aboutsummaryrefslogtreecommitdiffstats
path: root/arch/powerpc
diff options
context:
space:
mode:
Diffstat (limited to 'arch/powerpc')
-rw-r--r--arch/powerpc/Makefile4
-rw-r--r--arch/powerpc/boot/Makefile2
-rw-r--r--arch/powerpc/boot/crt0.S4
-rw-r--r--arch/powerpc/include/asm/hvcall.h1
-rw-r--r--arch/powerpc/include/asm/kvm_book3s.h23
-rw-r--r--arch/powerpc/include/asm/kvm_book3s_64.h18
-rw-r--r--arch/powerpc/include/asm/kvm_host.h5
-rw-r--r--arch/powerpc/include/asm/kvm_ppc.h10
-rw-r--r--arch/powerpc/include/asm/perf_event.h2
-rw-r--r--arch/powerpc/include/uapi/asm/Kbuild1
-rw-r--r--arch/powerpc/include/uapi/asm/bpf_perf_event.h9
-rw-r--r--arch/powerpc/kernel/exceptions-64s.S9
-rw-r--r--arch/powerpc/kernel/legacy_serial.c6
-rw-r--r--arch/powerpc/kernel/machine_kexec_file_64.c54
-rw-r--r--arch/powerpc/kernel/msi.c7
-rw-r--r--arch/powerpc/kernel/ptrace.c7
-rw-r--r--arch/powerpc/kvm/book3s.c8
-rw-r--r--arch/powerpc/kvm/book3s_64_mmu_hv.c12
-rw-r--r--arch/powerpc/kvm/book3s_64_mmu_radix.c160
-rw-r--r--arch/powerpc/kvm/book3s_hv.c95
-rw-r--r--arch/powerpc/kvm/book3s_hv_nested.c190
-rw-r--r--arch/powerpc/kvm/book3s_hv_rm_mmu.c2
-rw-r--r--arch/powerpc/kvm/book3s_pr.c4
-rw-r--r--arch/powerpc/kvm/book3s_xics.c12
-rw-r--r--arch/powerpc/kvm/book3s_xive.c12
-rw-r--r--arch/powerpc/kvm/booke.c3
-rw-r--r--arch/powerpc/kvm/e500_mmu_host.c3
-rw-r--r--arch/powerpc/kvm/powerpc.c47
-rw-r--r--arch/powerpc/mm/dump_linuxpagetables.c1
-rw-r--r--arch/powerpc/mm/fault.c1
-rw-r--r--arch/powerpc/mm/hugetlbpage.c2
-rw-r--r--arch/powerpc/mm/init_64.c19
-rw-r--r--arch/powerpc/net/bpf_jit_comp64.c66
-rw-r--r--arch/powerpc/platforms/pseries/Kconfig3
-rw-r--r--arch/powerpc/platforms/pseries/papr_scm.c39
35 files changed, 665 insertions, 176 deletions
diff --git a/arch/powerpc/Makefile b/arch/powerpc/Makefile
index 8a2ce14d68d0..854199c9ab7e 100644
--- a/arch/powerpc/Makefile
+++ b/arch/powerpc/Makefile
@@ -228,10 +228,6 @@ KBUILD_CFLAGS += $(call cc-option,-mno-vsx)
228KBUILD_CFLAGS += $(call cc-option,-mno-spe) 228KBUILD_CFLAGS += $(call cc-option,-mno-spe)
229KBUILD_CFLAGS += $(call cc-option,-mspe=no) 229KBUILD_CFLAGS += $(call cc-option,-mspe=no)
230 230
231# Enable unit-at-a-time mode when possible. It shrinks the
232# kernel considerably.
233KBUILD_CFLAGS += $(call cc-option,-funit-at-a-time)
234
235# FIXME: the module load should be taught about the additional relocs 231# FIXME: the module load should be taught about the additional relocs
236# generated by this. 232# generated by this.
237# revert to pre-gcc-4.4 behaviour of .eh_frame 233# revert to pre-gcc-4.4 behaviour of .eh_frame
diff --git a/arch/powerpc/boot/Makefile b/arch/powerpc/boot/Makefile
index 39354365f54a..ed9883169190 100644
--- a/arch/powerpc/boot/Makefile
+++ b/arch/powerpc/boot/Makefile
@@ -197,7 +197,7 @@ $(obj)/empty.c:
197$(obj)/zImage.coff.lds $(obj)/zImage.ps3.lds : $(obj)/%: $(srctree)/$(src)/%.S 197$(obj)/zImage.coff.lds $(obj)/zImage.ps3.lds : $(obj)/%: $(srctree)/$(src)/%.S
198 $(Q)cp $< $@ 198 $(Q)cp $< $@
199 199
200$(obj)/serial.c: $(obj)/autoconf.h 200$(srctree)/$(src)/serial.c: $(obj)/autoconf.h
201 201
202$(obj)/autoconf.h: $(obj)/%: $(objtree)/include/generated/% 202$(obj)/autoconf.h: $(obj)/%: $(objtree)/include/generated/%
203 $(Q)cp $< $@ 203 $(Q)cp $< $@
diff --git a/arch/powerpc/boot/crt0.S b/arch/powerpc/boot/crt0.S
index 32dfe6d083f3..9b9d17437373 100644
--- a/arch/powerpc/boot/crt0.S
+++ b/arch/powerpc/boot/crt0.S
@@ -15,7 +15,7 @@
15RELA = 7 15RELA = 7
16RELACOUNT = 0x6ffffff9 16RELACOUNT = 0x6ffffff9
17 17
18 .text 18 .data
19 /* A procedure descriptor used when booting this as a COFF file. 19 /* A procedure descriptor used when booting this as a COFF file.
20 * When making COFF, this comes first in the link and we're 20 * When making COFF, this comes first in the link and we're
21 * linked at 0x500000. 21 * linked at 0x500000.
@@ -23,6 +23,8 @@ RELACOUNT = 0x6ffffff9
23 .globl _zimage_start_opd 23 .globl _zimage_start_opd
24_zimage_start_opd: 24_zimage_start_opd:
25 .long 0x500000, 0, 0, 0 25 .long 0x500000, 0, 0, 0
26 .text
27 b _zimage_start
26 28
27#ifdef __powerpc64__ 29#ifdef __powerpc64__
28.balign 8 30.balign 8
diff --git a/arch/powerpc/include/asm/hvcall.h b/arch/powerpc/include/asm/hvcall.h
index 33a4fc891947..463c63a9fcf1 100644
--- a/arch/powerpc/include/asm/hvcall.h
+++ b/arch/powerpc/include/asm/hvcall.h
@@ -335,6 +335,7 @@
335#define H_SET_PARTITION_TABLE 0xF800 335#define H_SET_PARTITION_TABLE 0xF800
336#define H_ENTER_NESTED 0xF804 336#define H_ENTER_NESTED 0xF804
337#define H_TLB_INVALIDATE 0xF808 337#define H_TLB_INVALIDATE 0xF808
338#define H_COPY_TOFROM_GUEST 0xF80C
338 339
339/* Values for 2nd argument to H_SET_MODE */ 340/* Values for 2nd argument to H_SET_MODE */
340#define H_SET_MODE_RESOURCE_SET_CIABR 1 341#define H_SET_MODE_RESOURCE_SET_CIABR 1
diff --git a/arch/powerpc/include/asm/kvm_book3s.h b/arch/powerpc/include/asm/kvm_book3s.h
index 09f8e9ba69bc..38f1b879f569 100644
--- a/arch/powerpc/include/asm/kvm_book3s.h
+++ b/arch/powerpc/include/asm/kvm_book3s.h
@@ -188,6 +188,13 @@ extern int kvmppc_book3s_hcall_implemented(struct kvm *kvm, unsigned long hc);
188extern int kvmppc_book3s_radix_page_fault(struct kvm_run *run, 188extern int kvmppc_book3s_radix_page_fault(struct kvm_run *run,
189 struct kvm_vcpu *vcpu, 189 struct kvm_vcpu *vcpu,
190 unsigned long ea, unsigned long dsisr); 190 unsigned long ea, unsigned long dsisr);
191extern unsigned long __kvmhv_copy_tofrom_guest_radix(int lpid, int pid,
192 gva_t eaddr, void *to, void *from,
193 unsigned long n);
194extern long kvmhv_copy_from_guest_radix(struct kvm_vcpu *vcpu, gva_t eaddr,
195 void *to, unsigned long n);
196extern long kvmhv_copy_to_guest_radix(struct kvm_vcpu *vcpu, gva_t eaddr,
197 void *from, unsigned long n);
191extern int kvmppc_mmu_walk_radix_tree(struct kvm_vcpu *vcpu, gva_t eaddr, 198extern int kvmppc_mmu_walk_radix_tree(struct kvm_vcpu *vcpu, gva_t eaddr,
192 struct kvmppc_pte *gpte, u64 root, 199 struct kvmppc_pte *gpte, u64 root,
193 u64 *pte_ret_p); 200 u64 *pte_ret_p);
@@ -196,8 +203,11 @@ extern int kvmppc_mmu_radix_translate_table(struct kvm_vcpu *vcpu, gva_t eaddr,
196 int table_index, u64 *pte_ret_p); 203 int table_index, u64 *pte_ret_p);
197extern int kvmppc_mmu_radix_xlate(struct kvm_vcpu *vcpu, gva_t eaddr, 204extern int kvmppc_mmu_radix_xlate(struct kvm_vcpu *vcpu, gva_t eaddr,
198 struct kvmppc_pte *gpte, bool data, bool iswrite); 205 struct kvmppc_pte *gpte, bool data, bool iswrite);
206extern void kvmppc_radix_tlbie_page(struct kvm *kvm, unsigned long addr,
207 unsigned int pshift, unsigned int lpid);
199extern void kvmppc_unmap_pte(struct kvm *kvm, pte_t *pte, unsigned long gpa, 208extern void kvmppc_unmap_pte(struct kvm *kvm, pte_t *pte, unsigned long gpa,
200 unsigned int shift, struct kvm_memory_slot *memslot, 209 unsigned int shift,
210 const struct kvm_memory_slot *memslot,
201 unsigned int lpid); 211 unsigned int lpid);
202extern bool kvmppc_hv_handle_set_rc(struct kvm *kvm, pgd_t *pgtable, 212extern bool kvmppc_hv_handle_set_rc(struct kvm *kvm, pgd_t *pgtable,
203 bool writing, unsigned long gpa, 213 bool writing, unsigned long gpa,
@@ -215,16 +225,14 @@ extern int kvmppc_radix_init(void);
215extern void kvmppc_radix_exit(void); 225extern void kvmppc_radix_exit(void);
216extern int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot, 226extern int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
217 unsigned long gfn); 227 unsigned long gfn);
218extern void kvmppc_unmap_pte(struct kvm *kvm, pte_t *pte,
219 unsigned long gpa, unsigned int shift,
220 struct kvm_memory_slot *memslot,
221 unsigned int lpid);
222extern int kvm_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot, 228extern int kvm_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
223 unsigned long gfn); 229 unsigned long gfn);
224extern int kvm_test_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot, 230extern int kvm_test_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
225 unsigned long gfn); 231 unsigned long gfn);
226extern long kvmppc_hv_get_dirty_log_radix(struct kvm *kvm, 232extern long kvmppc_hv_get_dirty_log_radix(struct kvm *kvm,
227 struct kvm_memory_slot *memslot, unsigned long *map); 233 struct kvm_memory_slot *memslot, unsigned long *map);
234extern void kvmppc_radix_flush_memslot(struct kvm *kvm,
235 const struct kvm_memory_slot *memslot);
228extern int kvmhv_get_rmmu_info(struct kvm *kvm, struct kvm_ppc_rmmu_info *info); 236extern int kvmhv_get_rmmu_info(struct kvm *kvm, struct kvm_ppc_rmmu_info *info);
229 237
230/* XXX remove this export when load_last_inst() is generic */ 238/* XXX remove this export when load_last_inst() is generic */
@@ -242,7 +250,7 @@ extern kvm_pfn_t kvmppc_gpa_to_pfn(struct kvm_vcpu *vcpu, gpa_t gpa,
242 bool writing, bool *writable); 250 bool writing, bool *writable);
243extern void kvmppc_add_revmap_chain(struct kvm *kvm, struct revmap_entry *rev, 251extern void kvmppc_add_revmap_chain(struct kvm *kvm, struct revmap_entry *rev,
244 unsigned long *rmap, long pte_index, int realmode); 252 unsigned long *rmap, long pte_index, int realmode);
245extern void kvmppc_update_dirty_map(struct kvm_memory_slot *memslot, 253extern void kvmppc_update_dirty_map(const struct kvm_memory_slot *memslot,
246 unsigned long gfn, unsigned long psize); 254 unsigned long gfn, unsigned long psize);
247extern void kvmppc_invalidate_hpte(struct kvm *kvm, __be64 *hptep, 255extern void kvmppc_invalidate_hpte(struct kvm *kvm, __be64 *hptep,
248 unsigned long pte_index); 256 unsigned long pte_index);
@@ -298,6 +306,7 @@ long kvmhv_nested_init(void);
298void kvmhv_nested_exit(void); 306void kvmhv_nested_exit(void);
299void kvmhv_vm_nested_init(struct kvm *kvm); 307void kvmhv_vm_nested_init(struct kvm *kvm);
300long kvmhv_set_partition_table(struct kvm_vcpu *vcpu); 308long kvmhv_set_partition_table(struct kvm_vcpu *vcpu);
309long kvmhv_copy_tofrom_guest_nested(struct kvm_vcpu *vcpu);
301void kvmhv_set_ptbl_entry(unsigned int lpid, u64 dw0, u64 dw1); 310void kvmhv_set_ptbl_entry(unsigned int lpid, u64 dw0, u64 dw1);
302void kvmhv_release_all_nested(struct kvm *kvm); 311void kvmhv_release_all_nested(struct kvm *kvm);
303long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu); 312long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu);
@@ -307,7 +316,7 @@ int kvmhv_run_single_vcpu(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu,
307void kvmhv_save_hv_regs(struct kvm_vcpu *vcpu, struct hv_guest_state *hr); 316void kvmhv_save_hv_regs(struct kvm_vcpu *vcpu, struct hv_guest_state *hr);
308void kvmhv_restore_hv_return_state(struct kvm_vcpu *vcpu, 317void kvmhv_restore_hv_return_state(struct kvm_vcpu *vcpu,
309 struct hv_guest_state *hr); 318 struct hv_guest_state *hr);
310long int kvmhv_nested_page_fault(struct kvm_vcpu *vcpu); 319long int kvmhv_nested_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu);
311 320
