing_exit; 263 struct kvmppc_exit_timing timing_exit;
259 struct kvmppc_exit_timing timing_last_enter; 264 struct kvmppc_exit_timing timing_last_enter;
260 u32 last_exit_type; 265 u32 last_exit_type;
diff --git a/arch/powerpc/include/asm/kvm_ppc.h b/arch/powerpc/include/asm/kvm_ppc.h
index ecb3bc74c344..9345238edecf 100644
--- a/arch/powerpc/include/asm/kvm_ppc.h
+++ b/arch/powerpc/include/asm/kvm_ppc.h
@@ -61,6 +61,7 @@ extern int kvmppc_emulate_instruction(struct kvm_run *run,
61 struct kvm_vcpu *vcpu); 61 struct kvm_vcpu *vcpu);
62extern int kvmppc_emulate_mmio(struct kvm_run *run, struct kvm_vcpu *vcpu); 62extern int kvmppc_emulate_mmio(struct kvm_run *run, struct kvm_vcpu *vcpu);
63extern void kvmppc_emulate_dec(struct kvm_vcpu *vcpu); 63extern void kvmppc_emulate_dec(struct kvm_vcpu *vcpu);
64extern u32 kvmppc_get_dec(struct kvm_vcpu *vcpu, u64 tb);
64 65
65/* Core-specific hooks */ 66/* Core-specific hooks */
66 67
@@ -142,4 +143,12 @@ static inline u32 kvmppc_set_field(u64 inst, int msb, int lsb, int value)
142 return r; 143 return r;
143} 144}
144 145
146void kvmppc_core_get_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs);
147int kvmppc_core_set_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs);
148
149void kvmppc_get_sregs_ivor(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs);
150int kvmppc_set_sregs_ivor(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs);
151
152void kvmppc_set_pid(struct kvm_vcpu *vcpu, u32 pid);
153
145#endif /* __POWERPC_KVM_PPC_H__ */ 154#endif /* __POWERPC_KVM_PPC_H__ */
diff --git a/arch/powerpc/include/asm/pgalloc.h b/arch/powerpc/include/asm/pgalloc.h
index abe8532bd14e..bf301ac62f35 100644
--- a/arch/powerpc/include/asm/pgalloc.h
+++ b/arch/powerpc/include/asm/pgalloc.h
@@ -31,14 +31,29 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t ptepage)
31#endif 31#endif
32 32
33#ifdef CONFIG_SMP 33#ifdef CONFIG_SMP
34extern void pgtable_free_tlb(struct mmu_gather *tlb, void *table, unsigned shift); 34struct mmu_gather;
35extern void pte_free_finish(void); 35extern void tlb_remove_table(struct mmu_gather *, void *);
36
37static inline void pgtable_free_tlb(struct mmu_gather *tlb, void *table, int shift)
38{
39 unsigned long pgf = (unsigned long)table;
40 BUG_ON(shift > MAX_PGTABLE_INDEX_SIZE);
41 pgf |= shift;
42 tlb_remove_table(tlb, (void *)pgf);
43}
44
45static inline void __tlb_remove_table(void *_table)
46{
47 void *table = (void *)((unsigned long)_table & ~MAX_PGTABLE_INDEX_SIZE);
48 unsigned shift = (unsigned long)_table & MAX_PGTABLE_INDEX_SIZE;
49
50 pgtable_free(table, shift);
51}
36#else /* CONFIG_SMP */ 52#else /* CONFIG_SMP */
37static inline void pgtable_free_tlb(struct mmu_gather *tlb, void *table, unsigned shift) 53static inline void pgtable_free_tlb(struct mmu_gather *tlb, void *table, unsigned shift)
38{ 54{
39 pgtable_free(table, shift); 55 pgtable_free(table, shift);
40} 56}
41static inline void pte_free_finish(void) { }
42#endif /* !CONFIG_SMP */ 57#endif /* !CONFIG_SMP */
43 58
44static inline void __pte_free_tlb(struct mmu_gather *tlb, struct page *ptepage, 59static inline void __pte_free_tlb(struct mmu_gather *tlb, struct page *ptepage,
diff --git a/arch/powerpc/include/asm/pte-hash64-64k.h b/arch/powerpc/include/asm/pte-hash64-64k.h
index c4490f9c67c4..59247e816ac5 100644
--- a/arch/powerpc/include/asm/pte-hash64-64k.h
+++ b/arch/powerpc/include/asm/pte-hash64-64k.h
@@ -22,7 +22,7 @@
22#define _PAGE_HASHPTE _PAGE_HPTE_SUB 22#define _PAGE_HASHPTE _PAGE_HPTE_SUB
23 23
24/* Note the full page bits must be in the same location as for normal 24/* Note the full page bits must be in the same location as for normal
25 * 4k pages as the same asssembly will be used to insert 64K pages 25 * 4k pages as the same assembly will be used to insert 64K pages
26 * wether the kernel has CONFIG_PPC_64K_PAGES or not 26 * wether the kernel has CONFIG_PPC_64K_PAGES or not
27 */ 27 */
28#define _PAGE_F_SECOND 0x00008000 /* full page: hidx bits */ 28#define _PAGE_F_SECOND 0x00008000 /* full page: hidx bits */
diff --git a/arch/powerpc/include/asm/thread_info.h b/arch/powerpc/include/asm/thread_info.h
index d8529ef13b23..37c353e8af7c 100644
--- a/arch/powerpc/include/asm/thread_info.h
+++ b/arch/powerpc/include/asm/thread_info.h
@@ -139,10 +139,12 @@ static inline struct thread_info *current_thread_info(void)
139#define TLF_NAPPING 0 /* idle thread enabled NAP mode */ 139#define TLF_NAPPING 0 /* idle thread enabled NAP mode */
140#define TLF_SLEEPING 1 /* suspend code enabled SLEEP mode */ 140#define TLF_SLEEPING 1 /* suspend code enabled SLEEP mode */
