aboutsummaryrefslogtreecommitdiffstats
path: root/arch/powerpc/kernel
diff options
context:
space:
mode:
authorAndrea Bastoni <bastoni@cs.unc.edu>2010-05-30 19:16:45 -0400
committerAndrea Bastoni <bastoni@cs.unc.edu>2010-05-30 19:16:45 -0400
commitada47b5fe13d89735805b566185f4885f5a3f750 (patch)
tree644b88f8a71896307d71438e9b3af49126ffb22b /arch/powerpc/kernel
parent43e98717ad40a4ae64545b5ba047c7b86aa44f4f (diff)
parent3280f21d43ee541f97f8cda5792150d2dbec20d5 (diff)
Merge branch 'wip-2.6.34' into old-private-masterarchived-private-master
Diffstat (limited to 'arch/powerpc/kernel')
-rw-r--r--arch/powerpc/kernel/Makefile10
-rw-r--r--arch/powerpc/kernel/align.c75
-rw-r--r--arch/powerpc/kernel/asm-offsets.c53
-rw-r--r--arch/powerpc/kernel/cacheinfo.c3
-rw-r--r--arch/powerpc/kernel/cputable.c8
-rw-r--r--arch/powerpc/kernel/crash.c2
-rw-r--r--arch/powerpc/kernel/dma-swiotlb.c6
-rw-r--r--arch/powerpc/kernel/dma.c1
-rw-r--r--arch/powerpc/kernel/e500-pmu.c129
-rw-r--r--arch/powerpc/kernel/entry_64.S16
-rw-r--r--arch/powerpc/kernel/exceptions-64s.S11
-rw-r--r--arch/powerpc/kernel/firmware.c2
-rw-r--r--arch/powerpc/kernel/head_32.S25
-rw-r--r--arch/powerpc/kernel/head_64.S24
-rw-r--r--arch/powerpc/kernel/head_8xx.S321
-rw-r--r--arch/powerpc/kernel/head_fsl_booke.S33
-rw-r--r--arch/powerpc/kernel/ibmebus.c1
-rw-r--r--arch/powerpc/kernel/idle.c4
-rw-r--r--arch/powerpc/kernel/io.c4
-rw-r--r--arch/powerpc/kernel/iommu.c11
-rw-r--r--arch/powerpc/kernel/irq.c247
-rw-r--r--arch/powerpc/kernel/kgdb.c4
-rw-r--r--arch/powerpc/kernel/kprobes.c5
-rw-r--r--arch/powerpc/kernel/legacy_serial.c2
-rw-r--r--arch/powerpc/kernel/lparcfg.c15
-rw-r--r--arch/powerpc/kernel/misc.S26
-rw-r--r--arch/powerpc/kernel/misc_32.S18
-rw-r--r--arch/powerpc/kernel/nvram_64.c62
-rw-r--r--arch/powerpc/kernel/of_platform.c3
-rw-r--r--arch/powerpc/kernel/paca.c93
-rw-r--r--arch/powerpc/kernel/pci-common.c53
-rw-r--r--arch/powerpc/kernel/pci_32.c1
-rw-r--r--arch/powerpc/kernel/pci_64.c2
-rw-r--r--arch/powerpc/kernel/pci_dn.c1
-rw-r--r--arch/powerpc/kernel/pci_of_scan.c12
-rw-r--r--arch/powerpc/kernel/perf_callchain.c27
-rw-r--r--arch/powerpc/kernel/perf_event.c39
-rw-r--r--arch/powerpc/kernel/perf_event_fsl_emb.c654
-rw-r--r--arch/powerpc/kernel/pmc.c10
-rw-r--r--arch/powerpc/kernel/power5+-pmu.c4
-rw-r--r--arch/powerpc/kernel/power5-pmu.c6
-rw-r--r--arch/powerpc/kernel/power6-pmu.c2
-rw-r--r--arch/powerpc/kernel/power7-pmu.c6
-rw-r--r--arch/powerpc/kernel/ppc970-pmu.c4
-rw-r--r--arch/powerpc/kernel/ppc_ksyms.c4
-rw-r--r--arch/powerpc/kernel/proc_powerpc.c (renamed from arch/powerpc/kernel/proc_ppc64.c)103
-rw-r--r--arch/powerpc/kernel/process.c128
-rw-r--r--arch/powerpc/kernel/prom.c913
-rw-r--r--arch/powerpc/kernel/prom_init.c82
-rw-r--r--arch/powerpc/kernel/ptrace.c516
-rw-r--r--arch/powerpc/kernel/rtas-proc.c14
-rw-r--r--arch/powerpc/kernel/rtas.c17
-rw-r--r--arch/powerpc/kernel/rtas_flash.c11
-rw-r--r--arch/powerpc/kernel/rtasd.c540
-rw-r--r--arch/powerpc/kernel/setup-common.c8
-rw-r--r--arch/powerpc/kernel/setup_32.c6
-rw-r--r--arch/powerpc/kernel/setup_64.c23
-rw-r--r--arch/powerpc/kernel/signal.c6
-rw-r--r--arch/powerpc/kernel/signal_32.c16
-rw-r--r--arch/powerpc/kernel/smp-tbsync.c1
-rw-r--r--arch/powerpc/kernel/smp.c17
-rw-r--r--arch/powerpc/kernel/softemu8xx.c1
-rw-r--r--arch/powerpc/kernel/swsusp_32.S2
-rw-r--r--arch/powerpc/kernel/sys_ppc32.c53
-rw-r--r--arch/powerpc/kernel/syscalls.c179
-rw-r--r--arch/powerpc/kernel/sysfs.c19
-rw-r--r--arch/powerpc/kernel/tau_6xx.c2
-rw-r--r--arch/powerpc/kernel/time.c93
-rw-r--r--arch/powerpc/kernel/traps.c178
-rw-r--r--arch/powerpc/kernel/udbg.c2
-rw-r--r--arch/powerpc/kernel/vdso.c2
-rw-r--r--arch/powerpc/kernel/vector.S2
-rw-r--r--arch/powerpc/kernel/vio.c1
-rw-r--r--arch/powerpc/kernel/vmlinux.lds.S3
74 files changed, 3147 insertions, 1830 deletions
diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile
index b23664a0b86c..877326320e74 100644
--- a/arch/powerpc/kernel/Makefile
+++ b/arch/powerpc/kernel/Makefile
@@ -42,10 +42,11 @@ obj-$(CONFIG_ALTIVEC) += vecemu.o
42obj-$(CONFIG_PPC_970_NAP) += idle_power4.o 42obj-$(CONFIG_PPC_970_NAP) += idle_power4.o
43obj-$(CONFIG_PPC_OF) += of_device.o of_platform.o prom_parse.o 43obj-$(CONFIG_PPC_OF) += of_device.o of_platform.o prom_parse.o
44obj-$(CONFIG_PPC_CLOCK) += clock.o 44obj-$(CONFIG_PPC_CLOCK) += clock.o
45procfs-$(CONFIG_PPC64) := proc_ppc64.o 45procfs-y := proc_powerpc.o
46obj-$(CONFIG_PROC_FS) += $(procfs-y) 46obj-$(CONFIG_PROC_FS) += $(procfs-y)
47rtaspci-$(CONFIG_PPC64)-$(CONFIG_PCI) := rtas_pci.o 47rtaspci-$(CONFIG_PPC64)-$(CONFIG_PCI) := rtas_pci.o
48obj-$(CONFIG_PPC_RTAS) += rtas.o rtas-rtc.o $(rtaspci-y-y) 48obj-$(CONFIG_PPC_RTAS) += rtas.o rtas-rtc.o $(rtaspci-y-y)
49obj-$(CONFIG_PPC_RTAS_DAEMON) += rtasd.o
49obj-$(CONFIG_RTAS_FLASH) += rtas_flash.o 50obj-$(CONFIG_RTAS_FLASH) += rtas_flash.o
50obj-$(CONFIG_RTAS_PROC) += rtas-proc.o 51obj-$(CONFIG_RTAS_PROC) += rtas-proc.o
51obj-$(CONFIG_LPARCFG) += lparcfg.o 52obj-$(CONFIG_LPARCFG) += lparcfg.o
@@ -97,11 +98,16 @@ obj64-$(CONFIG_AUDIT) += compat_audit.o
97 98
98obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o 99obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o
99obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o 100obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
100obj-$(CONFIG_PPC_PERF_CTRS) += perf_event.o perf_callchain.o 101obj-$(CONFIG_PERF_EVENTS) += perf_callchain.o
102
103obj-$(CONFIG_PPC_PERF_CTRS) += perf_event.o
101obj64-$(CONFIG_PPC_PERF_CTRS) += power4-pmu.o ppc970-pmu.o power5-pmu.o \ 104obj64-$(CONFIG_PPC_PERF_CTRS) += power4-pmu.o ppc970-pmu.o power5-pmu.o \
102 power5+-pmu.o power6-pmu.o power7-pmu.o 105 power5+-pmu.o power6-pmu.o power7-pmu.o
103obj32-$(CONFIG_PPC_PERF_CTRS) += mpc7450-pmu.o 106obj32-$(CONFIG_PPC_PERF_CTRS) += mpc7450-pmu.o
104 107
108obj-$(CONFIG_FSL_EMB_PERF_EVENT) += perf_event_fsl_emb.o
109obj-$(CONFIG_FSL_EMB_PERF_EVENT_E500) += e500-pmu.o
110
105obj-$(CONFIG_8XX_MINIMAL_FPEMU) += softemu8xx.o 111obj-$(CONFIG_8XX_MINIMAL_FPEMU) += softemu8xx.o
106 112
107ifneq ($(CONFIG_PPC_INDIRECT_IO),y) 113ifneq ($(CONFIG_PPC_INDIRECT_IO),y)
diff --git a/arch/powerpc/kernel/align.c b/arch/powerpc/kernel/align.c
index a5b632e52fae..b876e989220b 100644
--- a/arch/powerpc/kernel/align.c
+++ b/arch/powerpc/kernel/align.c
@@ -642,10 +642,14 @@ static int emulate_spe(struct pt_regs *regs, unsigned int reg,
642 */ 642 */
643static int emulate_vsx(unsigned char __user *addr, unsigned int reg, 643static int emulate_vsx(unsigned char __user *addr, unsigned int reg,
644 unsigned int areg, struct pt_regs *regs, 644 unsigned int areg, struct pt_regs *regs,
645 unsigned int flags, unsigned int length) 645 unsigned int flags, unsigned int length,
646 unsigned int elsize)
646{ 647{
647 char *ptr; 648 char *ptr;
649 unsigned long *lptr;
648 int ret = 0; 650 int ret = 0;
651 int sw = 0;
652 int i, j;
649 653
650 flush_vsx_to_thread(current); 654 flush_vsx_to_thread(current);
651 655
@@ -654,19 +658,35 @@ static int emulate_vsx(unsigned char __user *addr, unsigned int reg,
654 else 658 else
655 ptr = (char *) &current->thread.vr[reg - 32]; 659 ptr = (char *) &current->thread.vr[reg - 32];
656 660
657 if (flags & ST) 661 lptr = (unsigned long *) ptr;
658 ret = __copy_to_user(addr, ptr, length); 662
659 else { 663 if (flags & SW)
660 if (flags & SPLT){ 664 sw = elsize-1;
661 ret = __copy_from_user(ptr, addr, length); 665
662 ptr += length; 666 for (j = 0; j < length; j += elsize) {
667 for (i = 0; i < elsize; ++i) {
668 if (flags & ST)
669 ret |= __put_user(ptr[i^sw], addr + i);
670 else
671 ret |= __get_user(ptr[i^sw], addr + i);
663 } 672 }
664 ret |= __copy_from_user(ptr, addr, length); 673 ptr += elsize;
674 addr += elsize;
665 } 675 }
666 if (flags & U) 676
667 regs->gpr[areg] = regs->dar; 677 if (!ret) {
668 if (ret) 678 if (flags & U)
679 regs->gpr[areg] = regs->dar;
680
681 /* Splat load copies the same data to top and bottom 8 bytes */
682 if (flags & SPLT)
683 lptr[1] = lptr[0];
684 /* For 8 byte loads, zero the top 8 bytes */
685 else if (!(flags & ST) && (8 == length))
686 lptr[1] = 0;
687 } else
669 return -EFAULT; 688 return -EFAULT;
689
670 return 1; 690 return 1;
671} 691}
672#endif 692#endif
@@ -732,7 +752,7 @@ int fix_alignment(struct pt_regs *regs)
732 752
733#ifdef CONFIG_SPE 753#ifdef CONFIG_SPE
734 if ((instr >> 26) == 0x4) { 754 if ((instr >> 26) == 0x4) {
735 PPC_WARN_EMULATED(spe); 755 PPC_WARN_ALIGNMENT(spe, regs);
736 return emulate_spe(regs, reg, instr); 756 return emulate_spe(regs, reg, instr);
737 } 757 }
738#endif 758#endif
@@ -767,16 +787,25 @@ int fix_alignment(struct pt_regs *regs)
767 787
768#ifdef CONFIG_VSX 788#ifdef CONFIG_VSX
769 if ((instruction & 0xfc00003e) == 0x7c000018) { 789 if ((instruction & 0xfc00003e) == 0x7c000018) {
770 /* Additional register addressing bit (64 VSX vs 32 FPR/GPR */ 790 unsigned int elsize;
791
792 /* Additional register addressing bit (64 VSX vs 32 FPR/GPR) */
771 reg |= (instruction & 0x1) << 5; 793 reg |= (instruction & 0x1) << 5;
772 /* Simple inline decoder instead of a table */ 794 /* Simple inline decoder instead of a table */
795 /* VSX has only 8 and 16 byte memory accesses */
796 nb = 8;
773 if (instruction & 0x200) 797 if (instruction & 0x200)
774 nb = 16; 798 nb = 16;
775 else if (instruction & 0x080) 799
776 nb = 8; 800 /* Vector stores in little-endian mode swap individual
777 else 801 elements, so process them separately */
778 nb = 4; 802 elsize = 4;
803 if (instruction & 0x80)
804 elsize = 8;
805
779 flags = 0; 806 flags = 0;
807 if (regs->msr & MSR_LE)
808 flags |= SW;
780 if (instruction & 0x100) 809 if (instruction & 0x100)
781 flags |= ST; 810 flags |= ST;
782 if (instruction & 0x040) 811 if (instruction & 0x040)
@@ -786,15 +815,15 @@ int fix_alignment(struct pt_regs *regs)
786 flags |= SPLT; 815 flags |= SPLT;
787 nb = 8; 816 nb = 8;
788 } 817 }
789 PPC_WARN_EMULATED(vsx); 818 PPC_WARN_ALIGNMENT(vsx, regs);
790 return emulate_vsx(addr, reg, areg, regs, flags, nb); 819 return emulate_vsx(addr, reg, areg, regs, flags, nb, elsize);
791 } 820 }
792#endif 821#endif
793 /* A size of 0 indicates an instruction we don't support, with 822 /* A size of 0 indicates an instruction we don't support, with
794 * the exception of DCBZ which is handled as a special case here 823 * the exception of DCBZ which is handled as a special case here
795 */ 824 */
796 if (instr == DCBZ) { 825 if (instr == DCBZ) {
797 PPC_WARN_EMULATED(dcbz); 826 PPC_WARN_ALIGNMENT(dcbz, regs);
798 return emulate_dcbz(regs, addr); 827 return emulate_dcbz(regs, addr);
799 } 828 }
800 if (unlikely(nb == 0)) 829 if (unlikely(nb == 0))
@@ -804,7 +833,7 @@ int fix_alignment(struct pt_regs *regs)
804 * function 833 * function
805 */ 834 */
806 if (flags & M) { 835 if (flags & M) {
807 PPC_WARN_EMULATED(multiple); 836 PPC_WARN_ALIGNMENT(multiple, regs);
808 return emulate_multiple(regs, addr, reg, nb, 837 return emulate_multiple(regs, addr, reg, nb,
809 flags, instr, swiz); 838 flags, instr, swiz);
810 } 839 }
@@ -825,11 +854,11 @@ int fix_alignment(struct pt_regs *regs)
825 854
826 /* Special case for 16-byte FP loads and stores */ 855 /* Special case for 16-byte FP loads and stores */
827 if (nb == 16) { 856 if (nb == 16) {
828 PPC_WARN_EMULATED(fp_pair); 857 PPC_WARN_ALIGNMENT(fp_pair, regs);
829 return emulate_fp_pair(addr, reg, flags); 858 return emulate_fp_pair(addr, reg, flags);
830 } 859 }
831 860
832 PPC_WARN_EMULATED(unaligned); 861 PPC_WARN_ALIGNMENT(unaligned, regs);
833 862
834 /* If we are loading, get the data from user space, else 863 /* If we are loading, get the data from user space, else
835 * get it from register values 864 * get it from register values
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index 0812b0f414bb..c09138d150d4 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -133,7 +133,6 @@ int main(void)
133 DEFINE(PACAKMSR, offsetof(struct paca_struct, kernel_msr)); 133 DEFINE(PACAKMSR, offsetof(struct paca_struct, kernel_msr));
134 DEFINE(PACASOFTIRQEN, offsetof(struct paca_struct, soft_enabled)); 134 DEFINE(PACASOFTIRQEN, offsetof(struct paca_struct, soft_enabled));
135 DEFINE(PACAHARDIRQEN, offsetof(struct paca_struct, hard_enabled)); 135 DEFINE(PACAHARDIRQEN, offsetof(struct paca_struct, hard_enabled));
136 DEFINE(PACAPERFPEND, offsetof(struct paca_struct, perf_event_pending));
137 DEFINE(PACACONTEXTID, offsetof(struct paca_struct, context.id)); 136 DEFINE(PACACONTEXTID, offsetof(struct paca_struct, context.id));
138#ifdef CONFIG_PPC_MM_SLICES 137#ifdef CONFIG_PPC_MM_SLICES
139 DEFINE(PACALOWSLICESPSIZE, offsetof(struct paca_struct, 138 DEFINE(PACALOWSLICESPSIZE, offsetof(struct paca_struct,
@@ -190,6 +189,35 @@ int main(void)
190 DEFINE(PACA_SYSTEM_TIME, offsetof(struct paca_struct, system_time)); 189 DEFINE(PACA_SYSTEM_TIME, offsetof(struct paca_struct, system_time));
191 DEFINE(PACA_DATA_OFFSET, offsetof(struct paca_struct, data_offset)); 190 DEFINE(PACA_DATA_OFFSET, offsetof(struct paca_struct, data_offset));
192 DEFINE(PACA_TRAP_SAVE, offsetof(struct paca_struct, trap_save)); 191 DEFINE(PACA_TRAP_SAVE, offsetof(struct paca_struct, trap_save));
192#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
193 DEFINE(PACA_KVM_IN_GUEST, offsetof(struct paca_struct, kvm_in_guest));
194 DEFINE(PACA_KVM_SLB, offsetof(struct paca_struct, kvm_slb));
195 DEFINE(PACA_KVM_SLB_MAX, offsetof(struct paca_struct, kvm_slb_max));
196 DEFINE(PACA_KVM_CR, offsetof(struct paca_struct, shadow_vcpu.cr));
197 DEFINE(PACA_KVM_XER, offsetof(struct paca_struct, shadow_vcpu.xer));
198 DEFINE(PACA_KVM_R0, offsetof(struct paca_struct, shadow_vcpu.gpr[0]));
199 DEFINE(PACA_KVM_R1, offsetof(struct paca_struct, shadow_vcpu.gpr[1]));
200 DEFINE(PACA_KVM_R2, offsetof(struct paca_struct, shadow_vcpu.gpr[2]));
201 DEFINE(PACA_KVM_R3, offsetof(struct paca_struct, shadow_vcpu.gpr[3]));
202 DEFINE(PACA_KVM_R4, offsetof(struct paca_struct, shadow_vcpu.gpr[4]));
203 DEFINE(PACA_KVM_R5, offsetof(struct paca_struct, shadow_vcpu.gpr[5]));
204 DEFINE(PACA_KVM_R6, offsetof(struct paca_struct, shadow_vcpu.gpr[6]));
205 DEFINE(PACA_KVM_R7, offsetof(struct paca_struct, shadow_vcpu.gpr[7]));
206 DEFINE(PACA_KVM_R8, offsetof(struct paca_struct, shadow_vcpu.gpr[8]));
207 DEFINE(PACA_KVM_R9, offsetof(struct paca_struct, shadow_vcpu.gpr[9]));
208 DEFINE(PACA_KVM_R10, offsetof(struct paca_struct, shadow_vcpu.gpr[10]));
209 DEFINE(PACA_KVM_R11, offsetof(struct paca_struct, shadow_vcpu.gpr[11]));
210 DEFINE(PACA_KVM_R12, offsetof(struct paca_struct, shadow_vcpu.gpr[12]));
211 DEFINE(PACA_KVM_R13, offsetof(struct paca_struct, shadow_vcpu.gpr[13]));
212 DEFINE(PACA_KVM_HOST_R1, offsetof(struct paca_struct, shadow_vcpu.host_r1));
213 DEFINE(PACA_KVM_HOST_R2, offsetof(struct paca_struct, shadow_vcpu.host_r2));
214 DEFINE(PACA_KVM_VMHANDLER, offsetof(struct paca_struct,
215 shadow_vcpu.vmhandler));
216 DEFINE(PACA_KVM_SCRATCH0, offsetof(struct paca_struct,
217 shadow_vcpu.scratch0));
218 DEFINE(PACA_KVM_SCRATCH1, offsetof(struct paca_struct,
219 shadow_vcpu.scratch1));
220#endif
193#endif /* CONFIG_PPC64 */ 221#endif /* CONFIG_PPC64 */
194 222
195 /* RTAS */ 223 /* RTAS */
@@ -384,8 +412,6 @@ int main(void)
384 DEFINE(VCPU_HOST_PID, offsetof(struct kvm_vcpu, arch.host_pid)); 412 DEFINE(VCPU_HOST_PID, offsetof(struct kvm_vcpu, arch.host_pid));
385 DEFINE(VCPU_GPRS, offsetof(struct kvm_vcpu, arch.gpr)); 413 DEFINE(VCPU_GPRS, offsetof(struct kvm_vcpu, arch.gpr));
386 DEFINE(VCPU_LR, offsetof(struct kvm_vcpu, arch.lr)); 414 DEFINE(VCPU_LR, offsetof(struct kvm_vcpu, arch.lr));
387 DEFINE(VCPU_CR, offsetof(struct kvm_vcpu, arch.cr));
388 DEFINE(VCPU_XER, offsetof(struct kvm_vcpu, arch.xer));
389 DEFINE(VCPU_CTR, offsetof(struct kvm_vcpu, arch.ctr)); 415 DEFINE(VCPU_CTR, offsetof(struct kvm_vcpu, arch.ctr));
390 DEFINE(VCPU_PC, offsetof(struct kvm_vcpu, arch.pc)); 416 DEFINE(VCPU_PC, offsetof(struct kvm_vcpu, arch.pc));
391 DEFINE(VCPU_MSR, offsetof(struct kvm_vcpu, arch.msr)); 417 DEFINE(VCPU_MSR, offsetof(struct kvm_vcpu, arch.msr));
@@ -398,14 +424,29 @@ int main(void)
398 DEFINE(VCPU_LAST_INST, offsetof(struct kvm_vcpu, arch.last_inst)); 424 DEFINE(VCPU_LAST_INST, offsetof(struct kvm_vcpu, arch.last_inst));
399 DEFINE(VCPU_FAULT_DEAR, offsetof(struct kvm_vcpu, arch.fault_dear)); 425 DEFINE(VCPU_FAULT_DEAR, offsetof(struct kvm_vcpu, arch.fault_dear));
400 DEFINE(VCPU_FAULT_ESR, offsetof(struct kvm_vcpu, arch.fault_esr)); 426 DEFINE(VCPU_FAULT_ESR, offsetof(struct kvm_vcpu, arch.fault_esr));
427
428 /* book3s_64 */
429#ifdef CONFIG_PPC64
430 DEFINE(VCPU_FAULT_DSISR, offsetof(struct kvm_vcpu, arch.fault_dsisr));
431 DEFINE(VCPU_HOST_RETIP, offsetof(struct kvm_vcpu, arch.host_retip));
432 DEFINE(VCPU_HOST_R2, offsetof(struct kvm_vcpu, arch.host_r2));
433 DEFINE(VCPU_HOST_MSR, offsetof(struct kvm_vcpu, arch.host_msr));
434 DEFINE(VCPU_SHADOW_MSR, offsetof(struct kvm_vcpu, arch.shadow_msr));
435 DEFINE(VCPU_SHADOW_SRR1, offsetof(struct kvm_vcpu, arch.shadow_srr1));
436 DEFINE(VCPU_TRAMPOLINE_LOWMEM, offsetof(struct kvm_vcpu, arch.trampoline_lowmem));
437 DEFINE(VCPU_TRAMPOLINE_ENTER, offsetof(struct kvm_vcpu, arch.trampoline_enter));
438 DEFINE(VCPU_HIGHMEM_HANDLER, offsetof(struct kvm_vcpu, arch.highmem_handler));
439 DEFINE(VCPU_RMCALL, offsetof(struct kvm_vcpu, arch.rmcall));
440 DEFINE(VCPU_HFLAGS, offsetof(struct kvm_vcpu, arch.hflags));
441#else
442 DEFINE(VCPU_CR, offsetof(struct kvm_vcpu, arch.cr));
443 DEFINE(VCPU_XER, offsetof(struct kvm_vcpu, arch.xer));
444#endif /* CONFIG_PPC64 */
401#endif 445#endif
402#ifdef CONFIG_44x 446#ifdef CONFIG_44x
403 DEFINE(PGD_T_LOG2, PGD_T_LOG2); 447 DEFINE(PGD_T_LOG2, PGD_T_LOG2);
404 DEFINE(PTE_T_LOG2, PTE_T_LOG2); 448 DEFINE(PTE_T_LOG2, PTE_T_LOG2);
405#endif 449#endif
406#ifdef CONFIG_FSL_BOOKE
407 DEFINE(TLBCAM_SIZE, sizeof(struct tlbcam));
408#endif
409 450
410#ifdef CONFIG_KVM_EXIT_TIMING 451#ifdef CONFIG_KVM_EXIT_TIMING
411 DEFINE(VCPU_TIMING_EXIT_TBU, offsetof(struct kvm_vcpu, 452 DEFINE(VCPU_TIMING_EXIT_TBU, offsetof(struct kvm_vcpu,
diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index bb37b1d19a58..a3c684b4c862 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -19,6 +19,7 @@
19#include <linux/notifier.h> 19#include <linux/notifier.h>
20#include <linux/of.h> 20#include <linux/of.h>
21#include <linux/percpu.h> 21#include <linux/percpu.h>
22#include <linux/slab.h>
22#include <asm/prom.h> 23#include <asm/prom.h>
23 24
24#include "cacheinfo.h" 25#include "cacheinfo.h"
@@ -642,7 +643,7 @@ static struct kobj_attribute *cache_index_opt_attrs[] = {
642 &cache_assoc_attr, 643 &cache_assoc_attr,
643}; 644};
644 645
645static struct sysfs_ops cache_index_ops = { 646static const struct sysfs_ops cache_index_ops = {
646 .show = cache_index_show, 647 .show = cache_index_show,
647}; 648};
648 649
diff --git a/arch/powerpc/kernel/cputable.c b/arch/powerpc/kernel/cputable.c
index 03c862b6a9c4..8af4949434b2 100644
--- a/arch/powerpc/kernel/cputable.c
+++ b/arch/powerpc/kernel/cputable.c
@@ -697,9 +697,9 @@ static struct cpu_spec __initdata cpu_specs[] = {
697 .machine_check = machine_check_generic, 697 .machine_check = machine_check_generic,
698 .platform = "ppc750", 698 .platform = "ppc750",
699 }, 699 },
700 { /* 750CL */ 700 { /* 750CL (and "Broadway") */
701 .pvr_mask = 0xfffff0f0, 701 .pvr_mask = 0xfffff0e0,
702 .pvr_value = 0x00087010, 702 .pvr_value = 0x00087000,
703 .cpu_name = "750CL", 703 .cpu_name = "750CL",
704 .cpu_features = CPU_FTRS_750CL, 704 .cpu_features = CPU_FTRS_750CL,
705 .cpu_user_features = COMMON_USER | PPC_FEATURE_PPC_LE, 705 .cpu_user_features = COMMON_USER | PPC_FEATURE_PPC_LE,
@@ -1808,7 +1808,7 @@ static struct cpu_spec __initdata cpu_specs[] = {
1808 .icache_bsize = 64, 1808 .icache_bsize = 64,
1809 .dcache_bsize = 64, 1809 .dcache_bsize = 64,
1810 .num_pmcs = 4, 1810 .num_pmcs = 4,
1811 .oprofile_cpu_type = "ppc/e500", /* xxx - galak, e500mc? */ 1811 .oprofile_cpu_type = "ppc/e500mc",
1812 .oprofile_type = PPC_OPROFILE_FSL_EMB, 1812 .oprofile_type = PPC_OPROFILE_FSL_EMB,
1813 .cpu_setup = __setup_cpu_e500mc, 1813 .cpu_setup = __setup_cpu_e500mc,
1814 .machine_check = machine_check_e500, 1814 .machine_check = machine_check_e500,
diff --git a/arch/powerpc/kernel/crash.c b/arch/powerpc/kernel/crash.c
index 0a8439aafdd1..6f4613dd05ef 100644
--- a/arch/powerpc/kernel/crash.c
+++ b/arch/powerpc/kernel/crash.c
@@ -373,7 +373,7 @@ void default_machine_crash_shutdown(struct pt_regs *regs)
373 hard_irq_disable(); 373 hard_irq_disable();
374 374
375 for_each_irq(i) { 375 for_each_irq(i) {
376 struct irq_desc *desc = irq_desc + i; 376 struct irq_desc *desc = irq_to_desc(i);
377 377
378 if (desc->status & IRQ_INPROGRESS) 378 if (desc->status & IRQ_INPROGRESS)
379 desc->chip->eoi(i); 379 desc->chip->eoi(i);
diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/dma-swiotlb.c
index e96cbbd9b449..4ff4da2c238b 100644
--- a/arch/powerpc/kernel/dma-swiotlb.c
+++ b/arch/powerpc/kernel/dma-swiotlb.c
@@ -1,7 +1,8 @@
1/* 1/*
2 * Contains routines needed to support swiotlb for ppc. 2 * Contains routines needed to support swiotlb for ppc.
3 * 3 *
4 * Copyright (C) 2009 Becky Bruce, Freescale Semiconductor 4 * Copyright (C) 2009-2010 Freescale Semiconductor, Inc.
5 * Author: Becky Bruce
5 * 6 *
6 * This program is free software; you can redistribute it and/or modify it 7 * This program is free software; you can redistribute it and/or modify it
7 * under the terms of the GNU General Public License as published by the 8 * under the terms of the GNU General Public License as published by the
@@ -21,7 +22,6 @@
21#include <asm/dma.h> 22#include <asm/dma.h>
22#include <asm/abs_addr.h> 23#include <asm/abs_addr.h>
23 24
24int swiotlb __read_mostly;
25unsigned int ppc_swiotlb_enable; 25unsigned int ppc_swiotlb_enable;
26 26
27/* 27/*
@@ -71,7 +71,7 @@ static int ppc_swiotlb_bus_notify(struct notifier_block *nb,
71 sd->max_direct_dma_addr = 0; 71 sd->max_direct_dma_addr = 0;
72 72
73 /* May need to bounce if the device can't address all of DRAM */ 73 /* May need to bounce if the device can't address all of DRAM */
74 if (dma_get_mask(dev) < lmb_end_of_DRAM()) 74 if ((dma_get_mask(dev) + 1) < lmb_end_of_DRAM())
75 set_dma_ops(dev, &swiotlb_dma_ops); 75 set_dma_ops(dev, &swiotlb_dma_ops);
76 76
77 return NOTIFY_DONE; 77 return NOTIFY_DONE;
diff --git a/arch/powerpc/kernel/dma.c b/arch/powerpc/kernel/dma.c
index 6215062caf8c..6c1df5757cd6 100644
--- a/arch/powerpc/kernel/dma.c
+++ b/arch/powerpc/kernel/dma.c
@@ -8,6 +8,7 @@
8#include <linux/device.h> 8#include <linux/device.h>
9#include <linux/dma-mapping.h> 9#include <linux/dma-mapping.h>
10#include <linux/dma-debug.h> 10#include <linux/dma-debug.h>
11#include <linux/gfp.h>
11#include <linux/lmb.h> 12#include <linux/lmb.h>
12#include <asm/bug.h> 13#include <asm/bug.h>
13#include <asm/abs_addr.h> 14#include <asm/abs_addr.h>
diff --git a/arch/powerpc/kernel/e500-pmu.c b/arch/powerpc/kernel/e500-pmu.c
new file mode 100644
index 000000000000..7c07de0d8943
--- /dev/null
+++ b/arch/powerpc/kernel/e500-pmu.c
@@ -0,0 +1,129 @@
1/*
2 * Performance counter support for e500 family processors.
3 *
4 * Copyright 2008-2009 Paul Mackerras, IBM Corporation.
5 * Copyright 2010 Freescale Semiconductor, Inc.
6 *
7 * This program is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation; either version
10 * 2 of the License, or (at your option) any later version.
11 */
12#include <linux/string.h>
13#include <linux/perf_event.h>
14#include <asm/reg.h>
15#include <asm/cputable.h>
16
17/*
18 * Map of generic hardware event types to hardware events
19 * Zero if unsupported
20 */
21static int e500_generic_events[] = {
22 [PERF_COUNT_HW_CPU_CYCLES] = 1,
23 [PERF_COUNT_HW_INSTRUCTIONS] = 2,
24 [PERF_COUNT_HW_CACHE_MISSES] = 41, /* Data L1 cache reloads */
25 [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 12,
26 [PERF_COUNT_HW_BRANCH_MISSES] = 15,
27};
28
29#define C(x) PERF_COUNT_HW_CACHE_##x
30
31/*
32 * Table of generalized cache-related events.
33 * 0 means not supported, -1 means nonsensical, other values
34 * are event codes.
35 */
36static int e500_cache_events[C(MAX)][C(OP_MAX)][C(RESULT_MAX)] = {
37 /*
38 * D-cache misses are not split into read/write/prefetch;
39 * use raw event 41.
40 */
41 [C(L1D)] = { /* RESULT_ACCESS RESULT_MISS */
42 [C(OP_READ)] = { 27, 0 },
43 [C(OP_WRITE)] = { 28, 0 },
44 [C(OP_PREFETCH)] = { 29, 0 },
45 },
46 [C(L1I)] = { /* RESULT_ACCESS RESULT_MISS */
47 [C(OP_READ)] = { 2, 60 },
48 [C(OP_WRITE)] = { -1, -1 },
49 [C(OP_PREFETCH)] = { 0, 0 },
50 },
51 /*
52 * Assuming LL means L2, it's not a good match for this model.
53 * It allocates only on L1 castout or explicit prefetch, and
54 * does not have separate read/write events (but it does have
55 * separate instruction/data events).
56 */
57 [C(LL)] = { /* RESULT_ACCESS RESULT_MISS */
58 [C(OP_READ)] = { 0, 0 },
59 [C(OP_WRITE)] = { 0, 0 },
60 [C(OP_PREFETCH)] = { 0, 0 },
61 },
62 /*
63 * There are data/instruction MMU misses, but that's a miss on
64 * the chip's internal level-one TLB which is probably not
65 * what the user wants. Instead, unified level-two TLB misses
66 * are reported here.