312void kvmppc_giveup_fac(struct kvm_vcpu *vcpu, ulong fac); 321void kvmppc_giveup_fac(struct kvm_vcpu *vcpu, ulong fac);
313 322
diff --git a/arch/powerpc/include/asm/kvm_book3s_64.h b/arch/powerpc/include/asm/kvm_book3s_64.h
index 6d298145d564..21b1ed5df888 100644
--- a/arch/powerpc/include/asm/kvm_book3s_64.h
+++ b/arch/powerpc/include/asm/kvm_book3s_64.h
@@ -55,6 +55,7 @@ struct kvm_nested_guest {
55 cpumask_t need_tlb_flush; 55 cpumask_t need_tlb_flush;
56 cpumask_t cpu_in_guest; 56 cpumask_t cpu_in_guest;
57 short prev_cpu[NR_CPUS]; 57 short prev_cpu[NR_CPUS];
58 u8 radix; /* is this nested guest radix */
58}; 59};
59 60
60/* 61/*
@@ -150,6 +151,18 @@ static inline bool kvm_is_radix(struct kvm *kvm)
150 return kvm->arch.radix; 151 return kvm->arch.radix;
151} 152}
152 153
154static inline bool kvmhv_vcpu_is_radix(struct kvm_vcpu *vcpu)
155{
156 bool radix;
157
158 if (vcpu->arch.nested)
159 radix = vcpu->arch.nested->radix;
160 else
161 radix = kvm_is_radix(vcpu->kvm);
162
163 return radix;
164}
165
153#define KVM_DEFAULT_HPT_ORDER 24 /* 16MB HPT by default */ 166#define KVM_DEFAULT_HPT_ORDER 24 /* 16MB HPT by default */
154#endif 167#endif
155 168
@@ -624,8 +637,11 @@ extern int kvmppc_create_pte(struct kvm *kvm, pgd_t *pgtable, pte_t pte,
624 unsigned long *rmapp, struct rmap_nested **n_rmap); 637 unsigned long *rmapp, struct rmap_nested **n_rmap);
625extern void kvmhv_insert_nest_rmap(struct kvm *kvm, unsigned long *rmapp, 638extern void kvmhv_insert_nest_rmap(struct kvm *kvm, unsigned long *rmapp,
626 struct rmap_nested **n_rmap); 639 struct rmap_nested **n_rmap);
640extern void kvmhv_update_nest_rmap_rc_list(struct kvm *kvm, unsigned long *rmapp,
641 unsigned long clr, unsigned long set,
642 unsigned long hpa, unsigned long nbytes);
627extern void kvmhv_remove_nest_rmap_range(struct kvm *kvm, 643extern void kvmhv_remove_nest_rmap_range(struct kvm *kvm,
628 struct kvm_memory_slot *memslot, 644 const struct kvm_memory_slot *memslot,
629 unsigned long gpa, unsigned long hpa, 645 unsigned long gpa, unsigned long hpa,
630 unsigned long nbytes); 646 unsigned long nbytes);
631 647
diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index fac6f631ed29..0f98f00da2ea 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h
@@ -72,7 +72,7 @@ extern int kvm_unmap_hva_range(struct kvm *kvm,
72 unsigned long start, unsigned long end); 72 unsigned long start, unsigned long end);
73extern int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end); 73extern int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
74extern int kvm_test_age_hva(struct kvm *kvm, unsigned long hva); 74extern int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
75extern void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte); 75extern int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
76 76
77#define HPTEG_CACHE_NUM (1 << 15) 77#define HPTEG_CACHE_NUM (1 << 15)
78#define HPTEG_HASH_BITS_PTE 13 78#define HPTEG_HASH_BITS_PTE 13
@@ -793,6 +793,7 @@ struct kvm_vcpu_arch {
793 /* For support of nested guests */ 793 /* For support of nested guests */
794 struct kvm_nested_guest *nested; 794 struct kvm_nested_guest *nested;
795 u32 nested_vcpu_id; 795 u32 nested_vcpu_id;
796 gpa_t nested_io_gpr;
796#endif 797#endif
797 798
798#ifdef CONFIG_KVM_BOOK3S_HV_EXIT_TIMING 799#ifdef CONFIG_KVM_BOOK3S_HV_EXIT_TIMING
@@ -827,6 +828,8 @@ struct kvm_vcpu_arch {
827#define KVM_MMIO_REG_FQPR 0x00c0 828#define KVM_MMIO_REG_FQPR 0x00c0
828#define KVM_MMIO_REG_VSX 0x0100 829#define KVM_MMIO_REG_VSX 0x0100
829#define KVM_MMIO_REG_VMX 0x0180 830#define KVM_MMIO_REG_VMX 0x0180
831#define KVM_MMIO_REG_NESTED_GPR 0xffc0
832
830 833
831#define __KVM_HAVE_ARCH_WQP 834#define __KVM_HAVE_ARCH_WQP
832#define __KVM_HAVE_CREATE_DEVICE 835#define __KVM_HAVE_CREATE_DEVICE
diff --git a/arch/powerpc/include/asm/kvm_ppc.h b/arch/powerpc/include/asm/kvm_ppc.h
index 9b89b1918dfc..eb0d79f0ca45 100644
--- a/arch/powerpc/include/asm/kvm_ppc.h
+++ b/arch/powerpc/include/asm/kvm_ppc.h
@@ -224,7 +224,8 @@ extern int kvmppc_core_prepare_memory_region(struct kvm *kvm,
224extern void kvmppc_core_commit_memory_region(struct kvm *kvm, 224extern void kvmppc_core_commit_memory_region(struct kvm *kvm,
225 const struct kvm_userspace_memory_region *mem, 225 const struct kvm_userspace_memory_region *mem,
226 const struct kvm_memory_slot *old, 226 const struct kvm_memory_slot *old,
227 const struct kvm_memory_slot *new); 227 const struct kvm_memory_slot *new,
228 enum kvm_mr_change change);
228extern int kvm_vm_ioctl_get_smmu_info(struct kvm *kvm, 229extern int kvm_vm_ioctl_get_smmu_info(struct kvm *kvm,
229 struct kvm_ppc_smmu_info *info); 230 struct kvm_ppc_smmu_info *info);
230extern void kvmppc_core_flush_memslot(struct kvm *kvm, 231extern void kvmppc_core_flush_memslot(struct kvm *kvm,
@@ -294,7 +295,8 @@ struct kvmppc_ops {
294 void (*commit_memory_region)(struct kvm *kvm, 295 void (*commit_memory_region)(struct kvm *kvm,
295 const struct kvm_userspace_memory_region *mem, 296 const struct kvm_userspace_memory_region *mem,
296 const struct kvm_memory_slot *old, 297 const struct kvm_memory_slot *old,
297 const struct kvm_memory_slot *new); 298 const struct kvm_memory_slot *new,
299 enum kvm_mr_change change);
298 int (*unmap_hva_range)(struct kvm *kvm, unsigned long start, 300 int (*unmap_hva_range)(struct kvm *kvm, unsigned long start,
299 unsigned long end); 301 unsigned long end);
300 int (*age_hva)(struct kvm *kvm, unsigned long start, unsigned long end); 302 int (*age_hva)(struct kvm *kvm, unsigned long start, unsigned long end);
@@ -326,6 +328,10 @@ struct kvmppc_ops {
326 unsigned long flags); 328 unsigned long flags);
327 void (*giveup_ext)(struct kvm_vcpu *vcpu, ulong msr); 329 void (*giveup_ext)(struct kvm_vcpu *vcpu, ulong msr);
328 int (*enable_nested)(struct kvm *kvm); 330 int (*enable_nested)(struct kvm *kvm);
331 int (*load_from_eaddr)(struct kvm_vcpu *vcpu, ulong *eaddr, void *ptr,
332 int size);
333 int (*store_to_eaddr)(struct kvm_vcpu *vcpu, ulong *eaddr, void *ptr,
334 int size);
329}; 335};
330 336
331extern struct kvmppc_ops *kvmppc_hv_ops; 337extern struct kvmppc_ops *kvmppc_hv_ops;
diff --git a/arch/powerpc/include/asm/perf_event.h b/arch/powerpc/include/asm/perf_event.h
index 8bf1b6351716..16a49819da9a 100644
--- a/arch/powerpc/include/asm/perf_event.h
+++ b/arch/powerpc/include/asm/perf_event.h
@@ -26,6 +26,8 @@
26#include <asm/ptrace.h> 26#include <asm/ptrace.h>
27#include <asm/reg.h> 27#include <asm/reg.h>
28 28
29#define perf_arch_bpf_user_pt_regs(regs) &regs->user_regs
30
29/* 31/*
30 * Overload regs->result to specify whether we should use the MSR (result 32 * Overload regs->result to specify whether we should use the MSR (result
31 * is zero) or the SIAR (result is non zero). 33 * is zero) or the SIAR (result is non zero).
diff --git a/arch/powerpc/include/uapi/asm/Kbuild b/arch/powerpc/include/uapi/asm/Kbuild
index a658091a19f9..3712152206f3 100644
--- a/arch/powerpc/include/uapi/asm/Kbuild
+++ b/arch/powerpc/include/uapi/asm/Kbuild
@@ -1,7 +1,6 @@
1# UAPI Header export list 1# UAPI Header export list
2include include/uapi/asm-generic/Kbuild.asm 2include include/uapi/asm-generic/Kbuild.asm
3 3
4generic-y += bpf_perf_event.h
5generic-y += param.h 4generic-y += param.h
6generic-y += poll.h 5generic-y += poll.h
7generic-y += resource.h 6generic-y += resource.h
diff --git a/arch/powerpc/include/uapi/asm/bpf_perf_event.h b/arch/powerpc/include/uapi/asm/bpf_perf_event.h
new file mode 100644
index 000000000000..b551b741653d
--- /dev/null
+++ b/arch/powerpc/include/uapi/asm/bpf_perf_event.h
@@ -0,0 +1,9 @@
1/* SPDX-License-Identifier: GPL-2.0 */
2#ifndef _UAPI__ASM_BPF_PERF_EVENT_H__
3#define _UAPI__ASM_BPF_PERF_EVENT_H__
4
5#include <asm/ptrace.h>
6
7typedef struct user_pt_regs bpf_user_pt_regs_t;
8
9#endif /* _UAPI__ASM_BPF_PERF_EVENT_H__ */
diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
index 89d32bb79d5e..db2691ff4c0b 100644
--- a/arch/powerpc/kernel/exceptions-64s.S
+++ b/arch/powerpc/kernel/exceptions-64s.S
@@ -995,7 +995,16 @@ EXC_COMMON_BEGIN(h_data_storage_common)
995 bl save_nvgprs 995 bl save_nvgprs
996 RECONCILE_IRQ_STATE(r10, r11) 996 RECONCILE_IRQ_STATE(r10, r11)
997 addi r3,r1,STACK_FRAME_OVERHEAD 997 addi r3,r1,STACK_FRAME_OVERHEAD
998BEGIN_MMU_FTR_SECTION
999 ld r4,PACA_EXGEN+EX_DAR(r13)
1000 lwz r5,PACA_EXGEN+EX_DSISR(r13)
1001 std r4,_DAR(r1)
1002 std r5,_DSISR(r1)
1003 li r5,SIGSEGV
1004 bl bad_page_fault
1005MMU_FTR_SECTION_ELSE
998 bl unknown_exception 1006 bl unknown_exception
1007ALT_MMU_FTR_SECTION_END_IFSET(MMU_FTR_TYPE_RADIX)
999 b ret_from_except 1008 b ret_from_except
1000 1009
1001 1010
diff --git a/arch/powerpc/kernel/legacy_serial.c b/arch/powerpc/kernel/legacy_serial.c
index 33b34a58fc62..5b9dce17f0c9 100644
--- a/arch/powerpc/kernel/legacy_serial.c
+++ b/arch/powerpc/kernel/legacy_serial.c
@@ -372,6 +372,8 @@ void __init find_legacy_serial_ports(void)
372 372
373 /* Now find out if one of these is out firmware console */ 373 /* Now find out if one of these is out firmware console */
374 path = of_get_property(of_chosen, "linux,stdout-path", NULL); 374 path = of_get_property(of_chosen, "linux,stdout-path", NULL);
375 if (path == NULL)
376 path = of_get_property(of_chosen, "stdout-path", NULL);
375 if (path != NULL) { 377 if (path != NULL) {
376 stdout = of_find_node_by_path(path); 378 stdout = of_find_node_by_path(path);
377 if (stdout) 379 if (stdout)
@@ -595,8 +597,10 @@ static int __init check_legacy_serial_console(void)
595 /* We are getting a weird phandle from OF ... */ 597 /* We are getting a weird phandle from OF ... */
596 /* ... So use the full path instead */ 598 /* ... So use the full path instead */
597 name = of_get_property(of_chosen, "linux,stdout-path", NULL); 599 name = of_get_property(of_chosen, "linux,stdout-path", NULL);
600 if (name == NULL)
601 name = of_get_property(of_chosen, "stdout-path", NULL);
598 if (name == NULL) { 602 if (name == NULL) {
599 DBG(" no linux,stdout-path !\n"); 603 DBG(" no stdout-path !\n");
600 return -ENODEV; 604 return -ENODEV;
601 } 605 }
602 prom_stdout = of_find_node_by_path(name); 606 prom_stdout = of_find_node_by_path(name);
diff --git a/arch/powerpc/kernel/machine_kexec_file_64.c b/arch/powerpc/kernel/machine_kexec_file_64.c
index c77e95e9b384..0d20c7ad40fa 100644
--- a/arch/powerpc/kernel/machine_kexec_file_64.c
+++ b/arch/powerpc/kernel/machine_kexec_file_64.c
@@ -24,7 +24,6 @@
24 24
25#include <linux/slab.h> 25#include <linux/slab.h>
26#include <linux/kexec.h> 26#include <linux/kexec.h>
27#include <linux/memblock.h>
28#include <linux/of_fdt.h> 27#include <linux/of_fdt.h>
29#include <linux/libfdt.h> 28#include <linux/libfdt.h>
30#include <asm/ima.h> 29#include <asm/ima.h>
@@ -47,59 +46,6 @@ int arch_kexec_kernel_image_probe(struct kimage *image, void *buf,
47} 46}
48 47
49/** 48/**
50 * arch_kexec_walk_mem - call func(data) for each unreserved memory block
51 * @kbuf: Context info for the search. Also passed to @func.