141#define TLF_RESTORE_SIGMASK 2 /* Restore signal mask in do_signal */ 141#define TLF_RESTORE_SIGMASK 2 /* Restore signal mask in do_signal */
142#define TLF_LAZY_MMU 3 /* tlb_batch is active */
142 143
143#define _TLF_NAPPING (1 << TLF_NAPPING) 144#define _TLF_NAPPING (1 << TLF_NAPPING)
144#define _TLF_SLEEPING (1 << TLF_SLEEPING) 145#define _TLF_SLEEPING (1 << TLF_SLEEPING)
145#define _TLF_RESTORE_SIGMASK (1 << TLF_RESTORE_SIGMASK) 146#define _TLF_RESTORE_SIGMASK (1 << TLF_RESTORE_SIGMASK)
147#define _TLF_LAZY_MMU (1 << TLF_LAZY_MMU)
146 148
147#ifndef __ASSEMBLY__ 149#ifndef __ASSEMBLY__
148#define HAVE_SET_RESTORE_SIGMASK 1 150#define HAVE_SET_RESTORE_SIGMASK 1
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index 6887661ac072..36e1c8a29be8 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -396,6 +396,7 @@ int main(void)
396 DEFINE(VCPU_HOST_STACK, offsetof(struct kvm_vcpu, arch.host_stack)); 396 DEFINE(VCPU_HOST_STACK, offsetof(struct kvm_vcpu, arch.host_stack));
397 DEFINE(VCPU_HOST_PID, offsetof(struct kvm_vcpu, arch.host_pid)); 397 DEFINE(VCPU_HOST_PID, offsetof(struct kvm_vcpu, arch.host_pid));
398 DEFINE(VCPU_GPRS, offsetof(struct kvm_vcpu, arch.gpr)); 398 DEFINE(VCPU_GPRS, offsetof(struct kvm_vcpu, arch.gpr));
399 DEFINE(VCPU_VRSAVE, offsetof(struct kvm_vcpu, arch.vrsave));
399 DEFINE(VCPU_SPRG4, offsetof(struct kvm_vcpu, arch.sprg4)); 400 DEFINE(VCPU_SPRG4, offsetof(struct kvm_vcpu, arch.sprg4));
400 DEFINE(VCPU_SPRG5, offsetof(struct kvm_vcpu, arch.sprg5)); 401 DEFINE(VCPU_SPRG5, offsetof(struct kvm_vcpu, arch.sprg5));
401 DEFINE(VCPU_SPRG6, offsetof(struct kvm_vcpu, arch.sprg6)); 402 DEFINE(VCPU_SPRG6, offsetof(struct kvm_vcpu, arch.sprg6));
diff --git a/arch/powerpc/kernel/kgdb.c b/arch/powerpc/kernel/kgdb.c
index bd9d35f59cf4..76a6e40a6f7c 100644
--- a/arch/powerpc/kernel/kgdb.c
+++ b/arch/powerpc/kernel/kgdb.c
@@ -142,7 +142,7 @@ static int kgdb_singlestep(struct pt_regs *regs)
142 return 0; 142 return 0;
143 143
144 /* 144 /*
145 * On Book E and perhaps other processsors, singlestep is handled on 145 * On Book E and perhaps other processors, singlestep is handled on
146 * the critical exception stack. This causes current_thread_info() 146 * the critical exception stack. This causes current_thread_info()
147 * to fail, since it it locates the thread_info by masking off 147 * to fail, since it it locates the thread_info by masking off
148 * the low bits of the current stack pointer. We work around 148 * the low bits of the current stack pointer. We work around
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 095043d79946..91e52df3d81d 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -395,6 +395,9 @@ struct task_struct *__switch_to(struct task_struct *prev,
395 struct thread_struct *new_thread, *old_thread; 395 struct thread_struct *new_thread, *old_thread;
396 unsigned long flags; 396 unsigned long flags;
397 struct task_struct *last; 397 struct task_struct *last;
398#ifdef CONFIG_PPC_BOOK3S_64
399 struct ppc64_tlb_batch *batch;
400#endif
398 401
399#ifdef CONFIG_SMP 402#ifdef CONFIG_SMP
400 /* avoid complexity of lazy save/restore of fpu 403 /* avoid complexity of lazy save/restore of fpu
@@ -513,7 +516,17 @@ struct task_struct *__switch_to(struct task_struct *prev,
513 old_thread->accum_tb += (current_tb - start_tb); 516 old_thread->accum_tb += (current_tb - start_tb);
514 new_thread->start_tb = current_tb; 517 new_thread->start_tb = current_tb;
515 } 518 }
516#endif 519#endif /* CONFIG_PPC64 */
520
521#ifdef CONFIG_PPC_BOOK3S_64
522 batch = &__get_cpu_var(ppc64_tlb_batch);
523 if (batch->active) {
524 current_thread_info()->local_flags |= _TLF_LAZY_MMU;
525 if (batch->index)
526 __flush_tlb_pending(batch);
527 batch->active = 0;
528 }
529#endif /* CONFIG_PPC_BOOK3S_64 */
517 530
518 local_irq_save(flags); 531 local_irq_save(flags);
519 532
@@ -528,6 +541,14 @@ struct task_struct *__switch_to(struct task_struct *prev,
528 hard_irq_disable(); 541 hard_irq_disable();
529 last = _switch(old_thread, new_thread); 542 last = _switch(old_thread, new_thread);
530 543
544#ifdef CONFIG_PPC_BOOK3S_64
545 if (current_thread_info()->local_flags & _TLF_LAZY_MMU) {
546 current_thread_info()->local_flags &= ~_TLF_LAZY_MMU;
547 batch = &__get_cpu_var(ppc64_tlb_batch);
548 batch->active = 1;
549 }
550#endif /* CONFIG_PPC_BOOK3S_64 */
551
531 local_irq_restore(flags); 552 local_irq_restore(flags);
532 553
533 return last; 554 return last;
diff --git a/arch/powerpc/kernel/vmlinux.lds.S b/arch/powerpc/kernel/vmlinux.lds.S
index b9150f07d266..920276c0f6a1 100644