67 */
68 [C(DTLB)] = { /* RESULT_ACCESS RESULT_MISS */
69 [C(OP_READ)] = { 26, 66 },
70 [C(OP_WRITE)] = { -1, -1 },
71 [C(OP_PREFETCH)] = { -1, -1 },
72 },
73 [C(BPU)] = { /* RESULT_ACCESS RESULT_MISS */
74 [C(OP_READ)] = { 12, 15 },
75 [C(OP_WRITE)] = { -1, -1 },
76 [C(OP_PREFETCH)] = { -1, -1 },
77 },
78};
79
80static int num_events = 128;
81
82/* Upper half of event id is PMLCb, for threshold events */
83static u64 e500_xlate_event(u64 event_id)
84{
85 u32 event_low = (u32)event_id;
86 u64 ret;
87
88 if (event_low >= num_events)
89 return 0;
90
91 ret = FSL_EMB_EVENT_VALID;
92
93 if (event_low >= 76 && event_low <= 81) {
94 ret |= FSL_EMB_EVENT_RESTRICTED;
95 ret |= event_id &
96 (FSL_EMB_EVENT_THRESHMUL | FSL_EMB_EVENT_THRESH);
97 } else if (event_id &
98 (FSL_EMB_EVENT_THRESHMUL | FSL_EMB_EVENT_THRESH)) {
99 /* Threshold requested on non-threshold event */
100 return 0;
101 }
102
103 return ret;
104}
105
106static struct fsl_emb_pmu e500_pmu = {
107 .name = "e500 family",
108 .n_counter = 4,
109 .n_restricted = 2,
110 .xlate_event = e500_xlate_event,
111 .n_generic = ARRAY_SIZE(e500_generic_events),
112 .generic_events = e500_generic_events,
113 .cache_events = &e500_cache_events,
114};
115
116static int init_e500_pmu(void)
117{
118 if (!cur_cpu_spec->oprofile_cpu_type)
119 return -ENODEV;
120
121 if (!strcmp(cur_cpu_spec->oprofile_cpu_type, "ppc/e500mc"))
122 num_events = 256;
123 else if (strcmp(cur_cpu_spec->oprofile_cpu_type, "ppc/e500"))
124 return -ENODEV;
125
126 return register_fsl_emb_pmu(&e500_pmu);
127}
128
129arch_initcall(init_e500_pmu);
diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S
index 9763267e38b4..42e9d908914a 100644
--- a/arch/powerpc/kernel/entry_64.S
+++ b/arch/powerpc/kernel/entry_64.S
@@ -551,20 +551,11 @@ restore:
551BEGIN_FW_FTR_SECTION 551BEGIN_FW_FTR_SECTION
552 ld r5,SOFTE(r1) 552 ld r5,SOFTE(r1)
553FW_FTR_SECTION_ELSE 553FW_FTR_SECTION_ELSE
554 b iseries_check_pending_irqs 554 b .Liseries_check_pending_irqs
555ALT_FW_FTR_SECTION_END_IFCLR(FW_FEATURE_ISERIES) 555ALT_FW_FTR_SECTION_END_IFCLR(FW_FEATURE_ISERIES)
5562: 5562:
557 TRACE_AND_RESTORE_IRQ(r5); 557 TRACE_AND_RESTORE_IRQ(r5);
558 558
559#ifdef CONFIG_PERF_EVENTS
560 /* check paca->perf_event_pending if we're enabling ints */
561 lbz r3,PACAPERFPEND(r13)
562 and. r3,r3,r5
563 beq 27f
564 bl .perf_event_do_pending
56527:
566#endif /* CONFIG_PERF_EVENTS */
567
568 /* extract EE bit and use it to restore paca->hard_enabled */ 559 /* extract EE bit and use it to restore paca->hard_enabled */
569 ld r3,_MSR(r1) 560 ld r3,_MSR(r1)
570 rldicl r4,r3,49,63 /* r0 = (r3 >> 15) & 1 */ 561 rldicl r4,r3,49,63 /* r0 = (r3 >> 15) & 1 */
@@ -623,7 +614,7 @@ ALT_FW_FTR_SECTION_END_IFCLR(FW_FEATURE_ISERIES)
623 614
624#endif /* CONFIG_PPC_BOOK3E */ 615#endif /* CONFIG_PPC_BOOK3E */
625 616
626iseries_check_pending_irqs: 617.Liseries_check_pending_irqs:
627#ifdef CONFIG_PPC_ISERIES 618#ifdef CONFIG_PPC_ISERIES
628 ld r5,SOFTE(r1) 619 ld r5,SOFTE(r1)
629 cmpdi 0,r5,0 620 cmpdi 0,r5,0
@@ -791,9 +782,8 @@ _GLOBAL(enter_rtas)
791 782
792 li r9,1 783 li r9,1
793 rldicr r9,r9,MSR_SF_LG,(63-MSR_SF_LG) 784 rldicr r9,r9,MSR_SF_LG,(63-MSR_SF_LG)
794 ori r9,r9,MSR_IR|MSR_DR|MSR_FE0|MSR_FE1|MSR_FP 785 ori r9,r9,MSR_IR|MSR_DR|MSR_FE0|MSR_FE1|MSR_FP|MSR_RI
795 andc r6,r0,r9 786 andc r6,r0,r9
796 ori r6,r6,MSR_RI
797 sync /* disable interrupts so SRR0/1 */ 787 sync /* disable interrupts so SRR0/1 */
798 mtmsrd r0 /* don't get trashed */ 788 mtmsrd r0 /* don't get trashed */
799 789
diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
index 1808876edcc9..e3be98ffe2a7 100644
--- a/arch/powerpc/kernel/exceptions-64s.S
+++ b/arch/powerpc/kernel/exceptions-64s.S
@@ -41,6 +41,7 @@ __start_interrupts:
41 . = 0x200 41 . = 0x200
42_machine_check_pSeries: 42_machine_check_pSeries:
43 HMT_MEDIUM 43 HMT_MEDIUM
44 DO_KVM 0x200
44 mtspr SPRN_SPRG_SCRATCH0,r13 /* save r13 */ 45 mtspr SPRN_SPRG_SCRATCH0,r13 /* save r13 */
45 EXCEPTION_PROLOG_PSERIES(PACA_EXMC, machine_check_common) 46 EXCEPTION_PROLOG_PSERIES(PACA_EXMC, machine_check_common)
46 47
@@ -48,6 +49,7 @@ _machine_check_pSeries:
48 .globl data_access_pSeries 49 .globl data_access_pSeries
49data_access_pSeries: 50data_access_pSeries:
50 HMT_MEDIUM 51 HMT_MEDIUM
52 DO_KVM 0x300
51 mtspr SPRN_SPRG_SCRATCH0,r13 53 mtspr SPRN_SPRG_SCRATCH0,r13
52BEGIN_FTR_SECTION 54BEGIN_FTR_SECTION
53 mfspr r13,SPRN_SPRG_PACA 55 mfspr r13,SPRN_SPRG_PACA
@@ -77,6 +79,7 @@ ALT_FTR_SECTION_END_IFCLR(CPU_FTR_SLB)
77 .globl data_access_slb_pSeries 79 .globl data_access_slb_pSeries
78data_access_slb_pSeries: 80data_access_slb_pSeries:
79 HMT_MEDIUM 81 HMT_MEDIUM
82 DO_KVM 0x380
80 mtspr SPRN_SPRG_SCRATCH0,r13 83 mtspr SPRN_SPRG_SCRATCH0,r13
81 mfspr r13,SPRN_SPRG_PACA /* get paca address into r13 */ 84 mfspr r13,SPRN_SPRG_PACA /* get paca address into r13 */
82 std r3,PACA_EXSLB+EX_R3(r13) 85 std r3,PACA_EXSLB+EX_R3(r13)
@@ -115,6 +118,7 @@ data_access_slb_pSeries:
115 .globl instruction_access_slb_pSeries 118 .globl instruction_access_slb_pSeries
116instruction_access_slb_pSeries: 119instruction_access_slb_pSeries:
117 HMT_MEDIUM 120 HMT_MEDIUM
121 DO_KVM 0x480
118 mtspr SPRN_SPRG_SCRATCH0,r13 122 mtspr SPRN_SPRG_SCRATCH0,r13
119 mfspr r13,SPRN_SPRG_PACA /* get paca address into r13 */ 123 mfspr r13,SPRN_SPRG_PACA /* get paca address into r13 */
120 std r3,PACA_EXSLB+EX_R3(r13) 124 std r3,PACA_EXSLB+EX_R3(r13)
@@ -154,6 +158,7 @@ instruction_access_slb_pSeries:
154 .globl system_call_pSeries 158 .globl system_call_pSeries
155system_call_pSeries: 159system_call_pSeries:
156 HMT_MEDIUM 160 HMT_MEDIUM
161 DO_KVM 0xc00
157BEGIN_FTR_SECTION 162BEGIN_FTR_SECTION
158 cmpdi r0,0x1ebe 163 cmpdi r0,0x1ebe
159 beq- 1f 164 beq- 1f
@@ -185,13 +190,19 @@ END_FTR_SECTION_IFSET(CPU_FTR_REAL_LE)
185 * prolog code of the PerformanceMonitor one. A little 190 * prolog code of the PerformanceMonitor one. A little
186 * trickery is thus necessary 191 * trickery is thus necessary
187 */ 192 */
193performance_monitor_pSeries_1:
188 . = 0xf00 194 . = 0xf00
195 DO_KVM 0xf00
189 b performance_monitor_pSeries 196 b performance_monitor_pSeries
190 197
198altivec_unavailable_pSeries_1:
191 . = 0xf20 199 . = 0xf20
200 DO_KVM 0xf20
192 b altivec_unavailable_pSeries 201 b altivec_unavailable_pSeries
193 202
203vsx_unavailable_pSeries_1:
194 . = 0xf40 204 . = 0xf40
205 DO_KVM 0xf40
195 b vsx_unavailable_pSeries 206 b vsx_unavailable_pSeries
196 207
197#ifdef CONFIG_CBE_RAS 208#ifdef CONFIG_CBE_RAS
diff --git a/arch/powerpc/kernel/firmware.c b/arch/powerpc/kernel/firmware.c
index 1679a70bbcad..6b1f4271eb53 100644
--- a/arch/powerpc/kernel/firmware.c
+++ b/arch/powerpc/kernel/firmware.c
@@ -17,5 +17,5 @@
17 17
18#include <asm/firmware.h> 18#include <asm/firmware.h>
19 19
20unsigned long powerpc_firmware_features; 20unsigned long powerpc_firmware_features __read_mostly;
21EXPORT_SYMBOL_GPL(powerpc_firmware_features); 21EXPORT_SYMBOL_GPL(powerpc_firmware_features);
diff --git a/arch/powerpc/kernel/head_32.S b/arch/powerpc/kernel/head_32.S
index 829c3fe7c5a2..e025e89fe93e 100644
--- a/arch/powerpc/kernel/head_32.S
+++ b/arch/powerpc/kernel/head_32.S
@@ -164,6 +164,9 @@ __after_mmu_off:
164#ifdef CONFIG_PPC_EARLY_DEBUG_CPM 164#ifdef CONFIG_PPC_EARLY_DEBUG_CPM
165 bl setup_cpm_bat 165 bl setup_cpm_bat
166#endif 166#endif
167#ifdef CONFIG_PPC_EARLY_DEBUG_USBGECKO
168 bl setup_usbgecko_bat
169#endif
167 170
168/* 171/*
169 * Call setup_cpu for CPU 0 and initialize 6xx Idle 172 * Call setup_cpu for CPU 0 and initialize 6xx Idle
@@ -1203,6 +1206,28 @@ setup_cpm_bat:
1203 blr 1206 blr
1204#endif 1207#endif
1205 1208
1209#ifdef CONFIG_PPC_EARLY_DEBUG_USBGECKO
1210setup_usbgecko_bat:
1211 /* prepare a BAT for early io */
1212#if defined(CONFIG_GAMECUBE)
1213 lis r8, 0x0c00
1214#elif defined(CONFIG_WII)
1215 lis r8, 0x0d00
1216#else
1217#error Invalid platform for USB Gecko based early debugging.
1218#endif
1219 /*
1220 * The virtual address used must match the virtual address
1221 * associated to the fixmap entry FIX_EARLY_DEBUG_BASE.
1222 */
1223 lis r11, 0xfffe /* top 128K */
1224 ori r8, r8, 0x002a /* uncached, guarded ,rw */
1225 ori r11, r11, 0x2 /* 128K, Vs=1, Vp=0 */
1226 mtspr SPRN_DBAT1L, r8
1227 mtspr SPRN_DBAT1U, r11
1228 blr
1229#endif
1230
1206#ifdef CONFIG_8260 1231#ifdef CONFIG_8260
1207/* Jump into the system reset for the rom. 1232/* Jump into the system reset for the rom.
1208 * We first disable the MMU, and then jump to the ROM reset address. 1233 * We first disable the MMU, and then jump to the ROM reset address.
diff --git a/arch/powerpc/kernel/head_64.S b/arch/powerpc/kernel/head_64.S
index c38afdb45d7b..bed9a29ee383 100644
--- a/arch/powerpc/kernel/head_64.S
+++ b/arch/powerpc/kernel/head_64.S
@@ -37,6 +37,7 @@
37#include <asm/firmware.h> 37#include <asm/firmware.h>
38#include <asm/page_64.h> 38#include <asm/page_64.h>
39#include <asm/irqflags.h> 39#include <asm/irqflags.h>
40#include <asm/kvm_book3s_64_asm.h>
40 41
41/* The physical memory is layed out such that the secondary processor 42/* The physical memory is layed out such that the secondary processor
42 * spin code sits at 0x0000...0x00ff. On server, the vectors follow 43 * spin code sits at 0x0000...0x00ff. On server, the vectors follow
@@ -165,6 +166,12 @@ exception_marker:
165#include "exceptions-64s.S" 166#include "exceptions-64s.S"
166#endif 167#endif
167 168
169/* KVM trampoline code needs to be close to the interrupt handlers */
170
171#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
172#include "../kvm/book3s_64_rmhandlers.S"
173#endif
174
168_GLOBAL(generic_secondary_thread_init) 175_GLOBAL(generic_secondary_thread_init)
169 mr r24,r3 176 mr r24,r3
170 177
@@ -212,7 +219,8 @@ generic_secondary_common_init:
212 * physical cpu id in r24, we need to search the pacas to find 219 * physical cpu id in r24, we need to search the pacas to find
213 * which logical id maps to our physical one. 220 * which logical id maps to our physical one.
214 */ 221 */
215 LOAD_REG_ADDR(r13, paca) /* Get base vaddr of paca array */ 222 LOAD_REG_ADDR(r13, paca) /* Load paca pointer */
223 ld r13,0(r13) /* Get base vaddr of paca array */
216 li r5,0 /* logical cpu id */ 224 li r5,0 /* logical cpu id */
2171: lhz r6,PACAHWCPUID(r13) /* Load HW procid from paca */ 2251: lhz r6,PACAHWCPUID(r13) /* Load HW procid from paca */
218 cmpw r6,r24 /* Compare to our id */ 226 cmpw r6,r24 /* Compare to our id */
@@ -529,7 +537,8 @@ _GLOBAL(pmac_secondary_start)
529 mtmsrd r3 /* RI on */ 537 mtmsrd r3 /* RI on */
530 538
531 /* Set up a paca value for this processor. */ 539 /* Set up a paca value for this processor. */
532 LOAD_REG_ADDR(r4,paca) /* Get base vaddr of paca array */ 540 LOAD_REG_ADDR(r4,paca) /* Load paca pointer */
541 ld r4,0(r4) /* Get base vaddr of paca array */
533 mulli r13,r24,PACA_SIZE /* Calculate vaddr of right paca */ 542 mulli r13,r24,PACA_SIZE /* Calculate vaddr of right paca */
534 add r13,r13,r4 /* for this processor. */ 543 add r13,r13,r4 /* for this processor. */
535 mtspr SPRN_SPRG_PACA,r13 /* Save vaddr of paca in an SPRG*/ 544 mtspr SPRN_SPRG_PACA,r13 /* Save vaddr of paca in an SPRG*/
@@ -608,6 +617,17 @@ _GLOBAL(start_secondary_prolog)
608 std r3,0(r1) /* Zero the stack frame pointer */ 617 std r3,0(r1) /* Zero the stack frame pointer */
609 bl .start_secondary 618 bl .start_secondary
610 b . 619 b .
620/*
621 * Reset stack pointer and call start_secondary
622 * to continue with online operation when woken up
623 * from cede in cpu offline.
624 */
625_GLOBAL(start_secondary_resume)
626 ld r1,PACAKSAVE(r13) /* Reload kernel stack pointer */
627 li r3,0
628 std r3,0(r1) /* Zero the stack frame pointer */
629 bl .start_secondary
630 b .
611#endif 631#endif
612 632
613/* 633/*
diff --git a/arch/powerpc/kernel/head_8xx.S b/arch/powerpc/kernel/head_8xx.S
index 6ded19d01891..3ef743fa5d7c 100644
--- a/arch/powerpc/kernel/head_8xx.S
+++ b/arch/powerpc/kernel/head_8xx.S
@@ -206,6 +206,8 @@ MachineCheck:
206 EXCEPTION_PROLOG 206 EXCEPTION_PROLOG
207 mfspr r4,SPRN_DAR 207 mfspr r4,SPRN_DAR
208 stw r4,_DAR(r11) 208 stw r4,_DAR(r11)
209 li r5,0x00f0
210 mtspr SPRN_DAR,r5 /* Tag DAR, to be used in DTLB Error */
209 mfspr r5,SPRN_DSISR 211 mfspr r5,SPRN_DSISR
210 stw r5,_DSISR(r11) 212 stw r5,_DSISR(r11)
211 addi r3,r1,STACK_FRAME_OVERHEAD 213 addi r3,r1,STACK_FRAME_OVERHEAD
@@ -222,6 +224,8 @@ DataAccess:
222 stw r10,_DSISR(r11) 224 stw r10,_DSISR(r11)
223 mr r5,r10 225 mr r5,r10
224 mfspr r4,SPRN_DAR 226 mfspr r4,SPRN_DAR
227 li r10,0x00f0
228 mtspr SPRN_DAR,r10 /* Tag DAR, to be used in DTLB Error */
225 EXC_XFER_EE_LITE(0x300, handle_page_fault) 229 EXC_XFER_EE_LITE(0x300, handle_page_fault)
226 230
227/* Instruction access exception. 231/* Instruction access exception.
@@ -244,6 +248,8 @@ Alignment:
244 EXCEPTION_PROLOG 248 EXCEPTION_PROLOG
245 mfspr r4,SPRN_DAR 249 mfspr r4,SPRN_DAR
246 stw r4,_DAR(r11) 250 stw r4,_DAR(r11)
251 li r5,0x00f0
252 mtspr SPRN_DAR,r5 /* Tag DAR, to be used in DTLB Error */
247 mfspr r5,SPRN_DSISR 253 mfspr r5,SPRN_DSISR
248 stw r5,_DSISR(r11) 254 stw r5,_DSISR(r11)
249 addi r3,r1,STACK_FRAME_OVERHEAD 255 addi r3,r1,STACK_FRAME_OVERHEAD
@@ -333,26 +339,20 @@ InstructionTLBMiss:
333 mfspr r11, SPRN_MD_TWC /* ....and get the pte address */ 339 mfspr r11, SPRN_MD_TWC /* ....and get the pte address */
334 lwz r10, 0(r11) /* Get the pte */ 340 lwz r10, 0(r11) /* Get the pte */
335 341
336#ifdef CONFIG_SWAP 342 andi. r11, r10, _PAGE_ACCESSED | _PAGE_PRESENT
337 /* do not set the _PAGE_ACCESSED bit of a non-present page */ 343 cmpwi cr0, r11, _PAGE_ACCESSED | _PAGE_PRESENT
338 andi. r11, r10, _PAGE_PRESENT 344 bne- cr0, 2f
339 beq 4f 345
340 ori r10, r10, _PAGE_ACCESSED 346 /* Clear PP lsb, 0x400 */
341 mfspr r11, SPRN_MD_TWC /* get the pte address again */ 347 rlwinm r10, r10, 0, 22, 20
342 stw r10, 0(r11)
3434:
344#else
345 ori r10, r10, _PAGE_ACCESSED
346 stw r10, 0(r11)
347#endif
348 348
349 /* The Linux PTE won't go exactly into the MMU TLB. 349 /* The Linux PTE won't go exactly into the MMU TLB.
350 * Software indicator bits 21, 22 and 28 must be clear. 350 * Software indicator bits 22 and 28 must be clear.
351 * Software indicator bits 24, 25, 26, and 27 must be 351 * Software indicator bits 24, 25, 26, and 27 must be
352 * set. All other Linux PTE bits control the behavior 352 * set. All other Linux PTE bits control the behavior
353 * of the MMU. 353 * of the MMU.
354 */ 354 */
3552: li r11, 0x00f0 355 li r11, 0x00f0
356 rlwimi r10, r11, 0, 24, 28 /* Set 24-27, clear 28 */ 356 rlwimi r10, r11, 0, 24, 28 /* Set 24-27, clear 28 */
357 DO_8xx_CPU6(0x2d80, r3) 357 DO_8xx_CPU6(0x2d80, r3)
358 mtspr SPRN_MI_RPN, r10 /* Update TLB entry */ 358 mtspr SPRN_MI_RPN, r10 /* Update TLB entry */
@@ -365,6 +365,22 @@ InstructionTLBMiss:
365 lwz r3, 8(r0) 365 lwz r3, 8(r0)
366#endif 366#endif
367 rfi 367 rfi
3682:
369 mfspr r11, SPRN_SRR1
370 /* clear all error bits as TLB Miss
371 * sets a few unconditionally
372 */
373 rlwinm r11, r11, 0, 0xffff
374 mtspr SPRN_SRR1, r11
375
376 mfspr r10, SPRN_M_TW /* Restore registers */
377 lwz r11, 0(r0)
378 mtcr r11
379 lwz r11, 4(r0)
380#ifdef CONFIG_8xx_CPU6
381 lwz r3, 8(r0)
382#endif
383 b InstructionAccess
368 384
369 . = 0x1200 385 . = 0x1200
370DataStoreTLBMiss: 386DataStoreTLBMiss:
@@ -406,29 +422,45 @@ DataStoreTLBMiss:
406 * above. 422 * above.
407 */ 423 */
408 rlwimi r11, r10, 0, 27, 27 424 rlwimi r11, r10, 0, 27, 27
425 /* Insert the WriteThru flag into the TWC from the Linux PTE.
426 * It is bit 25 in the Linux PTE and bit 30 in the TWC
427 */
428 rlwimi r11, r10, 32-5, 30, 30
409 DO_8xx_CPU6(0x3b80, r3) 429 DO_8xx_CPU6(0x3b80, r3)
410 mtspr SPRN_MD_TWC, r11 430 mtspr SPRN_MD_TWC, r11
411 431
412#ifdef CONFIG_SWAP 432 /* Both _PAGE_ACCESSED and _PAGE_PRESENT has to be set.
413 /* do not set the _PAGE_ACCESSED bit of a non-present page */ 433 * We also need to know if the insn is a load/store, so:
414 andi. r11, r10, _PAGE_PRESENT 434 * Clear _PAGE_PRESENT and load that which will
415 beq 4f 435 * trap into DTLB Error with store bit set accordinly.
416 ori r10, r10, _PAGE_ACCESSED 436 */
4174: 437 /* PRESENT=0x1, ACCESSED=0x20
418 /* and update pte in table */ 438 * r11 = ((r10 & PRESENT) & ((r10 & ACCESSED) >> 5));
419#else 439 * r10 = (r10 & ~PRESENT) | r11;
420 ori r10, r10, _PAGE_ACCESSED 440 */
421#endif 441 rlwinm r11, r10, 32-5, _PAGE_PRESENT
422 mfspr r11, SPRN_MD_TWC /* get the pte address again */ 442 and r11, r11, r10
423 stw r10, 0(r11) 443 rlwimi r10, r11, 0, _PAGE_PRESENT
444
445 /* Honour kernel RO, User NA */
446 /* 0x200 == Extended encoding, bit 22 */
447 /* r11 = (r10 & _PAGE_USER) >> 2 */
448 rlwinm r11, r10, 32-2, 0x200
449 or r10, r11, r10
450 /* r11 = (r10 & _PAGE_RW) >> 1 */
451 rlwinm r11, r10, 32-1, 0x200
452 or r10, r11, r10
453 /* invert RW and 0x200 bits */
454 xori r10, r10, _PAGE_RW | 0x200
424 455
425 /* The Linux PTE won't go exactly into the MMU TLB. 456 /* The Linux PTE won't go exactly into the MMU TLB.
426 * Software indicator bits 21, 22 and 28 must be clear. 457 * Software indicator bits 22 and 28 must be clear.
427 * Software indicator bits 24, 25, 26, and 27 must be 458 * Software indicator bits 24, 25, 26, and 27 must be
428 * set. All other Linux PTE bits control the behavior 459 * set. All other Linux PTE bits control the behavior
429 * of the MMU. 460 * of the MMU.
430 */ 461 */
4312: li r11, 0x00f0 4622: li r11, 0x00f0
463 mtspr SPRN_DAR,r11 /* Tag DAR */
432 rlwimi r10, r11, 0, 24, 28 /* Set 24-27, clear 28 */ 464 rlwimi r10, r11, 0, 24, 28 /* Set 24-27, clear 28 */
433 DO_8xx_CPU6(0x3d80, r3) 465 DO_8xx_CPU6(0x3d80, r3)
434 mtspr SPRN_MD_RPN, r10 /* Update TLB entry */ 466 mtspr SPRN_MD_RPN, r10 /* Update TLB entry */
@@ -469,97 +501,10 @@ DataTLBError:
469 stw r10, 0(r0) 501 stw r10, 0(r0)
470 stw r11, 4(r0) 502 stw r11, 4(r0)
471 503
472 /* First, make sure this was a store operation.
473 */
474 mfspr r10, SPRN_DSISR
475 andis. r11, r10, 0x0200 /* If set, indicates store op */
476 beq 2f
477
478 /* The EA of a data TLB miss is automatically stored in the MD_EPN
479 * register. The EA of a data TLB error is automatically stored in
480 * the DAR, but not the MD_EPN register. We must copy the 20 most
481 * significant bits of the EA from the DAR to MD_EPN before we
482 * start walking the page tables. We also need to copy the CASID
483 * value from the M_CASID register.
484 * Addendum: The EA of a data TLB error is _supposed_ to be stored
485 * in DAR, but it seems that this doesn't happen in some cases, such
486 * as when the error is due to a dcbi instruction to a page with a
487 * TLB that doesn't have the changed bit set. In such cases, there
488 * does not appear to be any way to recover the EA of the error
489 * since it is neither in DAR nor MD_EPN. As a workaround, the
490 * _PAGE_HWWRITE bit is set for all kernel data pages when the PTEs
491 * are initialized in mapin_ram(). This will avoid the problem,
492 * assuming we only use the dcbi instruction on kernel addresses.
493 */
494 mfspr r10, SPRN_DAR 504 mfspr r10, SPRN_DAR
495 rlwinm r11, r10, 0, 0, 19 505 cmpwi cr0, r10, 0x00f0
496 ori r11, r11, MD_EVALID 506 beq- FixupDAR /* must be a buggy dcbX, icbi insn. */
497 mfspr r10, SPRN_M_CASID 507DARFixed:/* Return from dcbx instruction bug workaround, r10 holds value of DAR */
498 rlwimi r11, r10, 0, 28, 31
499 DO_8xx_CPU6(0x3780, r3)
500 mtspr SPRN_MD_EPN, r11
501
502 mfspr r10, SPRN_M_TWB /* Get level 1 table entry address */
503
504 /* If we are faulting a kernel address, we have to use the
505 * kernel page tables.
506 */
507 andi. r11, r10, 0x0800
508 beq 3f
509 lis r11, swapper_pg_dir@h
510 ori r11, r11, swapper_pg_dir@l
511 rlwimi r10, r11, 0, 2, 19
5123:
513 lwz r11, 0(r10) /* Get the level 1 entry */
514 rlwinm. r10, r11,0,0,19 /* Extract page descriptor page address */
515 beq 2f /* If zero, bail */
516
517 /* We have a pte table, so fetch the pte from the table.
518 */
519 ori r11, r11, 1 /* Set valid bit in physical L2 page */
520 DO_8xx_CPU6(0x3b80, r3)
521 mtspr SPRN_MD_TWC, r11 /* Load pte table base address */
522 mfspr r11, SPRN_MD_TWC /* ....and get the pte address */
523 lwz r10, 0(r11) /* Get the pte */
524
525 andi. r11, r10, _PAGE_RW /* Is it writeable? */
526 beq 2f /* Bail out if not */
527
528 /* Update 'changed', among others.
529 */
530#ifdef CONFIG_SWAP
531 ori r10, r10, _PAGE_DIRTY|_PAGE_HWWRITE
532 /* do not set the _PAGE_ACCESSED bit of a non-present page */
533 andi. r11, r10, _PAGE_PRESENT
534 beq 4f
535 ori r10, r10, _PAGE_ACCESSED
5364:
537#else
538 ori r10, r10, _PAGE_DIRTY|_PAGE_ACCESSED|_PAGE_HWWRITE
539#endif
540 mfspr r11, SPRN_MD_TWC /* Get pte address again */
541 stw r10, 0(r11) /* and update pte in table */
542
543 /* The Linux PTE won't go exactly into the MMU TLB.
544 * Software indicator bits 21, 22 and 28 must be clear.
545 * Software indicator bits 24, 25, 26, and 27 must be
546 * set. All other Linux PTE bits control the behavior
547 * of the MMU.
548 */
549 li r11, 0x00f0
550 rlwimi r10, r11, 0, 24, 28 /* Set 24-27, clear 28 */
551 DO_8xx_CPU6(0x3d80, r3)
552 mtspr SPRN_MD_RPN, r10 /* Update TLB entry */
553
554 mfspr r10, SPRN_M_TW /* Restore registers */
555 lwz r11, 0(r0)
556 mtcr r11
557 lwz r11, 4(r0)
558#ifdef CONFIG_8xx_CPU6
559 lwz r3, 8(r0)
560#endif
561 rfi
5622:
563 mfspr r10, SPRN_M_TW /* Restore registers */ 508 mfspr r10, SPRN_M_TW /* Restore registers */
564 lwz r11, 0(r0) 509 lwz r11, 0(r0)
565 mtcr r11 510 mtcr r11
@@ -588,6 +533,140 @@ DataTLBError:
588 533
589 . = 0x2000 534 . = 0x2000
590 535
536/* This is the procedure to calculate the data EA for buggy dcbx,dcbi instructions
537 * by decoding the registers used by the dcbx instruction and adding them.
538 * DAR is set to the calculated address and r10 also holds the EA on exit.
539 */
540 /* define if you don't want to use self modifying code */
541#define NO_SELF_MODIFYING_CODE
542FixupDAR:/* Entry point for dcbx workaround. */
543 /* fetch instruction from memory. */
544 mfspr r10, SPRN_SRR0
545 andis. r11, r10, 0x8000 /* Address >= 0x80000000 */
546 DO_8xx_CPU6(0x3780, r3)
547 mtspr SPRN_MD_EPN, r10
548 mfspr r11, SPRN_M_TWB /* Get level 1 table entry address */
549 beq- 3f /* Branch if user space */
550 lis r11, (swapper_pg_dir-PAGE_OFFSET)@h
551 ori r11, r11, (swapper_pg_dir-PAGE_OFFSET)@l
552 rlwimi r11, r10, 32-20, 0xffc /* r11 = r11&~0xffc|(r10>>20)&0xffc */
5533: lwz r11, 0(r11) /* Get the level 1 entry */
554 DO_8xx_CPU6(0x3b80, r3)
555 mtspr SPRN_MD_TWC, r11 /* Load pte table base address */
556 mfspr r11, SPRN_MD_TWC /* ....and get the pte address */
557 lwz r11, 0(r11) /* Get the pte */
558 /* concat physical page address(r11) and page offset(r10) */
559 rlwimi r11, r10, 0, 20, 31
560 lwz r11,0(r11)
561/* Check if it really is a dcbx instruction. */
562/* dcbt and dcbtst does not generate DTLB Misses/Errors,
563 * no need to include them here */
564 srwi r10, r11, 26 /* check if major OP code is 31 */
565 cmpwi cr0, r10, 31
566 bne- 141f
567 rlwinm r10, r11, 0, 21, 30
568 cmpwi cr0, r10, 2028 /* Is dcbz? */
569 beq+ 142f
570 cmpwi cr0, r10, 940 /* Is dcbi? */
571 beq+ 142f
572 cmpwi cr0, r10, 108 /* Is dcbst? */
573 beq+ 144f /* Fix up store bit! */
574 cmpwi cr0, r10, 172 /* Is dcbf? */
575 beq+ 142f
576 cmpwi cr0, r10, 1964 /* Is icbi? */
577 beq+ 142f
578141: mfspr r10, SPRN_DAR /* r10 must hold DAR at exit */
579 b DARFixed /* Nope, go back to normal TLB processing */
580
581144: mfspr r10, SPRN_DSISR
582 rlwinm r10, r10,0,7,5 /* Clear store bit for buggy dcbst insn */
583 mtspr SPRN_DSISR, r10
584142: /* continue, it was a dcbx, dcbi instruction. */
585#ifdef CONFIG_8xx_CPU6
586 lwz r3, 8(r0) /* restore r3 from memory */
587#endif
588#ifndef NO_SELF_MODIFYING_CODE
589 andis. r10,r11,0x1f /* test if reg RA is r0 */
590 li r10,modified_instr@l
591 dcbtst r0,r10 /* touch for store */
592 rlwinm r11,r11,0,0,20 /* Zero lower 10 bits */
593 oris r11,r11,640 /* Transform instr. to a "add r10,RA,RB" */
594 ori r11,r11,532
595 stw r11,0(r10) /* store add/and instruction */
596 dcbf 0,r10 /* flush new instr. to memory. */
597 icbi 0,r10 /* invalidate instr. cache line */
598 lwz r11, 4(r0) /* restore r11 from memory */
599 mfspr r10, SPRN_M_TW /* restore r10 from M_TW */
600 isync /* Wait until new instr is loaded from memory */
601modified_instr:
602 .space 4 /* this is where the add instr. is stored */
603 bne+ 143f
604 subf r10,r0,r10 /* r10=r10-r0, only if reg RA is r0 */
605143: mtdar r10 /* store faulting EA in DAR */
606 b DARFixed /* Go back to normal TLB handling */
607#else
608 mfctr r10
609 mtdar r10 /* save ctr reg in DAR */
610 rlwinm r10, r11, 24, 24, 28 /* offset into jump table for reg RB */
611 addi r10, r10, 150f@l /* add start of table */
612 mtctr r10 /* load ctr with jump address */
613 xor r10, r10, r10 /* sum starts at zero */
614 bctr /* jump into table */
615150:
616 add r10, r10, r0 ;b 151f
617 add r10, r10, r1 ;b 151f
618 add r10, r10, r2 ;b 151f
619 add r10, r10, r3 ;b 151f
620 add r10, r10, r4 ;b 151f
621 add r10, r10, r5 ;b 151f
622 add r10, r10, r6 ;b 151f
623 add r10, r10, r7 ;b 151f
624 add r10, r10, r8 ;b 151f
625 add r10, r10, r9 ;b 151f
626 mtctr r11 ;b 154f /* r10 needs special handling */
627 mtctr r11 ;b 153f /* r11 needs special handling */
628 add r10, r10, r12 ;b 151f
629 add r10, r10, r13 ;b 151f
630 add r10, r10, r14 ;b 151f
631 add r10, r10, r15 ;b 151f
632 add r10, r10, r16 ;b 151f
633 add r10, r10, r17 ;b 151f
634 add r10, r10, r18 ;b 151f
635 add r10, r10, r19 ;b 151f
636 add r10, r10, r20 ;b 151f
637 add r10, r10, r21 ;b 151f
638 add r10, r10, r22 ;b 151f
639 add r10, r10, r23 ;b 151f
640 add r10, r10, r24 ;b 151f
641 add r10, r10, r25 ;b 151f
642 add r10, r10, r26 ;b 151f
643 add r10, r10, r27 ;b 151f
644 add r10, r10, r28 ;b 151f
645 add r10, r10, r29 ;b 151f
646 add r10, r10, r30 ;b 151f
647 add r10, r10, r31
648151:
649 rlwinm. r11,r11,19,24,28 /* offset into jump table for reg RA */
650 beq 152f /* if reg RA is zero, don't add it */
651 addi r11, r11, 150b@l /* add start of table */
652 mtctr r11 /* load ctr with jump address */
653 rlwinm r11,r11,0,16,10 /* make sure we don't execute this more than once */
654 bctr /* jump into table */
655152:
656 mfdar r11
657 mtctr r11 /* restore ctr reg from DAR */
658 mtdar r10 /* save fault EA to DAR */
659 b DARFixed /* Go back to normal TLB handling */
660
661 /* special handling for r10,r11 since these are modified already */
662153: lwz r11, 4(r0) /* load r11 from memory */
663 b 155f
664154: mfspr r11, SPRN_M_TW /* load r10 from M_TW */
665155: add r10, r10, r11 /* add it */
666 mfctr r11 /* restore r11 */
667 b 151b
668#endif
669
591 .globl giveup_fpu 670 .globl giveup_fpu
592giveup_fpu: 671giveup_fpu:
593 blr 672 blr
@@ -689,12 +768,12 @@ start_here:
689 */ 768 */
690initial_mmu: 769initial_mmu:
691 tlbia /* Invalidate all TLB entries */ 770 tlbia /* Invalidate all TLB entries */
692#ifdef CONFIG_PIN_TLB 771/* Always pin the first 8 MB ITLB to prevent ITLB
772 misses while mucking around with SRR0/SRR1 in asm
773*/
693 lis r8, MI_RSV4I@h 774 lis r8, MI_RSV4I@h
694 ori r8, r8, 0x1c00 775 ori r8, r8, 0x1c00
695#else 776
696 li r8, 0
697#endif
698 mtspr SPRN_MI_CTR, r8 /* Set instruction MMU control */ 777 mtspr SPRN_MI_CTR, r8 /* Set instruction MMU control */
699 778
700#ifdef CONFIG_PIN_TLB 779#ifdef CONFIG_PIN_TLB
diff --git a/arch/powerpc/kernel/head_fsl_booke.S b/arch/powerpc/kernel/head_fsl_booke.S
index 975788ca05d2..725526547994 100644
--- a/arch/powerpc/kernel/head_fsl_booke.S
+++ b/arch/powerpc/kernel/head_fsl_booke.S
@@ -214,11 +214,11 @@ skpinv: addi r6,r6,1 /* Increment */
214 bl 1f /* Find our address */ 214 bl 1f /* Find our address */
2151: mflr r9 2151: mflr r9
216 rlwimi r7,r9,0,20,31 216 rlwimi r7,r9,0,20,31
217 addi r7,r7,24 217 addi r7,r7,(2f - 1b)
218 mtspr SPRN_SRR0,r7 218 mtspr SPRN_SRR0,r7
219 mtspr SPRN_SRR1,r6 219 mtspr SPRN_SRR1,r6
220 rfi 220 rfi
221 2212:
222/* 4. Clear out PIDs & Search info */ 222/* 4. Clear out PIDs & Search info */
223 li r6,0 223 li r6,0
224 mtspr SPRN_MAS6,r6 224 mtspr SPRN_MAS6,r6
@@ -747,9 +747,6 @@ finish_tlb_load:
747#else 747#else
748 rlwimi r12, r11, 26, 27, 31 /* extract WIMGE from pte */ 748 rlwimi r12, r11, 26, 27, 31 /* extract WIMGE from pte */
749#endif 749#endif
750#ifdef CONFIG_SMP
751 ori r12, r12, MAS2_M
752#endif
753 mtspr SPRN_MAS2, r12 750 mtspr SPRN_MAS2, r12
754 751
755#ifdef CONFIG_PTE_64BIT 752#ifdef CONFIG_PTE_64BIT
@@ -887,13 +884,17 @@ KernelSPE:
887 lwz r3,_MSR(r1) 884 lwz r3,_MSR(r1)
888 oris r3,r3,MSR_SPE@h 885 oris r3,r3,MSR_SPE@h
889 stw r3,_MSR(r1) /* enable use of SPE after return */ 886 stw r3,_MSR(r1) /* enable use of SPE after return */
887#ifdef CONFIG_PRINTK
890 lis r3,87f@h 888 lis r3,87f@h
891 ori r3,r3,87f@l 889 ori r3,r3,87f@l
892 mr r4,r2 /* current */ 890 mr r4,r2 /* current */
893 lwz r5,_NIP(r1) 891 lwz r5,_NIP(r1)
894 bl printk 892 bl printk
893#endif
895 b ret_from_except 894 b ret_from_except
895#ifdef CONFIG_PRINTK
89687: .string "SPE used in kernel (task=%p, pc=%x) \n" 89687: .string "SPE used in kernel (task=%p, pc=%x) \n"
897#endif
897 .align 4,0 898 .align 4,0
898 899
899#endif /* CONFIG_SPE */ 900#endif /* CONFIG_SPE */
@@ -944,28 +945,6 @@ _GLOBAL(__setup_e500mc_ivors)
944 blr 945 blr
945 946
946/* 947/*
947 * extern void loadcam_entry(unsigned int index)
948 *
949 * Load TLBCAM[index] entry in to the L2 CAM MMU
950 */
951_GLOBAL(loadcam_entry)
952 lis r4,TLBCAM@ha
953 addi r4,r4,TLBCAM@l
954 mulli r5,r3,TLBCAM_SIZE
955 add r3,r5,r4
956 lwz r4,0(r3)
957 mtspr SPRN_MAS0,r4
958 lwz r4,4(r3)
959 mtspr SPRN_MAS1,r4
960 lwz r4,8(r3)
961 mtspr SPRN_MAS2,r4
962 lwz r4,12(r3)
963 mtspr SPRN_MAS3,r4
964 tlbwe
965 isync
966 blr
967
968/*
969 * extern void giveup_altivec(struct task_struct *prev) 948 * extern void giveup_altivec(struct task_struct *prev)
970 * 949 *
971 * The e500 core does not have an AltiVec unit. 950 * The e500 core does not have an AltiVec unit.
diff --git a/arch/powerpc/kernel/ibmebus.c b/arch/powerpc/kernel/ibmebus.c
index a4c8b38b0ba1..71cf280da184 100644
--- a/arch/powerpc/kernel/ibmebus.c
+++ b/arch/powerpc/kernel/ibmebus.c
@@ -42,6 +42,7 @@
42#include <linux/dma-mapping.h> 42#include <linux/dma-mapping.h>
43#include <linux/interrupt.h> 43#include <linux/interrupt.h>
44#include <linux/of.h> 44#include <linux/of.h>
45#include <linux/slab.h>
45#include <linux/of_platform.h> 46#include <linux/of_platform.h>
46#include <asm/ibmebus.h> 47#include <asm/ibmebus.h>
47#include <asm/abs_addr.h> 48#include <asm/abs_addr.h>
diff --git a/arch/powerpc/kernel/idle.c b/arch/powerpc/kernel/idle.c
index 88d9c1d5e5fb..049dda60e475 100644
--- a/arch/powerpc/kernel/idle.c
+++ b/arch/powerpc/kernel/idle.c
@@ -110,18 +110,16 @@ int powersave_nap;
110 */ 110 */
111static ctl_table powersave_nap_ctl_table[]={ 111static ctl_table powersave_nap_ctl_table[]={
112 { 112 {
113 .ctl_name = KERN_PPC_POWERSAVE_NAP,
114 .procname = "powersave-nap", 113 .procname = "powersave-nap",
115 .data = &powersave_nap, 114 .data = &powersave_nap,
116 .maxlen = sizeof(int), 115 .maxlen = sizeof(int),
117 .mode = 0644, 116 .mode = 0644,
118 .proc_handler = &proc_dointvec, 117 .proc_handler = proc_dointvec,
119 }, 118 },
120 {} 119 {}
121}; 120};
122static ctl_table powersave_nap_sysctl_root[] = { 121static ctl_table powersave_nap_sysctl_root[] = {
123 { 122 {
124 .ctl_name = CTL_KERN,
125 .procname = "kernel", 123 .procname = "kernel",
126 .mode = 0555, 124 .mode = 0555,
127 .child = powersave_nap_ctl_table, 125 .child = powersave_nap_ctl_table,
diff --git a/arch/powerpc/kernel/io.c b/arch/powerpc/kernel/io.c
index 1882bf419fa6..8dc7547c2377 100644
--- a/arch/powerpc/kernel/io.c
+++ b/arch/powerpc/kernel/io.c
@@ -161,7 +161,7 @@ void _memcpy_fromio(void *dest, const volatile void __iomem *src,
161 dest++; 161 dest++;
162 n--; 162 n--;
163 } 163 }
164 while(n > 4) { 164 while(n >= 4) {
165 *((u32 *)dest) = *((volatile u32 *)vsrc); 165 *((u32 *)dest) = *((volatile u32 *)vsrc);
166 eieio(); 166 eieio();
167 vsrc += 4; 167 vsrc += 4;
@@ -190,7 +190,7 @@ void _memcpy_toio(volatile void __iomem *dest, const void *src, unsigned long n)
190 vdest++; 190 vdest++;
191 n--; 191 n--;
192 } 192 }
193 while(n > 4) { 193 while(n >= 4) {
194 *((volatile u32 *)vdest) = *((volatile u32 *)src); 194 *((volatile u32 *)vdest) = *((volatile u32 *)src);
195 src += 4; 195 src += 4;
196 vdest += 4; 196 vdest += 4;
diff --git a/arch/powerpc/kernel/iommu.c b/arch/powerpc/kernel/iommu.c
index fd51578e29dd..ec94f906ea43 100644
--- a/arch/powerpc/kernel/iommu.c
+++ b/arch/powerpc/kernel/iommu.c
@@ -30,7 +30,7 @@
30#include <linux/spinlock.h> 30#include <linux/spinlock.h>
31#include <linux/string.h> 31#include <linux/string.h>
32#include <linux/dma-mapping.h> 32#include <linux/dma-mapping.h>
33#include <linux/bitops.h> 33#include <linux/bitmap.h>
34#include <linux/iommu-helper.h> 34#include <linux/iommu-helper.h>
35#include <linux/crash_dump.h> 35#include <linux/crash_dump.h>
36#include <asm/io.h> 36#include <asm/io.h>
@@ -42,12 +42,7 @@
42 42
43#define DBG(...) 43#define DBG(...)