52 * @func: Function to call for each memory block.
53 *
54 * This function is used by kexec_add_buffer and kexec_locate_mem_hole
55 * to find unreserved memory to load kexec segments into.
56 *
57 * Return: The memory walk will stop when func returns a non-zero value
58 * and that value will be returned. If all free regions are visited without
59 * func returning non-zero, then zero will be returned.
60 */
61int arch_kexec_walk_mem(struct kexec_buf *kbuf,
62 int (*func)(struct resource *, void *))
63{
64 int ret = 0;
65 u64 i;
66 phys_addr_t mstart, mend;
67 struct resource res = { };
68
69 if (kbuf->top_down) {
70 for_each_free_mem_range_reverse(i, NUMA_NO_NODE, 0,
71 &mstart, &mend, NULL) {
72 /*
73 * In memblock, end points to the first byte after the
74 * range while in kexec, end points to the last byte
75 * in the range.
76 */
77 res.start = mstart;
78 res.end = mend - 1;
79 ret = func(&res, kbuf);
80 if (ret)
81 break;
82 }
83 } else {
84 for_each_free_mem_range(i, NUMA_NO_NODE, 0, &mstart, &mend,
85 NULL) {
86 /*
87 * In memblock, end points to the first byte after the
88 * range while in kexec, end points to the last byte
89 * in the range.
90 */
91 res.start = mstart;
92 res.end = mend - 1;
93 ret = func(&res, kbuf);
94 if (ret)
95 break;
96 }
97 }
98
99 return ret;
100}
101
102/**
103 * setup_purgatory - initialize the purgatory's global variables 49 * setup_purgatory - initialize the purgatory's global variables
104 * @image: kexec image. 50 * @image: kexec image.
105 * @slave_code: Slave code for the purgatory. 51 * @slave_code: Slave code for the purgatory.
diff --git a/arch/powerpc/kernel/msi.c b/arch/powerpc/kernel/msi.c
index dab616a33b8d..f2197654be07 100644
--- a/arch/powerpc/kernel/msi.c
+++ b/arch/powerpc/kernel/msi.c
@@ -34,5 +34,10 @@ void arch_teardown_msi_irqs(struct pci_dev *dev)
34{ 34{
35 struct pci_controller *phb = pci_bus_to_host(dev->bus); 35 struct pci_controller *phb = pci_bus_to_host(dev->bus);
36 36
37 phb->controller_ops.teardown_msi_irqs(dev); 37 /*
38 * We can be called even when arch_setup_msi_irqs() returns -ENOSYS,
39 * so check the pointer again.
40 */
41 if (phb->controller_ops.teardown_msi_irqs)
42 phb->controller_ops.teardown_msi_irqs(dev);
38} 43}
diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
index afb819f4ca68..714c3480c52d 100644
--- a/arch/powerpc/kernel/ptrace.c
+++ b/arch/powerpc/kernel/ptrace.c
@@ -3266,12 +3266,17 @@ long do_syscall_trace_enter(struct pt_regs *regs)
3266 user_exit(); 3266 user_exit();
3267 3267
3268 if (test_thread_flag(TIF_SYSCALL_EMU)) { 3268 if (test_thread_flag(TIF_SYSCALL_EMU)) {
3269 ptrace_report_syscall(regs);
3270 /* 3269 /*
3270 * A nonzero return code from tracehook_report_syscall_entry()
3271 * tells us to prevent the syscall execution, but we are not
3272 * going to execute it anyway.
3273 *
3271 * Returning -1 will skip the syscall execution. We want to 3274 * Returning -1 will skip the syscall execution. We want to
3272 * avoid clobbering any register also, thus, not 'gotoing' 3275 * avoid clobbering any register also, thus, not 'gotoing'
3273 * skip label. 3276 * skip label.
3274 */ 3277 */
3278 if (tracehook_report_syscall_entry(regs))
3279 ;
3275 return -1; 3280 return -1;
3276 } 3281 }
3277 3282
diff --git a/arch/powerpc/kvm/book3s.c b/arch/powerpc/kvm/book3s.c
index fd9893bc7aa1..bd1a677dd9e4 100644
--- a/arch/powerpc/kvm/book3s.c
+++ b/arch/powerpc/kvm/book3s.c
@@ -830,9 +830,10 @@ int kvmppc_core_prepare_memory_region(struct kvm *kvm,
830void kvmppc_core_commit_memory_region(struct kvm *kvm, 830void kvmppc_core_commit_memory_region(struct kvm *kvm,
831 const struct kvm_userspace_memory_region *mem, 831 const struct kvm_userspace_memory_region *mem,
832 const struct kvm_memory_slot *old, 832 const struct kvm_memory_slot *old,
833 const struct kvm_memory_slot *new) 833 const struct kvm_memory_slot *new,
834 enum kvm_mr_change change)
834{ 835{
835 kvm->arch.kvm_ops->commit_memory_region(kvm, mem, old, new); 836 kvm->arch.kvm_ops->commit_memory_region(kvm, mem, old, new, change);
836} 837}
837 838
838int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end) 839int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end)
@@ -850,9 +851,10 @@ int kvm_test_age_hva(struct kvm *kvm, unsigned long hva)
850 return kvm->arch.kvm_ops->test_age_hva(kvm, hva); 851 return kvm->arch.kvm_ops->test_age_hva(kvm, hva);
851} 852}
852 853
853void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte) 854int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
854{ 855{
855 kvm->arch.kvm_ops->set_spte_hva(kvm, hva, pte); 856 kvm->arch.kvm_ops->set_spte_hva(kvm, hva, pte);
857 return 0;
856} 858}
857 859
858void kvmppc_mmu_destroy(struct kvm_vcpu *vcpu) 860void kvmppc_mmu_destroy(struct kvm_vcpu *vcpu)
diff --git a/arch/powerpc/kvm/book3s_64_mmu_hv.c b/arch/powerpc/kvm/book3s_64_mmu_hv.c
index c615617e78ac..6f2d2fb4e098 100644
--- a/arch/powerpc/kvm/book3s_64_mmu_hv.c
+++ b/arch/powerpc/kvm/book3s_64_mmu_hv.c
@@ -743,12 +743,15 @@ void kvmppc_rmap_reset(struct kvm *kvm)
743 srcu_idx = srcu_read_lock(&kvm->srcu); 743 srcu_idx = srcu_read_lock(&kvm->srcu);
744 slots = kvm_memslots(kvm); 744 slots = kvm_memslots(kvm);
745 kvm_for_each_memslot(memslot, slots) { 745 kvm_for_each_memslot(memslot, slots) {
746 /* Mutual exclusion with kvm_unmap_hva_range etc. */
747 spin_lock(&kvm->mmu_lock);
746 /* 748 /*
747 * This assumes it is acceptable to lose reference and 749 * This assumes it is acceptable to lose reference and
748 * change bits across a reset. 750 * change bits across a reset.
749 */ 751 */
750 memset(memslot->arch.rmap, 0, 752 memset(memslot->arch.rmap, 0,
751 memslot->npages * sizeof(*memslot->arch.rmap)); 753 memslot->npages * sizeof(*memslot->arch.rmap));
754 spin_unlock(&kvm->mmu_lock);
752 } 755 }
753 srcu_read_unlock(&kvm->srcu, srcu_idx); 756 srcu_read_unlock(&kvm->srcu, srcu_idx);
754} 757}
@@ -896,11 +899,12 @@ void kvmppc_core_flush_memslot_hv(struct kvm *kvm,
896 899
897 gfn = memslot->base_gfn; 900 gfn = memslot->base_gfn;
898 rmapp = memslot->arch.rmap; 901 rmapp = memslot->arch.rmap;
902 if (kvm_is_radix(kvm)) {
903 kvmppc_radix_flush_memslot(kvm, memslot);
904 return;
905 }
906
899 for (n = memslot->npages; n; --n, ++gfn) { 907 for (n = memslot->npages; n; --n, ++gfn) {
900 if (kvm_is_radix(kvm)) {
901 kvm_unmap_radix(kvm, memslot, gfn);
902 continue;
903 }
904 /* 908 /*
905 * Testing the present bit without locking is OK because 909 * Testing the present bit without locking is OK because
906 * the memslot has been marked invalid already, and hence 910 * the memslot has been marked invalid already, and hence
diff --git a/arch/powerpc/kvm/book3s_64_mmu_radix.c b/arch/powerpc/kvm/book3s_64_mmu_radix.c
index d68162ee159b..fb88167a402a 100644
--- a/arch/powerpc/kvm/book3s_64_mmu_radix.c
+++ b/arch/powerpc/kvm/book3s_64_mmu_radix.c
@@ -29,6 +29,103 @@
29 */ 29 */
30static int p9_supported_radix_bits[4] = { 5, 9, 9, 13 }; 30static int p9_supported_radix_bits[4] = { 5, 9, 9, 13 };
31 31
32unsigned long __kvmhv_copy_tofrom_guest_radix(int lpid, int pid,
33 gva_t eaddr, void *to, void *from,
34 unsigned long n)
35{
36 unsigned long quadrant, ret = n;
37 int old_pid, old_lpid;
38 bool is_load = !!to;
39
40 /* Can't access quadrants 1 or 2 in non-HV mode, call the HV to do it */
41 if (kvmhv_on_pseries())
42 return plpar_hcall_norets(H_COPY_TOFROM_GUEST, lpid, pid, eaddr,
43 __pa(to), __pa(from), n);
44
45 quadrant = 1;
46 if (!pid)
47 quadrant = 2;
48 if (is_load)
49 from = (void *) (eaddr | (quadrant << 62));
50 else
51 to = (void *) (eaddr | (quadrant << 62));
52
53 preempt_disable();
54
55 /* switch the lpid first to avoid running host with unallocated pid */
56 old_lpid = mfspr(SPRN_LPID);
57 if (old_lpid != lpid)
58 mtspr(SPRN_LPID, lpid);
59 if (quadrant == 1) {
60 old_pid = mfspr(SPRN_PID);
61 if (old_pid != pid)
62 mtspr(SPRN_PID, pid);
63 }
64 isync();
65
66 pagefault_disable();
67 if (is_load)
68 ret = raw_copy_from_user(to, from, n);
69 else
70 ret = raw_copy_to_user(to, from, n);
71 pagefault_enable();
72
73 /* switch the pid first to avoid running host with unallocated pid */
74 if (quadrant == 1 && pid != old_pid)
75 mtspr(SPRN_PID, old_pid);
76 if (lpid != old_lpid)
77 mtspr(SPRN_LPID, old_lpid);
78 isync();
79
80 preempt_enable();
81
82 return ret;
83}
84EXPORT_SYMBOL_GPL(__kvmhv_copy_tofrom_guest_radix);
85
86static long kvmhv_copy_tofrom_guest_radix(struct kvm_vcpu *vcpu, gva_t eaddr,
87 void *to, void *from, unsigned long n)
88{
89 int lpid = vcpu->kvm->arch.lpid;
90 int pid = vcpu->arch.pid;
91
92 /* This would cause a data segment intr so don't allow the access */
93 if (eaddr & (0x3FFUL << 52))
94 return -EINVAL;
95
96 /* Should we be using the nested lpid */
97 if (vcpu->arch.nested)
98 lpid = vcpu->arch.nested->shadow_lpid;
99
100 /* If accessing quadrant 3 then pid is expected to be 0 */
101 if (((eaddr >> 62) & 0x3) == 0x3)
102 pid = 0;
103
104 eaddr &= ~(0xFFFUL << 52);
105
106 return __kvmhv_copy_tofrom_guest_radix(lpid, pid, eaddr, to, from, n);
107}
108
109long kvmhv_copy_from_guest_radix(struct kvm_vcpu *vcpu, gva_t eaddr, void *to,
110 unsigned long n)
111{
112 long ret;
113
114 ret = kvmhv_copy_tofrom_guest_radix(vcpu, eaddr, to, NULL, n);
115 if (ret > 0)
116 memset(to + (n - ret), 0, ret);
117
118 return ret;
119}
120EXPORT_SYMBOL_GPL(kvmhv_copy_from_guest_radix);
121
122long kvmhv_copy_to_guest_radix(struct kvm_vcpu *vcpu, gva_t eaddr, void *from,
123 unsigned long n)
124{
125 return kvmhv_copy_tofrom_guest_radix(vcpu, eaddr, NULL, from, n);
126}
127EXPORT_SYMBOL_GPL(kvmhv_copy_to_guest_radix);
128
32int kvmppc_mmu_walk_radix_tree(struct kvm_vcpu *vcpu, gva_t eaddr, 129int kvmppc_mmu_walk_radix_tree(struct kvm_vcpu *vcpu, gva_t eaddr,
33 struct kvmppc_pte *gpte, u64 root, 130 struct kvmppc_pte *gpte, u64 root,
34 u64 *pte_ret_p) 131 u64 *pte_ret_p)
@@ -197,8 +294,8 @@ int kvmppc_mmu_radix_xlate(struct kvm_vcpu *vcpu, gva_t eaddr,
197 return 0; 294 return 0;
198} 295}
199 296
200static void kvmppc_radix_tlbie_page(struct kvm *kvm, unsigned long addr, 297void kvmppc_radix_tlbie_page(struct kvm *kvm, unsigned long addr,
201 unsigned int pshift, unsigned int lpid) 298 unsigned int pshift, unsigned int lpid)
202{ 299{
203 unsigned long psize = PAGE_SIZE; 300 unsigned long psize = PAGE_SIZE;
204 int psi; 301 int psi;
@@ -284,7 +381,8 @@ static void kvmppc_pmd_free(pmd_t *pmdp)