--- a/arch/powerpc/kernel/vmlinux.lds.S
+++ b/arch/powerpc/kernel/vmlinux.lds.S
@@ -160,7 +160,7 @@ SECTIONS
160 INIT_RAM_FS 160 INIT_RAM_FS
161 } 161 }
162 162
163 PERCPU(L1_CACHE_BYTES, PAGE_SIZE) 163 PERCPU_SECTION(L1_CACHE_BYTES)
164 164
165 . = ALIGN(8); 165 . = ALIGN(8);
166 .machine.desc : AT(ADDR(.machine.desc) - LOAD_OFFSET) { 166 .machine.desc : AT(ADDR(.machine.desc) - LOAD_OFFSET) {
diff --git a/arch/powerpc/kvm/44x.c b/arch/powerpc/kvm/44x.c
index 74d0e7421143..da3a1225c0ac 100644
--- a/arch/powerpc/kvm/44x.c
+++ b/arch/powerpc/kvm/44x.c
@@ -107,6 +107,16 @@ int kvmppc_core_vcpu_translate(struct kvm_vcpu *vcpu,
107 return 0; 107 return 0;
108} 108}
109 109
110void kvmppc_core_get_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
111{
112 kvmppc_get_sregs_ivor(vcpu, sregs);
113}
114
115int kvmppc_core_set_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
116{
117 return kvmppc_set_sregs_ivor(vcpu, sregs);
118}
119
110struct kvm_vcpu *kvmppc_core_vcpu_create(struct kvm *kvm, unsigned int id) 120struct kvm_vcpu *kvmppc_core_vcpu_create(struct kvm *kvm, unsigned int id)
111{ 121{
112 struct kvmppc_vcpu_44x *vcpu_44x; 122 struct kvmppc_vcpu_44x *vcpu_44x;
diff --git a/arch/powerpc/kvm/44x_emulate.c b/arch/powerpc/kvm/44x_emulate.c
index 65ea083a5b27..549bb2c9a47a 100644
--- a/arch/powerpc/kvm/44x_emulate.c
+++ b/arch/powerpc/kvm/44x_emulate.c
@@ -158,7 +158,6 @@ int kvmppc_core_emulate_mtspr(struct kvm_vcpu *vcpu, int sprn, int rs)
158 emulated = kvmppc_booke_emulate_mtspr(vcpu, sprn, rs); 158 emulated = kvmppc_booke_emulate_mtspr(vcpu, sprn, rs);
159 } 159 }
160 160
161 kvmppc_set_exit_type(vcpu, EMULATED_MTSPR_EXITS);
162 return emulated; 161 return emulated;
163} 162}
164 163
@@ -179,7 +178,6 @@ int kvmppc_core_emulate_mfspr(struct kvm_vcpu *vcpu, int sprn, int rt)
179 emulated = kvmppc_booke_emulate_mfspr(vcpu, sprn, rt); 178 emulated = kvmppc_booke_emulate_mfspr(vcpu, sprn, rt);
180 } 179 }
181 180
182 kvmppc_set_exit_type(vcpu, EMULATED_MFSPR_EXITS);
183 return emulated; 181 return emulated;
184} 182}
185 183
diff --git a/arch/powerpc/kvm/booke.c b/arch/powerpc/kvm/booke.c
index ef76acb455c3..8462b3a1c1c7 100644
--- a/arch/powerpc/kvm/booke.c
+++ b/arch/powerpc/kvm/booke.c
@@ -569,6 +569,7 @@ int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs)
569 kvmppc_set_msr(vcpu, regs->msr); 569 kvmppc_set_msr(vcpu, regs->msr);
570 vcpu->arch.shared->srr0 = regs->srr0; 570 vcpu->arch.shared->srr0 = regs->srr0;
571 vcpu->arch.shared->srr1 = regs->srr1; 571 vcpu->arch.shared->srr1 = regs->srr1;
572 kvmppc_set_pid(vcpu, regs->pid);
572 vcpu->arch.shared->sprg0 = regs->sprg0; 573 vcpu->arch.shared->sprg0 = regs->sprg0;
573 vcpu->arch.shared->sprg1 = regs->sprg1; 574 vcpu->arch.shared->sprg1 = regs->sprg1;
574 vcpu->arch.shared->sprg2 = regs->sprg2; 575 vcpu->arch.shared->sprg2 = regs->sprg2;
@@ -584,16 +585,165 @@ int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs)
584 return 0; 585 return 0;
585} 586}
586 587
588static void get_sregs_base(struct kvm_vcpu *vcpu,
589 struct kvm_sregs *sregs)
590{
591 u64 tb = get_tb();
592
593 sregs->u.e.features |= KVM_SREGS_E_BASE;
594
595 sregs->u.e.csrr0 = vcpu->arch.csrr0;
596 sregs->u.e.csrr1 = vcpu->arch.csrr1;
597 sregs->u.e.mcsr = vcpu->arch.mcsr;
598 sregs->u.e.esr = vcpu->arch.esr;
599 sregs->u.e.dear = vcpu->arch.shared->dar;
600 sregs->u.e.tsr = vcpu->arch.tsr;
601 sregs->u.e.tcr = vcpu->arch.tcr;
602 sregs->u.e.dec = kvmppc_get_dec(vcpu, tb);
603 sregs->u.e.tb = tb;
604 sregs->u.e.vrsave = vcpu->arch.vrsave;
605}
606
607static int set_sregs_base(struct kvm_vcpu *vcpu,
608 struct kvm_sregs *sregs)
609{
610 if (!(sregs->u.e.features & KVM_SREGS_E_BASE))
611 return 0;
612
613 vcpu->arch.csrr0 = sregs->u.e.csrr0;
614 vcpu->arch.csrr1 = sregs->u.e.csrr1;
615 vcpu->arch.mcsr = sregs->u.e.mcsr;
616 vcpu->arch.esr = sregs->u.e.esr;
617 vcpu->arch.shared->dar = sregs->u.e.dear;
618 vcpu->arch.vrsave = sregs->u.e.vrsave;
619 vcpu->arch.tcr = sregs->u.e.tcr;
620
621 if (sregs->u.e.update_special & KVM_SREGS_E_UPDATE_DEC)
622 vcpu->arch.dec = sregs->u.e.dec;
623
624 kvmppc_emulate_dec(vcpu);
625
626 if (sregs->u.e.update_special & KVM_SREGS_E_UPDATE_TSR) {
627 /*
628 * FIXME: existing KVM timer handling is incomplete.
629 * TSR cannot be read by the guest, and its value in
630 * vcpu->arch is always zero. For now, just handle
631 * the case where the caller is trying to inject a
632 * decrementer interrupt.
633 */
634
635 if ((sregs->u.e.tsr & TSR_DIS) &&
636 (vcpu->arch.tcr & TCR_DIE))
637 kvmppc_core_queue_dec(vcpu);
638 }
639
640 return 0;