44 44
45#ifdef CONFIG_IOMMU_VMERGE 45static int novmerge;
46static int novmerge = 0;
47#else
48static int novmerge = 1;
49#endif
50
51static int protect4gb = 1; 46static int protect4gb = 1;
52 47
53static void __iommu_free(struct iommu_table *, dma_addr_t, unsigned int); 48static void __iommu_free(struct iommu_table *, dma_addr_t, unsigned int);
@@ -251,7 +246,7 @@ static void __iommu_free(struct iommu_table *tbl, dma_addr_t dma_addr,
251 } 246 }
252 247
253 ppc_md.tce_free(tbl, entry, npages); 248 ppc_md.tce_free(tbl, entry, npages);
254 iommu_area_free(tbl->it_map, free_entry, npages); 249 bitmap_clear(tbl->it_map, free_entry, npages);
255} 250}
256 251
257static void iommu_free(struct iommu_table *tbl, dma_addr_t dma_addr, 252static void iommu_free(struct iommu_table *tbl, dma_addr_t dma_addr,
diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
index e5d121177984..066bd31551d5 100644
--- a/arch/powerpc/kernel/irq.c
+++ b/arch/powerpc/kernel/irq.c
@@ -53,7 +53,6 @@
53#include <linux/bootmem.h> 53#include <linux/bootmem.h>
54#include <linux/pci.h> 54#include <linux/pci.h>
55#include <linux/debugfs.h> 55#include <linux/debugfs.h>
56#include <linux/perf_event.h>
57 56
58#include <asm/uaccess.h> 57#include <asm/uaccess.h>
59#include <asm/system.h> 58#include <asm/system.h>
@@ -70,9 +69,13 @@
70#include <asm/firmware.h> 69#include <asm/firmware.h>
71#include <asm/lv1call.h> 70#include <asm/lv1call.h>
72#endif 71#endif
72#define CREATE_TRACE_POINTS
73#include <asm/trace.h>
74
75DEFINE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat);
76EXPORT_PER_CPU_SYMBOL(irq_stat);
73 77
74int __irq_offset_value; 78int __irq_offset_value;
75static int ppc_spurious_interrupts;
76 79
77#ifdef CONFIG_PPC32 80#ifdef CONFIG_PPC32
78EXPORT_SYMBOL(__irq_offset_value); 81EXPORT_SYMBOL(__irq_offset_value);
@@ -85,7 +88,10 @@ extern int tau_interrupts(int);
85#endif /* CONFIG_PPC32 */ 88#endif /* CONFIG_PPC32 */
86 89
87#ifdef CONFIG_PPC64 90#ifdef CONFIG_PPC64
91
92#ifndef CONFIG_SPARSE_IRQ
88EXPORT_SYMBOL(irq_desc); 93EXPORT_SYMBOL(irq_desc);
94#endif
89 95
90int distribute_irqs = 1; 96int distribute_irqs = 1;
91 97
@@ -138,11 +144,6 @@ notrace void raw_local_irq_restore(unsigned long en)
138 } 144 }
139#endif /* CONFIG_PPC_STD_MMU_64 */ 145#endif /* CONFIG_PPC_STD_MMU_64 */
140 146
141 if (test_perf_event_pending()) {
142 clear_perf_event_pending();
143 perf_event_do_pending();
144 }
145
146 /* 147 /*
147 * if (get_paca()->hard_enabled) return; 148 * if (get_paca()->hard_enabled) return;
148 * But again we need to take care that gcc gets hard_enabled directly 149 * But again we need to take care that gcc gets hard_enabled directly
@@ -175,78 +176,135 @@ notrace void raw_local_irq_restore(unsigned long en)
175EXPORT_SYMBOL(raw_local_irq_restore); 176EXPORT_SYMBOL(raw_local_irq_restore);
176#endif /* CONFIG_PPC64 */ 177#endif /* CONFIG_PPC64 */
177 178
179static int show_other_interrupts(struct seq_file *p, int prec)
180{
181 int j;
182
183#if defined(CONFIG_PPC32) && defined(CONFIG_TAU_INT)
184 if (tau_initialized) {
185 seq_printf(p, "%*s: ", prec, "TAU");
186 for_each_online_cpu(j)
187 seq_printf(p, "%10u ", tau_interrupts(j));
188 seq_puts(p, " PowerPC Thermal Assist (cpu temp)\n");
189 }
190#endif /* CONFIG_PPC32 && CONFIG_TAU_INT */
191
192 seq_printf(p, "%*s: ", prec, "LOC");
193 for_each_online_cpu(j)
194 seq_printf(p, "%10u ", per_cpu(irq_stat, j).timer_irqs);
195 seq_printf(p, " Local timer interrupts\n");
196
197 seq_printf(p, "%*s: ", prec, "SPU");
198 for_each_online_cpu(j)
199 seq_printf(p, "%10u ", per_cpu(irq_stat, j).spurious_irqs);
200 seq_printf(p, " Spurious interrupts\n");
201
202 seq_printf(p, "%*s: ", prec, "CNT");
203 for_each_online_cpu(j)
204 seq_printf(p, "%10u ", per_cpu(irq_stat, j).pmu_irqs);
205 seq_printf(p, " Performance monitoring interrupts\n");
206
207 seq_printf(p, "%*s: ", prec, "MCE");
208 for_each_online_cpu(j)
209 seq_printf(p, "%10u ", per_cpu(irq_stat, j).mce_exceptions);
210 seq_printf(p, " Machine check exceptions\n");
211
212 return 0;
213}
214
178int show_interrupts(struct seq_file *p, void *v) 215int show_interrupts(struct seq_file *p, void *v)
179{ 216{
180 int i = *(loff_t *)v, j; 217 unsigned long flags, any_count = 0;
218 int i = *(loff_t *) v, j, prec;
181 struct irqaction *action; 219 struct irqaction *action;
182 struct irq_desc *desc; 220 struct irq_desc *desc;
183 unsigned long flags;
184 221
222 if (i > nr_irqs)
223 return 0;
224
225 for (prec = 3, j = 1000; prec < 10 && j <= nr_irqs; ++prec)
226 j *= 10;
227
228 if (i == nr_irqs)
229 return show_other_interrupts(p, prec);
230
231 /* print header */
185 if (i == 0) { 232 if (i == 0) {
186 seq_puts(p, " "); 233 seq_printf(p, "%*s", prec + 8, "");
187 for_each_online_cpu(j) 234 for_each_online_cpu(j)
188 seq_printf(p, "CPU%d ", j); 235 seq_printf(p, "CPU%-8d", j);
189 seq_putc(p, '\n'); 236 seq_putc(p, '\n');
190 } 237 }
191 238
192 if (i < NR_IRQS) { 239 desc = irq_to_desc(i);
193 desc = get_irq_desc(i); 240 if (!desc)
194 spin_lock_irqsave(&desc->lock, flags); 241 return 0;
195 action = desc->action; 242
196 if (!action || !action->handler) 243 raw_spin_lock_irqsave(&desc->lock, flags);
197 goto skip; 244 for_each_online_cpu(j)
198 seq_printf(p, "%3d: ", i); 245 any_count |= kstat_irqs_cpu(i, j);
199#ifdef CONFIG_SMP 246 action = desc->action;
200 for_each_online_cpu(j) 247 if (!action && !any_count)
201 seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); 248 goto out;
202#else 249
203 seq_printf(p, "%10u ", kstat_irqs(i)); 250 seq_printf(p, "%*d: ", prec, i);
204#endif /* CONFIG_SMP */ 251 for_each_online_cpu(j)
205 if (desc->chip) 252 seq_printf(p, "%10u ", kstat_irqs_cpu(i, j));
206 seq_printf(p, " %s ", desc->chip->typename); 253
207 else 254 if (desc->chip)
208 seq_puts(p, " None "); 255 seq_printf(p, " %-16s", desc->chip->name);
209 seq_printf(p, "%s", (desc->status & IRQ_LEVEL) ? "Level " : "Edge "); 256 else
210 seq_printf(p, " %s", action->name); 257 seq_printf(p, " %-16s", "None");
211 for (action = action->next; action; action = action->next) 258 seq_printf(p, " %-8s", (desc->status & IRQ_LEVEL) ? "Level" : "Edge");
259
260 if (action) {
261 seq_printf(p, " %s", action->name);
262 while ((action = action->next) != NULL)
212 seq_printf(p, ", %s", action->name); 263 seq_printf(p, ", %s", action->name);
213 seq_putc(p, '\n');
214skip:
215 spin_unlock_irqrestore(&desc->lock, flags);
216 } else if (i == NR_IRQS) {
217#if defined(CONFIG_PPC32) && defined(CONFIG_TAU_INT)
218 if (tau_initialized){
219 seq_puts(p, "TAU: ");
220 for_each_online_cpu(j)
221 seq_printf(p, "%10u ", tau_interrupts(j));
222 seq_puts(p, " PowerPC Thermal Assist (cpu temp)\n");
223 }
224#endif /* CONFIG_PPC32 && CONFIG_TAU_INT*/
225 seq_printf(p, "BAD: %10u\n", ppc_spurious_interrupts);
226 } 264 }
265
266 seq_putc(p, '\n');
267out:
268 raw_spin_unlock_irqrestore(&desc->lock, flags);
227 return 0; 269 return 0;
228} 270}
229 271
272/*
273 * /proc/stat helpers
274 */
275u64 arch_irq_stat_cpu(unsigned int cpu)
276{
277 u64 sum = per_cpu(irq_stat, cpu).timer_irqs;
278
279 sum += per_cpu(irq_stat, cpu).pmu_irqs;
280 sum += per_cpu(irq_stat, cpu).mce_exceptions;
281 sum += per_cpu(irq_stat, cpu).spurious_irqs;
282
283 return sum;
284}
285
230#ifdef CONFIG_HOTPLUG_CPU 286#ifdef CONFIG_HOTPLUG_CPU
231void fixup_irqs(cpumask_t map) 287void fixup_irqs(cpumask_t map)
232{ 288{
289 struct irq_desc *desc;
233 unsigned int irq; 290 unsigned int irq;
234 static int warned; 291 static int warned;
235 292
236 for_each_irq(irq) { 293 for_each_irq(irq) {
237 cpumask_t mask; 294 cpumask_t mask;
238 295
239 if (irq_desc[irq].status & IRQ_PER_CPU) 296 desc = irq_to_desc(irq);
297 if (desc && desc->status & IRQ_PER_CPU)
240 continue; 298 continue;
241 299
242 cpumask_and(&mask, irq_desc[irq].affinity, &map); 300 cpumask_and(&mask, desc->affinity, &map);
243 if (any_online_cpu(mask) == NR_CPUS) { 301 if (any_online_cpu(mask) == NR_CPUS) {
244 printk("Breaking affinity for irq %i\n", irq); 302 printk("Breaking affinity for irq %i\n", irq);
245 mask = map; 303 mask = map;
246 } 304 }
247 if (irq_desc[irq].chip->set_affinity) 305 if (desc->chip->set_affinity)
248 irq_desc[irq].chip->set_affinity(irq, &mask); 306 desc->chip->set_affinity(irq, &mask);
249 else if (irq_desc[irq].action && !(warned++)) 307 else if (desc->action && !(warned++))
250 printk("Cannot set affinity for irq %i\n", irq); 308 printk("Cannot set affinity for irq %i\n", irq);
251 } 309 }
252 310
@@ -273,7 +331,7 @@ static inline void handle_one_irq(unsigned int irq)
273 return; 331 return;
274 } 332 }
275 333
276 desc = irq_desc + irq; 334 desc = irq_to_desc(irq);
277 saved_sp_limit = current->thread.ksp_limit; 335 saved_sp_limit = current->thread.ksp_limit;
278 336
279 irqtp->task = curtp->task; 337 irqtp->task = curtp->task;
@@ -325,6 +383,8 @@ void do_IRQ(struct pt_regs *regs)
325 struct pt_regs *old_regs = set_irq_regs(regs); 383 struct pt_regs *old_regs = set_irq_regs(regs);
326 unsigned int irq; 384 unsigned int irq;
327 385
386 trace_irq_entry(regs);
387
328 irq_enter(); 388 irq_enter();
329 389
330 check_stack_overflow(); 390 check_stack_overflow();
@@ -334,8 +394,7 @@ void do_IRQ(struct pt_regs *regs)
334 if (irq != NO_IRQ && irq != NO_IRQ_IGNORE) 394 if (irq != NO_IRQ && irq != NO_IRQ_IGNORE)
335 handle_one_irq(irq); 395 handle_one_irq(irq);
336 else if (irq != NO_IRQ_IGNORE) 396 else if (irq != NO_IRQ_IGNORE)
337 /* That's not SMP safe ... but who cares ? */ 397 __get_cpu_var(irq_stat).spurious_irqs++;
338 ppc_spurious_interrupts++;
339 398
340 irq_exit(); 399 irq_exit();
341 set_irq_regs(old_regs); 400 set_irq_regs(old_regs);
@@ -348,6 +407,8 @@ void do_IRQ(struct pt_regs *regs)
348 timer_interrupt(regs); 407 timer_interrupt(regs);
349 } 408 }
350#endif 409#endif
410
411 trace_irq_exit(regs);
351} 412}
352 413
353void __init init_IRQ(void) 414void __init init_IRQ(void)
@@ -453,7 +514,7 @@ void do_softirq(void)
453 */ 514 */
454 515
455static LIST_HEAD(irq_hosts); 516static LIST_HEAD(irq_hosts);
456static DEFINE_SPINLOCK(irq_big_lock); 517static DEFINE_RAW_SPINLOCK(irq_big_lock);
457static unsigned int revmap_trees_allocated; 518static unsigned int revmap_trees_allocated;
458static DEFINE_MUTEX(revmap_trees_mutex); 519static DEFINE_MUTEX(revmap_trees_mutex);
459struct irq_map_entry irq_map[NR_IRQS]; 520struct irq_map_entry irq_map[NR_IRQS];
@@ -499,14 +560,14 @@ struct irq_host *irq_alloc_host(struct device_node *of_node,
499 if (host->ops->match == NULL) 560 if (host->ops->match == NULL)
500 host->ops->match = default_irq_host_match; 561 host->ops->match = default_irq_host_match;
501 562
502 spin_lock_irqsave(&irq_big_lock, flags); 563 raw_spin_lock_irqsave(&irq_big_lock, flags);
503 564
504 /* If it's a legacy controller, check for duplicates and 565 /* If it's a legacy controller, check for duplicates and
505 * mark it as allocated (we use irq 0 host pointer for that 566 * mark it as allocated (we use irq 0 host pointer for that
506 */ 567 */
507 if (revmap_type == IRQ_HOST_MAP_LEGACY) { 568 if (revmap_type == IRQ_HOST_MAP_LEGACY) {
508 if (irq_map[0].host != NULL) { 569 if (irq_map[0].host != NULL) {
509 spin_unlock_irqrestore(&irq_big_lock, flags); 570 raw_spin_unlock_irqrestore(&irq_big_lock, flags);
510 /* If we are early boot, we can't free the structure, 571 /* If we are early boot, we can't free the structure,
511 * too bad... 572 * too bad...
512 * this will be fixed once slab is made available early 573 * this will be fixed once slab is made available early
@@ -520,7 +581,7 @@ struct irq_host *irq_alloc_host(struct device_node *of_node,
520 } 581 }
521 582
522 list_add(&host->link, &irq_hosts); 583 list_add(&host->link, &irq_hosts);
523 spin_unlock_irqrestore(&irq_big_lock, flags); 584 raw_spin_unlock_irqrestore(&irq_big_lock, flags);
524 585
525 /* Additional setups per revmap type */ 586 /* Additional setups per revmap type */
526 switch(revmap_type) { 587 switch(revmap_type) {
@@ -535,7 +596,7 @@ struct irq_host *irq_alloc_host(struct device_node *of_node,
535 smp_wmb(); 596 smp_wmb();
536 597
537 /* Clear norequest flags */ 598 /* Clear norequest flags */
538 get_irq_desc(i)->status &= ~IRQ_NOREQUEST; 599 irq_to_desc(i)->status &= ~IRQ_NOREQUEST;
539 600
540 /* Legacy flags are left to default at this point, 601 /* Legacy flags are left to default at this point,
541 * one can then use irq_create_mapping() to 602 * one can then use irq_create_mapping() to
@@ -571,13 +632,13 @@ struct irq_host *irq_find_host(struct device_node *node)
571 * the absence of a device node. This isn't a problem so far 632 * the absence of a device node. This isn't a problem so far
572 * yet though... 633 * yet though...
573 */ 634 */
574 spin_lock_irqsave(&irq_big_lock, flags); 635 raw_spin_lock_irqsave(&irq_big_lock, flags);
575 list_for_each_entry(h, &irq_hosts, link) 636 list_for_each_entry(h, &irq_hosts, link)
576 if (h->ops->match(h, node)) { 637 if (h->ops->match(h, node)) {
577 found = h; 638 found = h;
578 break; 639 break;
579 } 640 }
580 spin_unlock_irqrestore(&irq_big_lock, flags); 641 raw_spin_unlock_irqrestore(&irq_big_lock, flags);
581 return found; 642 return found;
582} 643}
583EXPORT_SYMBOL_GPL(irq_find_host); 644EXPORT_SYMBOL_GPL(irq_find_host);
@@ -601,8 +662,16 @@ void irq_set_virq_count(unsigned int count)
601static int irq_setup_virq(struct irq_host *host, unsigned int virq, 662static int irq_setup_virq(struct irq_host *host, unsigned int virq,
602 irq_hw_number_t hwirq) 663 irq_hw_number_t hwirq)
603{ 664{
665 struct irq_desc *desc;
666
667 desc = irq_to_desc_alloc_node(virq, 0);
668 if (!desc) {
669 pr_debug("irq: -> allocating desc failed\n");
670 goto error;
671 }
672
604 /* Clear IRQ_NOREQUEST flag */ 673 /* Clear IRQ_NOREQUEST flag */
605 get_irq_desc(virq)->status &= ~IRQ_NOREQUEST; 674 desc->status &= ~IRQ_NOREQUEST;
606 675
607 /* map it */ 676 /* map it */
608 smp_wmb(); 677 smp_wmb();
@@ -611,11 +680,14 @@ static int irq_setup_virq(struct irq_host *host, unsigned int virq,
611 680
612 if (host->ops->map(host, virq, hwirq)) { 681 if (host->ops->map(host, virq, hwirq)) {
613 pr_debug("irq: -> mapping failed, freeing\n"); 682 pr_debug("irq: -> mapping failed, freeing\n");
614 irq_free_virt(virq, 1); 683 goto error;
615 return -1;
616 } 684 }
617 685
618 return 0; 686 return 0;
687
688error:
689 irq_free_virt(virq, 1);
690 return -1;
619} 691}
620 692
621unsigned int irq_create_direct_mapping(struct irq_host *host) 693unsigned int irq_create_direct_mapping(struct irq_host *host)
@@ -699,7 +771,7 @@ unsigned int irq_create_mapping(struct irq_host *host,
699EXPORT_SYMBOL_GPL(irq_create_mapping); 771EXPORT_SYMBOL_GPL(irq_create_mapping);
700 772
701unsigned int irq_create_of_mapping(struct device_node *controller, 773unsigned int irq_create_of_mapping(struct device_node *controller,
702 u32 *intspec, unsigned int intsize) 774 const u32 *intspec, unsigned int intsize)
703{ 775{
704 struct irq_host *host; 776 struct irq_host *host;
705 irq_hw_number_t hwirq; 777 irq_hw_number_t hwirq;
@@ -732,7 +804,7 @@ unsigned int irq_create_of_mapping(struct device_node *controller,
732 804
733 /* Set type if specified and different than the current one */ 805 /* Set type if specified and different than the current one */
734 if (type != IRQ_TYPE_NONE && 806 if (type != IRQ_TYPE_NONE &&
735 type != (get_irq_desc(virq)->status & IRQF_TRIGGER_MASK)) 807 type != (irq_to_desc(virq)->status & IRQF_TRIGGER_MASK))
736 set_irq_type(virq, type); 808 set_irq_type(virq, type);
737 return virq; 809 return virq;
738} 810}
@@ -804,7 +876,7 @@ void irq_dispose_mapping(unsigned int virq)
804 irq_map[virq].hwirq = host->inval_irq; 876 irq_map[virq].hwirq = host->inval_irq;
805 877
806 /* Set some flags */ 878 /* Set some flags */
807 get_irq_desc(virq)->status |= IRQ_NOREQUEST; 879 irq_to_desc(virq)->status |= IRQ_NOREQUEST;
808 880
809 /* Free it */ 881 /* Free it */
810 irq_free_virt(virq, 1); 882 irq_free_virt(virq, 1);
@@ -935,7 +1007,7 @@ unsigned int irq_alloc_virt(struct irq_host *host,
935 if (count == 0 || count > (irq_virq_count - NUM_ISA_INTERRUPTS)) 1007 if (count == 0 || count > (irq_virq_count - NUM_ISA_INTERRUPTS))
936 return NO_IRQ; 1008 return NO_IRQ;
937 1009
938 spin_lock_irqsave(&irq_big_lock, flags); 1010 raw_spin_lock_irqsave(&irq_big_lock, flags);
939 1011
940 /* Use hint for 1 interrupt if any */ 1012 /* Use hint for 1 interrupt if any */
941 if (count == 1 && hint >= NUM_ISA_INTERRUPTS && 1013 if (count == 1 && hint >= NUM_ISA_INTERRUPTS &&
@@ -959,7 +1031,7 @@ unsigned int irq_alloc_virt(struct irq_host *host,
959 } 1031 }
960 } 1032 }
961 if (found == NO_IRQ) { 1033 if (found == NO_IRQ) {
962 spin_unlock_irqrestore(&irq_big_lock, flags); 1034 raw_spin_unlock_irqrestore(&irq_big_lock, flags);
963 return NO_IRQ; 1035 return NO_IRQ;
964 } 1036 }
965 hint_found: 1037 hint_found:
@@ -968,7 +1040,7 @@ unsigned int irq_alloc_virt(struct irq_host *host,
968 smp_wmb(); 1040 smp_wmb();
969 irq_map[i].host = host; 1041 irq_map[i].host = host;
970 } 1042 }
971 spin_unlock_irqrestore(&irq_big_lock, flags); 1043 raw_spin_unlock_irqrestore(&irq_big_lock, flags);
972 return found; 1044 return found;
973} 1045}
974 1046
@@ -980,7 +1052,7 @@ void irq_free_virt(unsigned int virq, unsigned int count)
980 WARN_ON (virq < NUM_ISA_INTERRUPTS); 1052 WARN_ON (virq < NUM_ISA_INTERRUPTS);
981 WARN_ON (count == 0 || (virq + count) > irq_virq_count); 1053 WARN_ON (count == 0 || (virq + count) > irq_virq_count);
982 1054
983 spin_lock_irqsave(&irq_big_lock, flags); 1055 raw_spin_lock_irqsave(&irq_big_lock, flags);
984 for (i = virq; i < (virq + count); i++) { 1056 for (i = virq; i < (virq + count); i++) {
985 struct irq_host *host; 1057 struct irq_host *host;
986 1058
@@ -993,15 +1065,27 @@ void irq_free_virt(unsigned int virq, unsigned int count)
993 smp_wmb(); 1065 smp_wmb();
994 irq_map[i].host = NULL; 1066 irq_map[i].host = NULL;
995 } 1067 }
996 spin_unlock_irqrestore(&irq_big_lock, flags); 1068 raw_spin_unlock_irqrestore(&irq_big_lock, flags);
997} 1069}
998 1070
999void irq_early_init(void) 1071int arch_early_irq_init(void)
1000{ 1072{
1001 unsigned int i; 1073 struct irq_desc *desc;
1074 int i;
1075
1076 for (i = 0; i < NR_IRQS; i++) {
1077 desc = irq_to_desc(i);
1078 if (desc)
1079 desc->status |= IRQ_NOREQUEST;
1080 }
1002 1081
1003 for (i = 0; i < NR_IRQS; i++) 1082 return 0;
1004 get_irq_desc(i)->status |= IRQ_NOREQUEST; 1083}
1084
1085int arch_init_chip_data(struct irq_desc *desc, int node)
1086{
1087 desc->status |= IRQ_NOREQUEST;
1088 return 0;
1005} 1089}
1006 1090
1007/* We need to create the radix trees late */ 1091/* We need to create the radix trees late */
@@ -1063,16 +1147,19 @@ static int virq_debug_show(struct seq_file *m, void *private)
1063 seq_printf(m, "%-5s %-7s %-15s %s\n", "virq", "hwirq", 1147 seq_printf(m, "%-5s %-7s %-15s %s\n", "virq", "hwirq",
1064 "chip name", "host name"); 1148 "chip name", "host name");
1065 1149
1066 for (i = 1; i < NR_IRQS; i++) { 1150 for (i = 1; i < nr_irqs; i++) {
1067 desc = get_irq_desc(i); 1151 desc = irq_to_desc(i);
1068 spin_lock_irqsave(&desc->lock, flags); 1152 if (!desc)
1153 continue;
1154
1155 raw_spin_lock_irqsave(&desc->lock, flags);
1069 1156
1070 if (desc->action && desc->action->handler) { 1157 if (desc->action && desc->action->handler) {
1071 seq_printf(m, "%5d ", i); 1158 seq_printf(m, "%5d ", i);
1072 seq_printf(m, "0x%05lx ", virq_to_hw(i)); 1159 seq_printf(m, "0x%05lx ", virq_to_hw(i));
1073 1160
1074 if (desc->chip && desc->chip->typename) 1161 if (desc->chip && desc->chip->name)
1075 p = desc->chip->typename; 1162 p = desc->chip->name;
1076 else 1163 else
1077 p = none; 1164 p = none;
1078 seq_printf(m, "%-15s ", p); 1165 seq_printf(m, "%-15s ", p);
@@ -1084,7 +1171,7 @@ static int virq_debug_show(struct seq_file *m, void *private)
1084 seq_printf(m, "%s\n", p); 1171 seq_printf(m, "%s\n", p);
1085 } 1172 }
1086 1173
1087 spin_unlock_irqrestore(&desc->lock, flags); 1174 raw_spin_unlock_irqrestore(&desc->lock, flags);
1088 } 1175 }
1089 1176
1090 return 0; 1177 return 0;
diff --git a/arch/powerpc/kernel/kgdb.c b/arch/powerpc/kernel/kgdb.c
index 641c74bb8e27..41bada0298c8 100644
--- a/arch/powerpc/kernel/kgdb.c
+++ b/arch/powerpc/kernel/kgdb.c
@@ -52,7 +52,7 @@ static struct hard_trap_info
52 { 0x2030, 0x08 /* SIGFPE */ }, /* spe fp data */ 52 { 0x2030, 0x08 /* SIGFPE */ }, /* spe fp data */
53 { 0x2040, 0x08 /* SIGFPE */ }, /* spe fp data */ 53 { 0x2040, 0x08 /* SIGFPE */ }, /* spe fp data */
54 { 0x2050, 0x08 /* SIGFPE */ }, /* spe fp round */ 54 { 0x2050, 0x08 /* SIGFPE */ }, /* spe fp round */
55 { 0x2060, 0x0e /* SIGILL */ }, /* performace monitor */ 55 { 0x2060, 0x0e /* SIGILL */ }, /* performance monitor */
56 { 0x2900, 0x08 /* SIGFPE */ }, /* apu unavailable */ 56 { 0x2900, 0x08 /* SIGFPE */ }, /* apu unavailable */
57 { 0x3100, 0x0e /* SIGALRM */ }, /* fixed interval timer */ 57 { 0x3100, 0x0e /* SIGALRM */ }, /* fixed interval timer */
58 { 0x3200, 0x02 /* SIGINT */ }, /* watchdog */ 58 { 0x3200, 0x02 /* SIGINT */ }, /* watchdog */
@@ -333,7 +333,7 @@ int kgdb_arch_handle_exception(int vector, int signo, int err_code,
333 atomic_set(&kgdb_cpu_doing_single_step, -1); 333 atomic_set(&kgdb_cpu_doing_single_step, -1);
334 /* set the trace bit if we're stepping */ 334 /* set the trace bit if we're stepping */
335 if (remcom_in_buffer[0] == 's') { 335 if (remcom_in_buffer[0] == 's') {
336#if defined(CONFIG_40x) || defined(CONFIG_BOOKE) 336#ifdef CONFIG_PPC_ADV_DEBUG_REGS
337 mtspr(SPRN_DBCR0, 337 mtspr(SPRN_DBCR0,
338 mfspr(SPRN_DBCR0) | DBCR0_IC | DBCR0_IDM); 338 mfspr(SPRN_DBCR0) | DBCR0_IC | DBCR0_IDM);
339 linux_regs->msr |= MSR_DE; 339 linux_regs->msr |= MSR_DE;
diff --git a/arch/powerpc/kernel/kprobes.c b/arch/powerpc/kernel/kprobes.c
index c9329786073b..b36f074524ad 100644
--- a/arch/powerpc/kernel/kprobes.c
+++ b/arch/powerpc/kernel/kprobes.c
@@ -31,12 +31,13 @@
31#include <linux/preempt.h> 31#include <linux/preempt.h>
32#include <linux/module.h> 32#include <linux/module.h>
33#include <linux/kdebug.h> 33#include <linux/kdebug.h>
34#include <linux/slab.h>
34#include <asm/cacheflush.h> 35#include <asm/cacheflush.h>
35#include <asm/sstep.h> 36#include <asm/sstep.h>
36#include <asm/uaccess.h> 37#include <asm/uaccess.h>
37#include <asm/system.h> 38#include <asm/system.h>
38 39
39#ifdef CONFIG_BOOKE 40#ifdef CONFIG_PPC_ADV_DEBUG_REGS
40#define MSR_SINGLESTEP (MSR_DE) 41#define MSR_SINGLESTEP (MSR_DE)
41#else 42#else
42#define MSR_SINGLESTEP (MSR_SE) 43#define MSR_SINGLESTEP (MSR_SE)
@@ -110,7 +111,7 @@ static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
110 * like Decrementer or External Interrupt */ 111 * like Decrementer or External Interrupt */
111 regs->msr &= ~MSR_EE; 112 regs->msr &= ~MSR_EE;
112 regs->msr |= MSR_SINGLESTEP; 113 regs->msr |= MSR_SINGLESTEP;
113#ifdef CONFIG_BOOKE 114#ifdef CONFIG_PPC_ADV_DEBUG_REGS
114 regs->msr &= ~MSR_CE; 115 regs->msr &= ~MSR_CE;
115 mtspr(SPRN_DBCR0, mfspr(SPRN_DBCR0) | DBCR0_IC | DBCR0_IDM); 116 mtspr(SPRN_DBCR0, mfspr(SPRN_DBCR0) | DBCR0_IC | DBCR0_IDM);
116#endif 117#endif
diff --git a/arch/powerpc/kernel/legacy_serial.c b/arch/powerpc/kernel/legacy_serial.c
index 9ddfaef1a184..035ada5443ee 100644
--- a/arch/powerpc/kernel/legacy_serial.c
+++ b/arch/powerpc/kernel/legacy_serial.c
@@ -469,7 +469,7 @@ static int __init serial_dev_init(void)
469 return -ENODEV; 469 return -ENODEV;
470 470
471 /* 471 /*
472 * Before we register the platfrom serial devices, we need 472 * Before we register the platform serial devices, we need
473 * to fixup their interrupts and their IO ports. 473 * to fixup their interrupts and their IO ports.
474 */ 474 */
475 DBG("Fixing serial ports interrupts and IO ports ...\n"); 475 DBG("Fixing serial ports interrupts and IO ports ...\n");
diff --git a/arch/powerpc/kernel/lparcfg.c b/arch/powerpc/kernel/lparcfg.c
index ed0ac4e4b8d8..c2c70e1b32cd 100644
--- a/arch/powerpc/kernel/lparcfg.c
+++ b/arch/powerpc/kernel/lparcfg.c
@@ -24,6 +24,7 @@
24#include <linux/proc_fs.h> 24#include <linux/proc_fs.h>
25#include <linux/init.h> 25#include <linux/init.h>
26#include <linux/seq_file.h> 26#include <linux/seq_file.h>
27#include <linux/slab.h>
27#include <asm/uaccess.h> 28#include <asm/uaccess.h>
28#include <asm/iseries/hv_lp_config.h> 29#include <asm/iseries/hv_lp_config.h>
29#include <asm/lppaca.h> 30#include <asm/lppaca.h>
@@ -359,7 +360,7 @@ static void parse_system_parameter_string(struct seq_file *m)
359 360
360 unsigned char *local_buffer = kmalloc(SPLPAR_MAXLENGTH, GFP_KERNEL); 361 unsigned char *local_buffer = kmalloc(SPLPAR_MAXLENGTH, GFP_KERNEL);
361 if (!local_buffer) { 362 if (!local_buffer) {
362 printk(KERN_ERR "%s %s kmalloc failure at line %d \n", 363 printk(KERN_ERR "%s %s kmalloc failure at line %d\n",
363 __FILE__, __func__, __LINE__); 364 __FILE__, __func__, __LINE__);
364 return; 365 return;
365 } 366 }
@@ -383,13 +384,13 @@ static void parse_system_parameter_string(struct seq_file *m)
383 int idx, w_idx; 384 int idx, w_idx;
384 char *workbuffer = kzalloc(SPLPAR_MAXLENGTH, GFP_KERNEL); 385 char *workbuffer = kzalloc(SPLPAR_MAXLENGTH, GFP_KERNEL);
385 if (!workbuffer) { 386 if (!workbuffer) {
386 printk(KERN_ERR "%s %s kmalloc failure at line %d \n", 387 printk(KERN_ERR "%s %s kmalloc failure at line %d\n",
387 __FILE__, __func__, __LINE__); 388 __FILE__, __func__, __LINE__);
388 kfree(local_buffer); 389 kfree(local_buffer);
389 return; 390 return;
390 } 391 }
391#ifdef LPARCFG_DEBUG 392#ifdef LPARCFG_DEBUG
392 printk(KERN_INFO "success calling get-system-parameter \n"); 393 printk(KERN_INFO "success calling get-system-parameter\n");
393#endif 394#endif
394 splpar_strlen = local_buffer[0] * 256 + local_buffer[1]; 395 splpar_strlen = local_buffer[0] * 256 + local_buffer[1];
395 local_buffer += 2; /* step over strlen value */ 396 local_buffer += 2; /* step over strlen value */
@@ -440,7 +441,7 @@ static int lparcfg_count_active_processors(void)
440 441
441 while ((cpus_dn = of_find_node_by_type(cpus_dn, "cpu"))) { 442 while ((cpus_dn = of_find_node_by_type(cpus_dn, "cpu"))) {
442#ifdef LPARCFG_DEBUG 443#ifdef LPARCFG_DEBUG
443 printk(KERN_ERR "cpus_dn %p \n", cpus_dn); 444 printk(KERN_ERR "cpus_dn %p\n", cpus_dn);
444#endif 445#endif
445 count++; 446 count++;
446 } 447 }
@@ -725,7 +726,7 @@ static int lparcfg_data(struct seq_file *m, void *v)
725 const unsigned int *lp_index_ptr; 726 const unsigned int *lp_index_ptr;
726 unsigned int lp_index = 0; 727 unsigned int lp_index = 0;
727 728
728 seq_printf(m, "%s %s \n", MODULE_NAME, MODULE_VERS); 729 seq_printf(m, "%s %s\n", MODULE_NAME, MODULE_VERS);
729 730
730 rootdn = of_find_node_by_path("/"); 731 rootdn = of_find_node_by_path("/");
731 if (rootdn) { 732 if (rootdn) {
@@ -781,9 +782,9 @@ static int __init lparcfg_init(void)
781 !firmware_has_feature(FW_FEATURE_ISERIES)) 782 !firmware_has_feature(FW_FEATURE_ISERIES))
782 mode |= S_IWUSR; 783 mode |= S_IWUSR;
783 784
784 ent = proc_create("ppc64/lparcfg", mode, NULL, &lparcfg_fops); 785 ent = proc_create("powerpc/lparcfg", mode, NULL, &lparcfg_fops);
785 if (!ent) { 786 if (!ent) {
786 printk(KERN_ERR "Failed to create ppc64/lparcfg\n"); 787 printk(KERN_ERR "Failed to create powerpc/lparcfg\n");
787 return -EIO; 788 return -EIO;
788 } 789 }
789 790
diff --git a/arch/powerpc/kernel/misc.S b/arch/powerpc/kernel/misc.S
index 2d29752cbe16..22e507c8a556 100644
--- a/arch/powerpc/kernel/misc.S
+++ b/arch/powerpc/kernel/misc.S
@@ -127,3 +127,29 @@ _GLOBAL(__setup_cpu_power7)
127_GLOBAL(__restore_cpu_power7) 127_GLOBAL(__restore_cpu_power7)
128 /* place holder */ 128 /* place holder */
129 blr 129 blr
130
131/*
132 * Get a minimal set of registers for our caller's nth caller.
133 * r3 = regs pointer, r5 = n.
134 *
135 * We only get R1 (stack pointer), NIP (next instruction pointer)
136 * and LR (link register). These are all we can get in the
137 * general case without doing complicated stack unwinding, but
138 * fortunately they are enough to do a stack backtrace, which
139 * is all we need them for.