284 381
285/* Called with kvm->mmu_lock held */ 382/* Called with kvm->mmu_lock held */
286void kvmppc_unmap_pte(struct kvm *kvm, pte_t *pte, unsigned long gpa, 383void kvmppc_unmap_pte(struct kvm *kvm, pte_t *pte, unsigned long gpa,
287 unsigned int shift, struct kvm_memory_slot *memslot, 384 unsigned int shift,
385 const struct kvm_memory_slot *memslot,
288 unsigned int lpid) 386 unsigned int lpid)
289 387
290{ 388{
@@ -683,6 +781,7 @@ int kvmppc_book3s_instantiate_page(struct kvm_vcpu *vcpu,
683 pte_t pte, *ptep; 781 pte_t pte, *ptep;
684 unsigned int shift, level; 782 unsigned int shift, level;
685 int ret; 783 int ret;
784 bool large_enable;
686 785
687 /* used to check for invalidations in progress */ 786 /* used to check for invalidations in progress */
688 mmu_seq = kvm->mmu_notifier_seq; 787 mmu_seq = kvm->mmu_notifier_seq;
@@ -732,12 +831,15 @@ int kvmppc_book3s_instantiate_page(struct kvm_vcpu *vcpu,
732 pte = *ptep; 831 pte = *ptep;
733 local_irq_enable(); 832 local_irq_enable();
734 833
834 /* If we're logging dirty pages, always map single pages */
835 large_enable = !(memslot->flags & KVM_MEM_LOG_DIRTY_PAGES);
836
735 /* Get pte level from shift/size */ 837 /* Get pte level from shift/size */
736 if (shift == PUD_SHIFT && 838 if (large_enable && shift == PUD_SHIFT &&
737 (gpa & (PUD_SIZE - PAGE_SIZE)) == 839 (gpa & (PUD_SIZE - PAGE_SIZE)) ==
738 (hva & (PUD_SIZE - PAGE_SIZE))) { 840 (hva & (PUD_SIZE - PAGE_SIZE))) {
739 level = 2; 841 level = 2;
740 } else if (shift == PMD_SHIFT && 842 } else if (large_enable && shift == PMD_SHIFT &&
741 (gpa & (PMD_SIZE - PAGE_SIZE)) == 843 (gpa & (PMD_SIZE - PAGE_SIZE)) ==
742 (hva & (PMD_SIZE - PAGE_SIZE))) { 844 (hva & (PMD_SIZE - PAGE_SIZE))) {
743 level = 1; 845 level = 1;
@@ -857,7 +959,7 @@ int kvmppc_book3s_radix_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu,
857 return ret; 959 return ret;
858} 960}
859 961
860/* Called with kvm->lock held */ 962/* Called with kvm->mmu_lock held */
861int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot, 963int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
862 unsigned long gfn) 964 unsigned long gfn)
863{ 965{
@@ -872,7 +974,7 @@ int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
872 return 0; 974 return 0;
873} 975}
874 976
875/* Called with kvm->lock held */ 977/* Called with kvm->mmu_lock held */
876int kvm_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot, 978int kvm_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
877 unsigned long gfn) 979 unsigned long gfn)
878{ 980{
@@ -880,18 +982,24 @@ int kvm_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
880 unsigned long gpa = gfn << PAGE_SHIFT; 982 unsigned long gpa = gfn << PAGE_SHIFT;
881 unsigned int shift; 983 unsigned int shift;
882 int ref = 0; 984 int ref = 0;
985 unsigned long old, *rmapp;
883 986
884 ptep = __find_linux_pte(kvm->arch.pgtable, gpa, NULL, &shift); 987 ptep = __find_linux_pte(kvm->arch.pgtable, gpa, NULL, &shift);
885 if (ptep && pte_present(*ptep) && pte_young(*ptep)) { 988 if (ptep && pte_present(*ptep) && pte_young(*ptep)) {
886 kvmppc_radix_update_pte(kvm, ptep, _PAGE_ACCESSED, 0, 989 old = kvmppc_radix_update_pte(kvm, ptep, _PAGE_ACCESSED, 0,
887 gpa, shift); 990 gpa, shift);
888 /* XXX need to flush tlb here? */ 991 /* XXX need to flush tlb here? */
992 /* Also clear bit in ptes in shadow pgtable for nested guests */
993 rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
994 kvmhv_update_nest_rmap_rc_list(kvm, rmapp, _PAGE_ACCESSED, 0,
995 old & PTE_RPN_MASK,
996 1UL << shift);
889 ref = 1; 997 ref = 1;
890 } 998 }
891 return ref; 999 return ref;
892} 1000}
893 1001
894/* Called with kvm->lock held */ 1002/* Called with kvm->mmu_lock held */
895int kvm_test_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot, 1003int kvm_test_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
896 unsigned long gfn) 1004 unsigned long gfn)
897{ 1005{
@@ -915,15 +1023,23 @@ static int kvm_radix_test_clear_dirty(struct kvm *kvm,
915 pte_t *ptep; 1023 pte_t *ptep;
916 unsigned int shift; 1024 unsigned int shift;
917 int ret = 0; 1025 int ret = 0;
1026 unsigned long old, *rmapp;
918 1027
919 ptep = __find_linux_pte(kvm->arch.pgtable, gpa, NULL, &shift); 1028 ptep = __find_linux_pte(kvm->arch.pgtable, gpa, NULL, &shift);
920 if (ptep && pte_present(*ptep) && pte_dirty(*ptep)) { 1029 if (ptep && pte_present(*ptep) && pte_dirty(*ptep)) {
921 ret = 1; 1030 ret = 1;
922 if (shift) 1031 if (shift)
923 ret = 1 << (shift - PAGE_SHIFT); 1032 ret = 1 << (shift - PAGE_SHIFT);
924 kvmppc_radix_update_pte(kvm, ptep, _PAGE_DIRTY, 0, 1033 spin_lock(&kvm->mmu_lock);
925 gpa, shift); 1034 old = kvmppc_radix_update_pte(kvm, ptep, _PAGE_DIRTY, 0,
1035 gpa, shift);
926 kvmppc_radix_tlbie_page(kvm, gpa, shift, kvm->arch.lpid); 1036 kvmppc_radix_tlbie_page(kvm, gpa, shift, kvm->arch.lpid);
1037 /* Also clear bit in ptes in shadow pgtable for nested guests */
1038 rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
1039 kvmhv_update_nest_rmap_rc_list(kvm, rmapp, _PAGE_DIRTY, 0,
1040 old & PTE_RPN_MASK,
1041 1UL << shift);
1042 spin_unlock(&kvm->mmu_lock);
927 } 1043 }
928 return ret; 1044 return ret;
929} 1045}
@@ -953,6 +1069,26 @@ long kvmppc_hv_get_dirty_log_radix(struct kvm *kvm,
953 return 0; 1069 return 0;
954} 1070}
955 1071
1072void kvmppc_radix_flush_memslot(struct kvm *kvm,
1073 const struct kvm_memory_slot *memslot)
1074{
1075 unsigned long n;
1076 pte_t *ptep;
1077 unsigned long gpa;
1078 unsigned int shift;
1079
1080 gpa = memslot->base_gfn << PAGE_SHIFT;
1081 spin_lock(&kvm->mmu_lock);
1082 for (n = memslot->npages; n; --n) {
1083 ptep = __find_linux_pte(kvm->arch.pgtable, gpa, NULL, &shift);
1084 if (ptep && pte_present(*ptep))
1085 kvmppc_unmap_pte(kvm, ptep, gpa, shift, memslot,
1086 kvm->arch.lpid);
1087 gpa += PAGE_SIZE;
1088 }
1089 spin_unlock(&kvm->mmu_lock);
1090}
1091
956static void add_rmmu_ap_encoding(struct kvm_ppc_rmmu_info *info, 1092static void add_rmmu_ap_encoding(struct kvm_ppc_rmmu_info *info,
957 int psize, int *indexp) 1093 int psize, int *indexp)
958{ 1094{
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index a56f8413758a..5a066fc299e1 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -985,6 +985,10 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
985 kvmppc_set_gpr(vcpu, 3, 0); 985 kvmppc_set_gpr(vcpu, 3, 0);
986 vcpu->arch.hcall_needed = 0; 986 vcpu->arch.hcall_needed = 0;
987 return -EINTR; 987 return -EINTR;
988 } else if (ret == H_TOO_HARD) {
989 kvmppc_set_gpr(vcpu, 3, 0);
990 vcpu->arch.hcall_needed = 0;
991 return RESUME_HOST;
988 } 992 }
989 break; 993 break;
990 case H_TLB_INVALIDATE: 994 case H_TLB_INVALIDATE:
@@ -992,7 +996,11 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
992 if (nesting_enabled(vcpu->kvm)) 996 if (nesting_enabled(vcpu->kvm))
993 ret = kvmhv_do_nested_tlbie(vcpu); 997 ret = kvmhv_do_nested_tlbie(vcpu);
994 break; 998 break;
995 999 case H_COPY_TOFROM_GUEST:
1000 ret = H_FUNCTION;
1001 if (nesting_enabled(vcpu->kvm))
1002 ret = kvmhv_copy_tofrom_guest_nested(vcpu);
1003 break;
996 default: 1004 default:
997 return RESUME_HOST; 1005 return RESUME_HOST;
998 } 1006 }
@@ -1336,7 +1344,7 @@ static int kvmppc_handle_exit_hv(struct kvm_run *run, struct kvm_vcpu *vcpu,
1336 return r; 1344 return r;
1337} 1345}
1338 1346
1339static int kvmppc_handle_nested_exit(struct kvm_vcpu *vcpu) 1347static int kvmppc_handle_nested_exit(struct kvm_run *run, struct kvm_vcpu *vcpu)
1340{ 1348{
1341 int r; 1349 int r;
1342 int srcu_idx; 1350 int srcu_idx;
@@ -1394,7 +1402,7 @@ static int kvmppc_handle_nested_exit(struct kvm_vcpu *vcpu)
1394 */ 1402 */
1395 case BOOK3S_INTERRUPT_H_DATA_STORAGE: 1403 case BOOK3S_INTERRUPT_H_DATA_STORAGE:
1396 srcu_idx = srcu_read_lock(&vcpu->kvm->srcu); 1404 srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
1397 r = kvmhv_nested_page_fault(vcpu); 1405 r = kvmhv_nested_page_fault(run, vcpu);
1398 srcu_read_unlock(&vcpu->kvm->srcu, srcu_idx); 1406 srcu_read_unlock(&vcpu->kvm->srcu, srcu_idx);
1399 break; 1407 break;
1400 case BOOK3S_INTERRUPT_H_INST_STORAGE: 1408 case BOOK3S_INTERRUPT_H_INST_STORAGE:
@@ -1404,7 +1412,7 @@ static int kvmppc_handle_nested_exit(struct kvm_vcpu *vcpu)
1404 if (vcpu->arch.shregs.msr & HSRR1_HISI_WRITE) 1412 if (vcpu->arch.shregs.msr & HSRR1_HISI_WRITE)
1405 vcpu->arch.fault_dsisr |= DSISR_ISSTORE; 1413 vcpu->arch.fault_dsisr |= DSISR_ISSTORE;
1406 srcu_idx = srcu_read_lock(&vcpu->kvm->srcu); 1414 srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
1407 r = kvmhv_nested_page_fault(vcpu); 1415 r = kvmhv_nested_page_fault(run, vcpu);
1408 srcu_read_unlock(&vcpu->kvm->srcu, srcu_idx); 1416 srcu_read_unlock(&vcpu->kvm->srcu, srcu_idx);
1409 break; 1417 break;
1410 1418
@@ -4059,7 +4067,7 @@ int kvmhv_run_single_vcpu(struct kvm_run *kvm_run,
4059 if (!nested) 4067 if (!nested)
4060 r = kvmppc_handle_exit_hv(kvm_run, vcpu, current); 4068 r = kvmppc_handle_exit_hv(kvm_run, vcpu, current);
4061 else 4069 else
4062 r = kvmppc_handle_nested_exit(vcpu); 4070 r = kvmppc_handle_nested_exit(kvm_run, vcpu);
4063 } 4071 }
4064 vcpu->arch.ret = r; 4072 vcpu->arch.ret = r;
4065 4073
@@ -4371,7 +4379,8 @@ static int kvmppc_core_prepare_memory_region_hv(struct kvm *kvm,
4371static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm, 4379static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,
4372 const struct kvm_userspace_memory_region *mem, 4380 const struct kvm_userspace_memory_region *mem,
4373 const struct kvm_memory_slot *old, 4381 const struct kvm_memory_slot *old,
4374 const struct kvm_memory_slot *new) 4382 const struct kvm_memory_slot *new,
4383 enum kvm_mr_change change)
4375{ 4384{
4376 unsigned long npages = mem->memory_size >> PAGE_SHIFT; 4385 unsigned long npages = mem->memory_size >> PAGE_SHIFT;
4377 4386
@@ -4383,6 +4392,23 @@ static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,
4383 */ 4392 */
4384 if (npages) 4393 if (npages)
4385 atomic64_inc(&kvm->arch.mmio_update); 4394 atomic64_inc(&kvm->arch.mmio_update);
4395
4396 /*
4397 * For change == KVM_MR_MOVE or KVM_MR_DELETE, higher levels
4398 * have already called kvm_arch_flush_shadow_memslot() to
4399 * flush shadow mappings. For KVM_MR_CREATE we have no
4400 * previous mappings. So the only case to handle is
4401 * KVM_MR_FLAGS_ONLY when the KVM_MEM_LOG_DIRTY_PAGES bit
4402 * has been changed.