641}
642
643static void get_sregs_arch206(struct kvm_vcpu *vcpu,
644 struct kvm_sregs *sregs)
645{
646 sregs->u.e.features |= KVM_SREGS_E_ARCH206;
647
648 sregs->u.e.pir = 0;
649 sregs->u.e.mcsrr0 = vcpu->arch.mcsrr0;
650 sregs->u.e.mcsrr1 = vcpu->arch.mcsrr1;
651 sregs->u.e.decar = vcpu->arch.decar;
652 sregs->u.e.ivpr = vcpu->arch.ivpr;
653}
654
655static int set_sregs_arch206(struct kvm_vcpu *vcpu,
656 struct kvm_sregs *sregs)
657{
658 if (!(sregs->u.e.features & KVM_SREGS_E_ARCH206))
659 return 0;
660
661 if (sregs->u.e.pir != 0)
662 return -EINVAL;
663
664 vcpu->arch.mcsrr0 = sregs->u.e.mcsrr0;
665 vcpu->arch.mcsrr1 = sregs->u.e.mcsrr1;
666 vcpu->arch.decar = sregs->u.e.decar;
667 vcpu->arch.ivpr = sregs->u.e.ivpr;
668
669 return 0;
670}
671
672void kvmppc_get_sregs_ivor(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
673{
674 sregs->u.e.features |= KVM_SREGS_E_IVOR;
675
676 sregs->u.e.ivor_low[0] = vcpu->arch.ivor[BOOKE_IRQPRIO_CRITICAL];
677 sregs->u.e.ivor_low[1] = vcpu->arch.ivor[BOOKE_IRQPRIO_MACHINE_CHECK];
678 sregs->u.e.ivor_low[2] = vcpu->arch.ivor[BOOKE_IRQPRIO_DATA_STORAGE];
679 sregs->u.e.ivor_low[3] = vcpu->arch.ivor[BOOKE_IRQPRIO_INST_STORAGE];
680 sregs->u.e.ivor_low[4] = vcpu->arch.ivor[BOOKE_IRQPRIO_EXTERNAL];
681 sregs->u.e.ivor_low[5] = vcpu->arch.ivor[BOOKE_IRQPRIO_ALIGNMENT];
682 sregs->u.e.ivor_low[6] = vcpu->arch.ivor[BOOKE_IRQPRIO_PROGRAM];
683 sregs->u.e.ivor_low[7] = vcpu->arch.ivor[BOOKE_IRQPRIO_FP_UNAVAIL];
684 sregs->u.e.ivor_low[8] = vcpu->arch.ivor[BOOKE_IRQPRIO_SYSCALL];
685 sregs->u.e.ivor_low[9] = vcpu->arch.ivor[BOOKE_IRQPRIO_AP_UNAVAIL];
686 sregs->u.e.ivor_low[10] = vcpu->arch.ivor[BOOKE_IRQPRIO_DECREMENTER];
687 sregs->u.e.ivor_low[11] = vcpu->arch.ivor[BOOKE_IRQPRIO_FIT];
688 sregs->u.e.ivor_low[12] = vcpu->arch.ivor[BOOKE_IRQPRIO_WATCHDOG];
689 sregs->u.e.ivor_low[13] = vcpu->arch.ivor[BOOKE_IRQPRIO_DTLB_MISS];
690 sregs->u.e.ivor_low[14] = vcpu->arch.ivor[BOOKE_IRQPRIO_ITLB_MISS];
691 sregs->u.e.ivor_low[15] = vcpu->arch.ivor[BOOKE_IRQPRIO_DEBUG];
692}
693
694int kvmppc_set_sregs_ivor(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
695{
696 if (!(sregs->u.e.features & KVM_SREGS_E_IVOR))
697 return 0;
698
699 vcpu->arch.ivor[BOOKE_IRQPRIO_CRITICAL] = sregs->u.e.ivor_low[0];
700 vcpu->arch.ivor[BOOKE_IRQPRIO_MACHINE_CHECK] = sregs->u.e.ivor_low[1];
701 vcpu->arch.ivor[BOOKE_IRQPRIO_DATA_STORAGE] = sregs->u.e.ivor_low[2];
702 vcpu->arch.ivor[BOOKE_IRQPRIO_INST_STORAGE] = sregs->u.e.ivor_low[3];
703 vcpu->arch.ivor[BOOKE_IRQPRIO_EXTERNAL] = sregs->u.e.ivor_low[4];
704 vcpu->arch.ivor[BOOKE_IRQPRIO_ALIGNMENT] = sregs->u.e.ivor_low[5];
705 vcpu->arch.ivor[BOOKE_IRQPRIO_PROGRAM] = sregs->u.e.ivor_low[6];
706 vcpu->arch.ivor[BOOKE_IRQPRIO_FP_UNAVAIL] = sregs->u.e.ivor_low[7];
707 vcpu->arch.ivor[BOOKE_IRQPRIO_SYSCALL] = sregs->u.e.ivor_low[8];
708 vcpu->arch.ivor[BOOKE_IRQPRIO_AP_UNAVAIL] = sregs->u.e.ivor_low[9];
709 vcpu->arch.ivor[BOOKE_IRQPRIO_DECREMENTER] = sregs->u.e.ivor_low[10];
710 vcpu->arch.ivor[BOOKE_IRQPRIO_FIT] = sregs->u.e.ivor_low[11];
711 vcpu->arch.ivor[BOOKE_IRQPRIO_WATCHDOG] = sregs->u.e.ivor_low[12];
712 vcpu->arch.ivor[BOOKE_IRQPRIO_DTLB_MISS] = sregs->u.e.ivor_low[13];
713 vcpu->arch.ivor[BOOKE_IRQPRIO_ITLB_MISS] = sregs->u.e.ivor_low[14];
714 vcpu->arch.ivor[BOOKE_IRQPRIO_DEBUG] = sregs->u.e.ivor_low[15];
715
716 return 0;
717}
718
587int kvm_arch_vcpu_ioctl_get_sregs(struct kvm_vcpu *vcpu, 719int kvm_arch_vcpu_ioctl_get_sregs(struct kvm_vcpu *vcpu,
588 struct kvm_sregs *sregs) 720 struct kvm_sregs *sregs)
589{ 721{
590 return -ENOTSUPP; 722 sregs->pvr = vcpu->arch.pvr;
723
724 get_sregs_base(vcpu, sregs);
725 get_sregs_arch206(vcpu, sregs);
726 kvmppc_core_get_sregs(vcpu, sregs);
727 return 0;
591} 728}
592 729
593int kvm_arch_vcpu_ioctl_set_sregs(struct kvm_vcpu *vcpu, 730int kvm_arch_vcpu_ioctl_set_sregs(struct kvm_vcpu *vcpu,
594 struct kvm_sregs *sregs) 731 struct kvm_sregs *sregs)
595{ 732{
596 return -ENOTSUPP; 733 int ret;
734
735 if (vcpu->arch.pvr != sregs->pvr)
736 return -EINVAL;
737
738 ret = set_sregs_base(vcpu, sregs);
739 if (ret < 0)
740 return ret;
741
742 ret = set_sregs_arch206(vcpu, sregs);
743 if (ret < 0)
744 return ret;
745
746 return kvmppc_core_set_sregs(vcpu, sregs);
597} 747}
598 748
599int kvm_arch_vcpu_ioctl_get_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu) 749int kvm_arch_vcpu_ioctl_get_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu)
diff --git a/arch/powerpc/kvm/booke_interrupts.S b/arch/powerpc/kvm/booke_interrupts.S
index 1cc471faac2d..b58ccae95904 100644