140 */
141_GLOBAL(perf_arch_fetch_caller_regs)
142 mr r6,r1
143 cmpwi r5,0
144 mflr r4
145 ble 2f
146 mtctr r5
1471: PPC_LL r6,0(r6)
148 bdnz 1b
149 PPC_LL r4,PPC_LR_STKOFF(r6)
1502: PPC_LL r7,0(r6)
151 PPC_LL r7,PPC_LR_STKOFF(r7)
152 PPC_STL r6,GPR1-STACK_FRAME_OVERHEAD(r3)
153 PPC_STL r4,_NIP-STACK_FRAME_OVERHEAD(r3)
154 PPC_STL r7,_LINK-STACK_FRAME_OVERHEAD(r3)
155 blr
diff --git a/arch/powerpc/kernel/misc_32.S b/arch/powerpc/kernel/misc_32.S
index da9c0c4c10f3..8649f536f8df 100644
--- a/arch/powerpc/kernel/misc_32.S
+++ b/arch/powerpc/kernel/misc_32.S
@@ -502,15 +502,7 @@ _GLOBAL(clear_pages)
502 li r0,PAGE_SIZE/L1_CACHE_BYTES 502 li r0,PAGE_SIZE/L1_CACHE_BYTES
503 slw r0,r0,r4 503 slw r0,r0,r4
504 mtctr r0 504 mtctr r0
505#ifdef CONFIG_8xx
506 li r4, 0
5071: stw r4, 0(r3)
508 stw r4, 4(r3)
509 stw r4, 8(r3)
510 stw r4, 12(r3)
511#else
5121: dcbz 0,r3 5051: dcbz 0,r3
513#endif
514 addi r3,r3,L1_CACHE_BYTES 506 addi r3,r3,L1_CACHE_BYTES
515 bdnz 1b 507 bdnz 1b
516 blr 508 blr
@@ -535,15 +527,6 @@ _GLOBAL(copy_page)
535 addi r3,r3,-4 527 addi r3,r3,-4
536 addi r4,r4,-4 528 addi r4,r4,-4
537 529
538#ifdef CONFIG_8xx
539 /* don't use prefetch on 8xx */
540 li r0,4096/L1_CACHE_BYTES
541 mtctr r0
5421: COPY_16_BYTES
543 bdnz 1b
544 blr
545
546#else /* not 8xx, we can prefetch */
547 li r5,4 530 li r5,4
548 531
549#if MAX_COPY_PREFETCH > 1 532#if MAX_COPY_PREFETCH > 1
@@ -584,7 +567,6 @@ _GLOBAL(copy_page)
584 li r0,MAX_COPY_PREFETCH 567 li r0,MAX_COPY_PREFETCH
585 li r11,4 568 li r11,4
586 b 2b 569 b 2b
587#endif /* CONFIG_8xx */
588 570
589/* 571/*
590 * void atomic_clear_mask(atomic_t mask, atomic_t *addr) 572 * void atomic_clear_mask(atomic_t mask, atomic_t *addr)
diff --git a/arch/powerpc/kernel/nvram_64.c b/arch/powerpc/kernel/nvram_64.c
index 0ed31f220482..9cf197f01e94 100644
--- a/arch/powerpc/kernel/nvram_64.c
+++ b/arch/powerpc/kernel/nvram_64.c
@@ -139,8 +139,8 @@ out:
139 139
140} 140}
141 141
142static int dev_nvram_ioctl(struct inode *inode, struct file *file, 142static long dev_nvram_ioctl(struct file *file, unsigned int cmd,
143 unsigned int cmd, unsigned long arg) 143 unsigned long arg)
144{ 144{
145 switch(cmd) { 145 switch(cmd) {
146#ifdef CONFIG_PPC_PMAC 146#ifdef CONFIG_PPC_PMAC
@@ -169,11 +169,11 @@ static int dev_nvram_ioctl(struct inode *inode, struct file *file,
169} 169}
170 170
171const struct file_operations nvram_fops = { 171const struct file_operations nvram_fops = {
172 .owner = THIS_MODULE, 172 .owner = THIS_MODULE,
173 .llseek = dev_nvram_llseek, 173 .llseek = dev_nvram_llseek,
174 .read = dev_nvram_read, 174 .read = dev_nvram_read,
175 .write = dev_nvram_write, 175 .write = dev_nvram_write,
176 .ioctl = dev_nvram_ioctl, 176 .unlocked_ioctl = dev_nvram_ioctl,
177}; 177};
178 178
179static struct miscdevice nvram_dev = { 179static struct miscdevice nvram_dev = {
@@ -184,7 +184,7 @@ static struct miscdevice nvram_dev = {
184 184
185 185
186#ifdef DEBUG_NVRAM 186#ifdef DEBUG_NVRAM
187static void nvram_print_partitions(char * label) 187static void __init nvram_print_partitions(char * label)
188{ 188{
189 struct list_head * p; 189 struct list_head * p;
190 struct nvram_partition * tmp_part; 190 struct nvram_partition * tmp_part;
@@ -202,7 +202,7 @@ static void nvram_print_partitions(char * label)
202#endif 202#endif
203 203
204 204
205static int nvram_write_header(struct nvram_partition * part) 205static int __init nvram_write_header(struct nvram_partition * part)
206{ 206{
207 loff_t tmp_index; 207 loff_t tmp_index;
208 int rc; 208 int rc;
@@ -214,7 +214,7 @@ static int nvram_write_header(struct nvram_partition * part)
214} 214}
215 215
216 216
217static unsigned char nvram_checksum(struct nvram_header *p) 217static unsigned char __init nvram_checksum(struct nvram_header *p)
218{ 218{
219 unsigned int c_sum, c_sum2; 219 unsigned int c_sum, c_sum2;
220 unsigned short *sp = (unsigned short *)p->name; /* assume 6 shorts */ 220 unsigned short *sp = (unsigned short *)p->name; /* assume 6 shorts */
@@ -228,32 +228,7 @@ static unsigned char nvram_checksum(struct nvram_header *p)
228 return c_sum; 228 return c_sum;
229} 229}
230 230
231 231static int __init nvram_remove_os_partition(void)
232/*
233 * Find an nvram partition, sig can be 0 for any
234 * partition or name can be NULL for any name, else
235 * tries to match both
236 */
237struct nvram_partition *nvram_find_partition(int sig, const char *name)
238{
239 struct nvram_partition * part;
240 struct list_head * p;
241
242 list_for_each(p, &nvram_part->partition) {
243 part = list_entry(p, struct nvram_partition, partition);
244
245 if (sig && part->header.signature != sig)
246 continue;
247 if (name && 0 != strncmp(name, part->header.name, 12))
248 continue;
249 return part;
250 }
251 return NULL;
252}
253EXPORT_SYMBOL(nvram_find_partition);
254
255
256static int nvram_remove_os_partition(void)
257{ 232{
258 struct list_head *i; 233 struct list_head *i;
259 struct list_head *j; 234 struct list_head *j;
@@ -319,7 +294,7 @@ static int nvram_remove_os_partition(void)
319 * Will create a partition starting at the first free 294 * Will create a partition starting at the first free
320 * space found if space has enough room. 295 * space found if space has enough room.
321 */ 296 */
322static int nvram_create_os_partition(void) 297static int __init nvram_create_os_partition(void)
323{ 298{
324 struct nvram_partition *part; 299 struct nvram_partition *part;
325 struct nvram_partition *new_part; 300 struct nvram_partition *new_part;
@@ -363,8 +338,8 @@ static int nvram_create_os_partition(void)
363 338
364 rc = nvram_write_header(new_part); 339 rc = nvram_write_header(new_part);
365 if (rc <= 0) { 340 if (rc <= 0) {
366 printk(KERN_ERR "nvram_create_os_partition: nvram_write_header \ 341 printk(KERN_ERR "nvram_create_os_partition: nvram_write_header "
367 failed (%d)\n", rc); 342 "failed (%d)\n", rc);
368 return rc; 343 return rc;
369 } 344 }
370 345
@@ -374,7 +349,7 @@ static int nvram_create_os_partition(void)
374 rc = ppc_md.nvram_write((char *)&seq_init, sizeof(seq_init), &tmp_index); 349 rc = ppc_md.nvram_write((char *)&seq_init, sizeof(seq_init), &tmp_index);
375 if (rc <= 0) { 350 if (rc <= 0) {
376 printk(KERN_ERR "nvram_create_os_partition: nvram_write " 351 printk(KERN_ERR "nvram_create_os_partition: nvram_write "
377 "failed (%d)\n", rc); 352 "failed (%d)\n", rc);
378 return rc; 353 return rc;
379 } 354 }
380 355
@@ -422,7 +397,7 @@ static int nvram_create_os_partition(void)
422 * 5.) If the max chunk cannot be allocated then try finding a chunk 397 * 5.) If the max chunk cannot be allocated then try finding a chunk
423 * that will satisfy the minum needed (NVRAM_MIN_REQ). 398 * that will satisfy the minum needed (NVRAM_MIN_REQ).
424 */ 399 */
425static int nvram_setup_partition(void) 400static int __init nvram_setup_partition(void)
426{ 401{
427 struct list_head * p; 402 struct list_head * p;
428 struct nvram_partition * part; 403 struct nvram_partition * part;
@@ -480,7 +455,7 @@ static int nvram_setup_partition(void)
480} 455}
481 456
482 457
483static int nvram_scan_partitions(void) 458static int __init nvram_scan_partitions(void)
484{ 459{
485 loff_t cur_index = 0; 460 loff_t cur_index = 0;
486 struct nvram_header phead; 461 struct nvram_header phead;
@@ -706,6 +681,9 @@ int nvram_clear_error_log(void)
706 int clear_word = ERR_FLAG_ALREADY_LOGGED; 681 int clear_word = ERR_FLAG_ALREADY_LOGGED;
707 int rc; 682 int rc;
708 683
684 if (nvram_error_log_index == -1)
685 return -1;
686
709 tmp_index = nvram_error_log_index; 687 tmp_index = nvram_error_log_index;
710 688
711 rc = ppc_md.nvram_write((char *)&clear_word, sizeof(int), &tmp_index); 689 rc = ppc_md.nvram_write((char *)&clear_word, sizeof(int), &tmp_index);
diff --git a/arch/powerpc/kernel/of_platform.c b/arch/powerpc/kernel/of_platform.c
index 1a4fc0d11a03..6c1dfc3ff8bc 100644
--- a/arch/powerpc/kernel/of_platform.c
+++ b/arch/powerpc/kernel/of_platform.c
@@ -17,7 +17,6 @@
17#include <linux/init.h> 17#include <linux/init.h>
18#include <linux/module.h> 18#include <linux/module.h>
19#include <linux/mod_devicetable.h> 19#include <linux/mod_devicetable.h>
20#include <linux/slab.h>
21#include <linux/pci.h> 20#include <linux/pci.h>
22#include <linux/of.h> 21#include <linux/of.h>
23#include <linux/of_device.h> 22#include <linux/of_device.h>
@@ -214,7 +213,7 @@ EXPORT_SYMBOL(of_find_device_by_node);
214static int of_dev_phandle_match(struct device *dev, void *data) 213static int of_dev_phandle_match(struct device *dev, void *data)
215{ 214{
216 phandle *ph = data; 215 phandle *ph = data;
217 return to_of_device(dev)->node->linux_phandle == *ph; 216 return to_of_device(dev)->node->phandle == *ph;
218} 217}
219 218
220struct of_device *of_find_device_by_phandle(phandle ph) 219struct of_device *of_find_device_by_phandle(phandle ph)
diff --git a/arch/powerpc/kernel/paca.c b/arch/powerpc/kernel/paca.c
index d16b1ea55d44..0c40c6f476fe 100644
--- a/arch/powerpc/kernel/paca.c
+++ b/arch/powerpc/kernel/paca.c
@@ -9,11 +9,15 @@
9 9
10#include <linux/threads.h> 10#include <linux/threads.h>
11#include <linux/module.h> 11#include <linux/module.h>
12#include <linux/lmb.h>
12 13
14#include <asm/firmware.h>
13#include <asm/lppaca.h> 15#include <asm/lppaca.h>
14#include <asm/paca.h> 16#include <asm/paca.h>
15#include <asm/sections.h> 17#include <asm/sections.h>
16#include <asm/pgtable.h> 18#include <asm/pgtable.h>
19#include <asm/iseries/lpar_map.h>
20#include <asm/iseries/hv_types.h>
17 21
18/* This symbol is provided by the linker - let it fill in the paca 22/* This symbol is provided by the linker - let it fill in the paca
19 * field correctly */ 23 * field correctly */
@@ -70,37 +74,82 @@ struct slb_shadow slb_shadow[] __cacheline_aligned = {
70 * processors. The processor VPD array needs one entry per physical 74 * processors. The processor VPD array needs one entry per physical
71 * processor (not thread). 75 * processor (not thread).
72 */ 76 */
73struct paca_struct paca[NR_CPUS]; 77struct paca_struct *paca;
74EXPORT_SYMBOL(paca); 78EXPORT_SYMBOL(paca);
75 79
76void __init initialise_pacas(void) 80struct paca_struct boot_paca;
77{
78 int cpu;
79 81
80 /* The TOC register (GPR2) points 32kB into the TOC, so that 64kB 82void __init initialise_paca(struct paca_struct *new_paca, int cpu)
81 * of the TOC can be addressed using a single machine instruction. 83{
82 */ 84 /* The TOC register (GPR2) points 32kB into the TOC, so that 64kB
85 * of the TOC can be addressed using a single machine instruction.
86 */
83 unsigned long kernel_toc = (unsigned long)(&__toc_start) + 0x8000UL; 87 unsigned long kernel_toc = (unsigned long)(&__toc_start) + 0x8000UL;
84 88
85 /* Can't use for_each_*_cpu, as they aren't functional yet */
86 for (cpu = 0; cpu < NR_CPUS; cpu++) {
87 struct paca_struct *new_paca = &paca[cpu];
88
89#ifdef CONFIG_PPC_BOOK3S 89#ifdef CONFIG_PPC_BOOK3S
90 new_paca->lppaca_ptr = &lppaca[cpu]; 90 new_paca->lppaca_ptr = &lppaca[cpu];
91#else 91#else
92 new_paca->kernel_pgd = swapper_pg_dir; 92 new_paca->kernel_pgd = swapper_pg_dir;
93#endif 93#endif
94 new_paca->lock_token = 0x8000; 94 new_paca->lock_token = 0x8000;
95 new_paca->paca_index = cpu; 95 new_paca->paca_index = cpu;
96 new_paca->kernel_toc = kernel_toc; 96 new_paca->kernel_toc = kernel_toc;
97 new_paca->kernelbase = (unsigned long) _stext; 97 new_paca->kernelbase = (unsigned long) _stext;
98 new_paca->kernel_msr = MSR_KERNEL; 98 new_paca->kernel_msr = MSR_KERNEL;
99 new_paca->hw_cpu_id = 0xffff; 99 new_paca->hw_cpu_id = 0xffff;
100 new_paca->__current = &init_task; 100 new_paca->__current = &init_task;
101#ifdef CONFIG_PPC_STD_MMU_64 101#ifdef CONFIG_PPC_STD_MMU_64
102 new_paca->slb_shadow_ptr = &slb_shadow[cpu]; 102 new_paca->slb_shadow_ptr = &slb_shadow[cpu];
103#endif /* CONFIG_PPC_STD_MMU_64 */ 103#endif /* CONFIG_PPC_STD_MMU_64 */
104}
105
106static int __initdata paca_size;
107
108void __init allocate_pacas(void)
109{
110 int nr_cpus, cpu, limit;
111
112 /*
113 * We can't take SLB misses on the paca, and we want to access them
114 * in real mode, so allocate them within the RMA and also within
115 * the first segment. On iSeries they must be within the area mapped
116 * by the HV, which is HvPagesToMap * HVPAGESIZE bytes.
117 */
118 limit = min(0x10000000ULL, lmb.rmo_size);
119 if (firmware_has_feature(FW_FEATURE_ISERIES))
120 limit = min(limit, HvPagesToMap * HVPAGESIZE);
121
122 nr_cpus = NR_CPUS;
123 /* On iSeries we know we can never have more than 64 cpus */
124 if (firmware_has_feature(FW_FEATURE_ISERIES))
125 nr_cpus = min(64, nr_cpus);
126
127 paca_size = PAGE_ALIGN(sizeof(struct paca_struct) * nr_cpus);
128
129 paca = __va(lmb_alloc_base(paca_size, PAGE_SIZE, limit));
130 memset(paca, 0, paca_size);
131
132 printk(KERN_DEBUG "Allocated %u bytes for %d pacas at %p\n",
133 paca_size, nr_cpus, paca);
134
135 /* Can't use for_each_*_cpu, as they aren't functional yet */
136 for (cpu = 0; cpu < nr_cpus; cpu++)
137 initialise_paca(&paca[cpu], cpu);
138}
139
140void __init free_unused_pacas(void)
141{
142 int new_size;
143
144 new_size = PAGE_ALIGN(sizeof(struct paca_struct) * num_possible_cpus());
145
146 if (new_size >= paca_size)
147 return;
148
149 lmb_free(__pa(paca) + new_size, paca_size - new_size);
150
151 printk(KERN_DEBUG "Freed %u bytes for unused pacas\n",
152 paca_size - new_size);
104 153
105 } 154 paca_size = new_size;
106} 155}
diff --git a/arch/powerpc/kernel/pci-common.c b/arch/powerpc/kernel/pci-common.c
index e8dfdbd9327a..0c0567e58409 100644
--- a/arch/powerpc/kernel/pci-common.c
+++ b/arch/powerpc/kernel/pci-common.c
@@ -26,6 +26,7 @@
26#include <linux/syscalls.h> 26#include <linux/syscalls.h>
27#include <linux/irq.h> 27#include <linux/irq.h>
28#include <linux/vmalloc.h> 28#include <linux/vmalloc.h>
29#include <linux/slab.h>
29 30
30#include <asm/processor.h> 31#include <asm/processor.h>
31#include <asm/io.h> 32#include <asm/io.h>
@@ -63,21 +64,6 @@ struct dma_map_ops *get_pci_dma_ops(void)
63} 64}
64EXPORT_SYMBOL(get_pci_dma_ops); 65EXPORT_SYMBOL(get_pci_dma_ops);
65 66
66int pci_set_dma_mask(struct pci_dev *dev, u64 mask)
67{
68 return dma_set_mask(&dev->dev, mask);
69}
70
71int pci_set_consistent_dma_mask(struct pci_dev *dev, u64 mask)
72{
73 int rc;
74
75 rc = dma_set_mask(&dev->dev, mask);
76 dev->dev.coherent_dma_mask = dev->dma_mask;
77
78 return rc;
79}
80
81struct pci_controller *pcibios_alloc_controller(struct device_node *dev) 67struct pci_controller *pcibios_alloc_controller(struct device_node *dev)
82{ 68{
83 struct pci_controller *phb; 69 struct pci_controller *phb;
@@ -1047,10 +1033,8 @@ static void __devinit pcibios_fixup_bridge(struct pci_bus *bus)
1047 1033
1048 struct pci_dev *dev = bus->self; 1034 struct pci_dev *dev = bus->self;
1049 1035
1050 for (i = 0; i < PCI_BUS_NUM_RESOURCES; ++i) { 1036 pci_bus_for_each_resource(bus, res, i) {
1051 if ((res = bus->resource[i]) == NULL) 1037 if (!res || !res->flags)
1052 continue;
1053 if (!res->flags)
1054 continue; 1038 continue;
1055 if (i >= 3 && bus->self->transparent) 1039 if (i >= 3 && bus->self->transparent)
1056 continue; 1040 continue;
@@ -1107,6 +1091,12 @@ void __devinit pcibios_setup_bus_devices(struct pci_bus *bus)
1107 list_for_each_entry(dev, &bus->devices, bus_list) { 1091 list_for_each_entry(dev, &bus->devices, bus_list) {
1108 struct dev_archdata *sd = &dev->dev.archdata; 1092 struct dev_archdata *sd = &dev->dev.archdata;
1109 1093
1094 /* Cardbus can call us to add new devices to a bus, so ignore
1095 * those who are already fully discovered
1096 */
1097 if (dev->is_added)
1098 continue;
1099
1110 /* Setup OF node pointer in archdata */ 1100 /* Setup OF node pointer in archdata */
1111 sd->of_node = pci_device_to_OF_node(dev); 1101 sd->of_node = pci_device_to_OF_node(dev);
1112 1102
@@ -1147,6 +1137,13 @@ void __devinit pcibios_fixup_bus(struct pci_bus *bus)
1147} 1137}
1148EXPORT_SYMBOL(pcibios_fixup_bus); 1138EXPORT_SYMBOL(pcibios_fixup_bus);
1149 1139
1140void __devinit pci_fixup_cardbus(struct pci_bus *bus)
1141{
1142 /* Now fixup devices on that bus */
1143 pcibios_setup_bus_devices(bus);
1144}
1145
1146
1150static int skip_isa_ioresource_align(struct pci_dev *dev) 1147static int skip_isa_ioresource_align(struct pci_dev *dev)
1151{ 1148{
1152 if ((ppc_pci_flags & PPC_PCI_CAN_SKIP_ISA_ALIGN) && 1149 if ((ppc_pci_flags & PPC_PCI_CAN_SKIP_ISA_ALIGN) &&
@@ -1168,21 +1165,20 @@ static int skip_isa_ioresource_align(struct pci_dev *dev)
1168 * but we want to try to avoid allocating at 0x2900-0x2bff 1165 * but we want to try to avoid allocating at 0x2900-0x2bff
1169 * which might have be mirrored at 0x0100-0x03ff.. 1166 * which might have be mirrored at 0x0100-0x03ff..
1170 */ 1167 */
1171void pcibios_align_resource(void *data, struct resource *res, 1168resource_size_t pcibios_align_resource(void *data, const struct resource *res,
1172 resource_size_t size, resource_size_t align) 1169 resource_size_t size, resource_size_t align)
1173{ 1170{
1174 struct pci_dev *dev = data; 1171 struct pci_dev *dev = data;
1172 resource_size_t start = res->start;
1175 1173
1176 if (res->flags & IORESOURCE_IO) { 1174 if (res->flags & IORESOURCE_IO) {
1177 resource_size_t start = res->start;
1178
1179 if (skip_isa_ioresource_align(dev)) 1175 if (skip_isa_ioresource_align(dev))
1180 return; 1176 return start;
1181 if (start & 0x300) { 1177 if (start & 0x300)
1182 start = (start + 0x3ff) & ~0x3ff; 1178 start = (start + 0x3ff) & ~0x3ff;
1183 res->start = start;
1184 }
1185 } 1179 }
1180
1181 return start;
1186} 1182}
1187EXPORT_SYMBOL(pcibios_align_resource); 1183EXPORT_SYMBOL(pcibios_align_resource);
1188 1184
@@ -1265,9 +1261,8 @@ void pcibios_allocate_bus_resources(struct pci_bus *bus)
1265 pr_debug("PCI: Allocating bus resources for %04x:%02x...\n", 1261 pr_debug("PCI: Allocating bus resources for %04x:%02x...\n",
1266 pci_domain_nr(bus), bus->number); 1262 pci_domain_nr(bus), bus->number);
1267 1263
1268 for (i = 0; i < PCI_BUS_NUM_RESOURCES; ++i) { 1264 pci_bus_for_each_resource(bus, res, i) {
1269 if ((res = bus->resource[i]) == NULL || !res->flags 1265 if (!res || !res->flags || res->start > res->end || res->parent)
1270 || res->start > res->end || res->parent)
1271 continue; 1266 continue;
1272 if (bus->parent == NULL) 1267 if (bus->parent == NULL)
1273 pr = (res->flags & IORESOURCE_IO) ? 1268 pr = (res->flags & IORESOURCE_IO) ?
diff --git a/arch/powerpc/kernel/pci_32.c b/arch/powerpc/kernel/pci_32.c
index c13668cf36d9..e7db5b48004a 100644
--- a/arch/powerpc/kernel/pci_32.c
+++ b/arch/powerpc/kernel/pci_32.c
@@ -14,6 +14,7 @@
14#include <linux/irq.h> 14#include <linux/irq.h>
15#include <linux/list.h> 15#include <linux/list.h>
16#include <linux/of.h> 16#include <linux/of.h>
17#include <linux/slab.h>
17 18
18#include <asm/processor.h> 19#include <asm/processor.h>
19#include <asm/io.h> 20#include <asm/io.h>
diff --git a/arch/powerpc/kernel/pci_64.c b/arch/powerpc/kernel/pci_64.c
index ccf56ac92de5..d43fc65749c1 100644
--- a/arch/powerpc/kernel/pci_64.c
+++ b/arch/powerpc/kernel/pci_64.c
@@ -224,7 +224,7 @@ long sys_pciconfig_iobase(long which, unsigned long in_bus,
224 * G5 machines... So when something asks for bus 0 io base 224 * G5 machines... So when something asks for bus 0 io base
225 * (bus 0 is HT root), we return the AGP one instead. 225 * (bus 0 is HT root), we return the AGP one instead.
226 */ 226 */
227 if (in_bus == 0 && machine_is_compatible("MacRISC4")) { 227 if (in_bus == 0 && of_machine_is_compatible("MacRISC4")) {
228 struct device_node *agp; 228 struct device_node *agp;
229 229
230 agp = of_find_compatible_node(NULL, NULL, "u3-agp"); 230 agp = of_find_compatible_node(NULL, NULL, "u3-agp");
diff --git a/arch/powerpc/kernel/pci_dn.c b/arch/powerpc/kernel/pci_dn.c
index d5e36e5dc7c2..d56b35ee7f74 100644
--- a/arch/powerpc/kernel/pci_dn.c
+++ b/arch/powerpc/kernel/pci_dn.c
@@ -23,6 +23,7 @@
23#include <linux/pci.h> 23#include <linux/pci.h>
24#include <linux/string.h> 24#include <linux/string.h>
25#include <linux/init.h> 25#include <linux/init.h>
26#include <linux/gfp.h>
26 27
27#include <asm/io.h> 28#include <asm/io.h>
28#include <asm/prom.h> 29#include <asm/prom.h>
diff --git a/arch/powerpc/kernel/pci_of_scan.c b/arch/powerpc/kernel/pci_of_scan.c
index 7311fdfb9bf8..cd11d5ca80df 100644
--- a/arch/powerpc/kernel/pci_of_scan.c
+++ b/arch/powerpc/kernel/pci_of_scan.c
@@ -123,6 +123,7 @@ struct pci_dev *of_create_pci_dev(struct device_node *node,
123{ 123{
124 struct pci_dev *dev; 124 struct pci_dev *dev;
125 const char *type; 125 const char *type;
126 struct pci_slot *slot;
126 127
127 dev = alloc_pci_dev(); 128 dev = alloc_pci_dev();
128 if (!dev) 129 if (!dev)
@@ -140,6 +141,11 @@ struct pci_dev *of_create_pci_dev(struct device_node *node,
140 dev->devfn = devfn; 141 dev->devfn = devfn;
141 dev->multifunction = 0; /* maybe a lie? */ 142 dev->multifunction = 0; /* maybe a lie? */
142 dev->needs_freset = 0; /* pcie fundamental reset required */ 143 dev->needs_freset = 0; /* pcie fundamental reset required */
144 set_pcie_port_type(dev);
145
146 list_for_each_entry(slot, &dev->bus->slots, list)
147 if (PCI_SLOT(dev->devfn) == slot->number)
148 dev->slot = slot;
143 149
144 dev->vendor = get_int_prop(node, "vendor-id", 0xffff); 150 dev->vendor = get_int_prop(node, "vendor-id", 0xffff);
145 dev->device = get_int_prop(node, "device-id", 0xffff); 151 dev->device = get_int_prop(node, "device-id", 0xffff);
@@ -160,10 +166,14 @@ struct pci_dev *of_create_pci_dev(struct device_node *node,
160 dev->error_state = pci_channel_io_normal; 166 dev->error_state = pci_channel_io_normal;
161 dev->dma_mask = 0xffffffff; 167 dev->dma_mask = 0xffffffff;
162 168
169 /* Early fixups, before probing the BARs */
170 pci_fixup_device(pci_fixup_early, dev);
171
163 if (!strcmp(type, "pci") || !strcmp(type, "pciex")) { 172 if (!strcmp(type, "pci") || !strcmp(type, "pciex")) {
164 /* a PCI-PCI bridge */ 173 /* a PCI-PCI bridge */
165 dev->hdr_type = PCI_HEADER_TYPE_BRIDGE; 174 dev->hdr_type = PCI_HEADER_TYPE_BRIDGE;
166 dev->rom_base_reg = PCI_ROM_ADDRESS1; 175 dev->rom_base_reg = PCI_ROM_ADDRESS1;
176 set_pcie_hotplug_bridge(dev);
167 } else if (!strcmp(type, "cardbus")) { 177 } else if (!strcmp(type, "cardbus")) {
168 dev->hdr_type = PCI_HEADER_TYPE_CARDBUS; 178 dev->hdr_type = PCI_HEADER_TYPE_CARDBUS;
169 } else { 179 } else {
@@ -294,7 +304,7 @@ static void __devinit __of_scan_bus(struct device_node *node,
294 int reglen, devfn; 304 int reglen, devfn;
295 struct pci_dev *dev; 305 struct pci_dev *dev;
296 306
297 pr_debug("of_scan_bus(%s) bus no %d... \n", 307 pr_debug("of_scan_bus(%s) bus no %d...\n",
298 node->full_name, bus->number); 308 node->full_name, bus->number);
299 309
300 /* Scan direct children */ 310 /* Scan direct children */
diff --git a/arch/powerpc/kernel/perf_callchain.c b/arch/powerpc/kernel/perf_callchain.c
index 0a03cf70d247..95ad9dad298e 100644
--- a/arch/powerpc/kernel/perf_callchain.c
+++ b/arch/powerpc/kernel/perf_callchain.c
@@ -119,13 +119,6 @@ static void perf_callchain_kernel(struct pt_regs *regs,
119} 119}
120 120
121#ifdef CONFIG_PPC64 121#ifdef CONFIG_PPC64
122
123#ifdef CONFIG_HUGETLB_PAGE
124#define is_huge_psize(pagesize) (HPAGE_SHIFT && mmu_huge_psizes[pagesize])
125#else
126#define is_huge_psize(pagesize) 0
127#endif
128
129/* 122/*
130 * On 64-bit we don't want to invoke hash_page on user addresses from 123 * On 64-bit we don't want to invoke hash_page on user addresses from
131 * interrupt context, so if the access faults, we read the page tables 124 * interrupt context, so if the access faults, we read the page tables
@@ -135,7 +128,7 @@ static int read_user_stack_slow(void __user *ptr, void *ret, int nb)
135{ 128{
136 pgd_t *pgdir; 129 pgd_t *pgdir;
137 pte_t *ptep, pte; 130 pte_t *ptep, pte;
138 int pagesize; 131 unsigned shift;
139 unsigned long addr = (unsigned long) ptr; 132 unsigned long addr = (unsigned long) ptr;
140 unsigned long offset; 133 unsigned long offset;
141 unsigned long pfn; 134 unsigned long pfn;
@@ -145,17 +138,14 @@ static int read_user_stack_slow(void __user *ptr, void *ret, int nb)
145 if (!pgdir) 138 if (!pgdir)
146 return -EFAULT; 139 return -EFAULT;
147 140
148 pagesize = get_slice_psize(current->mm, addr); 141 ptep = find_linux_pte_or_hugepte(pgdir, addr, &shift);
142 if (!shift)
143 shift = PAGE_SHIFT;
149 144
150 /* align address to page boundary */ 145 /* align address to page boundary */
151 offset = addr & ((1ul << mmu_psize_defs[pagesize].shift) - 1); 146 offset = addr & ((1UL << shift) - 1);
152 addr -= offset; 147 addr -= offset;
153 148
154 if (is_huge_psize(pagesize))
155 ptep = huge_pte_offset(current->mm, addr);
156 else
157 ptep = find_linux_pte(pgdir, addr);
158
159 if (ptep == NULL) 149 if (ptep == NULL)
160 return -EFAULT; 150 return -EFAULT;
161 pte = *ptep; 151 pte = *ptep;
@@ -497,17 +487,14 @@ static void perf_callchain_user_32(struct pt_regs *regs,
497 * Since we can't get PMU interrupts inside a PMU interrupt handler, 487 * Since we can't get PMU interrupts inside a PMU interrupt handler,
498 * we don't need separate irq and nmi entries here. 488 * we don't need separate irq and nmi entries here.
499 */ 489 */
500static DEFINE_PER_CPU(struct perf_callchain_entry, callchain); 490static DEFINE_PER_CPU(struct perf_callchain_entry, cpu_perf_callchain);
501 491
502struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) 492struct perf_callchain_entry *perf_callchain(struct pt_regs *regs)
503{ 493{
504 struct perf_callchain_entry *entry = &__get_cpu_var(callchain); 494 struct perf_callchain_entry *entry = &__get_cpu_var(cpu_perf_callchain);
505 495
506 entry->nr = 0; 496 entry->nr = 0;
507 497
508 if (current->pid == 0) /* idle task? */
509 return entry;
510
511 if (!user_mode(regs)) { 498 if (!user_mode(regs)) {
512 perf_callchain_kernel(regs, entry); 499 perf_callchain_kernel(regs, entry);
513 if (current->mm) 500 if (current->mm)
diff --git a/arch/powerpc/kernel/perf_event.c b/arch/powerpc/kernel/perf_event.c
index 87f1663584b0..08460a2e9f41 100644
--- a/arch/powerpc/kernel/perf_event.c
+++ b/arch/powerpc/kernel/perf_event.c
@@ -718,10 +718,10 @@ static int collect_events(struct perf_event *group, int max_count,
718 return n; 718 return n;
719} 719}
720 720
721static void event_sched_in(struct perf_event *event, int cpu) 721static void event_sched_in(struct perf_event *event)
722{ 722{
723 event->state = PERF_EVENT_STATE_ACTIVE; 723 event->state = PERF_EVENT_STATE_ACTIVE;
724 event->oncpu = cpu; 724 event->oncpu = smp_processor_id();
725 event->tstamp_running += event->ctx->time - event->tstamp_stopped; 725 event->tstamp_running += event->ctx->time - event->tstamp_stopped;
726 if (is_software_event(event)) 726 if (is_software_event(event))
727 event->pmu->enable(event); 727 event->pmu->enable(event);
@@ -735,7 +735,7 @@ static void event_sched_in(struct perf_event *event, int cpu)
735 */ 735 */
736int hw_perf_group_sched_in(struct perf_event *group_leader, 736int hw_perf_group_sched_in(struct perf_event *group_leader,
737 struct perf_cpu_context *cpuctx, 737 struct perf_cpu_context *cpuctx,
738 struct perf_event_context *ctx, int cpu) 738 struct perf_event_context *ctx)
739{ 739{
740 struct cpu_hw_events *cpuhw; 740 struct cpu_hw_events *cpuhw;
741 long i, n, n0; 741 long i, n, n0;
@@ -766,10 +766,10 @@ int hw_perf_group_sched_in(struct perf_event *group_leader,
766 cpuhw->event[i]->hw.config = cpuhw->events[i]; 766 cpuhw->event[i]->hw.config = cpuhw->events[i];
767 cpuctx->active_oncpu += n; 767 cpuctx->active_oncpu += n;
768 n = 1; 768 n = 1;
769 event_sched_in(group_leader, cpu); 769 event_sched_in(group_leader);
770 list_for_each_entry(sub, &group_leader->sibling_list, group_entry) { 770 list_for_each_entry(sub, &group_leader->sibling_list, group_entry) {
771 if (sub->state != PERF_EVENT_STATE_OFF) { 771 if (sub->state != PERF_EVENT_STATE_OFF) {
772 event_sched_in(sub, cpu); 772 event_sched_in(sub);
773 ++n; 773 ++n;
774 } 774 }
775 } 775 }
@@ -1164,10 +1164,10 @@ static void record_and_restart(struct perf_event *event, unsigned long val,
1164 * Finally record data if requested. 1164 * Finally record data if requested.
1165 */ 1165 */
1166 if (record) { 1166 if (record) {
1167 struct perf_sample_data data = { 1167 struct perf_sample_data data;
1168 .addr = 0, 1168
1169 .period = event->hw.last_period, 1169 perf_sample_data_init(&data, ~0ULL);
1170 }; 1170 data.period = event->hw.last_period;
1171 1171
1172 if (event->attr.sample_type & PERF_SAMPLE_ADDR) 1172 if (event->attr.sample_type & PERF_SAMPLE_ADDR)
1173 perf_get_data_addr(regs, &data.addr); 1173 perf_get_data_addr(regs, &data.addr);
@@ -1287,7 +1287,7 @@ static void perf_event_interrupt(struct pt_regs *regs)
1287 irq_exit(); 1287 irq_exit();
1288} 1288}
1289 1289
1290void hw_perf_event_setup(int cpu) 1290static void power_pmu_setup(int cpu)
1291{ 1291{
1292 struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu); 1292 struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
1293 1293
@@ -1297,6 +1297,23 @@ void hw_perf_event_setup(int cpu)
1297 cpuhw->mmcr[0] = MMCR0_FC; 1297 cpuhw->mmcr[0] = MMCR0_FC;
1298} 1298}
1299 1299
1300static int __cpuinit
1301power_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
1302{
1303 unsigned int cpu = (long)hcpu;
1304
1305 switch (action & ~CPU_TASKS_FROZEN) {
1306 case CPU_UP_PREPARE:
1307 power_pmu_setup(cpu);
1308 break;
1309
1310 default:
1311 break;
1312 }
1313
1314 return NOTIFY_OK;
1315}
1316
1300int register_power_pmu(struct power_pmu *pmu) 1317int register_power_pmu(struct power_pmu *pmu)
1301{ 1318{
1302 if (ppmu) 1319 if (ppmu)
@@ -1314,5 +1331,7 @@ int register_power_pmu(struct power_pmu *pmu)
1314 freeze_events_kernel = MMCR0_FCHV; 1331 freeze_events_kernel = MMCR0_FCHV;
1315#endif /* CONFIG_PPC64 */ 1332#endif /* CONFIG_PPC64 */
1316 1333
1334 perf_cpu_notifier(power_pmu_notifier);
1335
1317 return 0; 1336 return 0;
1318} 1337}
diff --git a/arch/powerpc/kernel/perf_event_fsl_emb.c b/arch/powerpc/kernel/perf_event_fsl_emb.c
new file mode 100644
index 000000000000..369872f6cf78
--- /dev/null
+++ b/arch/powerpc/kernel/perf_event_fsl_emb.c
@@ -0,0 +1,654 @@
1/*
2 * Performance event support - Freescale Embedded Performance Monitor
3 *
4 * Copyright 2008-2009 Paul Mackerras, IBM Corporation.
5 * Copyright 2010 Freescale Semiconductor, Inc.
6 *
7 * This program is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation; either version
10 * 2 of the License, or (at your option) any later version.
11 */
12#include <linux/kernel.h>
13#include <linux/sched.h>
14#include <linux/perf_event.h>
15#include <linux/percpu.h>
16#include <linux/hardirq.h>
17#include <asm/reg_fsl_emb.h>
18#include <asm/pmc.h>
19#include <asm/machdep.h>
20#include <asm/firmware.h>
21#include <asm/ptrace.h>
22
23struct cpu_hw_events {
24 int n_events;
25 int disabled;
26 u8 pmcs_enabled;
27 struct perf_event *event[MAX_HWEVENTS];
28};
29static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
30
31static struct fsl_emb_pmu *ppmu;
32
33/* Number of perf_events counting hardware events */
34static atomic_t num_events;
35/* Used to avoid races in calling reserve/release_pmc_hardware */
36static DEFINE_MUTEX(pmc_reserve_mutex);
37
38/*
39 * If interrupts were soft-disabled when a PMU interrupt occurs, treat
40 * it as an NMI.
41 */
42static inline int perf_intr_is_nmi(struct pt_regs *regs)
43{
44#ifdef __powerpc64__
45 return !regs->softe;
46#else
47 return 0;
48#endif
49}
50
51static void perf_event_interrupt(struct pt_regs *regs);
52
53/*
54 * Read one performance monitor counter (PMC).
55 */
56static unsigned long read_pmc(int idx)
57{
58 unsigned long val;
59
60 switch (idx) {
61 case 0:
62 val = mfpmr(PMRN_PMC0);
63 break;
64 case 1:
65 val = mfpmr(PMRN_PMC1);
66 break;
67 case 2:
68 val = mfpmr(PMRN_PMC2);
69 break;
70 case 3:
71 val = mfpmr(PMRN_PMC3);
72 break;
73 default:
74 printk(KERN_ERR "oops trying to read PMC%d\n", idx);
75 val = 0;
76 }
77 return val;
78}
79
80/*
81 * Write one PMC.
82 */
83static void write_pmc(int idx, unsigned long val)
84{
85 switch (idx) {
86 case 0:
87 mtpmr(PMRN_PMC0, val);
88 break;
89 case 1:
90 mtpmr(PMRN_PMC1, val);
91 break;
92 case 2:
93 mtpmr(PMRN_PMC2, val);
94 break;
95 case 3:
96 mtpmr(PMRN_PMC3, val);
97 break;
98 default:
99 printk(KERN_ERR "oops trying to write PMC%d\n", idx);
100 }
101
102 isync();
103}
104
105/*
106 * Write one local control A register
107 */
108static void write_pmlca(int idx, unsigned long val)
109{
110 switch (idx) {
111 case 0:
112 mtpmr(PMRN_PMLCA0, val);
113 break;
114 case 1:
115 mtpmr(PMRN_PMLCA1, val);
116 break;
117 case 2:
118 mtpmr(PMRN_PMLCA2, val);
119 break;
120 case 3:
121 mtpmr(PMRN_PMLCA3, val);
122 break;
123 default:
124 printk(KERN_ERR "oops trying to write PMLCA%d\n", idx);
125 }
126
127 isync();
128}
129
130/*
131 * Write one local control B register
132 */
133static void write_pmlcb(int idx, unsigned long val)
134{
135 switch (idx) {
136 case 0:
137 mtpmr(PMRN_PMLCB0, val);
138 break;
139 case 1:
140 mtpmr(PMRN_PMLCB1, val);
141 break;
142 case 2:
143 mtpmr(PMRN_PMLCB2, val);
144 break;
145 case 3:
146 mtpmr(PMRN_PMLCB3, val);
147 break;
148 default:
149 printk(KERN_ERR "oops trying to write PMLCB%d\n", idx);
150 }
151
152 isync();
153}
154
155static void fsl_emb_pmu_read(struct perf_event *event)
156{
157 s64 val, delta, prev;
158
159 /*
160 * Performance monitor interrupts come even when interrupts
161 * are soft-disabled, as long as interrupts are hard-enabled.