4403 * For radix guests, we flush on setting KVM_MEM_LOG_DIRTY_PAGES
4404 * to get rid of any THP PTEs in the partition-scoped page tables
4405 * so we can track dirtiness at the page level; we flush when
4406 * clearing KVM_MEM_LOG_DIRTY_PAGES so that we can go back to
4407 * using THP PTEs.
4408 */
4409 if (change == KVM_MR_FLAGS_ONLY && kvm_is_radix(kvm) &&
4410 ((new->flags ^ old->flags) & KVM_MEM_LOG_DIRTY_PAGES))
4411 kvmppc_radix_flush_memslot(kvm, old);
4386} 4412}
4387 4413
4388/* 4414/*
@@ -4532,12 +4558,15 @@ int kvmppc_switch_mmu_to_hpt(struct kvm *kvm)
4532{ 4558{
4533 if (nesting_enabled(kvm)) 4559 if (nesting_enabled(kvm))
4534 kvmhv_release_all_nested(kvm); 4560 kvmhv_release_all_nested(kvm);
4561 kvmppc_rmap_reset(kvm);
4562 kvm->arch.process_table = 0;
4563 /* Mutual exclusion with kvm_unmap_hva_range etc. */
4564 spin_lock(&kvm->mmu_lock);
4565 kvm->arch.radix = 0;
4566 spin_unlock(&kvm->mmu_lock);
4535 kvmppc_free_radix(kvm); 4567 kvmppc_free_radix(kvm);
4536 kvmppc_update_lpcr(kvm, LPCR_VPM1, 4568 kvmppc_update_lpcr(kvm, LPCR_VPM1,
4537 LPCR_VPM1 | LPCR_UPRT | LPCR_GTSE | LPCR_HR); 4569 LPCR_VPM1 | LPCR_UPRT | LPCR_GTSE | LPCR_HR);
4538 kvmppc_rmap_reset(kvm);
4539 kvm->arch.radix = 0;
4540 kvm->arch.process_table = 0;
4541 return 0; 4570 return 0;
4542} 4571}
4543 4572
@@ -4549,12 +4578,14 @@ int kvmppc_switch_mmu_to_radix(struct kvm *kvm)
4549 err = kvmppc_init_vm_radix(kvm); 4578 err = kvmppc_init_vm_radix(kvm);
4550 if (err) 4579 if (err)
4551 return err; 4580 return err;
4552 4581 kvmppc_rmap_reset(kvm);
4582 /* Mutual exclusion with kvm_unmap_hva_range etc. */
4583 spin_lock(&kvm->mmu_lock);
4584 kvm->arch.radix = 1;
4585 spin_unlock(&kvm->mmu_lock);
4553 kvmppc_free_hpt(&kvm->arch.hpt); 4586 kvmppc_free_hpt(&kvm->arch.hpt);
4554 kvmppc_update_lpcr(kvm, LPCR_UPRT | LPCR_GTSE | LPCR_HR, 4587 kvmppc_update_lpcr(kvm, LPCR_UPRT | LPCR_GTSE | LPCR_HR,
4555 LPCR_VPM1 | LPCR_UPRT | LPCR_GTSE | LPCR_HR); 4588 LPCR_VPM1 | LPCR_UPRT | LPCR_GTSE | LPCR_HR);
4556 kvmppc_rmap_reset(kvm);
4557 kvm->arch.radix = 1;
4558 return 0; 4589 return 0;
4559} 4590}
4560 4591
@@ -5214,6 +5245,44 @@ static int kvmhv_enable_nested(struct kvm *kvm)
5214 return 0; 5245 return 0;
5215} 5246}
5216 5247
5248static int kvmhv_load_from_eaddr(struct kvm_vcpu *vcpu, ulong *eaddr, void *ptr,
5249 int size)
5250{
5251 int rc = -EINVAL;
5252
5253 if (kvmhv_vcpu_is_radix(vcpu)) {
5254 rc = kvmhv_copy_from_guest_radix(vcpu, *eaddr, ptr, size);
5255
5256 if (rc > 0)
5257 rc = -EINVAL;
5258 }
5259
5260 /* For now quadrants are the only way to access nested guest memory */
5261 if (rc && vcpu->arch.nested)
5262 rc = -EAGAIN;
5263
5264 return rc;
5265}
5266
5267static int kvmhv_store_to_eaddr(struct kvm_vcpu *vcpu, ulong *eaddr, void *ptr,
5268 int size)
5269{
5270 int rc = -EINVAL;
5271
5272 if (kvmhv_vcpu_is_radix(vcpu)) {
5273 rc = kvmhv_copy_to_guest_radix(vcpu, *eaddr, ptr, size);
5274
5275 if (rc > 0)
5276 rc = -EINVAL;
5277 }
5278
5279 /* For now quadrants are the only way to access nested guest memory */
5280 if (rc && vcpu->arch.nested)
5281 rc = -EAGAIN;
5282
5283 return rc;
5284}
5285
5217static struct kvmppc_ops kvm_ops_hv = { 5286static struct kvmppc_ops kvm_ops_hv = {
5218 .get_sregs = kvm_arch_vcpu_ioctl_get_sregs_hv, 5287 .get_sregs = kvm_arch_vcpu_ioctl_get_sregs_hv,
5219 .set_sregs = kvm_arch_vcpu_ioctl_set_sregs_hv, 5288 .set_sregs = kvm_arch_vcpu_ioctl_set_sregs_hv,
@@ -5254,6 +5323,8 @@ static struct kvmppc_ops kvm_ops_hv = {
5254 .get_rmmu_info = kvmhv_get_rmmu_info, 5323 .get_rmmu_info = kvmhv_get_rmmu_info,
5255 .set_smt_mode = kvmhv_set_smt_mode, 5324 .set_smt_mode = kvmhv_set_smt_mode,
5256 .enable_nested = kvmhv_enable_nested, 5325 .enable_nested = kvmhv_enable_nested,
5326 .load_from_eaddr = kvmhv_load_from_eaddr,
5327 .store_to_eaddr = kvmhv_store_to_eaddr,
5257}; 5328};
5258 5329
5259static int kvm_init_subcore_bitmap(void) 5330static int kvm_init_subcore_bitmap(void)
diff --git a/arch/powerpc/kvm/book3s_hv_nested.c b/arch/powerpc/kvm/book3s_hv_nested.c
index 401d2ecbebc5..735e0ac6f5b2 100644
--- a/arch/powerpc/kvm/book3s_hv_nested.c
+++ b/arch/powerpc/kvm/book3s_hv_nested.c
@@ -195,6 +195,26 @@ void kvmhv_restore_hv_return_state(struct kvm_vcpu *vcpu,
195 vcpu->arch.ppr = hr->ppr; 195 vcpu->arch.ppr = hr->ppr;
196} 196}
197 197
198static void kvmhv_nested_mmio_needed(struct kvm_vcpu *vcpu, u64 regs_ptr)
199{
200 /* No need to reflect the page fault to L1, we've handled it */
201 vcpu->arch.trap = 0;
202
203 /*
204 * Since the L2 gprs have already been written back into L1 memory when
205 * we complete the mmio, store the L1 memory location of the L2 gpr
206 * being loaded into by the mmio so that the loaded value can be
207 * written there in kvmppc_complete_mmio_load()
208 */
209 if (((vcpu->arch.io_gpr & KVM_MMIO_REG_EXT_MASK) == KVM_MMIO_REG_GPR)
210 && (vcpu->mmio_is_write == 0)) {
211 vcpu->arch.nested_io_gpr = (gpa_t) regs_ptr +
212 offsetof(struct pt_regs,
213 gpr[vcpu->arch.io_gpr]);
214 vcpu->arch.io_gpr = KVM_MMIO_REG_NESTED_GPR;
215 }
216}
217
198long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu) 218long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu)
199{ 219{
200 long int err, r; 220 long int err, r;
@@ -316,6 +336,11 @@ long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu)
316 if (r == -EINTR) 336 if (r == -EINTR)
317 return H_INTERRUPT; 337 return H_INTERRUPT;
318 338
339 if (vcpu->mmio_needed) {
340 kvmhv_nested_mmio_needed(vcpu, regs_ptr);
341 return H_TOO_HARD;
342 }
343
319 return vcpu->arch.trap; 344 return vcpu->arch.trap;
320} 345}
321 346
@@ -437,6 +462,81 @@ long kvmhv_set_partition_table(struct kvm_vcpu *vcpu)
437} 462}
438 463
439/* 464/*
465 * Handle the H_COPY_TOFROM_GUEST hcall.
466 * r4 = L1 lpid of nested guest
467 * r5 = pid
468 * r6 = eaddr to access
469 * r7 = to buffer (L1 gpa)
470 * r8 = from buffer (L1 gpa)
471 * r9 = n bytes to copy
472 */
473long kvmhv_copy_tofrom_guest_nested(struct kvm_vcpu *vcpu)
474{
475 struct kvm_nested_guest *gp;
476 int l1_lpid = kvmppc_get_gpr(vcpu, 4);
477 int pid = kvmppc_get_gpr(vcpu, 5);
478 gva_t eaddr = kvmppc_get_gpr(vcpu, 6);
479 gpa_t gp_to = (gpa_t) kvmppc_get_gpr(vcpu, 7);
480 gpa_t gp_from = (gpa_t) kvmppc_get_gpr(vcpu, 8);
481 void *buf;
482 unsigned long n = kvmppc_get_gpr(vcpu, 9);
483 bool is_load = !!gp_to;
484 long rc;
485
486 if (gp_to && gp_from) /* One must be NULL to determine the direction */
487 return H_PARAMETER;
488
489 if (eaddr & (0xFFFUL << 52))
490 return H_PARAMETER;
491
492 buf = kzalloc(n, GFP_KERNEL);
493 if (!buf)
494 return H_NO_MEM;
495
496 gp = kvmhv_get_nested(vcpu->kvm, l1_lpid, false);
497 if (!gp) {
498 rc = H_PARAMETER;
499 goto out_free;
500 }
501
502 mutex_lock(&gp->tlb_lock);
503
504 if (is_load) {
505 /* Load from the nested guest into our buffer */
506 rc = __kvmhv_copy_tofrom_guest_radix(gp->shadow_lpid, pid,
507 eaddr, buf, NULL, n);
508 if (rc)
509 goto not_found;
510
511 /* Write what was loaded into our buffer back to the L1 guest */
512 rc = kvm_vcpu_write_guest(vcpu, gp_to, buf, n);
513 if (rc)
514 goto not_found;
515 } else {
516 /* Load the data to be stored from the L1 guest into our buf */
517 rc = kvm_vcpu_read_guest(vcpu, gp_from, buf, n);
518 if (rc)
519 goto not_found;
520
521 /* Store from our buffer into the nested guest */
522 rc = __kvmhv_copy_tofrom_guest_radix(gp->shadow_lpid, pid,
523 eaddr, NULL, buf, n);
524 if (rc)
525 goto not_found;
526 }
527
528out_unlock:
529 mutex_unlock(&gp->tlb_lock);
530 kvmhv_put_nested(gp);
531out_free:
532 kfree(buf);
533 return rc;
534not_found:
535 rc = H_NOT_FOUND;
536 goto out_unlock;
537}
538
539/*
440 * Reload the partition table entry for a guest. 540 * Reload the partition table entry for a guest.
441 * Caller must hold gp->tlb_lock. 541 * Caller must hold gp->tlb_lock.