--- a/arch/powerpc/kvm/booke_interrupts.S
+++ b/arch/powerpc/kvm/booke_interrupts.S
@@ -380,7 +380,6 @@ lightweight_exit:
380 * because host interrupt handlers would get confused. */ 380 * because host interrupt handlers would get confused. */
381 lwz r1, VCPU_GPR(r1)(r4) 381 lwz r1, VCPU_GPR(r1)(r4)
382 382
383 /* XXX handle USPRG0 */
384 /* Host interrupt handlers may have clobbered these guest-readable 383 /* Host interrupt handlers may have clobbered these guest-readable
385 * SPRGs, so we need to reload them here with the guest's values. */ 384 * SPRGs, so we need to reload them here with the guest's values. */
386 lwz r3, VCPU_SPRG4(r4) 385 lwz r3, VCPU_SPRG4(r4)
diff --git a/arch/powerpc/kvm/e500.c b/arch/powerpc/kvm/e500.c
index e3768ee9b595..318dbc61ba44 100644
--- a/arch/powerpc/kvm/e500.c
+++ b/arch/powerpc/kvm/e500.c
@@ -63,6 +63,7 @@ int kvmppc_core_vcpu_setup(struct kvm_vcpu *vcpu)
63 63
64 /* Registers init */ 64 /* Registers init */
65 vcpu->arch.pvr = mfspr(SPRN_PVR); 65 vcpu->arch.pvr = mfspr(SPRN_PVR);
66 vcpu_e500->svr = mfspr(SPRN_SVR);
66 67
67 /* Since booke kvm only support one core, update all vcpus' PIR to 0 */ 68 /* Since booke kvm only support one core, update all vcpus' PIR to 0 */
68 vcpu->vcpu_id = 0; 69 vcpu->vcpu_id = 0;
@@ -96,6 +97,81 @@ int kvmppc_core_vcpu_translate(struct kvm_vcpu *vcpu,
96 return 0; 97 return 0;
97} 98}
98 99
100void kvmppc_core_get_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
101{
102 struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
103
104 sregs->u.e.features |= KVM_SREGS_E_ARCH206_MMU | KVM_SREGS_E_SPE |
105 KVM_SREGS_E_PM;
106 sregs->u.e.impl_id = KVM_SREGS_E_IMPL_FSL;
107
108 sregs->u.e.impl.fsl.features = 0;
109 sregs->u.e.impl.fsl.svr = vcpu_e500->svr;
110 sregs->u.e.impl.fsl.hid0 = vcpu_e500->hid0;
111 sregs->u.e.impl.fsl.mcar = vcpu_e500->mcar;
112
113 sregs->u.e.mas0 = vcpu_e500->mas0;
114 sregs->u.e.mas1 = vcpu_e500->mas1;
115 sregs->u.e.mas2 = vcpu_e500->mas2;
116 sregs->u.e.mas7_3 = ((u64)vcpu_e500->mas7 << 32) | vcpu_e500->mas3;
117 sregs->u.e.mas4 = vcpu_e500->mas4;
118 sregs->u.e.mas6 = vcpu_e500->mas6;
119
120 sregs->u.e.mmucfg = mfspr(SPRN_MMUCFG);
121 sregs->u.e.tlbcfg[0] = vcpu_e500->tlb0cfg;
122 sregs->u.e.tlbcfg[1] = vcpu_e500->tlb1cfg;
123 sregs->u.e.tlbcfg[2] = 0;
124 sregs->u.e.tlbcfg[3] = 0;
125
126 sregs->u.e.ivor_high[0] = vcpu->arch.ivor[BOOKE_IRQPRIO_SPE_UNAVAIL];
127 sregs->u.e.ivor_high[1] = vcpu->arch.ivor[BOOKE_IRQPRIO_SPE_FP_DATA];
128 sregs->u.e.ivor_high[2] = vcpu->arch.ivor[BOOKE_IRQPRIO_SPE_FP_ROUND];
129 sregs->u.e.ivor_high[3] =
130 vcpu->arch.ivor[BOOKE_IRQPRIO_PERFORMANCE_MONITOR];
131
132 kvmppc_get_sregs_ivor(vcpu, sregs);
133}
134
135int kvmppc_core_set_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
136{
137 struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
138
139 if (sregs->u.e.impl_id == KVM_SREGS_E_IMPL_FSL) {
140 vcpu_e500->svr = sregs->u.e.impl.fsl.svr;
141 vcpu_e500->hid0 = sregs->u.e.impl.fsl.hid0;
142 vcpu_e500->mcar = sregs->u.e.impl.fsl.mcar;
143 }
144
145 if (sregs->u.e.features & KVM_SREGS_E_ARCH206_MMU) {
146 vcpu_e500->mas0 = sregs->u.e.mas0;
147 vcpu_e500->mas1 = sregs->u.e.mas1;
148 vcpu_e500->mas2 = sregs->u.e.mas2;
149 vcpu_e500->mas7 = sregs->u.e.mas7_3 >> 32;
150 vcpu_e500->mas3 = (u32)sregs->u.e.mas7_3;
151 vcpu_e500->mas4 = sregs->u.e.mas4;
152 vcpu_e500->mas6 = sregs->u.e.mas6;
153 }
154
155 if (!(sregs->u.e.features & KVM_SREGS_E_IVOR))
156 return 0;
157
158 if (sregs->u.e.features & KVM_SREGS_E_SPE) {
159 vcpu->arch.ivor[BOOKE_IRQPRIO_SPE_UNAVAIL] =
160 sregs->u.e.ivor_high[0];
161 vcpu->arch.ivor[BOOKE_IRQPRIO_SPE_FP_DATA] =
162 sregs->u.e.ivor_high[1];
163 vcpu->arch.ivor[BOOKE_IRQPRIO_SPE_FP_ROUND] =
164 sregs->u.e.ivor_high[2];
165 }
166
167 if (sregs->u.e.features & KVM_SREGS_E_PM) {
168 vcpu->arch.ivor[BOOKE_IRQPRIO_PERFORMANCE_MONITOR] =
169 sregs->u.e.ivor_high[3];
170 }
171
172 return kvmppc_set_sregs_ivor(vcpu, sregs);
173}
174
99struct kvm_vcpu *kvmppc_core_vcpu_create(struct kvm *kvm, unsigned int id) 175struct kvm_vcpu *kvmppc_core_vcpu_create(struct kvm *kvm, unsigned int id)
100{ 176{
101 struct kvmppc_vcpu_e500 *vcpu_e500; 177 struct kvmppc_vcpu_e500 *vcpu_e500;
diff --git a/arch/powerpc/kvm/e500_emulate.c b/arch/powerpc/kvm/e500_emulate.c
index 8e3edfbc9634..69cd665a0caf 100644
--- a/arch/powerpc/kvm/e500_emulate.c
+++ b/arch/powerpc/kvm/e500_emulate.c
@@ -1,5 +1,5 @@
1/* 1/*
2 * Copyright (C) 2008 Freescale Semiconductor, Inc. All rights reserved. 2 * Copyright (C) 2008-2011 Freescale Semiconductor, Inc. All rights reserved.