162 * Therefore we treat them like NMIs.
163 */
164 do {
165 prev = atomic64_read(&event->hw.prev_count);
166 barrier();
167 val = read_pmc(event->hw.idx);
168 } while (atomic64_cmpxchg(&event->hw.prev_count, prev, val) != prev);
169
170 /* The counters are only 32 bits wide */
171 delta = (val - prev) & 0xfffffffful;
172 atomic64_add(delta, &event->count);
173 atomic64_sub(delta, &event->hw.period_left);
174}
175
176/*
177 * Disable all events to prevent PMU interrupts and to allow
178 * events to be added or removed.
179 */
180void hw_perf_disable(void)
181{
182 struct cpu_hw_events *cpuhw;
183 unsigned long flags;
184
185 local_irq_save(flags);
186 cpuhw = &__get_cpu_var(cpu_hw_events);
187
188 if (!cpuhw->disabled) {
189 cpuhw->disabled = 1;
190
191 /*
192 * Check if we ever enabled the PMU on this cpu.
193 */
194 if (!cpuhw->pmcs_enabled) {
195 ppc_enable_pmcs();
196 cpuhw->pmcs_enabled = 1;
197 }
198
199 if (atomic_read(&num_events)) {
200 /*
201 * Set the 'freeze all counters' bit, and disable
202 * interrupts. The barrier is to make sure the
203 * mtpmr has been executed and the PMU has frozen
204 * the events before we return.
205 */
206
207 mtpmr(PMRN_PMGC0, PMGC0_FAC);
208 isync();
209 }
210 }
211 local_irq_restore(flags);
212}
213
214/*
215 * Re-enable all events if disable == 0.
216 * If we were previously disabled and events were added, then
217 * put the new config on the PMU.
218 */
219void hw_perf_enable(void)
220{
221 struct cpu_hw_events *cpuhw;
222 unsigned long flags;
223
224 local_irq_save(flags);
225 cpuhw = &__get_cpu_var(cpu_hw_events);
226 if (!cpuhw->disabled)
227 goto out;
228
229 cpuhw->disabled = 0;
230 ppc_set_pmu_inuse(cpuhw->n_events != 0);
231
232 if (cpuhw->n_events > 0) {
233 mtpmr(PMRN_PMGC0, PMGC0_PMIE | PMGC0_FCECE);
234 isync();
235 }
236
237 out:
238 local_irq_restore(flags);
239}
240
241static int collect_events(struct perf_event *group, int max_count,
242 struct perf_event *ctrs[])
243{
244 int n = 0;
245 struct perf_event *event;
246
247 if (!is_software_event(group)) {
248 if (n >= max_count)
249 return -1;
250 ctrs[n] = group;
251 n++;
252 }
253 list_for_each_entry(event, &group->sibling_list, group_entry) {
254 if (!is_software_event(event) &&
255 event->state != PERF_EVENT_STATE_OFF) {
256 if (n >= max_count)
257 return -1;
258 ctrs[n] = event;
259 n++;
260 }
261 }
262 return n;
263}
264
265/* perf must be disabled, context locked on entry */
266static int fsl_emb_pmu_enable(struct perf_event *event)
267{
268 struct cpu_hw_events *cpuhw;
269 int ret = -EAGAIN;
270 int num_counters = ppmu->n_counter;
271 u64 val;
272 int i;
273
274 cpuhw = &get_cpu_var(cpu_hw_events);
275
276 if (event->hw.config & FSL_EMB_EVENT_RESTRICTED)
277 num_counters = ppmu->n_restricted;
278
279 /*
280 * Allocate counters from top-down, so that restricted-capable
281 * counters are kept free as long as possible.
282 */
283 for (i = num_counters - 1; i >= 0; i--) {
284 if (cpuhw->event[i])
285 continue;
286
287 break;
288 }
289
290 if (i < 0)
291 goto out;
292
293 event->hw.idx = i;
294 cpuhw->event[i] = event;
295 ++cpuhw->n_events;
296
297 val = 0;
298 if (event->hw.sample_period) {
299 s64 left = atomic64_read(&event->hw.period_left);
300 if (left < 0x80000000L)
301 val = 0x80000000L - left;
302 }
303 atomic64_set(&event->hw.prev_count, val);
304 write_pmc(i, val);
305 perf_event_update_userpage(event);
306
307 write_pmlcb(i, event->hw.config >> 32);
308 write_pmlca(i, event->hw.config_base);
309
310 ret = 0;
311 out:
312 put_cpu_var(cpu_hw_events);
313 return ret;
314}
315
316/* perf must be disabled, context locked on entry */
317static void fsl_emb_pmu_disable(struct perf_event *event)
318{
319 struct cpu_hw_events *cpuhw;
320 int i = event->hw.idx;
321
322 if (i < 0)
323 goto out;
324
325 fsl_emb_pmu_read(event);
326
327 cpuhw = &get_cpu_var(cpu_hw_events);
328
329 WARN_ON(event != cpuhw->event[event->hw.idx]);
330
331 write_pmlca(i, 0);
332 write_pmlcb(i, 0);
333 write_pmc(i, 0);
334
335 cpuhw->event[i] = NULL;
336 event->hw.idx = -1;
337
338 /*
339 * TODO: if at least one restricted event exists, and we
340 * just freed up a non-restricted-capable counter, and
341 * there is a restricted-capable counter occupied by
342 * a non-restricted event, migrate that event to the
343 * vacated counter.
344 */
345
346 cpuhw->n_events--;
347
348 out:
349 put_cpu_var(cpu_hw_events);
350}
351
352/*
353 * Re-enable interrupts on a event after they were throttled
354 * because they were coming too fast.
355 *
356 * Context is locked on entry, but perf is not disabled.
357 */
358static void fsl_emb_pmu_unthrottle(struct perf_event *event)
359{
360 s64 val, left;
361 unsigned long flags;
362
363 if (event->hw.idx < 0 || !event->hw.sample_period)
364 return;
365 local_irq_save(flags);
366 perf_disable();
367 fsl_emb_pmu_read(event);
368 left = event->hw.sample_period;
369 event->hw.last_period = left;
370 val = 0;
371 if (left < 0x80000000L)
372 val = 0x80000000L - left;
373 write_pmc(event->hw.idx, val);
374 atomic64_set(&event->hw.prev_count, val);
375 atomic64_set(&event->hw.period_left, left);
376 perf_event_update_userpage(event);
377 perf_enable();
378 local_irq_restore(flags);
379}
380
381static struct pmu fsl_emb_pmu = {
382 .enable = fsl_emb_pmu_enable,
383 .disable = fsl_emb_pmu_disable,
384 .read = fsl_emb_pmu_read,
385 .unthrottle = fsl_emb_pmu_unthrottle,
386};
387
388/*
389 * Release the PMU if this is the last perf_event.
390 */
391static void hw_perf_event_destroy(struct perf_event *event)
392{
393 if (!atomic_add_unless(&num_events, -1, 1)) {
394 mutex_lock(&pmc_reserve_mutex);
395 if (atomic_dec_return(&num_events) == 0)
396 release_pmc_hardware();
397 mutex_unlock(&pmc_reserve_mutex);
398 }
399}
400
401/*
402 * Translate a generic cache event_id config to a raw event_id code.
403 */
404static int hw_perf_cache_event(u64 config, u64 *eventp)
405{
406 unsigned long type, op, result;
407 int ev;
408
409 if (!ppmu->cache_events)
410 return -EINVAL;
411
412 /* unpack config */
413 type = config & 0xff;
414 op = (config >> 8) & 0xff;
415 result = (config >> 16) & 0xff;
416
417 if (type >= PERF_COUNT_HW_CACHE_MAX ||
418 op >= PERF_COUNT_HW_CACHE_OP_MAX ||
419 result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
420 return -EINVAL;
421
422 ev = (*ppmu->cache_events)[type][op][result];
423 if (ev == 0)
424 return -EOPNOTSUPP;
425 if (ev == -1)
426 return -EINVAL;
427 *eventp = ev;
428 return 0;
429}
430
431const struct pmu *hw_perf_event_init(struct perf_event *event)
432{
433 u64 ev;
434 struct perf_event *events[MAX_HWEVENTS];
435 int n;
436 int err;
437 int num_restricted;
438 int i;
439
440 switch (event->attr.type) {
441 case PERF_TYPE_HARDWARE:
442 ev = event->attr.config;
443 if (ev >= ppmu->n_generic || ppmu->generic_events[ev] == 0)
444 return ERR_PTR(-EOPNOTSUPP);
445 ev = ppmu->generic_events[ev];
446 break;
447
448 case PERF_TYPE_HW_CACHE:
449 err = hw_perf_cache_event(event->attr.config, &ev);
450 if (err)
451 return ERR_PTR(err);
452 break;
453
454 case PERF_TYPE_RAW:
455 ev = event->attr.config;
456 break;
457
458 default:
459 return ERR_PTR(-EINVAL);
460 }
461
462 event->hw.config = ppmu->xlate_event(ev);
463 if (!(event->hw.config & FSL_EMB_EVENT_VALID))
464 return ERR_PTR(-EINVAL);
465
466 /*
467 * If this is in a group, check if it can go on with all the
468 * other hardware events in the group. We assume the event
469 * hasn't been linked into its leader's sibling list at this point.
470 */
471 n = 0;
472 if (event->group_leader != event) {
473 n = collect_events(event->group_leader,
474 ppmu->n_counter - 1, events);
475 if (n < 0)
476 return ERR_PTR(-EINVAL);
477 }
478
479 if (event->hw.config & FSL_EMB_EVENT_RESTRICTED) {
480 num_restricted = 0;
481 for (i = 0; i < n; i++) {
482 if (events[i]->hw.config & FSL_EMB_EVENT_RESTRICTED)
483 num_restricted++;
484 }
485
486 if (num_restricted >= ppmu->n_restricted)
487 return ERR_PTR(-EINVAL);
488 }
489
490 event->hw.idx = -1;
491
492 event->hw.config_base = PMLCA_CE | PMLCA_FCM1 |
493 (u32)((ev << 16) & PMLCA_EVENT_MASK);
494
495 if (event->attr.exclude_user)
496 event->hw.config_base |= PMLCA_FCU;
497 if (event->attr.exclude_kernel)
498 event->hw.config_base |= PMLCA_FCS;
499 if (event->attr.exclude_idle)
500 return ERR_PTR(-ENOTSUPP);
501
502 event->hw.last_period = event->hw.sample_period;
503 atomic64_set(&event->hw.period_left, event->hw.last_period);
504
505 /*
506 * See if we need to reserve the PMU.
507 * If no events are currently in use, then we have to take a
508 * mutex to ensure that we don't race with another task doing
509 * reserve_pmc_hardware or release_pmc_hardware.
510 */
511 err = 0;
512 if (!atomic_inc_not_zero(&num_events)) {
513 mutex_lock(&pmc_reserve_mutex);
514 if (atomic_read(&num_events) == 0 &&
515 reserve_pmc_hardware(perf_event_interrupt))
516 err = -EBUSY;
517 else
518 atomic_inc(&num_events);
519 mutex_unlock(&pmc_reserve_mutex);
520
521 mtpmr(PMRN_PMGC0, PMGC0_FAC);
522 isync();
523 }
524 event->destroy = hw_perf_event_destroy;
525
526 if (err)
527 return ERR_PTR(err);
528 return &fsl_emb_pmu;
529}
530
531/*
532 * A counter has overflowed; update its count and record
533 * things if requested. Note that interrupts are hard-disabled
534 * here so there is no possibility of being interrupted.
535 */
536static void record_and_restart(struct perf_event *event, unsigned long val,
537 struct pt_regs *regs, int nmi)
538{
539 u64 period = event->hw.sample_period;
540 s64 prev, delta, left;
541 int record = 0;
542
543 /* we don't have to worry about interrupts here */
544 prev = atomic64_read(&event->hw.prev_count);
545 delta = (val - prev) & 0xfffffffful;
546 atomic64_add(delta, &event->count);
547
548 /*
549 * See if the total period for this event has expired,
550 * and update for the next period.
551 */
552 val = 0;
553 left = atomic64_read(&event->hw.period_left) - delta;
554 if (period) {
555 if (left <= 0) {
556 left += period;
557 if (left <= 0)
558 left = period;
559 record = 1;
560 }
561 if (left < 0x80000000LL)
562 val = 0x80000000LL - left;
563 }
564
565 /*
566 * Finally record data if requested.
567 */
568 if (record) {
569 struct perf_sample_data data = {
570 .period = event->hw.last_period,
571 };
572
573 if (perf_event_overflow(event, nmi, &data, regs)) {
574 /*
575 * Interrupts are coming too fast - throttle them
576 * by setting the event to 0, so it will be
577 * at least 2^30 cycles until the next interrupt
578 * (assuming each event counts at most 2 counts
579 * per cycle).
580 */
581 val = 0;
582 left = ~0ULL >> 1;
583 }
584 }
585
586 write_pmc(event->hw.idx, val);
587 atomic64_set(&event->hw.prev_count, val);
588 atomic64_set(&event->hw.period_left, left);
589 perf_event_update_userpage(event);
590}
591
592static void perf_event_interrupt(struct pt_regs *regs)
593{
594 int i;
595 struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events);
596 struct perf_event *event;
597 unsigned long val;
598 int found = 0;
599 int nmi;
600
601 nmi = perf_intr_is_nmi(regs);
602 if (nmi)
603 nmi_enter();
604 else
605 irq_enter();
606
607 for (i = 0; i < ppmu->n_counter; ++i) {
608 event = cpuhw->event[i];
609
610 val = read_pmc(i);
611 if ((int)val < 0) {
612 if (event) {
613 /* event has overflowed */
614 found = 1;
615 record_and_restart(event, val, regs, nmi);
616 } else {
617 /*
618 * Disabled counter is negative,
619 * reset it just in case.
620 */
621 write_pmc(i, 0);
622 }
623 }
624 }
625
626 /* PMM will keep counters frozen until we return from the interrupt. */
627 mtmsr(mfmsr() | MSR_PMM);
628 mtpmr(PMRN_PMGC0, PMGC0_PMIE | PMGC0_FCECE);
629 isync();
630
631 if (nmi)
632 nmi_exit();
633 else
634 irq_exit();
635}
636
637void hw_perf_event_setup(int cpu)
638{
639 struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);
640
641 memset(cpuhw, 0, sizeof(*cpuhw));
642}
643
644int register_fsl_emb_pmu(struct fsl_emb_pmu *pmu)
645{
646 if (ppmu)
647 return -EBUSY; /* something's already registered */
648
649 ppmu = pmu;
650 pr_info("%s performance monitor hardware support registered\n",
651 pmu->name);
652
653 return 0;
654}
diff --git a/arch/powerpc/kernel/pmc.c b/arch/powerpc/kernel/pmc.c
index 0516e2d3e02e..461499b43cff 100644
--- a/arch/powerpc/kernel/pmc.c
+++ b/arch/powerpc/kernel/pmc.c
@@ -37,7 +37,7 @@ static void dummy_perf(struct pt_regs *regs)
37} 37}
38 38
39 39
40static DEFINE_SPINLOCK(pmc_owner_lock); 40static DEFINE_RAW_SPINLOCK(pmc_owner_lock);
41static void *pmc_owner_caller; /* mostly for debugging */ 41static void *pmc_owner_caller; /* mostly for debugging */
42perf_irq_t perf_irq = dummy_perf; 42perf_irq_t perf_irq = dummy_perf;
43 43
@@ -45,7 +45,7 @@ int reserve_pmc_hardware(perf_irq_t new_perf_irq)
45{ 45{
46 int err = 0; 46 int err = 0;
47 47
48 spin_lock(&pmc_owner_lock); 48 raw_spin_lock(&pmc_owner_lock);
49 49
50 if (pmc_owner_caller) { 50 if (pmc_owner_caller) {
51 printk(KERN_WARNING "reserve_pmc_hardware: " 51 printk(KERN_WARNING "reserve_pmc_hardware: "
@@ -59,21 +59,21 @@ int reserve_pmc_hardware(perf_irq_t new_perf_irq)
59 perf_irq = new_perf_irq ? new_perf_irq : dummy_perf; 59 perf_irq = new_perf_irq ? new_perf_irq : dummy_perf;
60 60
61 out: 61 out:
62 spin_unlock(&pmc_owner_lock); 62 raw_spin_unlock(&pmc_owner_lock);
63 return err; 63 return err;
64} 64}
65EXPORT_SYMBOL_GPL(reserve_pmc_hardware); 65EXPORT_SYMBOL_GPL(reserve_pmc_hardware);
66 66
67void release_pmc_hardware(void) 67void release_pmc_hardware(void)
68{ 68{
69 spin_lock(&pmc_owner_lock); 69 raw_spin_lock(&pmc_owner_lock);
70 70
71 WARN_ON(! pmc_owner_caller); 71 WARN_ON(! pmc_owner_caller);
72 72
73 pmc_owner_caller = NULL; 73 pmc_owner_caller = NULL;
74 perf_irq = dummy_perf; 74 perf_irq = dummy_perf;
75 75
76 spin_unlock(&pmc_owner_lock); 76 raw_spin_unlock(&pmc_owner_lock);
77} 77}
78EXPORT_SYMBOL_GPL(release_pmc_hardware); 78EXPORT_SYMBOL_GPL(release_pmc_hardware);
79 79
diff --git a/arch/powerpc/kernel/power5+-pmu.c b/arch/powerpc/kernel/power5+-pmu.c
index 0f4c1c73a6ad..199de527d411 100644
--- a/arch/powerpc/kernel/power5+-pmu.c
+++ b/arch/powerpc/kernel/power5+-pmu.c
@@ -73,10 +73,6 @@
73#define MMCR1_PMCSEL_MSK 0x7f 73#define MMCR1_PMCSEL_MSK 0x7f
74 74
75/* 75/*
76 * Bits in MMCRA
77 */
78
79/*
80 * Layout of constraint bits: 76 * Layout of constraint bits:
81 * 6666555555555544444444443333333333222222222211111111110000000000 77 * 6666555555555544444444443333333333222222222211111111110000000000
82 * 3210987654321098765432109876543210987654321098765432109876543210 78 * 3210987654321098765432109876543210987654321098765432109876543210
diff --git a/arch/powerpc/kernel/power5-pmu.c b/arch/powerpc/kernel/power5-pmu.c
index c351b3a57fbb..98b6a729a9dd 100644
--- a/arch/powerpc/kernel/power5-pmu.c
+++ b/arch/powerpc/kernel/power5-pmu.c
@@ -73,10 +73,6 @@
73#define MMCR1_PMCSEL_MSK 0x7f 73#define MMCR1_PMCSEL_MSK 0x7f
74 74
75/* 75/*
76 * Bits in MMCRA
77 */
78
79/*
80 * Layout of constraint bits: 76 * Layout of constraint bits:
81 * 6666555555555544444444443333333333222222222211111111110000000000 77 * 6666555555555544444444443333333333222222222211111111110000000000
82 * 3210987654321098765432109876543210987654321098765432109876543210 78 * 3210987654321098765432109876543210987654321098765432109876543210
@@ -390,7 +386,7 @@ static int power5_compute_mmcr(u64 event[], int n_ev,
390 unsigned int hwc[], unsigned long mmcr[]) 386 unsigned int hwc[], unsigned long mmcr[])
391{ 387{
392 unsigned long mmcr1 = 0; 388 unsigned long mmcr1 = 0;
393 unsigned long mmcra = 0; 389 unsigned long mmcra = MMCRA_SDAR_DCACHE_MISS | MMCRA_SDAR_ERAT_MISS;
394 unsigned int pmc, unit, byte, psel; 390 unsigned int pmc, unit, byte, psel;
395 unsigned int ttm, grp; 391 unsigned int ttm, grp;
396 int i, isbus, bit, grsel; 392 int i, isbus, bit, grsel;
diff --git a/arch/powerpc/kernel/power6-pmu.c b/arch/powerpc/kernel/power6-pmu.c
index ca399ba5034c..84a607bda8fb 100644
--- a/arch/powerpc/kernel/power6-pmu.c
+++ b/arch/powerpc/kernel/power6-pmu.c
@@ -178,7 +178,7 @@ static int p6_compute_mmcr(u64 event[], int n_ev,
178 unsigned int hwc[], unsigned long mmcr[]) 178 unsigned int hwc[], unsigned long mmcr[])
179{ 179{
180 unsigned long mmcr1 = 0; 180 unsigned long mmcr1 = 0;
181 unsigned long mmcra = 0; 181 unsigned long mmcra = MMCRA_SDAR_DCACHE_MISS | MMCRA_SDAR_ERAT_MISS;
182 int i; 182 int i;
183 unsigned int pmc, ev, b, u, s, psel; 183 unsigned int pmc, ev, b, u, s, psel;
184 unsigned int ttmset = 0; 184 unsigned int ttmset = 0;
diff --git a/arch/powerpc/kernel/power7-pmu.c b/arch/powerpc/kernel/power7-pmu.c
index 28a4daacdc02..852f7b7f6b40 100644
--- a/arch/powerpc/kernel/power7-pmu.c
+++ b/arch/powerpc/kernel/power7-pmu.c
@@ -51,10 +51,6 @@
51#define MMCR1_PMCSEL_MSK 0xff 51#define MMCR1_PMCSEL_MSK 0xff
52 52
53/* 53/*
54 * Bits in MMCRA
55 */
56
57/*
58 * Layout of constraint bits: 54 * Layout of constraint bits:
59 * 6666555555555544444444443333333333222222222211111111110000000000 55 * 6666555555555544444444443333333333222222222211111111110000000000
60 * 3210987654321098765432109876543210987654321098765432109876543210 56 * 3210987654321098765432109876543210987654321098765432109876543210
@@ -230,7 +226,7 @@ static int power7_compute_mmcr(u64 event[], int n_ev,
230 unsigned int hwc[], unsigned long mmcr[]) 226 unsigned int hwc[], unsigned long mmcr[])
231{ 227{
232 unsigned long mmcr1 = 0; 228 unsigned long mmcr1 = 0;
233 unsigned long mmcra = 0; 229 unsigned long mmcra = MMCRA_SDAR_DCACHE_MISS | MMCRA_SDAR_ERAT_MISS;
234 unsigned int pmc, unit, combine, l2sel, psel; 230 unsigned int pmc, unit, combine, l2sel, psel;
235 unsigned int pmc_inuse = 0; 231 unsigned int pmc_inuse = 0;
236 int i; 232 int i;
diff --git a/arch/powerpc/kernel/ppc970-pmu.c b/arch/powerpc/kernel/ppc970-pmu.c
index 479574413a93..8eff48e20dba 100644
--- a/arch/powerpc/kernel/ppc970-pmu.c
+++ b/arch/powerpc/kernel/ppc970-pmu.c
@@ -84,10 +84,6 @@ static short mmcr1_adder_bits[8] = {
84}; 84};
85 85
86/* 86/*
87 * Bits in MMCRA
88 */
89
90/*
91 * Layout of constraint bits: 87 * Layout of constraint bits:
92 * 6666555555555544444444443333333333222222222211111111110000000000 88 * 6666555555555544444444443333333333222222222211111111110000000000
93 * 3210987654321098765432109876543210987654321098765432109876543210 89 * 3210987654321098765432109876543210987654321098765432109876543210
diff --git a/arch/powerpc/kernel/ppc_ksyms.c b/arch/powerpc/kernel/ppc_ksyms.c
index c8b27bb4dbde..ab3e392ac63c 100644
--- a/arch/powerpc/kernel/ppc_ksyms.c
+++ b/arch/powerpc/kernel/ppc_ksyms.c
@@ -96,8 +96,6 @@ EXPORT_SYMBOL(copy_4K_page);
96EXPORT_SYMBOL(isa_io_base); 96EXPORT_SYMBOL(isa_io_base);
97EXPORT_SYMBOL(isa_mem_base); 97EXPORT_SYMBOL(isa_mem_base);
98EXPORT_SYMBOL(pci_dram_offset); 98EXPORT_SYMBOL(pci_dram_offset);
99EXPORT_SYMBOL(pci_alloc_consistent);
100EXPORT_SYMBOL(pci_free_consistent);
101#endif /* CONFIG_PCI */ 99#endif /* CONFIG_PCI */
102 100
103EXPORT_SYMBOL(start_thread); 101EXPORT_SYMBOL(start_thread);
@@ -109,6 +107,7 @@ EXPORT_SYMBOL(giveup_altivec);
109#endif /* CONFIG_ALTIVEC */ 107#endif /* CONFIG_ALTIVEC */
110#ifdef CONFIG_VSX 108#ifdef CONFIG_VSX
111EXPORT_SYMBOL(giveup_vsx); 109EXPORT_SYMBOL(giveup_vsx);
110EXPORT_SYMBOL_GPL(__giveup_vsx);
112#endif /* CONFIG_VSX */ 111#endif /* CONFIG_VSX */
113#ifdef CONFIG_SPE 112#ifdef CONFIG_SPE
114EXPORT_SYMBOL(giveup_spe); 113EXPORT_SYMBOL(giveup_spe);
@@ -162,7 +161,6 @@ EXPORT_SYMBOL(screen_info);
162 161
163#ifdef CONFIG_PPC32 162#ifdef CONFIG_PPC32
164EXPORT_SYMBOL(timer_interrupt); 163EXPORT_SYMBOL(timer_interrupt);
165EXPORT_SYMBOL(irq_desc);
166EXPORT_SYMBOL(tb_ticks_per_jiffy); 164EXPORT_SYMBOL(tb_ticks_per_jiffy);
167EXPORT_SYMBOL(cacheable_memcpy); 165EXPORT_SYMBOL(cacheable_memcpy);
168EXPORT_SYMBOL(cacheable_memzero); 166EXPORT_SYMBOL(cacheable_memzero);
diff --git a/arch/powerpc/kernel/proc_ppc64.c b/arch/powerpc/kernel/proc_powerpc.c
index c647ddef40dc..c8ae3714e79b 100644
--- a/arch/powerpc/kernel/proc_ppc64.c
+++ b/arch/powerpc/kernel/proc_powerpc.c
@@ -19,7 +19,6 @@
19#include <linux/init.h> 19#include <linux/init.h>
20#include <linux/mm.h> 20#include <linux/mm.h>
21#include <linux/proc_fs.h> 21#include <linux/proc_fs.h>
22#include <linux/slab.h>
23#include <linux/kernel.h> 22#include <linux/kernel.h>
24 23
25#include <asm/machdep.h> 24#include <asm/machdep.h>
@@ -28,55 +27,7 @@
28#include <asm/uaccess.h> 27#include <asm/uaccess.h>
29#include <asm/prom.h> 28#include <asm/prom.h>
30 29
31static loff_t page_map_seek( struct file *file, loff_t off, int whence); 30#ifdef CONFIG_PPC64
32static ssize_t page_map_read( struct file *file, char __user *buf, size_t nbytes,
33 loff_t *ppos);
34static int page_map_mmap( struct file *file, struct vm_area_struct *vma );
35
36static const struct file_operations page_map_fops = {
37 .llseek = page_map_seek,
38 .read = page_map_read,
39 .mmap = page_map_mmap
40};
41
42/*
43 * Create the ppc64 and ppc64/rtas directories early. This allows us to
44 * assume that they have been previously created in drivers.
45 */
46static int __init proc_ppc64_create(void)
47{
48 struct proc_dir_entry *root;
49
50 root = proc_mkdir("ppc64", NULL);
51 if (!root)
52 return 1;
53
54 if (!of_find_node_by_path("/rtas"))
55 return 0;
56
57 if (!proc_mkdir("rtas", root))
58 return 1;
59
60 if (!proc_symlink("rtas", NULL, "ppc64/rtas"))
61 return 1;
62
63 return 0;
64}
65core_initcall(proc_ppc64_create);
66
67static int __init proc_ppc64_init(void)
68{
69 struct proc_dir_entry *pde;
70
71 pde = proc_create_data("ppc64/systemcfg", S_IFREG|S_IRUGO, NULL,
72 &page_map_fops, vdso_data);
73 if (!pde)
74 return 1;
75 pde->size = PAGE_SIZE;
76
77 return 0;
78}
79__initcall(proc_ppc64_init);
80 31
81static loff_t page_map_seek( struct file *file, loff_t off, int whence) 32static loff_t page_map_seek( struct file *file, loff_t off, int whence)
82{ 33{
@@ -120,3 +71,55 @@ static int page_map_mmap( struct file *file, struct vm_area_struct *vma )
120 return 0; 71 return 0;
121} 72}
122 73
74static const struct file_operations page_map_fops = {
75 .llseek = page_map_seek,
76 .read = page_map_read,
77 .mmap = page_map_mmap
78};
79
80
81static int __init proc_ppc64_init(void)
82{
83 struct proc_dir_entry *pde;
84
85 pde = proc_create_data("powerpc/systemcfg", S_IFREG|S_IRUGO, NULL,
86 &page_map_fops, vdso_data);
87 if (!pde)
88 return 1;
89 pde->size = PAGE_SIZE;
90
91 return 0;
92}
93__initcall(proc_ppc64_init);
94
95#endif /* CONFIG_PPC64 */
96
97/*
98 * Create the ppc64 and ppc64/rtas directories early. This allows us to
99 * assume that they have been previously created in drivers.
100 */
101static int __init proc_ppc64_create(void)
102{
103 struct proc_dir_entry *root;
104
105 root = proc_mkdir("powerpc", NULL);
106 if (!root)
107 return 1;
108
109#ifdef CONFIG_PPC64
110 if (!proc_symlink("ppc64", NULL, "powerpc"))
111 pr_err("Failed to create link /proc/ppc64 -> /proc/powerpc\n");
112#endif
113
114 if (!of_find_node_by_path("/rtas"))
115 return 0;
116
117 if (!proc_mkdir("rtas", root))
118 return 1;
119
120 if (!proc_symlink("rtas", NULL, "powerpc/rtas"))
121 return 1;
122
123 return 0;
124}
125core_initcall(proc_ppc64_create);
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index c930ac38e59f..e4d71ced97ef 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -245,6 +245,24 @@ void discard_lazy_cpu_state(void)
245} 245}
246#endif /* CONFIG_SMP */ 246#endif /* CONFIG_SMP */
247 247
248#ifdef CONFIG_PPC_ADV_DEBUG_REGS
249void do_send_trap(struct pt_regs *regs, unsigned long address,
250 unsigned long error_code, int signal_code, int breakpt)
251{
252 siginfo_t info;
253
254 if (notify_die(DIE_DABR_MATCH, "dabr_match", regs, error_code,
255 11, SIGSEGV) == NOTIFY_STOP)
256 return;
257
258 /* Deliver the signal to userspace */
259 info.si_signo = SIGTRAP;
260 info.si_errno = breakpt; /* breakpoint or watchpoint id */
261 info.si_code = signal_code;
262 info.si_addr = (void __user *)address;
263 force_sig_info(SIGTRAP, &info, current);
264}
265#else /* !CONFIG_PPC_ADV_DEBUG_REGS */
248void do_dabr(struct pt_regs *regs, unsigned long address, 266void do_dabr(struct pt_regs *regs, unsigned long address,
249 unsigned long error_code) 267 unsigned long error_code)
250{ 268{
@@ -257,12 +275,6 @@ void do_dabr(struct pt_regs *regs, unsigned long address,
257 if (debugger_dabr_match(regs)) 275 if (debugger_dabr_match(regs))
258 return; 276 return;
259 277
260 /* Clear the DAC and struct entries. One shot trigger */
261#if defined(CONFIG_BOOKE)
262 mtspr(SPRN_DBCR0, mfspr(SPRN_DBCR0) & ~(DBSR_DAC1R | DBSR_DAC1W
263 | DBCR0_IDM));
264#endif
265
266 /* Clear the DABR */ 278 /* Clear the DABR */
267 set_dabr(0); 279 set_dabr(0);
268 280
@@ -273,9 +285,82 @@ void do_dabr(struct pt_regs *regs, unsigned long address,
273 info.si_addr = (void __user *)address; 285 info.si_addr = (void __user *)address;
274 force_sig_info(SIGTRAP, &info, current); 286 force_sig_info(SIGTRAP, &info, current);
275} 287}
288#endif /* CONFIG_PPC_ADV_DEBUG_REGS */
276 289
277static DEFINE_PER_CPU(unsigned long, current_dabr); 290static DEFINE_PER_CPU(unsigned long, current_dabr);
278 291
292#ifdef CONFIG_PPC_ADV_DEBUG_REGS
293/*
294 * Set the debug registers back to their default "safe" values.
295 */
296static void set_debug_reg_defaults(struct thread_struct *thread)
297{
298 thread->iac1 = thread->iac2 = 0;
299#if CONFIG_PPC_ADV_DEBUG_IACS > 2
300 thread->iac3 = thread->iac4 = 0;
301#endif
302 thread->dac1 = thread->dac2 = 0;
303#if CONFIG_PPC_ADV_DEBUG_DVCS > 0
304 thread->dvc1 = thread->dvc2 = 0;
305#endif
306 thread->dbcr0 = 0;
307#ifdef CONFIG_BOOKE
308 /*
309 * Force User/Supervisor bits to b11 (user-only MSR[PR]=1)
310 */
311 thread->dbcr1 = DBCR1_IAC1US | DBCR1_IAC2US | \
312 DBCR1_IAC3US | DBCR1_IAC4US;
313 /*
314 * Force Data Address Compare User/Supervisor bits to be User-only
315 * (0b11 MSR[PR]=1) and set all other bits in DBCR2 register to be 0.
316 */
317 thread->dbcr2 = DBCR2_DAC1US | DBCR2_DAC2US;
318#else
319 thread->dbcr1 = 0;
320#endif
321}
322
323static void prime_debug_regs(struct thread_struct *thread)
324{
325 mtspr(SPRN_IAC1, thread->iac1);
326 mtspr(SPRN_IAC2, thread->iac2);
327#if CONFIG_PPC_ADV_DEBUG_IACS > 2
328 mtspr(SPRN_IAC3, thread->iac3);
329 mtspr(SPRN_IAC4, thread->iac4);
330#endif
331 mtspr(SPRN_DAC1, thread->dac1);
332 mtspr(SPRN_DAC2, thread->dac2);
333#if CONFIG_PPC_ADV_DEBUG_DVCS > 0
334 mtspr(SPRN_DVC1, thread->dvc1);
335 mtspr(SPRN_DVC2, thread->dvc2);
336#endif
337 mtspr(SPRN_DBCR0, thread->dbcr0);
338 mtspr(SPRN_DBCR1, thread->dbcr1);
339#ifdef CONFIG_BOOKE
340 mtspr(SPRN_DBCR2, thread->dbcr2);
341#endif
342}
343/*
344 * Unless neither the old or new thread are making use of the
345 * debug registers, set the debug registers from the values
346 * stored in the new thread.
347 */
348static void switch_booke_debug_regs(struct thread_struct *new_thread)
349{
350 if ((current->thread.dbcr0 & DBCR0_IDM)
351 || (new_thread->dbcr0 & DBCR0_IDM))
352 prime_debug_regs(new_thread);
353}
354#else /* !CONFIG_PPC_ADV_DEBUG_REGS */
355static void set_debug_reg_defaults(struct thread_struct *thread)
356{
357 if (thread->dabr) {
358 thread->dabr = 0;
359 set_dabr(0);
360 }
361}
362#endif /* CONFIG_PPC_ADV_DEBUG_REGS */
363
279int set_dabr(unsigned long dabr) 364int set_dabr(unsigned long dabr)
280{ 365{
281 __get_cpu_var(current_dabr) = dabr; 366 __get_cpu_var(current_dabr) = dabr;
@@ -284,7 +369,7 @@ int set_dabr(unsigned long dabr)
284 return ppc_md.set_dabr(dabr); 369 return ppc_md.set_dabr(dabr);
285 370
286 /* XXX should we have a CPU_FTR_HAS_DABR ? */ 371 /* XXX should we have a CPU_FTR_HAS_DABR ? */
287#if defined(CONFIG_BOOKE) 372#ifdef CONFIG_PPC_ADV_DEBUG_REGS
288 mtspr(SPRN_DAC1, dabr); 373 mtspr(SPRN_DAC1, dabr);
289#elif defined(CONFIG_PPC_BOOK3S) 374#elif defined(CONFIG_PPC_BOOK3S)
290 mtspr(SPRN_DABR, dabr); 375 mtspr(SPRN_DABR, dabr);
@@ -371,10 +456,8 @@ struct task_struct *__switch_to(struct task_struct *prev,
371 456
372#endif /* CONFIG_SMP */ 457#endif /* CONFIG_SMP */
373 458
374#if defined(CONFIG_BOOKE) 459#ifdef CONFIG_PPC_ADV_DEBUG_REGS
375 /* If new thread DAC (HW breakpoint) is the same then leave it */ 460 switch_booke_debug_regs(&new->thread);
376 if (new->thread.dabr)
377 set_dabr(new->thread.dabr);
378#else 461#else
379 if (unlikely(__get_cpu_var(current_dabr) != new->thread.dabr)) 462 if (unlikely(__get_cpu_var(current_dabr) != new->thread.dabr))
380 set_dabr(new->thread.dabr); 463 set_dabr(new->thread.dabr);
@@ -514,7 +597,7 @@ void show_regs(struct pt_regs * regs)
514 printk(" CR: %08lx XER: %08lx\n", regs->ccr, regs->xer); 597 printk(" CR: %08lx XER: %08lx\n", regs->ccr, regs->xer);
515 trap = TRAP(regs); 598 trap = TRAP(regs);
516 if (trap == 0x300 || trap == 0x600) 599 if (trap == 0x300 || trap == 0x600)
517#if defined(CONFIG_4xx) || defined(CONFIG_BOOKE) 600#ifdef CONFIG_PPC_ADV_DEBUG_REGS
518 printk("DEAR: "REG", ESR: "REG"\n", regs->dar, regs->dsisr); 601 printk("DEAR: "REG", ESR: "REG"\n", regs->dar, regs->dsisr);
519#else 602#else
520 printk("DAR: "REG", DSISR: "REG"\n", regs->dar, regs->dsisr); 603 printk("DAR: "REG", DSISR: "REG"\n", regs->dar, regs->dsisr);
@@ -554,28 +637,9 @@ void exit_thread(void)
554 637
555void flush_thread(void) 638void flush_thread(void)
556{ 639{
557#ifdef CONFIG_PPC64
558 struct thread_info *t = current_thread_info();
559
560 if (test_ti_thread_flag(t, TIF_ABI_PENDING)) {
561 clear_ti_thread_flag(t, TIF_ABI_PENDING);
562 if (test_ti_thread_flag(t, TIF_32BIT))
563 clear_ti_thread_flag(t, TIF_32BIT);
564 else
565 set_ti_thread_flag(t, TIF_32BIT);
566 }
567#endif
568
569 discard_lazy_cpu_state(); 640 discard_lazy_cpu_state();
570 641
571 if (current->thread.dabr) { 642 set_debug_reg_defaults(&current->thread);
572 current->thread.dabr = 0;
573 set_dabr(0);
574
575#if defined(CONFIG_BOOKE)
576 current->thread.dbcr0 &= ~(DBSR_DAC1R | DBSR_DAC1W);
577#endif
578 }
579} 643}
580 644
581void 645void
diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index d4405b95bfaa..05131d634e73 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c
@@ -43,6 +43,7 @@
43#include <asm/smp.h> 43#include <asm/smp.h>
44#include <asm/system.h> 44#include <asm/system.h>
45#include <asm/mmu.h> 45#include <asm/mmu.h>
46#include <asm/paca.h>
46#include <asm/pgtable.h> 47#include <asm/pgtable.h>
47#include <asm/pci.h> 48#include <asm/pci.h>
48#include <asm/iommu.h> 49#include <asm/iommu.h>
@@ -61,365 +62,12 @@
61#define DBG(fmt...) 62#define DBG(fmt...)