442 */ 542 */
@@ -480,6 +580,7 @@ struct kvm_nested_guest *kvmhv_alloc_nested(struct kvm *kvm, unsigned int lpid)
480 if (shadow_lpid < 0) 580 if (shadow_lpid < 0)
481 goto out_free2; 581 goto out_free2;
482 gp->shadow_lpid = shadow_lpid; 582 gp->shadow_lpid = shadow_lpid;
583 gp->radix = 1;
483 584
484 memset(gp->prev_cpu, -1, sizeof(gp->prev_cpu)); 585 memset(gp->prev_cpu, -1, sizeof(gp->prev_cpu));
485 586
@@ -687,6 +788,57 @@ void kvmhv_insert_nest_rmap(struct kvm *kvm, unsigned long *rmapp,
687 *n_rmap = NULL; 788 *n_rmap = NULL;
688} 789}
689 790
791static void kvmhv_update_nest_rmap_rc(struct kvm *kvm, u64 n_rmap,
792 unsigned long clr, unsigned long set,
793 unsigned long hpa, unsigned long mask)
794{
795 struct kvm_nested_guest *gp;
796 unsigned long gpa;
797 unsigned int shift, lpid;
798 pte_t *ptep;
799
800 gpa = n_rmap & RMAP_NESTED_GPA_MASK;
801 lpid = (n_rmap & RMAP_NESTED_LPID_MASK) >> RMAP_NESTED_LPID_SHIFT;
802 gp = kvmhv_find_nested(kvm, lpid);
803 if (!gp)
804 return;
805
806 /* Find the pte */
807 ptep = __find_linux_pte(gp->shadow_pgtable, gpa, NULL, &shift);
808 /*
809 * If the pte is present and the pfn is still the same, update the pte.
810 * If the pfn has changed then this is a stale rmap entry, the nested
811 * gpa actually points somewhere else now, and there is nothing to do.
812 * XXX A future optimisation would be to remove the rmap entry here.
813 */
814 if (ptep && pte_present(*ptep) && ((pte_val(*ptep) & mask) == hpa)) {
815 __radix_pte_update(ptep, clr, set);
816 kvmppc_radix_tlbie_page(kvm, gpa, shift, lpid);
817 }
818}
819
820/*
821 * For a given list of rmap entries, update the rc bits in all ptes in shadow
822 * page tables for nested guests which are referenced by the rmap list.
823 */
824void kvmhv_update_nest_rmap_rc_list(struct kvm *kvm, unsigned long *rmapp,
825 unsigned long clr, unsigned long set,
826 unsigned long hpa, unsigned long nbytes)
827{
828 struct llist_node *entry = ((struct llist_head *) rmapp)->first;
829 struct rmap_nested *cursor;
830 unsigned long rmap, mask;
831
832 if ((clr | set) & ~(_PAGE_DIRTY | _PAGE_ACCESSED))
833 return;
834
835 mask = PTE_RPN_MASK & ~(nbytes - 1);
836 hpa &= mask;
837
838 for_each_nest_rmap_safe(cursor, entry, &rmap)
839 kvmhv_update_nest_rmap_rc(kvm, rmap, clr, set, hpa, mask);
840}
841
690static void kvmhv_remove_nest_rmap(struct kvm *kvm, u64 n_rmap, 842static void kvmhv_remove_nest_rmap(struct kvm *kvm, u64 n_rmap,
691 unsigned long hpa, unsigned long mask) 843 unsigned long hpa, unsigned long mask)
692{ 844{
@@ -723,7 +875,7 @@ static void kvmhv_remove_nest_rmap_list(struct kvm *kvm, unsigned long *rmapp,
723 875
724/* called with kvm->mmu_lock held */ 876/* called with kvm->mmu_lock held */
725void kvmhv_remove_nest_rmap_range(struct kvm *kvm, 877void kvmhv_remove_nest_rmap_range(struct kvm *kvm,
726 struct kvm_memory_slot *memslot, 878 const struct kvm_memory_slot *memslot,
727 unsigned long gpa, unsigned long hpa, 879 unsigned long gpa, unsigned long hpa,
728 unsigned long nbytes) 880 unsigned long nbytes)
729{ 881{
@@ -1049,7 +1201,7 @@ static long kvmhv_handle_nested_set_rc(struct kvm_vcpu *vcpu,
1049 struct kvm *kvm = vcpu->kvm; 1201 struct kvm *kvm = vcpu->kvm;
1050 bool writing = !!(dsisr & DSISR_ISSTORE); 1202 bool writing = !!(dsisr & DSISR_ISSTORE);
1051 u64 pgflags; 1203 u64 pgflags;
1052 bool ret; 1204 long ret;
1053 1205
1054 /* Are the rc bits set in the L1 partition scoped pte? */ 1206 /* Are the rc bits set in the L1 partition scoped pte? */
1055 pgflags = _PAGE_ACCESSED; 1207 pgflags = _PAGE_ACCESSED;
@@ -1062,16 +1214,22 @@ static long kvmhv_handle_nested_set_rc(struct kvm_vcpu *vcpu,
1062 /* Set the rc bit in the pte of our (L0) pgtable for the L1 guest */ 1214 /* Set the rc bit in the pte of our (L0) pgtable for the L1 guest */
1063 ret = kvmppc_hv_handle_set_rc(kvm, kvm->arch.pgtable, writing, 1215 ret = kvmppc_hv_handle_set_rc(kvm, kvm->arch.pgtable, writing,
1064 gpte.raddr, kvm->arch.lpid); 1216 gpte.raddr, kvm->arch.lpid);
1065 spin_unlock(&kvm->mmu_lock); 1217 if (!ret) {
1066 if (!ret) 1218 ret = -EINVAL;
1067 return -EINVAL; 1219 goto out_unlock;
1220 }
1068 1221
1069 /* Set the rc bit in the pte of the shadow_pgtable for the nest guest */ 1222 /* Set the rc bit in the pte of the shadow_pgtable for the nest guest */
1070 ret = kvmppc_hv_handle_set_rc(kvm, gp->shadow_pgtable, writing, n_gpa, 1223 ret = kvmppc_hv_handle_set_rc(kvm, gp->shadow_pgtable, writing, n_gpa,
1071 gp->shadow_lpid); 1224 gp->shadow_lpid);
1072 if (!ret) 1225 if (!ret)
1073 return -EINVAL; 1226 ret = -EINVAL;
1074 return 0; 1227 else
1228 ret = 0;
1229
1230out_unlock:
1231 spin_unlock(&kvm->mmu_lock);
1232 return ret;
1075} 1233}
1076 1234
1077static inline int kvmppc_radix_level_to_shift(int level) 1235static inline int kvmppc_radix_level_to_shift(int level)
@@ -1099,7 +1257,8 @@ static inline int kvmppc_radix_shift_to_level(int shift)
1099} 1257}
1100 1258
1101/* called with gp->tlb_lock held */ 1259/* called with gp->tlb_lock held */
1102static long int __kvmhv_nested_page_fault(struct kvm_vcpu *vcpu, 1260static long int __kvmhv_nested_page_fault(struct kvm_run *run,
1261 struct kvm_vcpu *vcpu,
1103 struct kvm_nested_guest *gp) 1262 struct kvm_nested_guest *gp)
1104{ 1263{
1105 struct kvm *kvm = vcpu->kvm; 1264 struct kvm *kvm = vcpu->kvm;
@@ -1180,9 +1339,9 @@ static long int __kvmhv_nested_page_fault(struct kvm_vcpu *vcpu,
1180 kvmppc_core_queue_data_storage(vcpu, ea, dsisr); 1339 kvmppc_core_queue_data_storage(vcpu, ea, dsisr);
1181 return RESUME_GUEST; 1340 return RESUME_GUEST;
1182 } 1341 }
1183 /* passthrough of emulated MMIO case... */ 1342
1184 pr_err("emulated MMIO passthrough?\n"); 1343 /* passthrough of emulated MMIO case */
1185 return -EINVAL; 1344 return kvmppc_hv_emulate_mmio(run, vcpu, gpa, ea, writing);
1186 } 1345 }
1187 if (memslot->flags & KVM_MEM_READONLY) { 1346 if (memslot->flags & KVM_MEM_READONLY) {
1188 if (writing) { 1347 if (writing) {
@@ -1220,6 +1379,8 @@ static long int __kvmhv_nested_page_fault(struct kvm_vcpu *vcpu,
1220 return ret; 1379 return ret;
1221 shift = kvmppc_radix_level_to_shift(level); 1380 shift = kvmppc_radix_level_to_shift(level);
1222 } 1381 }
1382 /* Align gfn to the start of the page */
1383 gfn = (gpa & ~((1UL << shift) - 1)) >> PAGE_SHIFT;
1223 1384
1224 /* 3. Compute the pte we need to insert for nest_gpa -> host r_addr */ 1385 /* 3. Compute the pte we need to insert for nest_gpa -> host r_addr */
1225 1386
@@ -1227,6 +1388,9 @@ static long int __kvmhv_nested_page_fault(struct kvm_vcpu *vcpu,
1227 perm |= gpte.may_read ? 0UL : _PAGE_READ; 1388 perm |= gpte.may_read ? 0UL : _PAGE_READ;
1228 perm |= gpte.may_write ? 0UL : _PAGE_WRITE; 1389 perm |= gpte.may_write ? 0UL : _PAGE_WRITE;
1229 perm |= gpte.may_execute ? 0UL : _PAGE_EXEC; 1390 perm |= gpte.may_execute ? 0UL : _PAGE_EXEC;
1391 /* Only set accessed/dirty (rc) bits if set in host and l1 guest ptes */
1392 perm |= (gpte.rc & _PAGE_ACCESSED) ? 0UL : _PAGE_ACCESSED;
1393 perm |= ((gpte.rc & _PAGE_DIRTY) && writing) ? 0UL : _PAGE_DIRTY;
1230 pte = __pte(pte_val(pte) & ~perm); 1394 pte = __pte(pte_val(pte) & ~perm);
1231 1395
1232 /* What size pte can we insert? */ 1396 /* What size pte can we insert? */
@@ -1264,13 +1428,13 @@ static long int __kvmhv_nested_page_fault(struct kvm_vcpu *vcpu,
1264 return RESUME_GUEST; 1428 return RESUME_GUEST;
1265} 1429}
1266 1430
1267long int kvmhv_nested_page_fault(struct kvm_vcpu *vcpu) 1431long int kvmhv_nested_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu)
1268{ 1432{
1269 struct kvm_nested_guest *gp = vcpu->arch.nested; 1433 struct kvm_nested_guest *gp = vcpu->arch.nested;
1270 long int ret; 1434 long int ret;
1271 1435
1272 mutex_lock(&gp->tlb_lock); 1436 mutex_lock(&gp->tlb_lock);
1273 ret = __kvmhv_nested_page_fault(vcpu, gp); 1437 ret = __kvmhv_nested_page_fault(run, vcpu, gp);
1274 mutex_unlock(&gp->tlb_lock); 1438 mutex_unlock(&gp->tlb_lock);
1275 return ret; 1439 return ret;
1276} 1440}
diff --git a/arch/powerpc/kvm/book3s_hv_rm_mmu.c b/arch/powerpc/kvm/book3s_hv_rm_mmu.c
index a67cf1cdeda4..3b3791ed74a6 100644
--- a/arch/powerpc/kvm/book3s_hv_rm_mmu.c
+++ b/arch/powerpc/kvm/book3s_hv_rm_mmu.c
@@ -107,7 +107,7 @@ void kvmppc_add_revmap_chain(struct kvm *kvm, struct revmap_entry *rev,
107EXPORT_SYMBOL_GPL(kvmppc_add_revmap_chain); 107EXPORT_SYMBOL_GPL(kvmppc_add_revmap_chain);
108 108
109/* Update the dirty bitmap of a memslot */ 109/* Update the dirty bitmap of a memslot */
110void kvmppc_update_dirty_map(struct kvm_memory_slot *memslot, 110void kvmppc_update_dirty_map(const struct kvm_memory_slot *memslot,
111 unsigned long gfn, unsigned long psize) 111 unsigned long gfn, unsigned long psize)
112{ 112{
113 unsigned long npages; 113 unsigned long npages;
diff --git a/arch/powerpc/kvm/book3s_pr.c b/arch/powerpc/kvm/book3s_pr.c
index 4efd65d9e828..811a3c2fb0e9 100644
--- a/arch/powerpc/kvm/book3s_pr.c
+++ b/arch/powerpc/kvm/book3s_pr.c
@@ -587,6 +587,7 @@ void kvmppc_set_pvr_pr(struct kvm_vcpu *vcpu, u32 pvr)
587 case PVR_POWER8: 587 case PVR_POWER8:
588 case PVR_POWER8E: 588 case PVR_POWER8E:
589 case PVR_POWER8NVL: 589 case PVR_POWER8NVL:
590 case PVR_POWER9:
590 vcpu->arch.hflags |= BOOK3S_HFLAG_MULTI_PGSIZE | 591 vcpu->arch.hflags |= BOOK3S_HFLAG_MULTI_PGSIZE |
591 BOOK3S_HFLAG_NEW_TLBIE; 592 BOOK3S_HFLAG_NEW_TLBIE;
592 break; 593 break;
@@ -1913,7 +1914,8 @@ static int kvmppc_core_prepare_memory_region_pr(struct kvm *kvm,
1913static void kvmppc_core_commit_memory_region_pr(struct kvm *kvm, 1914static void kvmppc_core_commit_memory_region_pr(struct kvm *kvm,
1914 const struct kvm_userspace_memory_region *mem, 1915 const struct kvm_userspace_memory_region *mem,
1915 const struct kvm_memory_slot *old, 1916 const struct kvm_memory_slot *old,
1916 const struct kvm_memory_slot *new) 1917 const struct kvm_memory_slot *new,
1918 enum kvm_mr_change change)
1917{ 1919{
1918 return; 1920 return;
1919} 1921}
diff --git a/arch/powerpc/kvm/book3s_xics.c b/arch/powerpc/kvm/book3s_xics.c
index b0b2bfc2ff51..f27ee57ab46e 100644
--- a/arch/powerpc/kvm/book3s_xics.c
+++ b/arch/powerpc/kvm/book3s_xics.c
@@ -1015,17 +1015,7 @@ static int xics_debug_show(struct seq_file *m, void *private)
1015 return 0; 1015 return 0;
1016} 1016}
1017 1017