3 * 3 *
4 * Author: Yu Liu, <yu.liu@freescale.com> 4 * Author: Yu Liu, <yu.liu@freescale.com>
5 * 5 *
@@ -78,8 +78,7 @@ int kvmppc_core_emulate_mtspr(struct kvm_vcpu *vcpu, int sprn, int rs)
78 78
79 switch (sprn) { 79 switch (sprn) {
80 case SPRN_PID: 80 case SPRN_PID:
81 vcpu_e500->pid[0] = vcpu->arch.shadow_pid = 81 kvmppc_set_pid(vcpu, spr_val);
82 vcpu->arch.pid = spr_val;
83 break; 82 break;
84 case SPRN_PID1: 83 case SPRN_PID1:
85 vcpu_e500->pid[1] = spr_val; break; 84 vcpu_e500->pid[1] = spr_val; break;
@@ -175,6 +174,8 @@ int kvmppc_core_emulate_mfspr(struct kvm_vcpu *vcpu, int sprn, int rt)
175 kvmppc_set_gpr(vcpu, rt, vcpu_e500->hid0); break; 174 kvmppc_set_gpr(vcpu, rt, vcpu_e500->hid0); break;
176 case SPRN_HID1: 175 case SPRN_HID1:
177 kvmppc_set_gpr(vcpu, rt, vcpu_e500->hid1); break; 176 kvmppc_set_gpr(vcpu, rt, vcpu_e500->hid1); break;
177 case SPRN_SVR:
178 kvmppc_set_gpr(vcpu, rt, vcpu_e500->svr); break;
178 179
179 case SPRN_MMUCSR0: 180 case SPRN_MMUCSR0:
180 kvmppc_set_gpr(vcpu, rt, 0); break; 181 kvmppc_set_gpr(vcpu, rt, 0); break;
diff --git a/arch/powerpc/kvm/e500_tlb.c b/arch/powerpc/kvm/e500_tlb.c
index d6d6d47a75a9..b18fe353397d 100644
--- a/arch/powerpc/kvm/e500_tlb.c
+++ b/arch/powerpc/kvm/e500_tlb.c
@@ -1,5 +1,5 @@
1/* 1/*
2 * Copyright (C) 2008 Freescale Semiconductor, Inc. All rights reserved. 2 * Copyright (C) 2008-2011 Freescale Semiconductor, Inc. All rights reserved.
3 * 3 *
4 * Author: Yu Liu, yu.liu@freescale.com 4 * Author: Yu Liu, yu.liu@freescale.com
5 * 5 *
@@ -24,6 +24,7 @@
24#include "../mm/mmu_decl.h" 24#include "../mm/mmu_decl.h"
25#include "e500_tlb.h" 25#include "e500_tlb.h"
26#include "trace.h" 26#include "trace.h"
27#include "timing.h"
27 28
28#define to_htlb1_esel(esel) (tlb1_entry_num - (esel) - 1) 29#define to_htlb1_esel(esel) (tlb1_entry_num - (esel) - 1)
29 30
@@ -506,6 +507,7 @@ int kvmppc_e500_emul_tlbsx(struct kvm_vcpu *vcpu, int rb)
506 vcpu_e500->mas7 = 0; 507 vcpu_e500->mas7 = 0;
507 } 508 }
508 509
510 kvmppc_set_exit_type(vcpu, EMULATED_TLBSX_EXITS);
509 return EMULATE_DONE; 511 return EMULATE_DONE;
510} 512}
511 513
@@ -571,6 +573,7 @@ int kvmppc_e500_emul_tlbwe(struct kvm_vcpu *vcpu)
571 write_host_tlbe(vcpu_e500, stlbsel, sesel); 573 write_host_tlbe(vcpu_e500, stlbsel, sesel);
572 } 574 }
573 575
576 kvmppc_set_exit_type(vcpu, EMULATED_TLBWE_EXITS);
574 return EMULATE_DONE; 577 return EMULATE_DONE;
575} 578}
576 579
@@ -672,6 +675,14 @@ int kvmppc_e500_tlb_search(struct kvm_vcpu *vcpu,
672 return -1; 675 return -1;
673} 676}
674 677
678void kvmppc_set_pid(struct kvm_vcpu *vcpu, u32 pid)
679{
680 struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
681
682 vcpu_e500->pid[0] = vcpu->arch.shadow_pid =
683 vcpu->arch.pid = pid;
684}
685
675void kvmppc_e500_tlb_setup(struct kvmppc_vcpu_e500 *vcpu_e500) 686void kvmppc_e500_tlb_setup(struct kvmppc_vcpu_e500 *vcpu_e500)
676{ 687{
677 struct tlbe *tlbe; 688 struct tlbe *tlbe;
diff --git a/arch/powerpc/kvm/emulate.c b/arch/powerpc/kvm/emulate.c
index c64fd2909bb2..141dce3c6810 100644
--- a/arch/powerpc/kvm/emulate.c
+++ b/arch/powerpc/kvm/emulate.c
@@ -114,6 +114,12 @@ void kvmppc_emulate_dec(struct kvm_vcpu *vcpu)
114 } 114 }
115} 115}
116 116
117u32 kvmppc_get_dec(struct kvm_vcpu *vcpu, u64 tb)
118{
119 u64 jd = tb - vcpu->arch.dec_jiffies;
120 return vcpu->arch.dec - jd;
121}
122
117/* XXX to do: 123/* XXX to do:
118 * lhax 124 * lhax
119 * lhaux 125 * lhaux
@@ -279,11 +285,8 @@ int kvmppc_emulate_instruction(struct kvm_run *run, struct kvm_vcpu *vcpu)
279 285
280 case SPRN_DEC: 286 case SPRN_DEC:
281 { 287 {
282 u64 jd = get_tb() - vcpu->arch.dec_jiffies; 288 kvmppc_set_gpr(vcpu, rt,
283 kvmppc_set_gpr(vcpu, rt, vcpu->arch.dec - jd); 289 kvmppc_get_dec(vcpu, get_tb()));
284 pr_debug("mfDEC: %x - %llx = %lx\n",
285 vcpu->arch.dec, jd,
286 kvmppc_get_gpr(vcpu, rt));
287 break; 290 break;
288 } 291 }
289 default: 292 default:
@@ -294,6 +297,7 @@ int kvmppc_emulate_instruction(struct kvm_run *run, struct kvm_vcpu *vcpu)
294 } 297 }
295 break; 298 break;
296 } 299 }
300 kvmppc_set_exit_type(vcpu, EMULATED_MFSPR_EXITS);
297 break; 301 break;
298 302
299 case OP_31_XOP_STHX: 303 case OP_31_XOP_STHX:
@@ -363,6 +367,7 @@ int kvmppc_emulate_instruction(struct kvm_run *run, struct kvm_vcpu *vcpu)
363 printk("mtspr: unknown spr %x\n", sprn); 367 printk("mtspr: unknown spr %x\n", sprn);
364 break; 368 break;
365 } 369 }
370 kvmppc_set_exit_type(vcpu, EMULATED_MTSPR_EXITS);
366 break; 371 break;
367 372
368 case OP_31_XOP_DCBI: 373 case OP_31_XOP_DCBI:
diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
index 99758460efde..616dd516ca1f 100644
--- a/arch/powerpc/kvm/powerpc.c
+++ b/arch/powerpc/kvm/powerpc.c
@@ -175,7 +175,11 @@ int kvm_dev_ioctl_check_extension(long ext)
175 int r; 175 int r;
176 176
177 switch (ext) { 177 switch (ext) {
178#ifdef CONFIG_BOOKE
179 case KVM_CAP_PPC_BOOKE_SREGS:
180#else
178 case KVM_CAP_PPC_SEGSTATE: 181 case KVM_CAP_PPC_SEGSTATE:
182#endif
179 case KVM_CAP_PPC_PAIRED_SINGLES: 183 case KVM_CAP_PPC_PAIRED_SINGLES:
180 case KVM_CAP_PPC_UNSET_IRQ: 184 case KVM_CAP_PPC_UNSET_IRQ:
181 case KVM_CAP_PPC_IRQ_LEVEL: 185 case KVM_CAP_PPC_IRQ_LEVEL:
@@ -284,6 +288,10 @@ int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
284 tasklet_init(&vcpu->arch.tasklet, kvmppc_decrementer_func, (ulong)vcpu); 288 tasklet_init(&vcpu->arch.tasklet, kvmppc_decrementer_func, (ulong)vcpu);
285 vcpu->arch.dec_timer.function = kvmppc_decrementer_wakeup; 289 vcpu->arch.dec_timer.function = kvmppc_decrementer_wakeup;
286 290
291#ifdef CONFIG_KVM_EXIT_TIMING
292 mutex_init(&vcpu->arch.exit_timing_lock);
293#endif
294
287 return 0; 295 return 0;
288} 296}
289 297
@@ -294,12 +302,25 @@ void kvm_arch_vcpu_uninit(struct kvm_vcpu *vcpu)
294 302
295void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu) 303void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
296{ 304{
305#ifdef CONFIG_BOOKE
306 /*
307 * vrsave (formerly usprg0) isn't used by Linux, but may
308 * be used by the guest.