62#endif 63#endif
63 64
64
65static int __initdata dt_root_addr_cells;
66static int __initdata dt_root_size_cells;
67
68#ifdef CONFIG_PPC64 65#ifdef CONFIG_PPC64
69int __initdata iommu_is_off; 66int __initdata iommu_is_off;
70int __initdata iommu_force_on; 67int __initdata iommu_force_on;
71unsigned long tce_alloc_start, tce_alloc_end; 68unsigned long tce_alloc_start, tce_alloc_end;
72#endif 69#endif
73 70
74typedef u32 cell_t;
75
76#if 0
77static struct boot_param_header *initial_boot_params __initdata;
78#else
79struct boot_param_header *initial_boot_params;
80#endif
81
82extern struct device_node *allnodes; /* temporary while merging */
83
84extern rwlock_t devtree_lock; /* temporary while merging */
85
86/* export that to outside world */
87struct device_node *of_chosen;
88
89static inline char *find_flat_dt_string(u32 offset)
90{
91 return ((char *)initial_boot_params) +
92 initial_boot_params->off_dt_strings + offset;
93}
94
95/**
96 * This function is used to scan the flattened device-tree, it is
97 * used to extract the memory informations at boot before we can
98 * unflatten the tree
99 */
100int __init of_scan_flat_dt(int (*it)(unsigned long node,
101 const char *uname, int depth,
102 void *data),
103 void *data)
104{
105 unsigned long p = ((unsigned long)initial_boot_params) +
106 initial_boot_params->off_dt_struct;
107 int rc = 0;
108 int depth = -1;
109
110 do {
111 u32 tag = *((u32 *)p);
112 char *pathp;
113
114 p += 4;
115 if (tag == OF_DT_END_NODE) {
116 depth --;
117 continue;
118 }
119 if (tag == OF_DT_NOP)
120 continue;
121 if (tag == OF_DT_END)
122 break;
123 if (tag == OF_DT_PROP) {
124 u32 sz = *((u32 *)p);
125 p += 8;
126 if (initial_boot_params->version < 0x10)
127 p = _ALIGN(p, sz >= 8 ? 8 : 4);
128 p += sz;
129 p = _ALIGN(p, 4);
130 continue;
131 }
132 if (tag != OF_DT_BEGIN_NODE) {
133 printk(KERN_WARNING "Invalid tag %x scanning flattened"
134 " device tree !\n", tag);
135 return -EINVAL;
136 }
137 depth++;
138 pathp = (char *)p;
139 p = _ALIGN(p + strlen(pathp) + 1, 4);
140 if ((*pathp) == '/') {
141 char *lp, *np;
142 for (lp = NULL, np = pathp; *np; np++)
143 if ((*np) == '/')
144 lp = np+1;
145 if (lp != NULL)
146 pathp = lp;
147 }
148 rc = it(p, pathp, depth, data);
149 if (rc != 0)
150 break;
151 } while(1);
152
153 return rc;
154}
155
156unsigned long __init of_get_flat_dt_root(void)
157{
158 unsigned long p = ((unsigned long)initial_boot_params) +
159 initial_boot_params->off_dt_struct;
160
161 while(*((u32 *)p) == OF_DT_NOP)
162 p += 4;
163 BUG_ON (*((u32 *)p) != OF_DT_BEGIN_NODE);
164 p += 4;
165 return _ALIGN(p + strlen((char *)p) + 1, 4);
166}
167
168/**
169 * This function can be used within scan_flattened_dt callback to get
170 * access to properties
171 */
172void* __init of_get_flat_dt_prop(unsigned long node, const char *name,
173 unsigned long *size)
174{
175 unsigned long p = node;
176
177 do {
178 u32 tag = *((u32 *)p);
179 u32 sz, noff;
180 const char *nstr;
181
182 p += 4;
183 if (tag == OF_DT_NOP)
184 continue;
185 if (tag != OF_DT_PROP)
186 return NULL;
187
188 sz = *((u32 *)p);
189 noff = *((u32 *)(p + 4));
190 p += 8;
191 if (initial_boot_params->version < 0x10)
192 p = _ALIGN(p, sz >= 8 ? 8 : 4);
193
194 nstr = find_flat_dt_string(noff);
195 if (nstr == NULL) {
196 printk(KERN_WARNING "Can't find property index"
197 " name !\n");
198 return NULL;
199 }
200 if (strcmp(name, nstr) == 0) {
201 if (size)
202 *size = sz;
203 return (void *)p;
204 }
205 p += sz;
206 p = _ALIGN(p, 4);
207 } while(1);
208}
209
210int __init of_flat_dt_is_compatible(unsigned long node, const char *compat)
211{
212 const char* cp;
213 unsigned long cplen, l;
214
215 cp = of_get_flat_dt_prop(node, "compatible", &cplen);
216 if (cp == NULL)
217 return 0;
218 while (cplen > 0) {
219 if (strncasecmp(cp, compat, strlen(compat)) == 0)
220 return 1;
221 l = strlen(cp) + 1;
222 cp += l;
223 cplen -= l;
224 }
225
226 return 0;
227}
228
229static void *__init unflatten_dt_alloc(unsigned long *mem, unsigned long size,
230 unsigned long align)
231{
232 void *res;
233
234 *mem = _ALIGN(*mem, align);
235 res = (void *)*mem;
236 *mem += size;
237
238 return res;
239}
240
241static unsigned long __init unflatten_dt_node(unsigned long mem,
242 unsigned long *p,
243 struct device_node *dad,
244 struct device_node ***allnextpp,
245 unsigned long fpsize)
246{
247 struct device_node *np;
248 struct property *pp, **prev_pp = NULL;
249 char *pathp;
250 u32 tag;
251 unsigned int l, allocl;
252 int has_name = 0;
253 int new_format = 0;
254
255 tag = *((u32 *)(*p));
256 if (tag != OF_DT_BEGIN_NODE) {
257 printk("Weird tag at start of node: %x\n", tag);
258 return mem;
259 }
260 *p += 4;
261 pathp = (char *)*p;
262 l = allocl = strlen(pathp) + 1;
263 *p = _ALIGN(*p + l, 4);
264
265 /* version 0x10 has a more compact unit name here instead of the full
266 * path. we accumulate the full path size using "fpsize", we'll rebuild
267 * it later. We detect this because the first character of the name is
268 * not '/'.
269 */
270 if ((*pathp) != '/') {
271 new_format = 1;
272 if (fpsize == 0) {
273 /* root node: special case. fpsize accounts for path
274 * plus terminating zero. root node only has '/', so
275 * fpsize should be 2, but we want to avoid the first
276 * level nodes to have two '/' so we use fpsize 1 here
277 */
278 fpsize = 1;
279 allocl = 2;
280 } else {
281 /* account for '/' and path size minus terminal 0
282 * already in 'l'
283 */
284 fpsize += l;
285 allocl = fpsize;
286 }
287 }
288
289
290 np = unflatten_dt_alloc(&mem, sizeof(struct device_node) + allocl,
291 __alignof__(struct device_node));
292 if (allnextpp) {
293 memset(np, 0, sizeof(*np));
294 np->full_name = ((char*)np) + sizeof(struct device_node);
295 if (new_format) {
296 char *p = np->full_name;
297 /* rebuild full path for new format */
298 if (dad && dad->parent) {
299 strcpy(p, dad->full_name);
300#ifdef DEBUG
301 if ((strlen(p) + l + 1) != allocl) {
302 DBG("%s: p: %d, l: %d, a: %d\n",
303 pathp, (int)strlen(p), l, allocl);
304 }
305#endif
306 p += strlen(p);
307 }
308 *(p++) = '/';
309 memcpy(p, pathp, l);
310 } else
311 memcpy(np->full_name, pathp, l);
312 prev_pp = &np->properties;
313 **allnextpp = np;
314 *allnextpp = &np->allnext;
315 if (dad != NULL) {
316 np->parent = dad;
317 /* we temporarily use the next field as `last_child'*/
318 if (dad->next == 0)
319 dad->child = np;
320 else
321 dad->next->sibling = np;
322 dad->next = np;
323 }
324 kref_init(&np->kref);
325 }
326 while(1) {
327 u32 sz, noff;
328 char *pname;
329
330 tag = *((u32 *)(*p));
331 if (tag == OF_DT_NOP) {
332 *p += 4;
333 continue;
334 }
335 if (tag != OF_DT_PROP)
336 break;
337 *p += 4;
338 sz = *((u32 *)(*p));
339 noff = *((u32 *)((*p) + 4));
340 *p += 8;
341 if (initial_boot_params->version < 0x10)
342 *p = _ALIGN(*p, sz >= 8 ? 8 : 4);
343
344 pname = find_flat_dt_string(noff);
345 if (pname == NULL) {
346 printk("Can't find property name in list !\n");
347 break;
348 }
349 if (strcmp(pname, "name") == 0)
350 has_name = 1;
351 l = strlen(pname) + 1;
352 pp = unflatten_dt_alloc(&mem, sizeof(struct property),
353 __alignof__(struct property));
354 if (allnextpp) {
355 if (strcmp(pname, "linux,phandle") == 0) {
356 np->node = *((u32 *)*p);
357 if (np->linux_phandle == 0)
358 np->linux_phandle = np->node;
359 }
360 if (strcmp(pname, "ibm,phandle") == 0)
361 np->linux_phandle = *((u32 *)*p);
362 pp->name = pname;
363 pp->length = sz;
364 pp->value = (void *)*p;
365 *prev_pp = pp;
366 prev_pp = &pp->next;
367 }
368 *p = _ALIGN((*p) + sz, 4);
369 }
370 /* with version 0x10 we may not have the name property, recreate
371 * it here from the unit name if absent
372 */
373 if (!has_name) {
374 char *p = pathp, *ps = pathp, *pa = NULL;
375 int sz;
376
377 while (*p) {
378 if ((*p) == '@')
379 pa = p;
380 if ((*p) == '/')
381 ps = p + 1;
382 p++;
383 }
384 if (pa < ps)
385 pa = p;
386 sz = (pa - ps) + 1;
387 pp = unflatten_dt_alloc(&mem, sizeof(struct property) + sz,
388 __alignof__(struct property));
389 if (allnextpp) {
390 pp->name = "name";
391 pp->length = sz;
392 pp->value = pp + 1;
393 *prev_pp = pp;
394 prev_pp = &pp->next;
395 memcpy(pp->value, ps, sz - 1);
396 ((char *)pp->value)[sz - 1] = 0;
397 DBG("fixed up name for %s -> %s\n", pathp,
398 (char *)pp->value);
399 }
400 }
401 if (allnextpp) {
402 *prev_pp = NULL;
403 np->name = of_get_property(np, "name", NULL);
404 np->type = of_get_property(np, "device_type", NULL);
405
406 if (!np->name)
407 np->name = "<NULL>";
408 if (!np->type)
409 np->type = "<NULL>";
410 }
411 while (tag == OF_DT_BEGIN_NODE) {
412 mem = unflatten_dt_node(mem, p, np, allnextpp, fpsize);
413 tag = *((u32 *)(*p));
414 }
415 if (tag != OF_DT_END_NODE) {
416 printk("Weird tag at end of node: %x\n", tag);
417 return mem;
418 }
419 *p += 4;
420 return mem;
421}
422
423static int __init early_parse_mem(char *p) 71static int __init early_parse_mem(char *p)
424{ 72{
425 if (!p) 73 if (!p)
@@ -446,7 +94,7 @@ static void __init move_device_tree(void)
446 DBG("-> move_device_tree\n"); 94 DBG("-> move_device_tree\n");
447 95
448 start = __pa(initial_boot_params); 96 start = __pa(initial_boot_params);
449 size = initial_boot_params->totalsize; 97 size = be32_to_cpu(initial_boot_params->totalsize);
450 98
451 if ((memory_limit && (start + size) > memory_limit) || 99 if ((memory_limit && (start + size) > memory_limit) ||
452 overlaps_crashkernel(start, size)) { 100 overlaps_crashkernel(start, size)) {
@@ -459,54 +107,6 @@ static void __init move_device_tree(void)
459 DBG("<- move_device_tree\n"); 107 DBG("<- move_device_tree\n");
460} 108}
461 109
462/**
463 * unflattens the device-tree passed by the firmware, creating the
464 * tree of struct device_node. It also fills the "name" and "type"
465 * pointers of the nodes so the normal device-tree walking functions
466 * can be used (this used to be done by finish_device_tree)
467 */
468void __init unflatten_device_tree(void)
469{
470 unsigned long start, mem, size;
471 struct device_node **allnextp = &allnodes;
472
473 DBG(" -> unflatten_device_tree()\n");
474
475 /* First pass, scan for size */
476 start = ((unsigned long)initial_boot_params) +
477 initial_boot_params->off_dt_struct;
478 size = unflatten_dt_node(0, &start, NULL, NULL, 0);
479 size = (size | 3) + 1;
480
481 DBG(" size is %lx, allocating...\n", size);
482
483 /* Allocate memory for the expanded device tree */
484 mem = lmb_alloc(size + 4, __alignof__(struct device_node));
485 mem = (unsigned long) __va(mem);
486
487 ((u32 *)mem)[size / 4] = 0xdeadbeef;
488
489 DBG(" unflattening %lx...\n", mem);
490
491 /* Second pass, do actual unflattening */
492 start = ((unsigned long)initial_boot_params) +
493 initial_boot_params->off_dt_struct;
494 unflatten_dt_node(mem, &start, NULL, &allnextp, 0);
495 if (*((u32 *)start) != OF_DT_END)
496 printk(KERN_WARNING "Weird tag at end of tree: %08x\n", *((u32 *)start));
497 if (((u32 *)mem)[size / 4] != 0xdeadbeef)
498 printk(KERN_WARNING "End of tree marker overwritten: %08x\n",
499 ((u32 *)mem)[size / 4] );
500 *allnextp = NULL;
501
502 /* Get pointer to OF "/chosen" node for use everywhere */
503 of_chosen = of_find_node_by_path("/chosen");
504 if (of_chosen == NULL)
505 of_chosen = of_find_node_by_path("/chosen@0");
506
507 DBG(" <- unflatten_device_tree()\n");
508}
509
510/* 110/*
511 * ibm,pa-features is a per-cpu property that contains a string of 111 * ibm,pa-features is a per-cpu property that contains a string of
512 * attribute descriptors, each of which has a 2 byte header plus up 112 * attribute descriptors, each of which has a 2 byte header plus up
@@ -763,48 +363,9 @@ static int __init early_init_dt_scan_cpus(unsigned long node,
763 return 0; 363 return 0;
764} 364}
765 365
766#ifdef CONFIG_BLK_DEV_INITRD 366void __init early_init_dt_scan_chosen_arch(unsigned long node)
767static void __init early_init_dt_check_for_initrd(unsigned long node)
768{
769 unsigned long l;
770 u32 *prop;
771
772 DBG("Looking for initrd properties... ");
773
774 prop = of_get_flat_dt_prop(node, "linux,initrd-start", &l);
775 if (prop) {
776 initrd_start = (unsigned long)__va(of_read_ulong(prop, l/4));
777
778 prop = of_get_flat_dt_prop(node, "linux,initrd-end", &l);
779 if (prop) {
780 initrd_end = (unsigned long)
781 __va(of_read_ulong(prop, l/4));
782 initrd_below_start_ok = 1;
783 } else {
784 initrd_start = 0;
785 }
786 }
787
788 DBG("initrd_start=0x%lx initrd_end=0x%lx\n", initrd_start, initrd_end);
789}
790#else
791static inline void early_init_dt_check_for_initrd(unsigned long node)
792{
793}
794#endif /* CONFIG_BLK_DEV_INITRD */
795
796static int __init early_init_dt_scan_chosen(unsigned long node,
797 const char *uname, int depth, void *data)
798{ 367{
799 unsigned long *lprop; 368 unsigned long *lprop;
800 unsigned long l;
801 char *p;
802
803 DBG("search \"chosen\", depth: %d, uname: %s\n", depth, uname);
804
805 if (depth != 1 ||
806 (strcmp(uname, "chosen") != 0 && strcmp(uname, "chosen@0") != 0))
807 return 0;
808 369
809#ifdef CONFIG_PPC64 370#ifdef CONFIG_PPC64
810 /* check if iommu is forced on or off */ 371 /* check if iommu is forced on or off */
@@ -815,17 +376,17 @@ static int __init early_init_dt_scan_chosen(unsigned long node,
815#endif 376#endif
816 377
817 /* mem=x on the command line is the preferred mechanism */ 378 /* mem=x on the command line is the preferred mechanism */
818 lprop = of_get_flat_dt_prop(node, "linux,memory-limit", NULL); 379 lprop = of_get_flat_dt_prop(node, "linux,memory-limit", NULL);
819 if (lprop) 380 if (lprop)
820 memory_limit = *lprop; 381 memory_limit = *lprop;
821 382
822#ifdef CONFIG_PPC64 383#ifdef CONFIG_PPC64
823 lprop = of_get_flat_dt_prop(node, "linux,tce-alloc-start", NULL); 384 lprop = of_get_flat_dt_prop(node, "linux,tce-alloc-start", NULL);
824 if (lprop) 385 if (lprop)
825 tce_alloc_start = *lprop; 386 tce_alloc_start = *lprop;
826 lprop = of_get_flat_dt_prop(node, "linux,tce-alloc-end", NULL); 387 lprop = of_get_flat_dt_prop(node, "linux,tce-alloc-end", NULL);
827 if (lprop) 388 if (lprop)
828 tce_alloc_end = *lprop; 389 tce_alloc_end = *lprop;
829#endif 390#endif
830 391
831#ifdef CONFIG_KEXEC 392#ifdef CONFIG_KEXEC
@@ -837,51 +398,6 @@ static int __init early_init_dt_scan_chosen(unsigned long node,
837 if (lprop) 398 if (lprop)
838 crashk_res.end = crashk_res.start + *lprop - 1; 399 crashk_res.end = crashk_res.start + *lprop - 1;
839#endif 400#endif
840
841 early_init_dt_check_for_initrd(node);
842
843 /* Retreive command line */
844 p = of_get_flat_dt_prop(node, "bootargs", &l);
845 if (p != NULL && l > 0)
846 strlcpy(cmd_line, p, min((int)l, COMMAND_LINE_SIZE));
847
848#ifdef CONFIG_CMDLINE
849 if (p == NULL || l == 0 || (l == 1 && (*p) == 0))
850 strlcpy(cmd_line, CONFIG_CMDLINE, COMMAND_LINE_SIZE);
851#endif /* CONFIG_CMDLINE */
852
853 DBG("Command line is: %s\n", cmd_line);
854
855 /* break now */
856 return 1;
857}
858
859static int __init early_init_dt_scan_root(unsigned long node,
860 const char *uname, int depth, void *data)
861{
862 u32 *prop;
863
864 if (depth != 0)
865 return 0;
866
867 prop = of_get_flat_dt_prop(node, "#size-cells", NULL);
868 dt_root_size_cells = (prop == NULL) ? 1 : *prop;
869 DBG("dt_root_size_cells = %x\n", dt_root_size_cells);
870
871 prop = of_get_flat_dt_prop(node, "#address-cells", NULL);
872 dt_root_addr_cells = (prop == NULL) ? 2 : *prop;
873 DBG("dt_root_addr_cells = %x\n", dt_root_addr_cells);
874
875 /* break now */
876 return 1;
877}
878
879static u64 __init dt_mem_next_cell(int s, cell_t **cellp)
880{
881 cell_t *p = *cellp;
882
883 *cellp = p + s;
884 return of_read_number(p, s);
885} 401}
886 402
887#ifdef CONFIG_PPC_PSERIES 403#ifdef CONFIG_PPC_PSERIES
@@ -893,22 +409,22 @@ static u64 __init dt_mem_next_cell(int s, cell_t **cellp)
893 */ 409 */
894static int __init early_init_dt_scan_drconf_memory(unsigned long node) 410static int __init early_init_dt_scan_drconf_memory(unsigned long node)
895{ 411{
896 cell_t *dm, *ls, *usm; 412 __be32 *dm, *ls, *usm;
897 unsigned long l, n, flags; 413 unsigned long l, n, flags;
898 u64 base, size, lmb_size; 414 u64 base, size, lmb_size;
899 unsigned int is_kexec_kdump = 0, rngs; 415 unsigned int is_kexec_kdump = 0, rngs;
900 416
901 ls = of_get_flat_dt_prop(node, "ibm,lmb-size", &l); 417 ls = of_get_flat_dt_prop(node, "ibm,lmb-size", &l);
902 if (ls == NULL || l < dt_root_size_cells * sizeof(cell_t)) 418 if (ls == NULL || l < dt_root_size_cells * sizeof(__be32))
903 return 0; 419 return 0;
904 lmb_size = dt_mem_next_cell(dt_root_size_cells, &ls); 420 lmb_size = dt_mem_next_cell(dt_root_size_cells, &ls);
905 421
906 dm = of_get_flat_dt_prop(node, "ibm,dynamic-memory", &l); 422 dm = of_get_flat_dt_prop(node, "ibm,dynamic-memory", &l);
907 if (dm == NULL || l < sizeof(cell_t)) 423 if (dm == NULL || l < sizeof(__be32))
908 return 0; 424 return 0;
909 425
910 n = *dm++; /* number of entries */ 426 n = *dm++; /* number of entries */
911 if (l < (n * (dt_root_addr_cells + 4) + 1) * sizeof(cell_t)) 427 if (l < (n * (dt_root_addr_cells + 4) + 1) * sizeof(__be32))
912 return 0; 428 return 0;
913 429
914 /* check if this is a kexec/kdump kernel. */ 430 /* check if this is a kexec/kdump kernel. */
@@ -963,65 +479,47 @@ static int __init early_init_dt_scan_drconf_memory(unsigned long node)
963#define early_init_dt_scan_drconf_memory(node) 0 479#define early_init_dt_scan_drconf_memory(node) 0
964#endif /* CONFIG_PPC_PSERIES */ 480#endif /* CONFIG_PPC_PSERIES */
965 481
966static int __init early_init_dt_scan_memory(unsigned long node, 482static int __init early_init_dt_scan_memory_ppc(unsigned long node,
967 const char *uname, int depth, void *data) 483 const char *uname,
484 int depth, void *data)
968{ 485{
969 char *type = of_get_flat_dt_prop(node, "device_type", NULL);
970 cell_t *reg, *endp;
971 unsigned long l;
972
973 /* Look for the ibm,dynamic-reconfiguration-memory node */
974 if (depth == 1 && 486 if (depth == 1 &&
975 strcmp(uname, "ibm,dynamic-reconfiguration-memory") == 0) 487 strcmp(uname, "ibm,dynamic-reconfiguration-memory") == 0)
976 return early_init_dt_scan_drconf_memory(node); 488 return early_init_dt_scan_drconf_memory(node);
489
490 return early_init_dt_scan_memory(node, uname, depth, data);
491}
977 492
978 /* We are scanning "memory" nodes only */ 493void __init early_init_dt_add_memory_arch(u64 base, u64 size)
979 if (type == NULL) { 494{
980 /* 495#if defined(CONFIG_PPC64)
981 * The longtrail doesn't have a device_type on the 496 if (iommu_is_off) {
982 * /memory node, so look for the node called /memory@0. 497 if (base >= 0x80000000ul)
983 */ 498 return;
984 if (depth != 1 || strcmp(uname, "memory@0") != 0) 499 if ((base + size) > 0x80000000ul)
985 return 0; 500 size = 0x80000000ul - base;
986 } else if (strcmp(type, "memory") != 0) 501 }
987 return 0; 502#endif
988
989 reg = of_get_flat_dt_prop(node, "linux,usable-memory", &l);
990 if (reg == NULL)
991 reg = of_get_flat_dt_prop(node, "reg", &l);
992 if (reg == NULL)
993 return 0;
994
995 endp = reg + (l / sizeof(cell_t));
996
997 DBG("memory scan node %s, reg size %ld, data: %x %x %x %x,\n",
998 uname, l, reg[0], reg[1], reg[2], reg[3]);
999
1000 while ((endp - reg) >= (dt_root_addr_cells + dt_root_size_cells)) {
1001 u64 base, size;
1002 503
1003 base = dt_mem_next_cell(dt_root_addr_cells, &reg); 504 lmb_add(base, size);
1004 size = dt_mem_next_cell(dt_root_size_cells, &reg);
1005 505
1006 if (size == 0) 506 memstart_addr = min((u64)memstart_addr, base);
1007 continue; 507}
1008 DBG(" - %llx , %llx\n", (unsigned long long)base,
1009 (unsigned long long)size);
1010#ifdef CONFIG_PPC64
1011 if (iommu_is_off) {
1012 if (base >= 0x80000000ul)
1013 continue;
1014 if ((base + size) > 0x80000000ul)
1015 size = 0x80000000ul - base;
1016 }
1017#endif
1018 lmb_add(base, size);
1019 508
1020 memstart_addr = min((u64)memstart_addr, base); 509u64 __init early_init_dt_alloc_memory_arch(u64 size, u64 align)
1021 } 510{
511 return lmb_alloc(size, align);
512}
1022 513
1023 return 0; 514#ifdef CONFIG_BLK_DEV_INITRD
515void __init early_init_dt_setup_initrd_arch(unsigned long start,
516 unsigned long end)
517{
518 initrd_start = (unsigned long)__va(start);
519 initrd_end = (unsigned long)__va(end);
520 initrd_below_start_ok = 1;
1024} 521}
522#endif
1025 523
1026static void __init early_reserve_mem(void) 524static void __init early_reserve_mem(void)
1027{ 525{
@@ -1186,7 +684,7 @@ void __init early_init_devtree(void *params)
1186 /* Scan memory nodes and rebuild LMBs */ 684 /* Scan memory nodes and rebuild LMBs */
1187 lmb_init(); 685 lmb_init();
1188 of_scan_flat_dt(early_init_dt_scan_root, NULL); 686 of_scan_flat_dt(early_init_dt_scan_root, NULL);
1189 of_scan_flat_dt(early_init_dt_scan_memory, NULL); 687 of_scan_flat_dt(early_init_dt_scan_memory_ppc, NULL);
1190 688
1191 /* Save command line for /proc/cmdline and then parse parameters */ 689 /* Save command line for /proc/cmdline and then parse parameters */
1192 strlcpy(boot_command_line, cmd_line, COMMAND_LINE_SIZE); 690 strlcpy(boot_command_line, cmd_line, COMMAND_LINE_SIZE);
@@ -1224,6 +722,8 @@ void __init early_init_devtree(void *params)
1224 * FIXME .. and the initrd too? */ 722 * FIXME .. and the initrd too? */
1225 move_device_tree(); 723 move_device_tree();
1226 724
725 allocate_pacas();
726
1227 DBG("Scanning CPUs ...\n"); 727 DBG("Scanning CPUs ...\n");
1228 728
1229 /* Retreive CPU related informations from the flat tree 729 /* Retreive CPU related informations from the flat tree
@@ -1234,25 +734,6 @@ void __init early_init_devtree(void *params)
1234 DBG(" <- early_init_devtree()\n"); 734 DBG(" <- early_init_devtree()\n");
1235} 735}
1236 736
1237
1238/**
1239 * Indicates whether the root node has a given value in its
1240 * compatible property.
1241 */
1242int machine_is_compatible(const char *compat)
1243{
1244 struct device_node *root;
1245 int rc = 0;
1246
1247 root = of_find_node_by_path("/");
1248 if (root) {
1249 rc = of_device_is_compatible(root, compat);
1250 of_node_put(root);
1251 }
1252 return rc;
1253}
1254EXPORT_SYMBOL(machine_is_compatible);
1255
1256/******* 737/*******
1257 * 738 *
1258 * New implementation of the OF "find" APIs, return a refcounted 739 * New implementation of the OF "find" APIs, return a refcounted
@@ -1265,27 +746,6 @@ EXPORT_SYMBOL(machine_is_compatible);
1265 *******/ 746 *******/
1266 747
1267/** 748/**
1268 * of_find_node_by_phandle - Find a node given a phandle
1269 * @handle: phandle of the node to find
1270 *
1271 * Returns a node pointer with refcount incremented, use
1272 * of_node_put() on it when done.
1273 */
1274struct device_node *of_find_node_by_phandle(phandle handle)
1275{
1276 struct device_node *np;
1277
1278 read_lock(&devtree_lock);
1279 for (np = allnodes; np != 0; np = np->allnext)
1280 if (np->linux_phandle == handle)
1281 break;
1282 of_node_get(np);
1283 read_unlock(&devtree_lock);
1284 return np;
1285}
1286EXPORT_SYMBOL(of_find_node_by_phandle);
1287
1288/**
1289 * of_find_next_cache_node - Find a node's subsidiary cache 749 * of_find_next_cache_node - Find a node's subsidiary cache
1290 * @np: node of type "cpu" or "cache" 750 * @np: node of type "cpu" or "cache"
1291 * 751 *
@@ -1316,161 +776,6 @@ struct device_node *of_find_next_cache_node(struct device_node *np)
1316 return NULL; 776 return NULL;
1317} 777}
1318 778
1319/**
1320 * of_find_all_nodes - Get next node in global list
1321 * @prev: Previous node or NULL to start iteration
1322 * of_node_put() will be called on it
1323 *
1324 * Returns a node pointer with refcount incremented, use
1325 * of_node_put() on it when done.
1326 */
1327struct device_node *of_find_all_nodes(struct device_node *prev)
1328{
1329 struct device_node *np;
1330
1331 read_lock(&devtree_lock);
1332 np = prev ? prev->allnext : allnodes;
1333 for (; np != 0; np = np->allnext)
1334 if (of_node_get(np))
1335 break;
1336 of_node_put(prev);
1337 read_unlock(&devtree_lock);
1338 return np;
1339}
1340EXPORT_SYMBOL(of_find_all_nodes);
1341
1342/**
1343 * of_node_get - Increment refcount of a node
1344 * @node: Node to inc refcount, NULL is supported to
1345 * simplify writing of callers
1346 *
1347 * Returns node.
1348 */
1349struct device_node *of_node_get(struct device_node *node)
1350{
1351 if (node)
1352 kref_get(&node->kref);
1353 return node;
1354}
1355EXPORT_SYMBOL(of_node_get);
1356
1357static inline struct device_node * kref_to_device_node(struct kref *kref)
1358{
1359 return container_of(kref, struct device_node, kref);
1360}
1361
1362/**
1363 * of_node_release - release a dynamically allocated node
1364 * @kref: kref element of the node to be released
1365 *
1366 * In of_node_put() this function is passed to kref_put()
1367 * as the destructor.
1368 */
1369static void of_node_release(struct kref *kref)
1370{
1371 struct device_node *node = kref_to_device_node(kref);
1372 struct property *prop = node->properties;
1373
1374 /* We should never be releasing nodes that haven't been detached. */
1375 if (!of_node_check_flag(node, OF_DETACHED)) {
1376 printk("WARNING: Bad of_node_put() on %s\n", node->full_name);
1377 dump_stack();
1378 kref_init(&node->kref);
1379 return;
1380 }
1381
1382 if (!of_node_check_flag(node, OF_DYNAMIC))
1383 return;
1384
1385 while (prop) {
1386 struct property *next = prop->next;
1387 kfree(prop->name);
1388 kfree(prop->value);
1389 kfree(prop);
1390 prop = next;
1391
1392 if (!prop) {
1393 prop = node->deadprops;
1394 node->deadprops = NULL;
1395 }
1396 }
1397 kfree(node->full_name);
1398 kfree(node->data);
1399 kfree(node);
1400}
1401
1402/**
1403 * of_node_put - Decrement refcount of a node
1404 * @node: Node to dec refcount, NULL is supported to
1405 * simplify writing of callers
1406 *
1407 */
1408void of_node_put(struct device_node *node)
1409{
1410 if (node)
1411 kref_put(&node->kref, of_node_release);
1412}
1413EXPORT_SYMBOL(of_node_put);
1414
1415/*
1416 * Plug a device node into the tree and global list.
1417 */
1418void of_attach_node(struct device_node *np)
1419{
1420 unsigned long flags;
1421
1422 write_lock_irqsave(&devtree_lock, flags);
1423 np->sibling = np->parent->child;
1424 np->allnext = allnodes;
1425 np->parent->child = np;
1426 allnodes = np;
1427 write_unlock_irqrestore(&devtree_lock, flags);
1428}
1429
1430/*
1431 * "Unplug" a node from the device tree. The caller must hold
1432 * a reference to the node. The memory associated with the node
1433 * is not freed until its refcount goes to zero.
1434 */
1435void of_detach_node(struct device_node *np)
1436{
1437 struct device_node *parent;
1438 unsigned long flags;
1439
1440 write_lock_irqsave(&devtree_lock, flags);
1441
1442 parent = np->parent;
1443 if (!parent)
1444 goto out_unlock;
1445
1446 if (allnodes == np)
1447 allnodes = np->allnext;
1448 else {
1449 struct device_node *prev;
1450 for (prev = allnodes;
1451 prev->allnext != np;
1452 prev = prev->allnext)
1453 ;
1454 prev->allnext = np->allnext;
1455 }
1456
1457 if (parent->child == np)
1458 parent->child = np->sibling;
1459 else {
1460 struct device_node *prevsib;
1461 for (prevsib = np->parent->child;
1462 prevsib->sibling != np;
1463 prevsib = prevsib->sibling)
1464 ;
1465 prevsib->sibling = np->sibling;
1466 }
1467
1468 of_node_set_flag(np, OF_DETACHED);
1469
1470out_unlock:
1471 write_unlock_irqrestore(&devtree_lock, flags);
1472}
1473
1474#ifdef CONFIG_PPC_PSERIES 779#ifdef CONFIG_PPC_PSERIES
1475/* 780/*
1476 * Fix up the uninitialized fields in a new device node: 781 * Fix up the uninitialized fields in a new device node:
@@ -1502,9 +807,9 @@ static int of_finish_dynamic_node(struct device_node *node)
1502 if (machine_is(powermac)) 807 if (machine_is(powermac))
1503 return -ENODEV; 808 return -ENODEV;
1504 809
1505 /* fix up new node's linux_phandle field */ 810 /* fix up new node's phandle field */
1506 if ((ibm_phandle = of_get_property(node, "ibm,phandle", NULL))) 811 if ((ibm_phandle = of_get_property(node, "ibm,phandle", NULL)))
1507 node->linux_phandle = *ibm_phandle; 812 node->phandle = *ibm_phandle;
1508 813
1509out: 814out:
1510 of_node_put(parent); 815 of_node_put(parent);
@@ -1543,120 +848,6 @@ static int __init prom_reconfig_setup(void)
1543__initcall(prom_reconfig_setup); 848__initcall(prom_reconfig_setup);
1544#endif 849#endif
1545 850
1546/*
1547 * Add a property to a node
1548 */
1549int prom_add_property(struct device_node* np, struct property* prop)
1550{
1551 struct property **next;
1552 unsigned long flags;
1553
1554 prop->next = NULL;
1555 write_lock_irqsave(&devtree_lock, flags);
1556 next = &np->properties;
1557 while (*next) {
1558 if (strcmp(prop->name, (*next)->name) == 0) {
1559 /* duplicate ! don't insert it */
1560 write_unlock_irqrestore(&devtree_lock, flags);
1561 return -1;
1562 }
1563 next = &(*next)->next;
1564 }
1565 *next = prop;
1566 write_unlock_irqrestore(&devtree_lock, flags);
1567
1568#ifdef CONFIG_PROC_DEVICETREE
1569 /* try to add to proc as well if it was initialized */
1570 if (np->pde)
1571 proc_device_tree_add_prop(np->pde, prop);
1572#endif /* CONFIG_PROC_DEVICETREE */
1573
1574 return 0;
1575}
1576
1577/*
1578 * Remove a property from a node. Note that we don't actually
1579 * remove it, since we have given out who-knows-how-many pointers
1580 * to the data using get-property. Instead we just move the property
1581 * to the "dead properties" list, so it won't be found any more.
1582 */
1583int prom_remove_property(struct device_node *np, struct property *prop)
1584{
1585 struct property **next;
1586 unsigned long flags;
1587 int found = 0;
1588
1589 write_lock_irqsave(&devtree_lock, flags);
1590 next = &np->properties;
1591 while (*next) {
1592 if (*next == prop) {
1593 /* found the node */
1594 *next = prop->next;
1595 prop->next = np->deadprops;
1596 np->deadprops = prop;
1597 found = 1;
1598 break;
1599 }
1600 next = &(*next)->next;
1601 }
1602 write_unlock_irqrestore(&devtree_lock, flags);
1603
1604 if (!found)
1605 return -ENODEV;
1606
1607#ifdef CONFIG_PROC_DEVICETREE
1608 /* try to remove the proc node as well */
1609 if (np->pde)
1610 proc_device_tree_remove_prop(np->pde, prop);
1611#endif /* CONFIG_PROC_DEVICETREE */
1612
1613 return 0;
1614}
1615
1616/*
1617 * Update a property in a node. Note that we don't actually
1618 * remove it, since we have given out who-knows-how-many pointers
1619 * to the data using get-property. Instead we just move the property
1620 * to the "dead properties" list, and add the new property to the
1621 * property list
1622 */
1623int prom_update_property(struct device_node *np,
1624 struct property *newprop,
1625 struct property *oldprop)
1626{
1627 struct property **next;
1628 unsigned long flags;
1629 int found = 0;
1630
1631 write_lock_irqsave(&devtree_lock, flags);
1632 next = &np->properties;
1633 while (*next) {
1634 if (*next == oldprop) {
1635 /* found the node */
1636 newprop->next = oldprop->next;
1637 *next = newprop;
1638 oldprop->next = np->deadprops;
1639 np->deadprops = oldprop;
1640 found = 1;
1641 break;
1642 }
1643 next = &(*next)->next;
1644 }
1645 write_unlock_irqrestore(&devtree_lock, flags);
1646
1647 if (!found)
1648 return -ENODEV;
1649
1650#ifdef CONFIG_PROC_DEVICETREE
1651 /* try to add to proc as well if it was initialized */
1652 if (np->pde)
1653 proc_device_tree_update_prop(np->pde, newprop, oldprop);
1654#endif /* CONFIG_PROC_DEVICETREE */
1655
1656 return 0;
1657}
1658
1659
1660/* Find the device node for a given logical cpu number, also returns the cpu 851/* Find the device node for a given logical cpu number, also returns the cpu
1661 * local thread number (index in ibm,interrupt-server#s) if relevant and 852 * local thread number (index in ibm,interrupt-server#s) if relevant and
1662 * asked for (non NULL) 853 * asked for (non NULL)
diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c
index bafac2e41ae1..97d4bd9442d3 100644
--- a/arch/powerpc/kernel/prom_init.c
+++ b/arch/powerpc/kernel/prom_init.c
@@ -653,6 +653,10 @@ static void __init early_cmdline_parse(void)
653#else 653#else
654#define OV5_CMO 0x00 654#define OV5_CMO 0x00
655#endif 655#endif
656#define OV5_TYPE1_AFFINITY 0x80 /* Type 1 NUMA affinity */
657
658/* Option Vector 6: IBM PAPR hints */
659#define OV6_LINUX 0x02 /* Linux is our OS */
656 660
657/* 661/*
658 * The architecture vector has an array of PVR mask/value pairs, 662 * The architecture vector has an array of PVR mask/value pairs,
@@ -665,7 +669,7 @@ static unsigned char ibm_architecture_vec[] = {
665 W(0xffffffff), W(0x0f000003), /* all 2.06-compliant */ 669 W(0xffffffff), W(0x0f000003), /* all 2.06-compliant */
666 W(0xffffffff), W(0x0f000002), /* all 2.05-compliant */ 670 W(0xffffffff), W(0x0f000002), /* all 2.05-compliant */
667 W(0xfffffffe), W(0x0f000001), /* all 2.04-compliant and earlier */ 671 W(0xfffffffe), W(0x0f000001), /* all 2.04-compliant and earlier */
668 5 - 1, /* 5 option vectors */ 672 6 - 1, /* 6 option vectors */
669 673
670 /* option vector 1: processor architectures supported */ 674 /* option vector 1: processor architectures supported */
671 3 - 2, /* length */ 675 3 - 2, /* length */
@@ -697,12 +701,29 @@ static unsigned char ibm_architecture_vec[] = {
697 0, /* don't halt */ 701 0, /* don't halt */
698 702
699 /* option vector 5: PAPR/OF options */ 703 /* option vector 5: PAPR/OF options */
700 5 - 2, /* length */ 704 13 - 2, /* length */
701 0, /* don't ignore, don't halt */ 705 0, /* don't ignore, don't halt */
702 OV5_LPAR | OV5_SPLPAR | OV5_LARGE_PAGES | OV5_DRCONF_MEMORY | 706 OV5_LPAR | OV5_SPLPAR | OV5_LARGE_PAGES | OV5_DRCONF_MEMORY |
703 OV5_DONATE_DEDICATE_CPU | OV5_MSI, 707 OV5_DONATE_DEDICATE_CPU | OV5_MSI,
704 0, 708 0,
705 OV5_CMO, 709 OV5_CMO,
710 OV5_TYPE1_AFFINITY,
711 0,
712 0,
713 0,
714 /* WARNING: The offset of the "number of cores" field below
715 * must match by the macro below. Update the definition if
716 * the structure layout changes.