1018static int xics_debug_open(struct inode *inode, struct file *file) 1018DEFINE_SHOW_ATTRIBUTE(xics_debug);
1019{
1020 return single_open(file, xics_debug_show, inode->i_private);
1021}
1022
1023static const struct file_operations xics_debug_fops = {
1024 .open = xics_debug_open,
1025 .read = seq_read,
1026 .llseek = seq_lseek,
1027 .release = single_release,
1028};
1029 1019
1030static void xics_debugfs_init(struct kvmppc_xics *xics) 1020static void xics_debugfs_init(struct kvmppc_xics *xics)
1031{ 1021{
diff --git a/arch/powerpc/kvm/book3s_xive.c b/arch/powerpc/kvm/book3s_xive.c
index ad4a370703d3..f78d002f0fe0 100644
--- a/arch/powerpc/kvm/book3s_xive.c
+++ b/arch/powerpc/kvm/book3s_xive.c
@@ -1968,17 +1968,7 @@ static int xive_debug_show(struct seq_file *m, void *private)
1968 return 0; 1968 return 0;
1969} 1969}
1970 1970
1971static int xive_debug_open(struct inode *inode, struct file *file) 1971DEFINE_SHOW_ATTRIBUTE(xive_debug);
1972{
1973 return single_open(file, xive_debug_show, inode->i_private);
1974}
1975
1976static const struct file_operations xive_debug_fops = {
1977 .open = xive_debug_open,
1978 .read = seq_read,
1979 .llseek = seq_lseek,
1980 .release = single_release,
1981};
1982 1972
1983static void xive_debugfs_init(struct kvmppc_xive *xive) 1973static void xive_debugfs_init(struct kvmppc_xive *xive)
1984{ 1974{
diff --git a/arch/powerpc/kvm/booke.c b/arch/powerpc/kvm/booke.c
index a9ca016da670..dbec4128bb51 100644
--- a/arch/powerpc/kvm/booke.c
+++ b/arch/powerpc/kvm/booke.c
@@ -1833,7 +1833,8 @@ int kvmppc_core_prepare_memory_region(struct kvm *kvm,
1833void kvmppc_core_commit_memory_region(struct kvm *kvm, 1833void kvmppc_core_commit_memory_region(struct kvm *kvm,
1834 const struct kvm_userspace_memory_region *mem, 1834 const struct kvm_userspace_memory_region *mem,
1835 const struct kvm_memory_slot *old, 1835 const struct kvm_memory_slot *old,
1836 const struct kvm_memory_slot *new) 1836 const struct kvm_memory_slot *new,
1837 enum kvm_mr_change change)
1837{ 1838{
1838} 1839}
1839 1840
diff --git a/arch/powerpc/kvm/e500_mmu_host.c b/arch/powerpc/kvm/e500_mmu_host.c
index 8f2985e46f6f..c3f312b2bcb3 100644
--- a/arch/powerpc/kvm/e500_mmu_host.c
+++ b/arch/powerpc/kvm/e500_mmu_host.c
@@ -757,10 +757,11 @@ int kvm_test_age_hva(struct kvm *kvm, unsigned long hva)
757 return 0; 757 return 0;
758} 758}
759 759
760void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte) 760int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
761{ 761{
762 /* The page will get remapped properly on its next fault */ 762 /* The page will get remapped properly on its next fault */
763 kvm_unmap_hva(kvm, hva); 763 kvm_unmap_hva(kvm, hva);
764 return 0;
764} 765}
765 766
766/*****************************************/ 767/*****************************************/
diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
index 2869a299c4ed..b90a7d154180 100644
--- a/arch/powerpc/kvm/powerpc.c
+++ b/arch/powerpc/kvm/powerpc.c
@@ -331,10 +331,17 @@ int kvmppc_st(struct kvm_vcpu *vcpu, ulong *eaddr, int size, void *ptr,
331{ 331{
332 ulong mp_pa = vcpu->arch.magic_page_pa & KVM_PAM & PAGE_MASK; 332 ulong mp_pa = vcpu->arch.magic_page_pa & KVM_PAM & PAGE_MASK;
333 struct kvmppc_pte pte; 333 struct kvmppc_pte pte;
334 int r; 334 int r = -EINVAL;
335 335
336 vcpu->stat.st++; 336 vcpu->stat.st++;
337 337
338 if (vcpu->kvm->arch.kvm_ops && vcpu->kvm->arch.kvm_ops->store_to_eaddr)
339 r = vcpu->kvm->arch.kvm_ops->store_to_eaddr(vcpu, eaddr, ptr,
340 size);
341
342 if ((!r) || (r == -EAGAIN))
343 return r;
344
338 r = kvmppc_xlate(vcpu, *eaddr, data ? XLATE_DATA : XLATE_INST, 345 r = kvmppc_xlate(vcpu, *eaddr, data ? XLATE_DATA : XLATE_INST,
339 XLATE_WRITE, &pte); 346 XLATE_WRITE, &pte);
340 if (r < 0) 347 if (r < 0)
@@ -367,10 +374,17 @@ int kvmppc_ld(struct kvm_vcpu *vcpu, ulong *eaddr, int size, void *ptr,
367{ 374{
368 ulong mp_pa = vcpu->arch.magic_page_pa & KVM_PAM & PAGE_MASK; 375 ulong mp_pa = vcpu->arch.magic_page_pa & KVM_PAM & PAGE_MASK;
369 struct kvmppc_pte pte; 376 struct kvmppc_pte pte;
370 int rc; 377 int rc = -EINVAL;
371 378
372 vcpu->stat.ld++; 379 vcpu->stat.ld++;
373 380
381 if (vcpu->kvm->arch.kvm_ops && vcpu->kvm->arch.kvm_ops->load_from_eaddr)
382 rc = vcpu->kvm->arch.kvm_ops->load_from_eaddr(vcpu, eaddr, ptr,
383 size);
384
385 if ((!rc) || (rc == -EAGAIN))
386 return rc;
387
374 rc = kvmppc_xlate(vcpu, *eaddr, data ? XLATE_DATA : XLATE_INST, 388 rc = kvmppc_xlate(vcpu, *eaddr, data ? XLATE_DATA : XLATE_INST,
375 XLATE_READ, &pte); 389 XLATE_READ, &pte);
376 if (rc) 390 if (rc)
@@ -518,7 +532,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
518 case KVM_CAP_PPC_UNSET_IRQ: 532 case KVM_CAP_PPC_UNSET_IRQ:
519 case KVM_CAP_PPC_IRQ_LEVEL: 533 case KVM_CAP_PPC_IRQ_LEVEL:
520 case KVM_CAP_ENABLE_CAP: 534 case KVM_CAP_ENABLE_CAP:
521 case KVM_CAP_ENABLE_CAP_VM:
522 case KVM_CAP_ONE_REG: 535 case KVM_CAP_ONE_REG:
523 case KVM_CAP_IOEVENTFD: 536 case KVM_CAP_IOEVENTFD:
524 case KVM_CAP_DEVICE_CTRL: 537 case KVM_CAP_DEVICE_CTRL:
@@ -543,8 +556,11 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
543#ifdef CONFIG_PPC_BOOK3S_64 556#ifdef CONFIG_PPC_BOOK3S_64
544 case KVM_CAP_SPAPR_TCE: 557 case KVM_CAP_SPAPR_TCE:
545 case KVM_CAP_SPAPR_TCE_64: 558 case KVM_CAP_SPAPR_TCE_64:
546 /* fallthrough */ 559 r = 1;
560 break;
547 case KVM_CAP_SPAPR_TCE_VFIO: 561 case KVM_CAP_SPAPR_TCE_VFIO:
562 r = !!cpu_has_feature(CPU_FTR_HVMODE);
563 break;
548 case KVM_CAP_PPC_RTAS: 564 case KVM_CAP_PPC_RTAS:
549 case KVM_CAP_PPC_FIXUP_HCALL: 565 case KVM_CAP_PPC_FIXUP_HCALL:
550 case KVM_CAP_PPC_ENABLE_HCALL: 566 case KVM_CAP_PPC_ENABLE_HCALL:
@@ -696,7 +712,7 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
696 const struct kvm_memory_slot *new, 712 const struct kvm_memory_slot *new,
697 enum kvm_mr_change change) 713 enum kvm_mr_change change)
698{ 714{
699 kvmppc_core_commit_memory_region(kvm, mem, old, new); 715 kvmppc_core_commit_memory_region(kvm, mem, old, new, change);
700} 716}
701 717
702void kvm_arch_flush_shadow_memslot(struct kvm *kvm, 718void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
@@ -1192,6 +1208,14 @@ static void kvmppc_complete_mmio_load(struct kvm_vcpu *vcpu,
1192 kvmppc_set_vmx_byte(vcpu, gpr); 1208 kvmppc_set_vmx_byte(vcpu, gpr);
1193 break; 1209 break;
1194#endif 1210#endif
1211#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
1212 case KVM_MMIO_REG_NESTED_GPR:
1213 if (kvmppc_need_byteswap(vcpu))
1214 gpr = swab64(gpr);
1215 kvm_vcpu_write_guest(vcpu, vcpu->arch.nested_io_gpr, &gpr,
1216 sizeof(gpr));
1217 break;
1218#endif
1195 default: 1219 default:
1196 BUG(); 1220 BUG();
1197 } 1221 }
@@ -2084,8 +2108,8 @@ int kvm_vm_ioctl_irq_line(struct kvm *kvm, struct kvm_irq_level *irq_event,
2084} 2108}
2085 2109
2086 2110
2087static int kvm_vm_ioctl_enable_cap(struct kvm *kvm, 2111int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
2088 struct kvm_enable_cap *cap) 2112 struct kvm_enable_cap *cap)
2089{ 2113{
2090 int r; 2114 int r;
2091 2115
@@ -2273,15 +2297,6 @@ long kvm_arch_vm_ioctl(struct file *filp,
2273 2297
2274 break; 2298 break;
2275 } 2299 }
2276 case KVM_ENABLE_CAP:
2277 {
2278 struct kvm_enable_cap cap;
2279 r = -EFAULT;
2280 if (copy_from_user(&cap, argp, sizeof(cap)))
2281 goto out;
2282 r = kvm_vm_ioctl_enable_cap(kvm, &cap);
2283 break;
2284 }
2285#ifdef CONFIG_SPAPR_TCE_IOMMU 2300#ifdef CONFIG_SPAPR_TCE_IOMMU
2286 case KVM_CREATE_SPAPR_TCE_64: { 2301 case KVM_CREATE_SPAPR_TCE_64: {
2287 struct kvm_create_spapr_tce_64 create_tce_64; 2302 struct kvm_create_spapr_tce_64 create_tce_64;
diff --git a/arch/powerpc/mm/dump_linuxpagetables.c b/arch/powerpc/mm/dump_linuxpagetables.c
index 2b74f8adf4d0..6aa41669ac1a 100644
--- a/arch/powerpc/mm/dump_linuxpagetables.c
+++ b/arch/powerpc/mm/dump_linuxpagetables.c
@@ -19,6 +19,7 @@
19#include <linux/hugetlb.h> 19#include <linux/hugetlb.h>
20#include <linux/io.h> 20#include <linux/io.h>
21#include <linux/mm.h> 21#include <linux/mm.h>
22#include <linux/highmem.h>
22#include <linux/sched.h> 23#include <linux/sched.h>
23#include <linux/seq_file.h> 24#include <linux/seq_file.h>
24#include <asm/fixmap.h> 25#include <asm/fixmap.h>
diff --git a/arch/powerpc/mm/fault.c b/arch/powerpc/mm/fault.c
index 1697e903bbf2..2e6fb1d758c3 100644
--- a/arch/powerpc/mm/fault.c
+++ b/arch/powerpc/mm/fault.c
@@ -636,6 +636,7 @@ void bad_page_fault(struct pt_regs *regs, unsigned long address, int sig)
636 switch (TRAP(regs)) { 636 switch (TRAP(regs)) {
637 case 0x300: 637 case 0x300:
638 case 0x380: 638 case 0x380:
639 case 0xe00:
639 printk(KERN_ALERT "Unable to handle kernel paging request for " 640 printk(KERN_ALERT "Unable to handle kernel paging request for "
640 "data at address 0x%08lx\n", regs->dar); 641 "data at address 0x%08lx\n", regs->dar);
641 break; 642 break;
diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c
index 8cf035e68378..4c01e9a01a74 100644
--- a/arch/powerpc/mm/hugetlbpage.c
+++ b/arch/powerpc/mm/hugetlbpage.c
@@ -289,7 +289,7 @@ static void hugepd_free(struct mmu_gather *tlb, void *hugepte)
289 289
290 (*batchp)->ptes[(*batchp)->index++] = hugepte; 290 (*batchp)->ptes[(*batchp)->index++] = hugepte;
291 if ((*batchp)->index == HUGEPD_FREELIST_SIZE) { 291 if ((*batchp)->index == HUGEPD_FREELIST_SIZE) {
292 call_rcu_sched(&(*batchp)->rcu, hugepd_free_rcu_callback); 292 call_rcu(&(*batchp)->rcu, hugepd_free_rcu_callback);
293 *batchp = NULL; 293 *batchp = NULL;
294 } 294 }
295 put_cpu_var(hugepd_freelist_cur); 295 put_cpu_var(hugepd_freelist_cur);
diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index 7a9886f98b0c..a5091c034747 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c
@@ -188,15 +188,20 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
188 pr_debug("vmemmap_populate %lx..%lx, node %d\n", start, end, node); 188 pr_debug("vmemmap_populate %lx..%lx, node %d\n", start, end, node);
189 189
190 for (; start < end; start += page_size) { 190 for (; start < end; start += page_size) {
191 void *p; 191 void *p = NULL;
192 int rc; 192 int rc;
193 193
194 if (vmemmap_populated(start, page_size)) 194 if (vmemmap_populated(start, page_size))
195 continue; 195 continue;
196 196
197 /*
198 * Allocate from the altmap first if we have one. This may
199 * fail due to alignment issues when using 16MB hugepages, so
200 * fall back to system memory if the altmap allocation fail.