309 *
310 * On non-booke this is associated with Altivec and
311 * is handled by code in book3s.c.
312 */
313 mtspr(SPRN_VRSAVE, vcpu->arch.vrsave);
314#endif
297 kvmppc_core_vcpu_load(vcpu, cpu); 315 kvmppc_core_vcpu_load(vcpu, cpu);
298} 316}
299 317
300void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu) 318void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
301{ 319{
302 kvmppc_core_vcpu_put(vcpu); 320 kvmppc_core_vcpu_put(vcpu);
321#ifdef CONFIG_BOOKE
322 vcpu->arch.vrsave = mfspr(SPRN_VRSAVE);
323#endif
303} 324}
304 325
305int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu, 326int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu,
diff --git a/arch/powerpc/kvm/timing.c b/arch/powerpc/kvm/timing.c
index a021f5827a33..319177df9587 100644
--- a/arch/powerpc/kvm/timing.c
+++ b/arch/powerpc/kvm/timing.c
@@ -34,8 +34,8 @@ void kvmppc_init_timing_stats(struct kvm_vcpu *vcpu)
34{ 34{
35 int i; 35 int i;
36 36
37 /* pause guest execution to avoid concurrent updates */ 37 /* Take a lock to avoid concurrent updates */
38 mutex_lock(&vcpu->mutex); 38 mutex_lock(&vcpu->arch.exit_timing_lock);
39 39
40 vcpu->arch.last_exit_type = 0xDEAD; 40 vcpu->arch.last_exit_type = 0xDEAD;
41 for (i = 0; i < __NUMBER_OF_KVM_EXIT_TYPES; i++) { 41 for (i = 0; i < __NUMBER_OF_KVM_EXIT_TYPES; i++) {
@@ -49,7 +49,7 @@ void kvmppc_init_timing_stats(struct kvm_vcpu *vcpu)
49 vcpu->arch.timing_exit.tv64 = 0; 49 vcpu->arch.timing_exit.tv64 = 0;
50 vcpu->arch.timing_last_enter.tv64 = 0; 50 vcpu->arch.timing_last_enter.tv64 = 0;
51 51
52 mutex_unlock(&vcpu->mutex); 52 mutex_unlock(&vcpu->arch.exit_timing_lock);
53} 53}
54 54
55static void add_exit_timing(struct kvm_vcpu *vcpu, u64 duration, int type) 55static void add_exit_timing(struct kvm_vcpu *vcpu, u64 duration, int type)
@@ -65,6 +65,8 @@ static void add_exit_timing(struct kvm_vcpu *vcpu, u64 duration, int type)
65 return; 65 return;
66 } 66 }
67 67
68 mutex_lock(&vcpu->arch.exit_timing_lock);
69
68 vcpu->arch.timing_count_type[type]++; 70 vcpu->arch.timing_count_type[type]++;
69 71
70 /* sum */ 72 /* sum */
@@ -93,6 +95,8 @@ static void add_exit_timing(struct kvm_vcpu *vcpu, u64 duration, int type)
93 vcpu->arch.timing_min_duration[type] = duration; 95 vcpu->arch.timing_min_duration[type] = duration;
94 if (unlikely(duration > vcpu->arch.timing_max_duration[type])) 96 if (unlikely(duration > vcpu->arch.timing_max_duration[type]))
95 vcpu->arch.timing_max_duration[type] = duration; 97 vcpu->arch.timing_max_duration[type] = duration;
98
99 mutex_unlock(&vcpu->arch.exit_timing_lock);
96} 100}
97 101
98void kvmppc_update_timing_stats(struct kvm_vcpu *vcpu) 102void kvmppc_update_timing_stats(struct kvm_vcpu *vcpu)
@@ -147,17 +151,30 @@ static int kvmppc_exit_timing_show(struct seq_file *m, void *private)
147{ 151{
148 struct kvm_vcpu *vcpu = m->private; 152 struct kvm_vcpu *vcpu = m->private;
149 int i; 153 int i;
154 u64 min, max, sum, sum_quad;
150 155
151 seq_printf(m, "%s", "type count min max sum sum_squared\n"); 156 seq_printf(m, "%s", "type count min max sum sum_squared\n");
152 157
158
153 for (i = 0; i < __NUMBER_OF_KVM_EXIT_TYPES; i++) { 159 for (i = 0; i < __NUMBER_OF_KVM_EXIT_TYPES; i++) {
160
161 min = vcpu->arch.timing_min_duration[i];
162 do_div(min, tb_ticks_per_usec);
163 max = vcpu->arch.timing_max_duration[i];
164 do_div(max, tb_ticks_per_usec);
165 sum = vcpu->arch.timing_sum_duration[i];
166 do_div(sum, tb_ticks_per_usec);
167 sum_quad = vcpu->arch.timing_sum_quad_duration[i];
168 do_div(sum_quad, tb_ticks_per_usec);
169
154 seq_printf(m, "%12s %10d %10lld %10lld %20lld %20lld\n", 170 seq_printf(m, "%12s %10d %10lld %10lld %20lld %20lld\n",
155 kvm_exit_names[i], 171 kvm_exit_names[i],
156 vcpu->arch.timing_count_type[i], 172 vcpu->arch.timing_count_type[i],
157 vcpu->arch.timing_min_duration[i], 173 min,
158 vcpu->arch.timing_max_duration[i], 174 max,
159 vcpu->arch.timing_sum_duration[i], 175 sum,
160 vcpu->arch.timing_sum_quad_duration[i]); 176 sum_quad);
177
161 } 178 }
162 return 0; 179 return 0;
163} 180}
diff --git a/arch/powerpc/mm/pgtable.c b/arch/powerpc/mm/pgtable.c
index 6a3997f98dfb..af40c8768a78 100644
--- a/arch/powerpc/mm/pgtable.c
+++ b/arch/powerpc/mm/pgtable.c
@@ -33,110 +33,6 @@
33 33
34#include "mmu_decl.h" 34#include "mmu_decl.h"
35 35
36DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
37
38#ifdef CONFIG_SMP
39
40/*
41 * Handle batching of page table freeing on SMP. Page tables are
42 * queued up and send to be freed later by RCU in order to avoid
43 * freeing a page table page that is being walked without locks
44 */
45
46static DEFINE_PER_CPU(struct pte_freelist_batch *, pte_freelist_cur);
47static unsigned long pte_freelist_forced_free;
48
49struct pte_freelist_batch
50{
51 struct rcu_head rcu;
52 unsigned int index;
53 unsigned long tables[0];
54};
55
56#define PTE_FREELIST_SIZE \
57 ((PAGE_SIZE - sizeof(struct pte_freelist_batch)) \
58 / sizeof(unsigned long))
59
60static void pte_free_smp_sync(void *arg)
61{
62 /* Do nothing, just ensure we sync with all CPUs */
63}
64
65/* This is only called when we are critically out of memory
66 * (and fail to get a page in pte_free_tlb).