717 */
718#define IBM_ARCH_VEC_NRCORES_OFFSET 100
719 W(NR_CPUS), /* number of cores supported */
720
721 /* option vector 6: IBM PAPR hints */
722 4 - 2, /* length */
723 0,
724 0,
725 OV6_LINUX,
726
706}; 727};
707 728
708/* Old method - ELF header with PT_NOTE sections */ 729/* Old method - ELF header with PT_NOTE sections */
@@ -792,13 +813,70 @@ static struct fake_elf {
792 } 813 }
793}; 814};
794 815
816static int __init prom_count_smt_threads(void)
817{
818 phandle node;
819 char type[64];
820 unsigned int plen;
821
822 /* Pick up th first CPU node we can find */
823 for (node = 0; prom_next_node(&node); ) {
824 type[0] = 0;
825 prom_getprop(node, "device_type", type, sizeof(type));
826
827 if (strcmp(type, RELOC("cpu")))
828 continue;
829 /*
830 * There is an entry for each smt thread, each entry being
831 * 4 bytes long. All cpus should have the same number of
832 * smt threads, so return after finding the first.
833 */
834 plen = prom_getproplen(node, "ibm,ppc-interrupt-server#s");
835 if (plen == PROM_ERROR)
836 break;
837 plen >>= 2;
838 prom_debug("Found 0x%x smt threads per core\n", (unsigned long)plen);
839
840 /* Sanity check */
841 if (plen < 1 || plen > 64) {
842 prom_printf("Threads per core 0x%x out of bounds, assuming 1\n",
843 (unsigned long)plen);
844 return 1;
845 }
846 return plen;
847 }
848 prom_debug("No threads found, assuming 1 per core\n");
849
850 return 1;
851
852}
853
854
795static void __init prom_send_capabilities(void) 855static void __init prom_send_capabilities(void)
796{ 856{
797 ihandle elfloader, root; 857 ihandle elfloader, root;
798 prom_arg_t ret; 858 prom_arg_t ret;
859 u32 *cores;
799 860
800 root = call_prom("open", 1, 1, ADDR("/")); 861 root = call_prom("open", 1, 1, ADDR("/"));
801 if (root != 0) { 862 if (root != 0) {
863 /* We need to tell the FW about the number of cores we support.
864 *
865 * To do that, we count the number of threads on the first core
866 * (we assume this is the same for all cores) and use it to
867 * divide NR_CPUS.
868 */
869 cores = (u32 *)PTRRELOC(&ibm_architecture_vec[IBM_ARCH_VEC_NRCORES_OFFSET]);
870 if (*cores != NR_CPUS) {
871 prom_printf("WARNING ! "
872 "ibm_architecture_vec structure inconsistent: 0x%x !\n",
873 *cores);
874 } else {
875 *cores = NR_CPUS / prom_count_smt_threads();
876 prom_printf("Max number of cores passed to firmware: 0x%x\n",
877 (unsigned long)*cores);
878 }
879
802 /* try calling the ibm,client-architecture-support method */ 880 /* try calling the ibm,client-architecture-support method */
803 prom_printf("Calling ibm,client-architecture-support..."); 881 prom_printf("Calling ibm,client-architecture-support...");
804 if (call_prom_ret("call-method", 3, 2, &ret, 882 if (call_prom_ret("call-method", 3, 2, &ret,
diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
index ef149880c145..ed2cfe17d25e 100644
--- a/arch/powerpc/kernel/ptrace.c
+++ b/arch/powerpc/kernel/ptrace.c
@@ -46,7 +46,7 @@
46/* 46/*
47 * Set of msr bits that gdb can change on behalf of a process. 47 * Set of msr bits that gdb can change on behalf of a process.
48 */ 48 */
49#if defined(CONFIG_40x) || defined(CONFIG_BOOKE) 49#ifdef CONFIG_PPC_ADV_DEBUG_REGS
50#define MSR_DEBUGCHANGE 0 50#define MSR_DEBUGCHANGE 0
51#else 51#else
52#define MSR_DEBUGCHANGE (MSR_SE | MSR_BE) 52#define MSR_DEBUGCHANGE (MSR_SE | MSR_BE)
@@ -703,7 +703,7 @@ void user_enable_single_step(struct task_struct *task)
703 struct pt_regs *regs = task->thread.regs; 703 struct pt_regs *regs = task->thread.regs;
704 704
705 if (regs != NULL) { 705 if (regs != NULL) {
706#if defined(CONFIG_40x) || defined(CONFIG_BOOKE) 706#ifdef CONFIG_PPC_ADV_DEBUG_REGS
707 task->thread.dbcr0 &= ~DBCR0_BT; 707 task->thread.dbcr0 &= ~DBCR0_BT;
708 task->thread.dbcr0 |= DBCR0_IDM | DBCR0_IC; 708 task->thread.dbcr0 |= DBCR0_IDM | DBCR0_IC;
709 regs->msr |= MSR_DE; 709 regs->msr |= MSR_DE;
@@ -720,7 +720,7 @@ void user_enable_block_step(struct task_struct *task)
720 struct pt_regs *regs = task->thread.regs; 720 struct pt_regs *regs = task->thread.regs;
721 721
722 if (regs != NULL) { 722 if (regs != NULL) {
723#if defined(CONFIG_40x) || defined(CONFIG_BOOKE) 723#ifdef CONFIG_PPC_ADV_DEBUG_REGS
724 task->thread.dbcr0 &= ~DBCR0_IC; 724 task->thread.dbcr0 &= ~DBCR0_IC;
725 task->thread.dbcr0 = DBCR0_IDM | DBCR0_BT; 725 task->thread.dbcr0 = DBCR0_IDM | DBCR0_BT;
726 regs->msr |= MSR_DE; 726 regs->msr |= MSR_DE;
@@ -737,17 +737,25 @@ void user_disable_single_step(struct task_struct *task)
737 struct pt_regs *regs = task->thread.regs; 737 struct pt_regs *regs = task->thread.regs;
738 738
739 if (regs != NULL) { 739 if (regs != NULL) {
740#if defined(CONFIG_BOOKE) 740#ifdef CONFIG_PPC_ADV_DEBUG_REGS
741 /* If DAC don't clear DBCRO_IDM or MSR_DE */ 741 /*
742 if (task->thread.dabr) 742 * The logic to disable single stepping should be as
743 task->thread.dbcr0 &= ~(DBCR0_IC | DBCR0_BT); 743 * simple as turning off the Instruction Complete flag.
744 else { 744 * And, after doing so, if all debug flags are off, turn
745 task->thread.dbcr0 &= ~(DBCR0_IC | DBCR0_BT | DBCR0_IDM); 745 * off DBCR0(IDM) and MSR(DE) .... Torez
746 */
747 task->thread.dbcr0 &= ~DBCR0_IC;
748 /*
749 * Test to see if any of the DBCR_ACTIVE_EVENTS bits are set.
750 */
751 if (!DBCR_ACTIVE_EVENTS(task->thread.dbcr0,
752 task->thread.dbcr1)) {
753 /*
754 * All debug events were off.....
755 */
756 task->thread.dbcr0 &= ~DBCR0_IDM;
746 regs->msr &= ~MSR_DE; 757 regs->msr &= ~MSR_DE;
747 } 758 }
748#elif defined(CONFIG_40x)
749 task->thread.dbcr0 &= ~(DBCR0_IC | DBCR0_BT | DBCR0_IDM);
750 regs->msr &= ~MSR_DE;
751#else 759#else
752 regs->msr &= ~(MSR_SE | MSR_BE); 760 regs->msr &= ~(MSR_SE | MSR_BE);
753#endif 761#endif
@@ -769,8 +777,7 @@ int ptrace_set_debugreg(struct task_struct *task, unsigned long addr,
769 if ((data & ~0x7UL) >= TASK_SIZE) 777 if ((data & ~0x7UL) >= TASK_SIZE)
770 return -EIO; 778 return -EIO;
771 779
772#ifndef CONFIG_BOOKE 780#ifndef CONFIG_PPC_ADV_DEBUG_REGS
773
774 /* For processors using DABR (i.e. 970), the bottom 3 bits are flags. 781 /* For processors using DABR (i.e. 970), the bottom 3 bits are flags.
775 * It was assumed, on previous implementations, that 3 bits were 782 * It was assumed, on previous implementations, that 3 bits were
776 * passed together with the data address, fitting the design of the 783 * passed together with the data address, fitting the design of the
@@ -789,21 +796,22 @@ int ptrace_set_debugreg(struct task_struct *task, unsigned long addr,
789 796
790 /* Move contents to the DABR register */ 797 /* Move contents to the DABR register */
791 task->thread.dabr = data; 798 task->thread.dabr = data;
792 799#else /* CONFIG_PPC_ADV_DEBUG_REGS */
793#endif
794#if defined(CONFIG_BOOKE)
795
796 /* As described above, it was assumed 3 bits were passed with the data 800 /* As described above, it was assumed 3 bits were passed with the data
797 * address, but we will assume only the mode bits will be passed 801 * address, but we will assume only the mode bits will be passed
798 * as to not cause alignment restrictions for DAC-based processors. 802 * as to not cause alignment restrictions for DAC-based processors.
799 */ 803 */
800 804
801 /* DAC's hold the whole address without any mode flags */ 805 /* DAC's hold the whole address without any mode flags */
802 task->thread.dabr = data & ~0x3UL; 806 task->thread.dac1 = data & ~0x3UL;
803 807
804 if (task->thread.dabr == 0) { 808 if (task->thread.dac1 == 0) {
805 task->thread.dbcr0 &= ~(DBSR_DAC1R | DBSR_DAC1W | DBCR0_IDM); 809 dbcr_dac(task) &= ~(DBCR_DAC1R | DBCR_DAC1W);
806 task->thread.regs->msr &= ~MSR_DE; 810 if (!DBCR_ACTIVE_EVENTS(task->thread.dbcr0,
811 task->thread.dbcr1)) {
812 task->thread.regs->msr &= ~MSR_DE;
813 task->thread.dbcr0 &= ~DBCR0_IDM;
814 }
807 return 0; 815 return 0;
808 } 816 }
809 817
@@ -814,17 +822,17 @@ int ptrace_set_debugreg(struct task_struct *task, unsigned long addr,
814 822
815 /* Set the Internal Debugging flag (IDM bit 1) for the DBCR0 823 /* Set the Internal Debugging flag (IDM bit 1) for the DBCR0
816 register */ 824 register */
817 task->thread.dbcr0 = DBCR0_IDM; 825 task->thread.dbcr0 |= DBCR0_IDM;
818 826
819 /* Check for write and read flags and set DBCR0 827 /* Check for write and read flags and set DBCR0
820 accordingly */ 828 accordingly */
829 dbcr_dac(task) &= ~(DBCR_DAC1R|DBCR_DAC1W);
821 if (data & 0x1UL) 830 if (data & 0x1UL)
822 task->thread.dbcr0 |= DBSR_DAC1R; 831 dbcr_dac(task) |= DBCR_DAC1R;
823 if (data & 0x2UL) 832 if (data & 0x2UL)
824 task->thread.dbcr0 |= DBSR_DAC1W; 833 dbcr_dac(task) |= DBCR_DAC1W;
825
826 task->thread.regs->msr |= MSR_DE; 834 task->thread.regs->msr |= MSR_DE;
827#endif 835#endif /* CONFIG_PPC_ADV_DEBUG_REGS */
828 return 0; 836 return 0;
829} 837}
830 838
@@ -839,6 +847,394 @@ void ptrace_disable(struct task_struct *child)
839 user_disable_single_step(child); 847 user_disable_single_step(child);
840} 848}
841 849
850#ifdef CONFIG_PPC_ADV_DEBUG_REGS
851static long set_intruction_bp(struct task_struct *child,
852 struct ppc_hw_breakpoint *bp_info)
853{
854 int slot;
855 int slot1_in_use = ((child->thread.dbcr0 & DBCR0_IAC1) != 0);
856 int slot2_in_use = ((child->thread.dbcr0 & DBCR0_IAC2) != 0);
857 int slot3_in_use = ((child->thread.dbcr0 & DBCR0_IAC3) != 0);
858 int slot4_in_use = ((child->thread.dbcr0 & DBCR0_IAC4) != 0);
859
860 if (dbcr_iac_range(child) & DBCR_IAC12MODE)
861 slot2_in_use = 1;
862 if (dbcr_iac_range(child) & DBCR_IAC34MODE)
863 slot4_in_use = 1;
864
865 if (bp_info->addr >= TASK_SIZE)
866 return -EIO;
867
868 if (bp_info->addr_mode != PPC_BREAKPOINT_MODE_EXACT) {
869
870 /* Make sure range is valid. */
871 if (bp_info->addr2 >= TASK_SIZE)
872 return -EIO;
873
874 /* We need a pair of IAC regsisters */
875 if ((!slot1_in_use) && (!slot2_in_use)) {
876 slot = 1;
877 child->thread.iac1 = bp_info->addr;
878 child->thread.iac2 = bp_info->addr2;
879 child->thread.dbcr0 |= DBCR0_IAC1;
880 if (bp_info->addr_mode ==
881 PPC_BREAKPOINT_MODE_RANGE_EXCLUSIVE)
882 dbcr_iac_range(child) |= DBCR_IAC12X;
883 else
884 dbcr_iac_range(child) |= DBCR_IAC12I;
885#if CONFIG_PPC_ADV_DEBUG_IACS > 2
886 } else if ((!slot3_in_use) && (!slot4_in_use)) {
887 slot = 3;
888 child->thread.iac3 = bp_info->addr;
889 child->thread.iac4 = bp_info->addr2;
890 child->thread.dbcr0 |= DBCR0_IAC3;
891 if (bp_info->addr_mode ==
892 PPC_BREAKPOINT_MODE_RANGE_EXCLUSIVE)
893 dbcr_iac_range(child) |= DBCR_IAC34X;
894 else
895 dbcr_iac_range(child) |= DBCR_IAC34I;
896#endif
897 } else
898 return -ENOSPC;
899 } else {
900 /* We only need one. If possible leave a pair free in
901 * case a range is needed later
902 */
903 if (!slot1_in_use) {
904 /*
905 * Don't use iac1 if iac1-iac2 are free and either
906 * iac3 or iac4 (but not both) are free
907 */
908 if (slot2_in_use || (slot3_in_use == slot4_in_use)) {
909 slot = 1;
910 child->thread.iac1 = bp_info->addr;
911 child->thread.dbcr0 |= DBCR0_IAC1;
912 goto out;
913 }
914 }
915 if (!slot2_in_use) {
916 slot = 2;
917 child->thread.iac2 = bp_info->addr;
918 child->thread.dbcr0 |= DBCR0_IAC2;
919#if CONFIG_PPC_ADV_DEBUG_IACS > 2
920 } else if (!slot3_in_use) {
921 slot = 3;
922 child->thread.iac3 = bp_info->addr;
923 child->thread.dbcr0 |= DBCR0_IAC3;
924 } else if (!slot4_in_use) {
925 slot = 4;
926 child->thread.iac4 = bp_info->addr;
927 child->thread.dbcr0 |= DBCR0_IAC4;
928#endif
929 } else
930 return -ENOSPC;
931 }
932out:
933 child->thread.dbcr0 |= DBCR0_IDM;
934 child->thread.regs->msr |= MSR_DE;
935
936 return slot;
937}
938
939static int del_instruction_bp(struct task_struct *child, int slot)
940{
941 switch (slot) {
942 case 1:
943 if ((child->thread.dbcr0 & DBCR0_IAC1) == 0)
944 return -ENOENT;
945
946 if (dbcr_iac_range(child) & DBCR_IAC12MODE) {
947 /* address range - clear slots 1 & 2 */
948 child->thread.iac2 = 0;
949 dbcr_iac_range(child) &= ~DBCR_IAC12MODE;
950 }
951 child->thread.iac1 = 0;
952 child->thread.dbcr0 &= ~DBCR0_IAC1;
953 break;
954 case 2:
955 if ((child->thread.dbcr0 & DBCR0_IAC2) == 0)
956 return -ENOENT;
957
958 if (dbcr_iac_range(child) & DBCR_IAC12MODE)
959 /* used in a range */
960 return -EINVAL;
961 child->thread.iac2 = 0;
962 child->thread.dbcr0 &= ~DBCR0_IAC2;
963 break;
964#if CONFIG_PPC_ADV_DEBUG_IACS > 2
965 case 3:
966 if ((child->thread.dbcr0 & DBCR0_IAC3) == 0)
967 return -ENOENT;
968
969 if (dbcr_iac_range(child) & DBCR_IAC34MODE) {
970 /* address range - clear slots 3 & 4 */
971 child->thread.iac4 = 0;
972 dbcr_iac_range(child) &= ~DBCR_IAC34MODE;
973 }
974 child->thread.iac3 = 0;
975 child->thread.dbcr0 &= ~DBCR0_IAC3;
976 break;
977 case 4:
978 if ((child->thread.dbcr0 & DBCR0_IAC4) == 0)
979 return -ENOENT;
980
981 if (dbcr_iac_range(child) & DBCR_IAC34MODE)
982 /* Used in a range */
983 return -EINVAL;
984 child->thread.iac4 = 0;
985 child->thread.dbcr0 &= ~DBCR0_IAC4;
986 break;
987#endif
988 default:
989 return -EINVAL;
990 }
991 return 0;
992}
993
994static int set_dac(struct task_struct *child, struct ppc_hw_breakpoint *bp_info)
995{
996 int byte_enable =
997 (bp_info->condition_mode >> PPC_BREAKPOINT_CONDITION_BE_SHIFT)
998 & 0xf;
999 int condition_mode =
1000 bp_info->condition_mode & PPC_BREAKPOINT_CONDITION_MODE;
1001 int slot;
1002
1003 if (byte_enable && (condition_mode == 0))
1004 return -EINVAL;
1005
1006 if (bp_info->addr >= TASK_SIZE)
1007 return -EIO;
1008
1009 if ((dbcr_dac(child) & (DBCR_DAC1R | DBCR_DAC1W)) == 0) {
1010 slot = 1;
1011 if (bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_READ)
1012 dbcr_dac(child) |= DBCR_DAC1R;
1013 if (bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_WRITE)
1014 dbcr_dac(child) |= DBCR_DAC1W;
1015 child->thread.dac1 = (unsigned long)bp_info->addr;
1016#if CONFIG_PPC_ADV_DEBUG_DVCS > 0
1017 if (byte_enable) {
1018 child->thread.dvc1 =
1019 (unsigned long)bp_info->condition_value;
1020 child->thread.dbcr2 |=
1021 ((byte_enable << DBCR2_DVC1BE_SHIFT) |
1022 (condition_mode << DBCR2_DVC1M_SHIFT));
1023 }
1024#endif
1025#ifdef CONFIG_PPC_ADV_DEBUG_DAC_RANGE
1026 } else if (child->thread.dbcr2 & DBCR2_DAC12MODE) {
1027 /* Both dac1 and dac2 are part of a range */
1028 return -ENOSPC;
1029#endif
1030 } else if ((dbcr_dac(child) & (DBCR_DAC2R | DBCR_DAC2W)) == 0) {
1031 slot = 2;
1032 if (bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_READ)
1033 dbcr_dac(child) |= DBCR_DAC2R;
1034 if (bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_WRITE)
1035 dbcr_dac(child) |= DBCR_DAC2W;
1036 child->thread.dac2 = (unsigned long)bp_info->addr;
1037#if CONFIG_PPC_ADV_DEBUG_DVCS > 0
1038 if (byte_enable) {
1039 child->thread.dvc2 =
1040 (unsigned long)bp_info->condition_value;
1041 child->thread.dbcr2 |=
1042 ((byte_enable << DBCR2_DVC2BE_SHIFT) |
1043 (condition_mode << DBCR2_DVC2M_SHIFT));
1044 }
1045#endif
1046 } else
1047 return -ENOSPC;
1048 child->thread.dbcr0 |= DBCR0_IDM;
1049 child->thread.regs->msr |= MSR_DE;
1050
1051 return slot + 4;
1052}
1053
1054static int del_dac(struct task_struct *child, int slot)
1055{
1056 if (slot == 1) {
1057 if ((dbcr_dac(child) & (DBCR_DAC1R | DBCR_DAC1W)) == 0)
1058 return -ENOENT;
1059
1060 child->thread.dac1 = 0;
1061 dbcr_dac(child) &= ~(DBCR_DAC1R | DBCR_DAC1W);
1062#ifdef CONFIG_PPC_ADV_DEBUG_DAC_RANGE
1063 if (child->thread.dbcr2 & DBCR2_DAC12MODE) {
1064 child->thread.dac2 = 0;
1065 child->thread.dbcr2 &= ~DBCR2_DAC12MODE;
1066 }
1067 child->thread.dbcr2 &= ~(DBCR2_DVC1M | DBCR2_DVC1BE);
1068#endif
1069#if CONFIG_PPC_ADV_DEBUG_DVCS > 0
1070 child->thread.dvc1 = 0;
1071#endif
1072 } else if (slot == 2) {
1073 if ((dbcr_dac(child) & (DBCR_DAC2R | DBCR_DAC2W)) == 0)
1074 return -ENOENT;
1075
1076#ifdef CONFIG_PPC_ADV_DEBUG_DAC_RANGE
1077 if (child->thread.dbcr2 & DBCR2_DAC12MODE)
1078 /* Part of a range */
1079 return -EINVAL;
1080 child->thread.dbcr2 &= ~(DBCR2_DVC2M | DBCR2_DVC2BE);
1081#endif
1082#if CONFIG_PPC_ADV_DEBUG_DVCS > 0
1083 child->thread.dvc2 = 0;
1084#endif
1085 child->thread.dac2 = 0;
1086 dbcr_dac(child) &= ~(DBCR_DAC2R | DBCR_DAC2W);
1087 } else
1088 return -EINVAL;
1089
1090 return 0;
1091}
1092#endif /* CONFIG_PPC_ADV_DEBUG_REGS */
1093
1094#ifdef CONFIG_PPC_ADV_DEBUG_DAC_RANGE
1095static int set_dac_range(struct task_struct *child,
1096 struct ppc_hw_breakpoint *bp_info)
1097{
1098 int mode = bp_info->addr_mode & PPC_BREAKPOINT_MODE_MASK;
1099
1100 /* We don't allow range watchpoints to be used with DVC */
1101 if (bp_info->condition_mode)
1102 return -EINVAL;
1103
1104 /*
1105 * Best effort to verify the address range. The user/supervisor bits
1106 * prevent trapping in kernel space, but let's fail on an obvious bad
1107 * range. The simple test on the mask is not fool-proof, and any
1108 * exclusive range will spill over into kernel space.
1109 */
1110 if (bp_info->addr >= TASK_SIZE)
1111 return -EIO;
1112 if (mode == PPC_BREAKPOINT_MODE_MASK) {
1113 /*
1114 * dac2 is a bitmask. Don't allow a mask that makes a
1115 * kernel space address from a valid dac1 value
1116 */
1117 if (~((unsigned long)bp_info->addr2) >= TASK_SIZE)
1118 return -EIO;
1119 } else {
1120 /*
1121 * For range breakpoints, addr2 must also be a valid address
1122 */
1123 if (bp_info->addr2 >= TASK_SIZE)
1124 return -EIO;
1125 }
1126
1127 if (child->thread.dbcr0 &
1128 (DBCR0_DAC1R | DBCR0_DAC1W | DBCR0_DAC2R | DBCR0_DAC2W))
1129 return -ENOSPC;
1130
1131 if (bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_READ)
1132 child->thread.dbcr0 |= (DBCR0_DAC1R | DBCR0_IDM);
1133 if (bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_WRITE)
1134 child->thread.dbcr0 |= (DBCR0_DAC1W | DBCR0_IDM);
1135 child->thread.dac1 = bp_info->addr;
1136 child->thread.dac2 = bp_info->addr2;
1137 if (mode == PPC_BREAKPOINT_MODE_RANGE_INCLUSIVE)
1138 child->thread.dbcr2 |= DBCR2_DAC12M;
1139 else if (mode == PPC_BREAKPOINT_MODE_RANGE_EXCLUSIVE)
1140 child->thread.dbcr2 |= DBCR2_DAC12MX;
1141 else /* PPC_BREAKPOINT_MODE_MASK */
1142 child->thread.dbcr2 |= DBCR2_DAC12MM;
1143 child->thread.regs->msr |= MSR_DE;
1144
1145 return 5;
1146}
1147#endif /* CONFIG_PPC_ADV_DEBUG_DAC_RANGE */
1148
1149static long ppc_set_hwdebug(struct task_struct *child,
1150 struct ppc_hw_breakpoint *bp_info)
1151{
1152 if (bp_info->version != 1)
1153 return -ENOTSUPP;
1154#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1155 /*
1156 * Check for invalid flags and combinations
1157 */
1158 if ((bp_info->trigger_type == 0) ||
1159 (bp_info->trigger_type & ~(PPC_BREAKPOINT_TRIGGER_EXECUTE |
1160 PPC_BREAKPOINT_TRIGGER_RW)) ||
1161 (bp_info->addr_mode & ~PPC_BREAKPOINT_MODE_MASK) ||
1162 (bp_info->condition_mode &
1163 ~(PPC_BREAKPOINT_CONDITION_MODE |
1164 PPC_BREAKPOINT_CONDITION_BE_ALL)))
1165 return -EINVAL;
1166#if CONFIG_PPC_ADV_DEBUG_DVCS == 0
1167 if (bp_info->condition_mode != PPC_BREAKPOINT_CONDITION_NONE)
1168 return -EINVAL;
1169#endif
1170
1171 if (bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_EXECUTE) {
1172 if ((bp_info->trigger_type != PPC_BREAKPOINT_TRIGGER_EXECUTE) ||
1173 (bp_info->condition_mode != PPC_BREAKPOINT_CONDITION_NONE))
1174 return -EINVAL;
1175 return set_intruction_bp(child, bp_info);
1176 }
1177 if (bp_info->addr_mode == PPC_BREAKPOINT_MODE_EXACT)
1178 return set_dac(child, bp_info);
1179
1180#ifdef CONFIG_PPC_ADV_DEBUG_DAC_RANGE
1181 return set_dac_range(child, bp_info);
1182#else
1183 return -EINVAL;
1184#endif
1185#else /* !CONFIG_PPC_ADV_DEBUG_DVCS */
1186 /*
1187 * We only support one data breakpoint
1188 */
1189 if (((bp_info->trigger_type & PPC_BREAKPOINT_TRIGGER_RW) == 0) ||
1190 ((bp_info->trigger_type & ~PPC_BREAKPOINT_TRIGGER_RW) != 0) ||
1191 (bp_info->trigger_type != PPC_BREAKPOINT_TRIGGER_WRITE) ||
1192 (bp_info->addr_mode != PPC_BREAKPOINT_MODE_EXACT) ||
1193 (bp_info->condition_mode != PPC_BREAKPOINT_CONDITION_NONE))
1194 return -EINVAL;
1195
1196 if (child->thread.dabr)
1197 return -ENOSPC;
1198
1199 if ((unsigned long)bp_info->addr >= TASK_SIZE)
1200 return -EIO;
1201
1202 child->thread.dabr = (unsigned long)bp_info->addr;
1203
1204 return 1;
1205#endif /* !CONFIG_PPC_ADV_DEBUG_DVCS */
1206}
1207
1208static long ppc_del_hwdebug(struct task_struct *child, long addr, long data)
1209{
1210#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1211 int rc;
1212
1213 if (data <= 4)
1214 rc = del_instruction_bp(child, (int)data);
1215 else
1216 rc = del_dac(child, (int)data - 4);
1217
1218 if (!rc) {
1219 if (!DBCR_ACTIVE_EVENTS(child->thread.dbcr0,
1220 child->thread.dbcr1)) {
1221 child->thread.dbcr0 &= ~DBCR0_IDM;
1222 child->thread.regs->msr &= ~MSR_DE;
1223 }
1224 }
1225 return rc;
1226#else
1227 if (data != 1)
1228 return -EINVAL;
1229 if (child->thread.dabr == 0)
1230 return -ENOENT;
1231
1232 child->thread.dabr = 0;
1233
1234 return 0;
1235#endif
1236}
1237
842/* 1238/*
843 * Here are the old "legacy" powerpc specific getregs/setregs ptrace calls, 1239 * Here are the old "legacy" powerpc specific getregs/setregs ptrace calls,
844 * we mark them as obsolete now, they will be removed in a future version 1240 * we mark them as obsolete now, they will be removed in a future version
@@ -932,13 +1328,77 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
932 break; 1328 break;
933 } 1329 }
934 1330
1331 case PPC_PTRACE_GETHWDBGINFO: {
1332 struct ppc_debug_info dbginfo;
1333
1334 dbginfo.version = 1;
1335#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1336 dbginfo.num_instruction_bps = CONFIG_PPC_ADV_DEBUG_IACS;
1337 dbginfo.num_data_bps = CONFIG_PPC_ADV_DEBUG_DACS;
1338 dbginfo.num_condition_regs = CONFIG_PPC_ADV_DEBUG_DVCS;
1339 dbginfo.data_bp_alignment = 4;
1340 dbginfo.sizeof_condition = 4;
1341 dbginfo.features = PPC_DEBUG_FEATURE_INSN_BP_RANGE |
1342 PPC_DEBUG_FEATURE_INSN_BP_MASK;
1343#ifdef CONFIG_PPC_ADV_DEBUG_DAC_RANGE
1344 dbginfo.features |=
1345 PPC_DEBUG_FEATURE_DATA_BP_RANGE |
1346 PPC_DEBUG_FEATURE_DATA_BP_MASK;
1347#endif
1348#else /* !CONFIG_PPC_ADV_DEBUG_REGS */
1349 dbginfo.num_instruction_bps = 0;
1350 dbginfo.num_data_bps = 1;
1351 dbginfo.num_condition_regs = 0;
1352#ifdef CONFIG_PPC64
1353 dbginfo.data_bp_alignment = 8;
1354#else
1355 dbginfo.data_bp_alignment = 4;
1356#endif
1357 dbginfo.sizeof_condition = 0;
1358 dbginfo.features = 0;
1359#endif /* CONFIG_PPC_ADV_DEBUG_REGS */
1360
1361 if (!access_ok(VERIFY_WRITE, data,
1362 sizeof(struct ppc_debug_info)))
1363 return -EFAULT;
1364 ret = __copy_to_user((struct ppc_debug_info __user *)data,
1365 &dbginfo, sizeof(struct ppc_debug_info)) ?
1366 -EFAULT : 0;
1367 break;
1368 }
1369
1370 case PPC_PTRACE_SETHWDEBUG: {
1371 struct ppc_hw_breakpoint bp_info;
1372
1373 if (!access_ok(VERIFY_READ, data,
1374 sizeof(struct ppc_hw_breakpoint)))
1375 return -EFAULT;
1376 ret = __copy_from_user(&bp_info,
1377 (struct ppc_hw_breakpoint __user *)data,
1378 sizeof(struct ppc_hw_breakpoint)) ?
1379 -EFAULT : 0;
1380 if (!ret)
1381 ret = ppc_set_hwdebug(child, &bp_info);
1382 break;
1383 }
1384
1385 case PPC_PTRACE_DELHWDEBUG: {
1386 ret = ppc_del_hwdebug(child, addr, data);
1387 break;
1388 }
1389
935 case PTRACE_GET_DEBUGREG: { 1390 case PTRACE_GET_DEBUGREG: {
936 ret = -EINVAL; 1391 ret = -EINVAL;
937 /* We only support one DABR and no IABRS at the moment */ 1392 /* We only support one DABR and no IABRS at the moment */
938 if (addr > 0) 1393 if (addr > 0)
939 break; 1394 break;
1395#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1396 ret = put_user(child->thread.dac1,
1397 (unsigned long __user *)data);
1398#else
940 ret = put_user(child->thread.dabr, 1399 ret = put_user(child->thread.dabr,
941 (unsigned long __user *)data); 1400 (unsigned long __user *)data);
1401#endif
942 break; 1402 break;
943 } 1403 }
944 1404
diff --git a/arch/powerpc/kernel/rtas-proc.c b/arch/powerpc/kernel/rtas-proc.c
index 1be9fe38bcb5..8777fb02349f 100644
--- a/arch/powerpc/kernel/rtas-proc.c
+++ b/arch/powerpc/kernel/rtas-proc.c
@@ -262,19 +262,19 @@ static int __init proc_rtas_init(void)
262 if (rtas_node == NULL) 262 if (rtas_node == NULL)
263 return -ENODEV; 263 return -ENODEV;
264 264
265 proc_create("ppc64/rtas/progress", S_IRUGO|S_IWUSR, NULL, 265 proc_create("powerpc/rtas/progress", S_IRUGO|S_IWUSR, NULL,
266 &ppc_rtas_progress_operations); 266 &ppc_rtas_progress_operations);
267 proc_create("ppc64/rtas/clock", S_IRUGO|S_IWUSR, NULL, 267 proc_create("powerpc/rtas/clock", S_IRUGO|S_IWUSR, NULL,
268 &ppc_rtas_clock_operations); 268 &ppc_rtas_clock_operations);
269 proc_create("ppc64/rtas/poweron", S_IWUSR|S_IRUGO, NULL, 269 proc_create("powerpc/rtas/poweron", S_IWUSR|S_IRUGO, NULL,
270 &ppc_rtas_poweron_operations); 270 &ppc_rtas_poweron_operations);
271 proc_create("ppc64/rtas/sensors", S_IRUGO, NULL, 271 proc_create("powerpc/rtas/sensors", S_IRUGO, NULL,
272 &ppc_rtas_sensors_operations); 272 &ppc_rtas_sensors_operations);
273 proc_create("ppc64/rtas/frequency", S_IWUSR|S_IRUGO, NULL, 273 proc_create("powerpc/rtas/frequency", S_IWUSR|S_IRUGO, NULL,
274 &ppc_rtas_tone_freq_operations); 274 &ppc_rtas_tone_freq_operations);
275 proc_create("ppc64/rtas/volume", S_IWUSR|S_IRUGO, NULL, 275 proc_create("powerpc/rtas/volume", S_IWUSR|S_IRUGO, NULL,
276 &ppc_rtas_tone_volume_operations); 276 &ppc_rtas_tone_volume_operations);
277 proc_create("ppc64/rtas/rmo_buffer", S_IRUSR, NULL, 277 proc_create("powerpc/rtas/rmo_buffer", S_IRUSR, NULL,
278 &ppc_rtas_rmo_buf_ops); 278 &ppc_rtas_rmo_buf_ops);
279 return 0; 279 return 0;
280} 280}
diff --git a/arch/powerpc/kernel/rtas.c b/arch/powerpc/kernel/rtas.c
index bf90361bb70f..74367841615a 100644
--- a/arch/powerpc/kernel/rtas.c
+++ b/arch/powerpc/kernel/rtas.c
@@ -23,6 +23,7 @@
23#include <linux/completion.h> 23#include <linux/completion.h>
24#include <linux/cpumask.h> 24#include <linux/cpumask.h>
25#include <linux/lmb.h> 25#include <linux/lmb.h>
26#include <linux/slab.h>
26 27
27#include <asm/prom.h> 28#include <asm/prom.h>
28#include <asm/rtas.h> 29#include <asm/rtas.h>
@@ -42,7 +43,7 @@
42#include <asm/mmu.h> 43#include <asm/mmu.h>
43 44
44struct rtas_t rtas = { 45struct rtas_t rtas = {
45 .lock = __RAW_SPIN_LOCK_UNLOCKED 46 .lock = __ARCH_SPIN_LOCK_UNLOCKED
46}; 47};
47EXPORT_SYMBOL(rtas); 48EXPORT_SYMBOL(rtas);
48 49
@@ -80,13 +81,13 @@ static unsigned long lock_rtas(void)
80 81
81 local_irq_save(flags); 82 local_irq_save(flags);
82 preempt_disable(); 83 preempt_disable();
83 __raw_spin_lock_flags(&rtas.lock, flags); 84 arch_spin_lock_flags(&rtas.lock, flags);
84 return flags; 85 return flags;
85} 86}
86 87
87static void unlock_rtas(unsigned long flags) 88static void unlock_rtas(unsigned long flags)
88{ 89{
89 __raw_spin_unlock(&rtas.lock); 90 arch_spin_unlock(&rtas.lock);
90 local_irq_restore(flags); 91 local_irq_restore(flags);
91 preempt_enable(); 92 preempt_enable();
92} 93}
@@ -978,7 +979,7 @@ int __init early_init_dt_scan_rtas(unsigned long node,
978 return 1; 979 return 1;
979} 980}
980 981
981static raw_spinlock_t timebase_lock; 982static arch_spinlock_t timebase_lock;
982static u64 timebase = 0; 983static u64 timebase = 0;
983 984
984void __cpuinit rtas_give_timebase(void) 985void __cpuinit rtas_give_timebase(void)
@@ -987,10 +988,10 @@ void __cpuinit rtas_give_timebase(void)
987 988
988 local_irq_save(flags); 989 local_irq_save(flags);
989 hard_irq_disable(); 990 hard_irq_disable();
990 __raw_spin_lock(&timebase_lock); 991 arch_spin_lock(&timebase_lock);
991 rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL); 992 rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL);
992 timebase = get_tb(); 993 timebase = get_tb();
993 __raw_spin_unlock(&timebase_lock); 994 arch_spin_unlock(&timebase_lock);
994 995
995 while (timebase) 996 while (timebase)
996 barrier(); 997 barrier();
@@ -1002,8 +1003,8 @@ void __cpuinit rtas_take_timebase(void)
1002{ 1003{
1003 while (!timebase) 1004 while (!timebase)
1004 barrier(); 1005 barrier();
1005 __raw_spin_lock(&timebase_lock); 1006 arch_spin_lock(&timebase_lock);
1006 set_tb(timebase >> 32, timebase & 0xffffffff); 1007 set_tb(timebase >> 32, timebase & 0xffffffff);
1007 timebase = 0; 1008 timebase = 0;
1008 __raw_spin_unlock(&timebase_lock); 1009 arch_spin_unlock(&timebase_lock);
1009} 1010}
diff --git a/arch/powerpc/kernel/rtas_flash.c b/arch/powerpc/kernel/rtas_flash.c
index 13011a96a977..bfc2abafac44 100644
--- a/arch/powerpc/kernel/rtas_flash.c
+++ b/arch/powerpc/kernel/rtas_flash.c
@@ -6,7 +6,7 @@
6 * as published by the Free Software Foundation; either version 6 * as published by the Free Software Foundation; either version
7 * 2 of the License, or (at your option) any later version. 7 * 2 of the License, or (at your option) any later version.