201 */
197 if (altmap) 202 if (altmap)
198 p = altmap_alloc_block_buf(page_size, altmap); 203 p = altmap_alloc_block_buf(page_size, altmap);
199 else 204 if (!p)
200 p = vmemmap_alloc_block_buf(page_size, node); 205 p = vmemmap_alloc_block_buf(page_size, node);
201 if (!p) 206 if (!p)
202 return -ENOMEM; 207 return -ENOMEM;
@@ -255,8 +260,15 @@ void __ref vmemmap_free(unsigned long start, unsigned long end,
255{ 260{
256 unsigned long page_size = 1 << mmu_psize_defs[mmu_vmemmap_psize].shift; 261 unsigned long page_size = 1 << mmu_psize_defs[mmu_vmemmap_psize].shift;
257 unsigned long page_order = get_order(page_size); 262 unsigned long page_order = get_order(page_size);
263 unsigned long alt_start = ~0, alt_end = ~0;
264 unsigned long base_pfn;
258 265
259 start = _ALIGN_DOWN(start, page_size); 266 start = _ALIGN_DOWN(start, page_size);
267 if (altmap) {
268 alt_start = altmap->base_pfn;
269 alt_end = altmap->base_pfn + altmap->reserve +
270 altmap->free + altmap->alloc + altmap->align;
271 }
260 272
261 pr_debug("vmemmap_free %lx...%lx\n", start, end); 273 pr_debug("vmemmap_free %lx...%lx\n", start, end);
262 274
@@ -280,8 +292,9 @@ void __ref vmemmap_free(unsigned long start, unsigned long end,
280 page = pfn_to_page(addr >> PAGE_SHIFT); 292 page = pfn_to_page(addr >> PAGE_SHIFT);
281 section_base = pfn_to_page(vmemmap_section_start(start)); 293 section_base = pfn_to_page(vmemmap_section_start(start));
282 nr_pages = 1 << page_order; 294 nr_pages = 1 << page_order;
295 base_pfn = PHYS_PFN(addr);
283 296
284 if (altmap) { 297 if (base_pfn >= alt_start && base_pfn < alt_end) {
285 vmem_altmap_free(altmap, nr_pages); 298 vmem_altmap_free(altmap, nr_pages);
286 } else if (PageReserved(page)) { 299 } else if (PageReserved(page)) {
287 /* allocated from bootmem */ 300 /* allocated from bootmem */
diff --git a/arch/powerpc/net/bpf_jit_comp64.c b/arch/powerpc/net/bpf_jit_comp64.c
index 17482f5de3e2..9393e231cbc2 100644
--- a/arch/powerpc/net/bpf_jit_comp64.c
+++ b/arch/powerpc/net/bpf_jit_comp64.c
@@ -891,6 +891,55 @@ cond_branch:
891 return 0; 891 return 0;
892} 892}
893 893
894/* Fix the branch target addresses for subprog calls */
895static int bpf_jit_fixup_subprog_calls(struct bpf_prog *fp, u32 *image,
896 struct codegen_context *ctx, u32 *addrs)
897{
898 const struct bpf_insn *insn = fp->insnsi;
899 bool func_addr_fixed;
900 u64 func_addr;
901 u32 tmp_idx;
902 int i, ret;
903
904 for (i = 0; i < fp->len; i++) {
905 /*
906 * During the extra pass, only the branch target addresses for
907 * the subprog calls need to be fixed. All other instructions
908 * can left untouched.
909 *
910 * The JITed image length does not change because we already
911 * ensure that the JITed instruction sequence for these calls
912 * are of fixed length by padding them with NOPs.
913 */
914 if (insn[i].code == (BPF_JMP | BPF_CALL) &&
915 insn[i].src_reg == BPF_PSEUDO_CALL) {
916 ret = bpf_jit_get_func_addr(fp, &insn[i], true,
917 &func_addr,
918 &func_addr_fixed);
919 if (ret < 0)
920 return ret;
921
922 /*
923 * Save ctx->idx as this would currently point to the
924 * end of the JITed image and set it to the offset of
925 * the instruction sequence corresponding to the
926 * subprog call temporarily.
927 */
928 tmp_idx = ctx->idx;
929 ctx->idx = addrs[i] / 4;
930 bpf_jit_emit_func_call_rel(image, ctx, func_addr);
931
932 /*
933 * Restore ctx->idx here. This is safe as the length
934 * of the JITed sequence remains unchanged.
935 */
936 ctx->idx = tmp_idx;
937 }
938 }
939
940 return 0;
941}
942
894struct powerpc64_jit_data { 943struct powerpc64_jit_data {
895 struct bpf_binary_header *header; 944 struct bpf_binary_header *header;
896 u32 *addrs; 945 u32 *addrs;
@@ -989,6 +1038,22 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
989skip_init_ctx: 1038skip_init_ctx:
990 code_base = (u32 *)(image + FUNCTION_DESCR_SIZE); 1039 code_base = (u32 *)(image + FUNCTION_DESCR_SIZE);
991 1040
1041 if (extra_pass) {
1042 /*
1043 * Do not touch the prologue and epilogue as they will remain
1044 * unchanged. Only fix the branch target address for subprog
1045 * calls in the body.
1046 *
1047 * This does not change the offsets and lengths of the subprog
1048 * call instruction sequences and hence, the size of the JITed
1049 * image as well.
1050 */
1051 bpf_jit_fixup_subprog_calls(fp, code_base, &cgctx, addrs);
1052
1053 /* There is no need to perform the usual passes. */
1054 goto skip_codegen_passes;
1055 }
1056
992 /* Code generation passes 1-2 */ 1057 /* Code generation passes 1-2 */
993 for (pass = 1; pass < 3; pass++) { 1058 for (pass = 1; pass < 3; pass++) {
994 /* Now build the prologue, body code & epilogue for real. */ 1059 /* Now build the prologue, body code & epilogue for real. */
@@ -1002,6 +1067,7 @@ skip_init_ctx:
1002 proglen - (cgctx.idx * 4), cgctx.seen); 1067 proglen - (cgctx.idx * 4), cgctx.seen);
1003 } 1068 }
1004 1069
1070skip_codegen_passes:
1005 if (bpf_jit_enable > 1) 1071 if (bpf_jit_enable > 1)
1006 /* 1072 /*
1007 * Note that we output the base address of the code_base 1073 * Note that we output the base address of the code_base
diff --git a/arch/powerpc/platforms/pseries/Kconfig b/arch/powerpc/platforms/pseries/Kconfig
index 2e4bd32154b5..472b784f01eb 100644
--- a/arch/powerpc/platforms/pseries/Kconfig
+++ b/arch/powerpc/platforms/pseries/Kconfig
@@ -140,8 +140,7 @@ config IBMEBUS
140 Bus device driver for GX bus based adapters. 140 Bus device driver for GX bus based adapters.
141 141
142config PAPR_SCM 142config PAPR_SCM
143 depends on PPC_PSERIES && MEMORY_HOTPLUG 143 depends on PPC_PSERIES && MEMORY_HOTPLUG && LIBNVDIMM
144 select LIBNVDIMM
145 tristate "Support for the PAPR Storage Class Memory interface" 144 tristate "Support for the PAPR Storage Class Memory interface"
146 help 145 help
147 Enable access to hypervisor provided storage class memory. 146 Enable access to hypervisor provided storage class memory.
diff --git a/arch/powerpc/platforms/pseries/papr_scm.c b/arch/powerpc/platforms/pseries/papr_scm.c
index ee9372b65ca5..7d6457ab5d34 100644
--- a/arch/powerpc/platforms/pseries/papr_scm.c
+++ b/arch/powerpc/platforms/pseries/papr_scm.c
@@ -55,7 +55,7 @@ static int drc_pmem_bind(struct papr_scm_priv *p)
55 do { 55 do {
56 rc = plpar_hcall(H_SCM_BIND_MEM, ret, p->drc_index, 0, 56 rc = plpar_hcall(H_SCM_BIND_MEM, ret, p->drc_index, 0,
57 p->blocks, BIND_ANY_ADDR, token); 57 p->blocks, BIND_ANY_ADDR, token);
58 token = be64_to_cpu(ret[0]); 58 token = ret[0];
59 cond_resched(); 59 cond_resched();
60 } while (rc == H_BUSY); 60 } while (rc == H_BUSY);
61 61
@@ -64,7 +64,7 @@ static int drc_pmem_bind(struct papr_scm_priv *p)
64 return -ENXIO; 64 return -ENXIO;
65 } 65 }
66 66
67 p->bound_addr = be64_to_cpu(ret[1]); 67 p->bound_addr = ret[1];
68 68
69 dev_dbg(&p->pdev->dev, "bound drc %x to %pR\n", p->drc_index, &p->res); 69 dev_dbg(&p->pdev->dev, "bound drc %x to %pR\n", p->drc_index, &p->res);
70 70
@@ -82,7 +82,7 @@ static int drc_pmem_unbind(struct papr_scm_priv *p)
82 do { 82 do {
83 rc = plpar_hcall(H_SCM_UNBIND_MEM, ret, p->drc_index, 83 rc = plpar_hcall(H_SCM_UNBIND_MEM, ret, p->drc_index,
84 p->bound_addr, p->blocks, token); 84 p->bound_addr, p->blocks, token);
85 token = be64_to_cpu(ret); 85 token = ret[0];
86 cond_resched(); 86 cond_resched();
87 } while (rc == H_BUSY); 87 } while (rc == H_BUSY);
88 88
@@ -223,6 +223,9 @@ static int papr_scm_nvdimm_init(struct papr_scm_priv *p)
223 goto err; 223 goto err;
224 } 224 }
225 225
226 if (nvdimm_bus_check_dimm_count(p->bus, 1))
227 goto err;
228
226 /* now add the region */ 229 /* now add the region */
227 230
228 memset(&mapping, 0, sizeof(mapping)); 231 memset(&mapping, 0, sizeof(mapping));
@@ -257,9 +260,12 @@ err: nvdimm_bus_unregister(p->bus);
257 260
258static int papr_scm_probe(struct platform_device *pdev) 261static int papr_scm_probe(struct platform_device *pdev)
259{ 262{
260 uint32_t drc_index, metadata_size, unit_cap[2];
261 struct device_node *dn = pdev->dev.of_node; 263 struct device_node *dn = pdev->dev.of_node;
264 u32 drc_index, metadata_size;
265 u64 blocks, block_size;
262 struct papr_scm_priv *p; 266 struct papr_scm_priv *p;
267 const char *uuid_str;
268 u64 uuid[2];
263 int rc; 269 int rc;
264 270
265 /* check we have all the required DT properties */ 271 /* check we have all the required DT properties */
@@ -268,8 +274,18 @@ static int papr_scm_probe(struct platform_device *pdev)
268 return -ENODEV; 274 return -ENODEV;
269 } 275 }
270 276
271 if (of_property_read_u32_array(dn, "ibm,unit-capacity", unit_cap, 2)) { 277 if (of_property_read_u64(dn, "ibm,block-size", &block_size)) {
272 dev_err(&pdev->dev, "%pOF: missing unit-capacity!\n", dn); 278 dev_err(&pdev->dev, "%pOF: missing block-size!\n", dn);
279 return -ENODEV;
280 }
281
282 if (of_property_read_u64(dn, "ibm,number-of-blocks", &blocks)) {
283 dev_err(&pdev->dev, "%pOF: missing number-of-blocks!\n", dn);
284 return -ENODEV;
285 }
286
287 if (of_property_read_string(dn, "ibm,unit-guid", &uuid_str)) {
288 dev_err(&pdev->dev, "%pOF: missing unit-guid!\n", dn);
273 return -ENODEV; 289 return -ENODEV;
274 } 290 }
275 291
@@ -282,8 +298,13 @@ static int papr_scm_probe(struct platform_device *pdev)
282 298
283 p->dn = dn; 299 p->dn = dn;
284 p->drc_index = drc_index; 300 p->drc_index = drc_index;
285 p->block_size = unit_cap[0]; 301 p->block_size = block_size;
286 p->blocks = unit_cap[1]; 302 p->blocks = blocks;
303
304 /* We just need to ensure that set cookies are unique across */
305 uuid_parse(uuid_str, (uuid_t *) uuid);
306 p->nd_set.cookie1 = uuid[0];
307 p->nd_set.cookie2 = uuid[1];
287 308
288 /* might be zero */ 309 /* might be zero */
289 p->metadata_size = metadata_size; 310 p->metadata_size = metadata_size;
@@ -296,7 +317,7 @@ static int papr_scm_probe(struct platform_device *pdev)
296 317
297 /* setup the resource for the newly bound range */ 318 /* setup the resource for the newly bound range */
298 p->res.start = p->bound_addr; 319 p->res.start = p->bound_addr;
299 p->res.end = p->bound_addr + p->blocks * p->block_size; 320 p->res.end = p->bound_addr + p->blocks * p->block_size - 1;
300 p->res.name = pdev->name; 321 p->res.name = pdev->name;
301 p->res.flags = IORESOURCE_MEM; 322 p->res.flags = IORESOURCE_MEM;
302 323
generated by cgit v1.2.2 (git 2.25.0) at 2025-10-13 05:41:43 -0400