67 */
68static void pgtable_free_now(void *table, unsigned shift)
69{
70 pte_freelist_forced_free++;
71
72 smp_call_function(pte_free_smp_sync, NULL, 1);
73
74 pgtable_free(table, shift);
75}
76
77static void pte_free_rcu_callback(struct rcu_head *head)
78{
79 struct pte_freelist_batch *batch =
80 container_of(head, struct pte_freelist_batch, rcu);
81 unsigned int i;
82
83 for (i = 0; i < batch->index; i++) {
84 void *table = (void *)(batch->tables[i] & ~MAX_PGTABLE_INDEX_SIZE);
85 unsigned shift = batch->tables[i] & MAX_PGTABLE_INDEX_SIZE;
86
87 pgtable_free(table, shift);
88 }
89
90 free_page((unsigned long)batch);
91}
92
93static void pte_free_submit(struct pte_freelist_batch *batch)
94{
95 call_rcu_sched(&batch->rcu, pte_free_rcu_callback);
96}
97
98void pgtable_free_tlb(struct mmu_gather *tlb, void *table, unsigned shift)
99{
100 /* This is safe since tlb_gather_mmu has disabled preemption */
101 struct pte_freelist_batch **batchp = &__get_cpu_var(pte_freelist_cur);
102 unsigned long pgf;
103
104 if (atomic_read(&tlb->mm->mm_users) < 2 ||
105 cpumask_equal(mm_cpumask(tlb->mm), cpumask_of(smp_processor_id()))){
106 pgtable_free(table, shift);
107 return;
108 }
109
110 if (*batchp == NULL) {
111 *batchp = (struct pte_freelist_batch *)__get_free_page(GFP_ATOMIC);
112 if (*batchp == NULL) {
113 pgtable_free_now(table, shift);
114 return;
115 }
116 (*batchp)->index = 0;
117 }
118 BUG_ON(shift > MAX_PGTABLE_INDEX_SIZE);
119 pgf = (unsigned long)table | shift;
120 (*batchp)->tables[(*batchp)->index++] = pgf;
121 if ((*batchp)->index == PTE_FREELIST_SIZE) {
122 pte_free_submit(*batchp);
123 *batchp = NULL;
124 }
125}
126
127void pte_free_finish(void)
128{
129 /* This is safe since tlb_gather_mmu has disabled preemption */
130 struct pte_freelist_batch **batchp = &__get_cpu_var(pte_freelist_cur);
131
132 if (*batchp == NULL)
133 return;
134 pte_free_submit(*batchp);
135 *batchp = NULL;
136}
137
138#endif /* CONFIG_SMP */
139
140static inline int is_exec_fault(void) 36static inline int is_exec_fault(void)
141{ 37{
142 return current->thread.regs && TRAP(current->thread.regs) == 0x400; 38 return current->thread.regs && TRAP(current->thread.regs) == 0x400;
diff --git a/arch/powerpc/mm/tlb_hash32.c b/arch/powerpc/mm/tlb_hash32.c
index 690566b66e8e..27b863c14941 100644
--- a/arch/powerpc/mm/tlb_hash32.c
+++ b/arch/powerpc/mm/tlb_hash32.c
@@ -71,9 +71,6 @@ void tlb_flush(struct mmu_gather *tlb)
71 */ 71 */
72 _tlbia(); 72 _tlbia();
73 } 73 }
74
75 /* Push out batch of freed page tables */
76 pte_free_finish();
77} 74}
78 75
79/* 76/*
diff --git a/arch/powerpc/mm/tlb_hash64.c b/arch/powerpc/mm/tlb_hash64.c
index c14d09f614f3..31f18207970b 100644
--- a/arch/powerpc/mm/tlb_hash64.c
+++ b/arch/powerpc/mm/tlb_hash64.c
@@ -155,7 +155,7 @@ void __flush_tlb_pending(struct ppc64_tlb_batch *batch)
155 155
156void tlb_flush(struct mmu_gather *tlb) 156void tlb_flush(struct mmu_gather *tlb)
157{ 157{
158 struct ppc64_tlb_batch *tlbbatch = &__get_cpu_var(ppc64_tlb_batch); 158 struct ppc64_tlb_batch *tlbbatch = &get_cpu_var(ppc64_tlb_batch);
159 159
160 /* If there's a TLB batch pending, then we must flush it because the 160 /* If there's a TLB batch pending, then we must flush it because the
161 * pages are going to be freed and we really don't want to have a CPU 161 * pages are going to be freed and we really don't want to have a CPU
@@ -164,8 +164,7 @@ void tlb_flush(struct mmu_gather *tlb)
164 if (tlbbatch->index) 164 if (tlbbatch->index)
165 __flush_tlb_pending(tlbbatch); 165 __flush_tlb_pending(tlbbatch);
166 166
167 /* Push out batch of freed page tables */ 167 put_cpu_var(ppc64_tlb_batch);
168 pte_free_finish();
169} 168}
170 169
171/** 170/**
diff --git a/arch/powerpc/mm/tlb_nohash.c b/arch/powerpc/mm/tlb_nohash.c
index 2a030d89bbc6..0bdad3aecc67 100644
--- a/arch/powerpc/mm/tlb_nohash.c
+++ b/arch/powerpc/mm/tlb_nohash.c
@@ -299,9 +299,6 @@ EXPORT_SYMBOL(flush_tlb_range);
299void tlb_flush(struct mmu_gather *tlb) 299void tlb_flush(struct mmu_gather *tlb)
300{ 300{
301 flush_tlb_mm(tlb->mm); 301 flush_tlb_mm(tlb->mm);
302
303 /* Push out batch of freed page tables */
304 pte_free_finish();
305} 302}
306 303
307/* 304/*
generated by cgit v1.2.2 (git 2.25.0) at 2025-09-21 06:04:29 -0400