8 * 8 *
9 * /proc/ppc64/rtas/firmware_flash interface 9 * /proc/powerpc/rtas/firmware_flash interface
10 * 10 *
11 * This file implements a firmware_flash interface to pump a firmware 11 * This file implements a firmware_flash interface to pump a firmware
12 * image into the kernel. At reboot time rtas_restart() will see the 12 * image into the kernel. At reboot time rtas_restart() will see the
@@ -15,6 +15,7 @@
15 15
16#include <linux/module.h> 16#include <linux/module.h>
17#include <linux/init.h> 17#include <linux/init.h>
18#include <linux/slab.h>
18#include <linux/proc_fs.h> 19#include <linux/proc_fs.h>
19#include <asm/delay.h> 20#include <asm/delay.h>
20#include <asm/uaccess.h> 21#include <asm/uaccess.h>
@@ -740,7 +741,7 @@ static int __init rtas_flash_init(void)
740 return 1; 741 return 1;
741 } 742 }
742 743
743 firmware_flash_pde = create_flash_pde("ppc64/rtas/" 744 firmware_flash_pde = create_flash_pde("powerpc/rtas/"
744 FIRMWARE_FLASH_NAME, 745 FIRMWARE_FLASH_NAME,
745 &rtas_flash_operations); 746 &rtas_flash_operations);
746 if (firmware_flash_pde == NULL) { 747 if (firmware_flash_pde == NULL) {
@@ -754,7 +755,7 @@ static int __init rtas_flash_init(void)
754 if (rc != 0) 755 if (rc != 0)
755 goto cleanup; 756 goto cleanup;
756 757
757 firmware_update_pde = create_flash_pde("ppc64/rtas/" 758 firmware_update_pde = create_flash_pde("powerpc/rtas/"
758 FIRMWARE_UPDATE_NAME, 759 FIRMWARE_UPDATE_NAME,
759 &rtas_flash_operations); 760 &rtas_flash_operations);
760 if (firmware_update_pde == NULL) { 761 if (firmware_update_pde == NULL) {
@@ -768,7 +769,7 @@ static int __init rtas_flash_init(void)
768 if (rc != 0) 769 if (rc != 0)
769 goto cleanup; 770 goto cleanup;
770 771
771 validate_pde = create_flash_pde("ppc64/rtas/" VALIDATE_FLASH_NAME, 772 validate_pde = create_flash_pde("powerpc/rtas/" VALIDATE_FLASH_NAME,
772 &validate_flash_operations); 773 &validate_flash_operations);
773 if (validate_pde == NULL) { 774 if (validate_pde == NULL) {
774 rc = -ENOMEM; 775 rc = -ENOMEM;
@@ -781,7 +782,7 @@ static int __init rtas_flash_init(void)
781 if (rc != 0) 782 if (rc != 0)
782 goto cleanup; 783 goto cleanup;
783 784
784 manage_pde = create_flash_pde("ppc64/rtas/" MANAGE_FLASH_NAME, 785 manage_pde = create_flash_pde("powerpc/rtas/" MANAGE_FLASH_NAME,
785 &manage_flash_operations); 786 &manage_flash_operations);
786 if (manage_pde == NULL) { 787 if (manage_pde == NULL) {
787 rc = -ENOMEM; 788 rc = -ENOMEM;
diff --git a/arch/powerpc/kernel/rtasd.c b/arch/powerpc/kernel/rtasd.c
new file mode 100644
index 000000000000..4190eae7850a
--- /dev/null
+++ b/arch/powerpc/kernel/rtasd.c
@@ -0,0 +1,540 @@
1/*
2 * Copyright (C) 2001 Anton Blanchard <anton@au.ibm.com>, IBM
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License
6 * as published by the Free Software Foundation; either version
7 * 2 of the License, or (at your option) any later version.
8 *
9 * Communication to userspace based on kernel/printk.c
10 */
11
12#include <linux/types.h>
13#include <linux/errno.h>
14#include <linux/sched.h>
15#include <linux/kernel.h>
16#include <linux/poll.h>
17#include <linux/proc_fs.h>
18#include <linux/init.h>
19#include <linux/vmalloc.h>
20#include <linux/spinlock.h>
21#include <linux/cpu.h>
22#include <linux/workqueue.h>
23#include <linux/slab.h>
24
25#include <asm/uaccess.h>
26#include <asm/io.h>
27#include <asm/rtas.h>
28#include <asm/prom.h>
29#include <asm/nvram.h>
30#include <asm/atomic.h>
31#include <asm/machdep.h>
32
33
34static DEFINE_SPINLOCK(rtasd_log_lock);
35
36static DECLARE_WAIT_QUEUE_HEAD(rtas_log_wait);
37
38static char *rtas_log_buf;
39static unsigned long rtas_log_start;
40static unsigned long rtas_log_size;
41
42static int surveillance_timeout = -1;
43
44static unsigned int rtas_error_log_max;
45static unsigned int rtas_error_log_buffer_max;
46
47/* RTAS service tokens */
48static unsigned int event_scan;
49static unsigned int rtas_event_scan_rate;
50
51static int full_rtas_msgs = 0;
52
53/* Stop logging to nvram after first fatal error */
54static int logging_enabled; /* Until we initialize everything,
55 * make sure we don't try logging
56 * anything */
57static int error_log_cnt;
58
59/*
60 * Since we use 32 bit RTAS, the physical address of this must be below
61 * 4G or else bad things happen. Allocate this in the kernel data and
62 * make it big enough.
63 */
64static unsigned char logdata[RTAS_ERROR_LOG_MAX];
65
66static char *rtas_type[] = {
67 "Unknown", "Retry", "TCE Error", "Internal Device Failure",
68 "Timeout", "Data Parity", "Address Parity", "Cache Parity",
69 "Address Invalid", "ECC Uncorrected", "ECC Corrupted",
70};
71
72static char *rtas_event_type(int type)
73{
74 if ((type > 0) && (type < 11))
75 return rtas_type[type];
76
77 switch (type) {
78 case RTAS_TYPE_EPOW:
79 return "EPOW";
80 case RTAS_TYPE_PLATFORM:
81 return "Platform Error";
82 case RTAS_TYPE_IO:
83 return "I/O Event";
84 case RTAS_TYPE_INFO:
85 return "Platform Information Event";
86 case RTAS_TYPE_DEALLOC:
87 return "Resource Deallocation Event";
88 case RTAS_TYPE_DUMP:
89 return "Dump Notification Event";
90 }
91
92 return rtas_type[0];
93}
94
95/* To see this info, grep RTAS /var/log/messages and each entry
96 * will be collected together with obvious begin/end.
97 * There will be a unique identifier on the begin and end lines.
98 * This will persist across reboots.
99 *
100 * format of error logs returned from RTAS:
101 * bytes (size) : contents
102 * --------------------------------------------------------
103 * 0-7 (8) : rtas_error_log
104 * 8-47 (40) : extended info
105 * 48-51 (4) : vendor id
106 * 52-1023 (vendor specific) : location code and debug data
107 */
108static void printk_log_rtas(char *buf, int len)
109{
110
111 int i,j,n = 0;
112 int perline = 16;
113 char buffer[64];
114 char * str = "RTAS event";
115
116 if (full_rtas_msgs) {
117 printk(RTAS_DEBUG "%d -------- %s begin --------\n",
118 error_log_cnt, str);
119
120 /*
121 * Print perline bytes on each line, each line will start
122 * with RTAS and a changing number, so syslogd will
123 * print lines that are otherwise the same. Separate every
124 * 4 bytes with a space.
125 */
126 for (i = 0; i < len; i++) {
127 j = i % perline;
128 if (j == 0) {
129 memset(buffer, 0, sizeof(buffer));
130 n = sprintf(buffer, "RTAS %d:", i/perline);
131 }
132
133 if ((i % 4) == 0)
134 n += sprintf(buffer+n, " ");
135
136 n += sprintf(buffer+n, "%02x", (unsigned char)buf[i]);
137
138 if (j == (perline-1))
139 printk(KERN_DEBUG "%s\n", buffer);
140 }
141 if ((i % perline) != 0)
142 printk(KERN_DEBUG "%s\n", buffer);
143
144 printk(RTAS_DEBUG "%d -------- %s end ----------\n",
145 error_log_cnt, str);
146 } else {
147 struct rtas_error_log *errlog = (struct rtas_error_log *)buf;
148
149 printk(RTAS_DEBUG "event: %d, Type: %s, Severity: %d\n",
150 error_log_cnt, rtas_event_type(errlog->type),
151 errlog->severity);
152 }
153}
154
155static int log_rtas_len(char * buf)
156{
157 int len;
158 struct rtas_error_log *err;
159
160 /* rtas fixed header */
161 len = 8;
162 err = (struct rtas_error_log *)buf;
163 if (err->extended_log_length) {
164
165 /* extended header */
166 len += err->extended_log_length;
167 }
168
169 if (rtas_error_log_max == 0)
170 rtas_error_log_max = rtas_get_error_log_max();
171
172 if (len > rtas_error_log_max)
173 len = rtas_error_log_max;
174
175 return len;
176}
177
178/*
179 * First write to nvram, if fatal error, that is the only
180 * place we log the info. The error will be picked up
181 * on the next reboot by rtasd. If not fatal, run the
182 * method for the type of error. Currently, only RTAS
183 * errors have methods implemented, but in the future
184 * there might be a need to store data in nvram before a
185 * call to panic().
186 *
187 * XXX We write to nvram periodically, to indicate error has
188 * been written and sync'd, but there is a possibility
189 * that if we don't shutdown correctly, a duplicate error
190 * record will be created on next reboot.
191 */
192void pSeries_log_error(char *buf, unsigned int err_type, int fatal)
193{
194 unsigned long offset;
195 unsigned long s;
196 int len = 0;
197
198 pr_debug("rtasd: logging event\n");
199 if (buf == NULL)
200 return;
201
202 spin_lock_irqsave(&rtasd_log_lock, s);
203
204 /* get length and increase count */
205 switch (err_type & ERR_TYPE_MASK) {
206 case ERR_TYPE_RTAS_LOG:
207 len = log_rtas_len(buf);
208 if (!(err_type & ERR_FLAG_BOOT))
209 error_log_cnt++;
210 break;
211 case ERR_TYPE_KERNEL_PANIC:
212 default:
213 WARN_ON_ONCE(!irqs_disabled()); /* @@@ DEBUG @@@ */
214 spin_unlock_irqrestore(&rtasd_log_lock, s);
215 return;
216 }
217
218#ifdef CONFIG_PPC64
219 /* Write error to NVRAM */
220 if (logging_enabled && !(err_type & ERR_FLAG_BOOT))
221 nvram_write_error_log(buf, len, err_type, error_log_cnt);
222#endif /* CONFIG_PPC64 */
223
224 /*
225 * rtas errors can occur during boot, and we do want to capture
226 * those somewhere, even if nvram isn't ready (why not?), and even
227 * if rtasd isn't ready. Put them into the boot log, at least.
228 */
229 if ((err_type & ERR_TYPE_MASK) == ERR_TYPE_RTAS_LOG)
230 printk_log_rtas(buf, len);
231
232 /* Check to see if we need to or have stopped logging */
233 if (fatal || !logging_enabled) {
234 logging_enabled = 0;
235 WARN_ON_ONCE(!irqs_disabled()); /* @@@ DEBUG @@@ */
236 spin_unlock_irqrestore(&rtasd_log_lock, s);
237 return;
238 }
239
240 /* call type specific method for error */
241 switch (err_type & ERR_TYPE_MASK) {
242 case ERR_TYPE_RTAS_LOG:
243 offset = rtas_error_log_buffer_max *
244 ((rtas_log_start+rtas_log_size) & LOG_NUMBER_MASK);
245
246 /* First copy over sequence number */
247 memcpy(&rtas_log_buf[offset], (void *) &error_log_cnt, sizeof(int));
248
249 /* Second copy over error log data */
250 offset += sizeof(int);
251 memcpy(&rtas_log_buf[offset], buf, len);
252
253 if (rtas_log_size < LOG_NUMBER)
254 rtas_log_size += 1;
255 else
256 rtas_log_start += 1;
257
258 WARN_ON_ONCE(!irqs_disabled()); /* @@@ DEBUG @@@ */
259 spin_unlock_irqrestore(&rtasd_log_lock, s);
260 wake_up_interruptible(&rtas_log_wait);
261 break;
262 case ERR_TYPE_KERNEL_PANIC:
263 default:
264 WARN_ON_ONCE(!irqs_disabled()); /* @@@ DEBUG @@@ */
265 spin_unlock_irqrestore(&rtasd_log_lock, s);
266 return;
267 }
268
269}
270
271static int rtas_log_open(struct inode * inode, struct file * file)
272{
273 return 0;
274}
275
276static int rtas_log_release(struct inode * inode, struct file * file)
277{
278 return 0;
279}
280
281/* This will check if all events are logged, if they are then, we
282 * know that we can safely clear the events in NVRAM.
283 * Next we'll sit and wait for something else to log.
284 */
285static ssize_t rtas_log_read(struct file * file, char __user * buf,
286 size_t count, loff_t *ppos)
287{
288 int error;
289 char *tmp;
290 unsigned long s;
291 unsigned long offset;
292
293 if (!buf || count < rtas_error_log_buffer_max)
294 return -EINVAL;
295
296 count = rtas_error_log_buffer_max;
297
298 if (!access_ok(VERIFY_WRITE, buf, count))
299 return -EFAULT;
300
301 tmp = kmalloc(count, GFP_KERNEL);
302 if (!tmp)
303 return -ENOMEM;
304
305 spin_lock_irqsave(&rtasd_log_lock, s);
306
307 /* if it's 0, then we know we got the last one (the one in NVRAM) */
308 while (rtas_log_size == 0) {
309 if (file->f_flags & O_NONBLOCK) {
310 spin_unlock_irqrestore(&rtasd_log_lock, s);
311 error = -EAGAIN;
312 goto out;
313 }
314
315 if (!logging_enabled) {
316 spin_unlock_irqrestore(&rtasd_log_lock, s);
317 error = -ENODATA;
318 goto out;
319 }
320#ifdef CONFIG_PPC64
321 nvram_clear_error_log();
322#endif /* CONFIG_PPC64 */
323
324 spin_unlock_irqrestore(&rtasd_log_lock, s);
325 error = wait_event_interruptible(rtas_log_wait, rtas_log_size);
326 if (error)
327 goto out;
328 spin_lock_irqsave(&rtasd_log_lock, s);
329 }
330
331 offset = rtas_error_log_buffer_max * (rtas_log_start & LOG_NUMBER_MASK);
332 memcpy(tmp, &rtas_log_buf[offset], count);
333
334 rtas_log_start += 1;
335 rtas_log_size -= 1;
336 spin_unlock_irqrestore(&rtasd_log_lock, s);
337
338 error = copy_to_user(buf, tmp, count) ? -EFAULT : count;
339out:
340 kfree(tmp);
341 return error;
342}
343
344static unsigned int rtas_log_poll(struct file *file, poll_table * wait)
345{
346 poll_wait(file, &rtas_log_wait, wait);
347 if (rtas_log_size)
348 return POLLIN | POLLRDNORM;
349 return 0;
350}
351
352static const struct file_operations proc_rtas_log_operations = {
353 .read = rtas_log_read,
354 .poll = rtas_log_poll,
355 .open = rtas_log_open,
356 .release = rtas_log_release,
357};
358
359static int enable_surveillance(int timeout)
360{
361 int error;
362
363 error = rtas_set_indicator(SURVEILLANCE_TOKEN, 0, timeout);
364
365 if (error == 0)
366 return 0;
367
368 if (error == -EINVAL) {
369 printk(KERN_DEBUG "rtasd: surveillance not supported\n");
370 return 0;
371 }
372
373 printk(KERN_ERR "rtasd: could not update surveillance\n");
374 return -1;
375}
376
377static void do_event_scan(void)
378{
379 int error;
380 do {
381 memset(logdata, 0, rtas_error_log_max);
382 error = rtas_call(event_scan, 4, 1, NULL,
383 RTAS_EVENT_SCAN_ALL_EVENTS, 0,
384 __pa(logdata), rtas_error_log_max);
385 if (error == -1) {
386 printk(KERN_ERR "event-scan failed\n");
387 break;
388 }
389
390 if (error == 0)
391 pSeries_log_error(logdata, ERR_TYPE_RTAS_LOG, 0);
392
393 } while(error == 0);
394}
395
396static void rtas_event_scan(struct work_struct *w);
397DECLARE_DELAYED_WORK(event_scan_work, rtas_event_scan);
398
399/*
400 * Delay should be at least one second since some machines have problems if
401 * we call event-scan too quickly.
402 */
403static unsigned long event_scan_delay = 1*HZ;
404static int first_pass = 1;
405
406static void rtas_event_scan(struct work_struct *w)
407{
408 unsigned int cpu;
409
410 do_event_scan();
411
412 get_online_cpus();
413
414 cpu = next_cpu(smp_processor_id(), cpu_online_map);
415 if (cpu == NR_CPUS) {
416 cpu = first_cpu(cpu_online_map);
417
418 if (first_pass) {
419 first_pass = 0;
420 event_scan_delay = 30*HZ/rtas_event_scan_rate;
421
422 if (surveillance_timeout != -1) {
423 pr_debug("rtasd: enabling surveillance\n");
424 enable_surveillance(surveillance_timeout);
425 pr_debug("rtasd: surveillance enabled\n");
426 }
427 }
428 }
429
430 schedule_delayed_work_on(cpu, &event_scan_work,
431 __round_jiffies_relative(event_scan_delay, cpu));
432
433 put_online_cpus();
434}
435
436#ifdef CONFIG_PPC64
437static void retreive_nvram_error_log(void)
438{
439 unsigned int err_type ;
440 int rc ;
441
442 /* See if we have any error stored in NVRAM */
443 memset(logdata, 0, rtas_error_log_max);
444 rc = nvram_read_error_log(logdata, rtas_error_log_max,
445 &err_type, &error_log_cnt);
446 /* We can use rtas_log_buf now */
447 logging_enabled = 1;
448 if (!rc) {
449 if (err_type != ERR_FLAG_ALREADY_LOGGED) {
450 pSeries_log_error(logdata, err_type | ERR_FLAG_BOOT, 0);
451 }
452 }
453}
454#else /* CONFIG_PPC64 */
455static void retreive_nvram_error_log(void)
456{
457}
458#endif /* CONFIG_PPC64 */
459
460static void start_event_scan(void)
461{
462 printk(KERN_DEBUG "RTAS daemon started\n");
463 pr_debug("rtasd: will sleep for %d milliseconds\n",
464 (30000 / rtas_event_scan_rate));
465
466 /* Retreive errors from nvram if any */
467 retreive_nvram_error_log();
468
469 schedule_delayed_work_on(first_cpu(cpu_online_map), &event_scan_work,
470 event_scan_delay);
471}
472
473static int __init rtas_init(void)
474{
475 struct proc_dir_entry *entry;
476
477 if (!machine_is(pseries) && !machine_is(chrp))
478 return 0;
479
480 /* No RTAS */
481 event_scan = rtas_token("event-scan");
482 if (event_scan == RTAS_UNKNOWN_SERVICE) {
483 printk(KERN_INFO "rtasd: No event-scan on system\n");
484 return -ENODEV;
485 }
486
487 rtas_event_scan_rate = rtas_token("rtas-event-scan-rate");
488 if (rtas_event_scan_rate == RTAS_UNKNOWN_SERVICE) {
489 printk(KERN_ERR "rtasd: no rtas-event-scan-rate on system\n");
490 return -ENODEV;
491 }
492
493 /* Make room for the sequence number */
494 rtas_error_log_max = rtas_get_error_log_max();
495 rtas_error_log_buffer_max = rtas_error_log_max + sizeof(int);
496
497 rtas_log_buf = vmalloc(rtas_error_log_buffer_max*LOG_NUMBER);
498 if (!rtas_log_buf) {
499 printk(KERN_ERR "rtasd: no memory\n");
500 return -ENOMEM;
501 }
502
503 entry = proc_create("powerpc/rtas/error_log", S_IRUSR, NULL,
504 &proc_rtas_log_operations);
505 if (!entry)
506 printk(KERN_ERR "Failed to create error_log proc entry\n");
507
508 start_event_scan();
509
510 return 0;
511}
512__initcall(rtas_init);
513
514static int __init surveillance_setup(char *str)
515{
516 int i;
517
518 /* We only do surveillance on pseries */
519 if (!machine_is(pseries))
520 return 0;
521
522 if (get_option(&str,&i)) {
523 if (i >= 0 && i <= 255)
524 surveillance_timeout = i;
525 }
526
527 return 1;
528}
529__setup("surveillance=", surveillance_setup);
530
531static int __init rtasmsgs_setup(char *str)
532{
533 if (strcmp(str, "on") == 0)
534 full_rtas_msgs = 1;
535 else if (strcmp(str, "off") == 0)
536 full_rtas_msgs = 0;
537
538 return 1;
539}
540__setup("rtasmsgs=", rtasmsgs_setup);
diff --git a/arch/powerpc/kernel/setup-common.c b/arch/powerpc/kernel/setup-common.c
index 4271f7a655a3..48f0a008b20b 100644
--- a/arch/powerpc/kernel/setup-common.c
+++ b/arch/powerpc/kernel/setup-common.c
@@ -36,6 +36,7 @@
36#include <linux/lmb.h> 36#include <linux/lmb.h>
37#include <linux/of_platform.h> 37#include <linux/of_platform.h>
38#include <asm/io.h> 38#include <asm/io.h>
39#include <asm/paca.h>
39#include <asm/prom.h> 40#include <asm/prom.h>
40#include <asm/processor.h> 41#include <asm/processor.h>
41#include <asm/vdso_datapage.h> 42#include <asm/vdso_datapage.h>
@@ -157,7 +158,7 @@ extern u32 cpu_temp_both(unsigned long cpu);
157#endif /* CONFIG_TAU */ 158#endif /* CONFIG_TAU */
158 159
159#ifdef CONFIG_SMP 160#ifdef CONFIG_SMP
160DEFINE_PER_CPU(unsigned int, pvr); 161DEFINE_PER_CPU(unsigned int, cpu_pvr);
161#endif 162#endif
162 163
163static int show_cpuinfo(struct seq_file *m, void *v) 164static int show_cpuinfo(struct seq_file *m, void *v)
@@ -209,7 +210,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
209 } 210 }
210 211
211#ifdef CONFIG_SMP 212#ifdef CONFIG_SMP
212 pvr = per_cpu(pvr, cpu_id); 213 pvr = per_cpu(cpu_pvr, cpu_id);
213#else 214#else
214 pvr = mfspr(SPRN_PVR); 215 pvr = mfspr(SPRN_PVR);
215#endif 216#endif
@@ -493,6 +494,8 @@ void __init smp_setup_cpu_maps(void)
493 * here will have to be reworked 494 * here will have to be reworked
494 */ 495 */
495 cpu_init_thread_core_maps(nthreads); 496 cpu_init_thread_core_maps(nthreads);
497
498 free_unused_pacas();
496} 499}
497#endif /* CONFIG_SMP */ 500#endif /* CONFIG_SMP */
498 501
@@ -660,6 +663,7 @@ late_initcall(check_cache_coherency);
660 663
661#ifdef CONFIG_DEBUG_FS 664#ifdef CONFIG_DEBUG_FS
662struct dentry *powerpc_debugfs_root; 665struct dentry *powerpc_debugfs_root;
666EXPORT_SYMBOL(powerpc_debugfs_root);
663 667
664static int powerpc_debugfs_init(void) 668static int powerpc_debugfs_init(void)
665{ 669{
diff --git a/arch/powerpc/kernel/setup_32.c b/arch/powerpc/kernel/setup_32.c
index 53bcf3d792db..8f58986c2ad9 100644
--- a/arch/powerpc/kernel/setup_32.c
+++ b/arch/powerpc/kernel/setup_32.c
@@ -39,7 +39,6 @@
39#include <asm/serial.h> 39#include <asm/serial.h>
40#include <asm/udbg.h> 40#include <asm/udbg.h>
41#include <asm/mmu_context.h> 41#include <asm/mmu_context.h>
42#include <asm/swiotlb.h>
43 42
44#include "setup.h" 43#include "setup.h"
45 44
@@ -343,11 +342,6 @@ void __init setup_arch(char **cmdline_p)
343 ppc_md.setup_arch(); 342 ppc_md.setup_arch();
344 if ( ppc_md.progress ) ppc_md.progress("arch: exit", 0x3eab); 343 if ( ppc_md.progress ) ppc_md.progress("arch: exit", 0x3eab);
345 344
346#ifdef CONFIG_SWIOTLB
347 if (ppc_swiotlb_enable)
348 swiotlb_init();
349#endif
350
351 paging_init(); 345 paging_init();
352 346
353 /* Initialize the MMU context management stuff */ 347 /* Initialize the MMU context management stuff */
diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c
index 04f638d82fb3..914389158a9b 100644
--- a/arch/powerpc/kernel/setup_64.c
+++ b/arch/powerpc/kernel/setup_64.c
@@ -61,7 +61,6 @@
61#include <asm/xmon.h> 61#include <asm/xmon.h>
62#include <asm/udbg.h> 62#include <asm/udbg.h>
63#include <asm/kexec.h> 63#include <asm/kexec.h>
64#include <asm/swiotlb.h>
65#include <asm/mmu_context.h> 64#include <asm/mmu_context.h>
66 65
67#include "setup.h" 66#include "setup.h"
@@ -144,9 +143,9 @@ early_param("smt-enabled", early_smt_enabled);
144#endif /* CONFIG_SMP */ 143#endif /* CONFIG_SMP */
145 144
146/* Put the paca pointer into r13 and SPRG_PACA */ 145/* Put the paca pointer into r13 and SPRG_PACA */
147void __init setup_paca(int cpu) 146static void __init setup_paca(struct paca_struct *new_paca)
148{ 147{
149 local_paca = &paca[cpu]; 148 local_paca = new_paca;
150 mtspr(SPRN_SPRG_PACA, local_paca); 149 mtspr(SPRN_SPRG_PACA, local_paca);
151#ifdef CONFIG_PPC_BOOK3E 150#ifdef CONFIG_PPC_BOOK3E
152 mtspr(SPRN_SPRG_TLB_EXFRAME, local_paca->extlb); 151 mtspr(SPRN_SPRG_TLB_EXFRAME, local_paca->extlb);
@@ -176,14 +175,12 @@ void __init early_setup(unsigned long dt_ptr)
176{ 175{
177 /* -------- printk is _NOT_ safe to use here ! ------- */ 176 /* -------- printk is _NOT_ safe to use here ! ------- */
178 177
179 /* Fill in any unititialised pacas */
180 initialise_pacas();
181
182 /* Identify CPU type */ 178 /* Identify CPU type */
183 identify_cpu(0, mfspr(SPRN_PVR)); 179 identify_cpu(0, mfspr(SPRN_PVR));
184 180
185 /* Assume we're on cpu 0 for now. Don't write to the paca yet! */ 181 /* Assume we're on cpu 0 for now. Don't write to the paca yet! */
186 setup_paca(0); 182 initialise_paca(&boot_paca, 0);
183 setup_paca(&boot_paca);
187 184
188 /* Initialize lockdep early or else spinlocks will blow */ 185 /* Initialize lockdep early or else spinlocks will blow */
189 lockdep_init(); 186 lockdep_init();
@@ -203,7 +200,7 @@ void __init early_setup(unsigned long dt_ptr)
203 early_init_devtree(__va(dt_ptr)); 200 early_init_devtree(__va(dt_ptr));
204 201
205 /* Now we know the logical id of our boot cpu, setup the paca. */ 202 /* Now we know the logical id of our boot cpu, setup the paca. */
206 setup_paca(boot_cpuid); 203 setup_paca(&paca[boot_cpuid]);
207 204
208 /* Fix up paca fields required for the boot cpu */ 205 /* Fix up paca fields required for the boot cpu */
209 get_paca()->cpu_start = 1; 206 get_paca()->cpu_start = 1;
@@ -356,11 +353,6 @@ void __init setup_system(void)
356 */ 353 */
357 initialize_cache_info(); 354 initialize_cache_info();
358 355
359 /*
360 * Initialize irq remapping subsystem
361 */
362 irq_early_init();
363
364#ifdef CONFIG_PPC_RTAS 356#ifdef CONFIG_PPC_RTAS
365 /* 357 /*
366 * Initialize RTAS if available 358 * Initialize RTAS if available
@@ -548,11 +540,6 @@ void __init setup_arch(char **cmdline_p)
548 if (ppc_md.setup_arch) 540 if (ppc_md.setup_arch)
549 ppc_md.setup_arch(); 541 ppc_md.setup_arch();
550 542
551#ifdef CONFIG_SWIOTLB
552 if (ppc_swiotlb_enable)
553 swiotlb_init();
554#endif
555
556 paging_init(); 543 paging_init();
557 544
558 /* Initialize the MMU context management stuff */ 545 /* Initialize the MMU context management stuff */
diff --git a/arch/powerpc/kernel/signal.c b/arch/powerpc/kernel/signal.c
index 00b5078da9a3..a0afb555a7c9 100644
--- a/arch/powerpc/kernel/signal.c
+++ b/arch/powerpc/kernel/signal.c
@@ -140,17 +140,15 @@ static int do_signal_pending(sigset_t *oldset, struct pt_regs *regs)
140 return 0; /* no signals delivered */ 140 return 0; /* no signals delivered */
141 } 141 }
142 142
143#ifndef CONFIG_PPC_ADV_DEBUG_REGS
143 /* 144 /*
144 * Reenable the DABR before delivering the signal to 145 * Reenable the DABR before delivering the signal to
145 * user space. The DABR will have been cleared if it 146 * user space. The DABR will have been cleared if it
146 * triggered inside the kernel. 147 * triggered inside the kernel.
147 */ 148 */
148 if (current->thread.dabr) { 149 if (current->thread.dabr)
149 set_dabr(current->thread.dabr); 150 set_dabr(current->thread.dabr);
150#if defined(CONFIG_BOOKE)
151 mtspr(SPRN_DBCR0, current->thread.dbcr0);
152#endif 151#endif
153 }
154 152
155 if (is32) { 153 if (is32) {
156 if (ka.sa.sa_flags & SA_SIGINFO) 154 if (ka.sa.sa_flags & SA_SIGINFO)
diff --git a/arch/powerpc/kernel/signal_32.c b/arch/powerpc/kernel/signal_32.c
index d670429a1608..266610119f66 100644
--- a/arch/powerpc/kernel/signal_32.c
+++ b/arch/powerpc/kernel/signal_32.c
@@ -1078,7 +1078,7 @@ int sys_debug_setcontext(struct ucontext __user *ctx,
1078 int i; 1078 int i;
1079 unsigned char tmp; 1079 unsigned char tmp;
1080 unsigned long new_msr = regs->msr; 1080 unsigned long new_msr = regs->msr;
1081#if defined(CONFIG_4xx) || defined(CONFIG_BOOKE) 1081#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1082 unsigned long new_dbcr0 = current->thread.dbcr0; 1082 unsigned long new_dbcr0 = current->thread.dbcr0;
1083#endif 1083#endif
1084 1084
@@ -1087,13 +1087,17 @@ int sys_debug_setcontext(struct ucontext __user *ctx,
1087 return -EFAULT; 1087 return -EFAULT;
1088 switch (op.dbg_type) { 1088 switch (op.dbg_type) {
1089 case SIG_DBG_SINGLE_STEPPING: 1089 case SIG_DBG_SINGLE_STEPPING:
1090#if defined(CONFIG_4xx) || defined(CONFIG_BOOKE) 1090#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1091 if (op.dbg_value) { 1091 if (op.dbg_value) {
1092 new_msr |= MSR_DE; 1092 new_msr |= MSR_DE;
1093 new_dbcr0 |= (DBCR0_IDM | DBCR0_IC); 1093 new_dbcr0 |= (DBCR0_IDM | DBCR0_IC);
1094 } else { 1094 } else {
1095 new_msr &= ~MSR_DE; 1095 new_dbcr0 &= ~DBCR0_IC;
1096 new_dbcr0 &= ~(DBCR0_IDM | DBCR0_IC); 1096 if (!DBCR_ACTIVE_EVENTS(new_dbcr0,
1097 current->thread.dbcr1)) {
1098 new_msr &= ~MSR_DE;
1099 new_dbcr0 &= ~DBCR0_IDM;
1100 }
1097 } 1101 }
1098#else 1102#else
1099 if (op.dbg_value) 1103 if (op.dbg_value)
@@ -1103,7 +1107,7 @@ int sys_debug_setcontext(struct ucontext __user *ctx,
1103#endif 1107#endif
1104 break; 1108 break;
1105 case SIG_DBG_BRANCH_TRACING: 1109 case SIG_DBG_BRANCH_TRACING:
1106#if defined(CONFIG_4xx) || defined(CONFIG_BOOKE) 1110#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1107 return -EINVAL; 1111 return -EINVAL;
1108#else 1112#else
1109 if (op.dbg_value) 1113 if (op.dbg_value)
@@ -1124,7 +1128,7 @@ int sys_debug_setcontext(struct ucontext __user *ctx,
1124 failure is a problem, anyway, and it's very unlikely unless 1128 failure is a problem, anyway, and it's very unlikely unless
1125 the user is really doing something wrong. */ 1129 the user is really doing something wrong. */
1126 regs->msr = new_msr; 1130 regs->msr = new_msr;
1127#if defined(CONFIG_4xx) || defined(CONFIG_BOOKE) 1131#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1128 current->thread.dbcr0 = new_dbcr0; 1132 current->thread.dbcr0 = new_dbcr0;
1129#endif 1133#endif
1130 1134
diff --git a/arch/powerpc/kernel/smp-tbsync.c b/arch/powerpc/kernel/smp-tbsync.c
index a5e54526403d..03e45c4a9ef1 100644
--- a/arch/powerpc/kernel/smp-tbsync.c
+++ b/arch/powerpc/kernel/smp-tbsync.c
@@ -10,6 +10,7 @@
10#include <linux/smp.h> 10#include <linux/smp.h>
11#include <linux/unistd.h> 11#include <linux/unistd.h>
12#include <linux/init.h> 12#include <linux/init.h>
13#include <linux/slab.h>
13#include <asm/atomic.h> 14#include <asm/atomic.h>
14#include <asm/smp.h> 15#include <asm/smp.h>
15#include <asm/time.h> 16#include <asm/time.h>
diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c
index 9b86a74d2815..c2ee14498077 100644
--- a/arch/powerpc/kernel/smp.c
+++ b/arch/powerpc/kernel/smp.c
@@ -218,6 +218,9 @@ void crash_send_ipi(void (*crash_ipi_callback)(struct pt_regs *))
218 218
219static void stop_this_cpu(void *dummy) 219static void stop_this_cpu(void *dummy)
220{ 220{
221 /* Remove this CPU */
222 set_cpu_online(smp_processor_id(), false);
223
221 local_irq_disable(); 224 local_irq_disable();
222 while (1) 225 while (1)
223 ; 226 ;
@@ -232,7 +235,7 @@ struct thread_info *current_set[NR_CPUS];
232 235
233static void __devinit smp_store_cpu_info(int id) 236static void __devinit smp_store_cpu_info(int id)
234{ 237{
235 per_cpu(pvr, id) = mfspr(SPRN_PVR); 238 per_cpu(cpu_pvr, id) = mfspr(SPRN_PVR);
236} 239}
237 240
238static void __init smp_create_idle(unsigned int cpu) 241static void __init smp_create_idle(unsigned int cpu)
@@ -616,4 +619,16 @@ void __cpu_die(unsigned int cpu)
616 if (smp_ops->cpu_die) 619 if (smp_ops->cpu_die)
617 smp_ops->cpu_die(cpu); 620 smp_ops->cpu_die(cpu);
618} 621}
622
623static DEFINE_MUTEX(powerpc_cpu_hotplug_driver_mutex);
624
625void cpu_hotplug_driver_lock()
626{
627 mutex_lock(&powerpc_cpu_hotplug_driver_mutex);
628}
629
630void cpu_hotplug_driver_unlock()
631{
632 mutex_unlock(&powerpc_cpu_hotplug_driver_mutex);
633}
619#endif 634#endif
diff --git a/arch/powerpc/kernel/softemu8xx.c b/arch/powerpc/kernel/softemu8xx.c
index 23c8c5e7dc4d..af0e8290b4fc 100644
--- a/arch/powerpc/kernel/softemu8xx.c
+++ b/arch/powerpc/kernel/softemu8xx.c
@@ -21,7 +21,6 @@
21#include <linux/stddef.h> 21#include <linux/stddef.h>
22#include <linux/unistd.h> 22#include <linux/unistd.h>
23#include <linux/ptrace.h> 23#include <linux/ptrace.h>
24#include <linux/slab.h>
25#include <linux/user.h> 24#include <linux/user.h>
26#include <linux/interrupt.h> 25#include <linux/interrupt.h>
27 26
diff --git a/arch/powerpc/kernel/swsusp_32.S b/arch/powerpc/kernel/swsusp_32.S
index b47d8ceffb52..b0754e237438 100644
--- a/arch/powerpc/kernel/swsusp_32.S
+++ b/arch/powerpc/kernel/swsusp_32.S
@@ -303,7 +303,7 @@ END_MMU_FTR_SECTION_IFSET(MMU_FTR_USE_HIGH_BATS)
303 lis r4,0x1000 303 lis r4,0x1000
3041: addic. r4,r4,-0x1000 3041: addic. r4,r4,-0x1000
305 tlbie r4 305 tlbie r4
306 blt 1b 306 bgt 1b
307 sync 307 sync
308 308
309 /* restore the MSR and turn on the MMU */ 309 /* restore the MSR and turn on the MMU */
diff --git a/arch/powerpc/kernel/sys_ppc32.c b/arch/powerpc/kernel/sys_ppc32.c
index b97c2d67f4ac..19471a1cef1a 100644
--- a/arch/powerpc/kernel/sys_ppc32.c
+++ b/arch/powerpc/kernel/sys_ppc32.c
@@ -41,6 +41,7 @@
41#include <linux/ptrace.h> 41#include <linux/ptrace.h>
42#include <linux/elf.h> 42#include <linux/elf.h>
43#include <linux/ipc.h> 43#include <linux/ipc.h>
44#include <linux/slab.h>
44 45
45#include <asm/ptrace.h> 46#include <asm/ptrace.h>
46#include <asm/types.h> 47#include <asm/types.h>
@@ -520,58 +521,6 @@ asmlinkage long compat_sys_umask(u32 mask)
520 return sys_umask((int)mask); 521 return sys_umask((int)mask);
521} 522}
522 523
523#ifdef CONFIG_SYSCTL_SYSCALL
524struct __sysctl_args32 {
525 u32 name;
526 int nlen;
527 u32 oldval;
528 u32 oldlenp;
529 u32 newval;
530 u32 newlen;
531 u32 __unused[4];
532};
533
534asmlinkage long compat_sys_sysctl(struct __sysctl_args32 __user *args)
535{
536 struct __sysctl_args32 tmp;
537 int error;
538 size_t oldlen;
539 size_t __user *oldlenp = NULL;
540 unsigned long addr = (((unsigned long)&args->__unused[0]) + 7) & ~7;
541
542 if (copy_from_user(&tmp, args, sizeof(tmp)))
543 return -EFAULT;
544
545 if (tmp.oldval && tmp.oldlenp) {
546 /* Duh, this is ugly and might not work if sysctl_args
547 is in read-only memory, but do_sysctl does indirectly
548 a lot of uaccess in both directions and we'd have to
549 basically copy the whole sysctl.c here, and
550 glibc's __sysctl uses rw memory for the structure
551 anyway. */
552 oldlenp = (size_t __user *)addr;
553 if (get_user(oldlen, (compat_size_t __user *)compat_ptr(tmp.oldlenp)) ||
554 put_user(oldlen, oldlenp))
555 return -EFAULT;
556 }
557
558 lock_kernel();
559 error = do_sysctl(compat_ptr(tmp.name), tmp.nlen,
560 compat_ptr(tmp.oldval), oldlenp,
561 compat_ptr(tmp.newval), tmp.newlen);
562 unlock_kernel();
563 if (oldlenp) {
564 if (!error) {
565 if (get_user(oldlen, oldlenp) ||
566 put_user(oldlen, (compat_size_t __user *)compat_ptr(tmp.oldlenp)))
567 error = -EFAULT;
568 }
569 copy_to_user(args->__unused, tmp.__unused, sizeof(tmp.__unused));
570 }
571 return error;
572}
573#endif
574
575unsigned long compat_sys_mmap2(unsigned long addr, size_t len, 524unsigned long compat_sys_mmap2(unsigned long addr, size_t len,
576 unsigned long prot, unsigned long flags, 525 unsigned long prot, unsigned long flags,
577 unsigned long fd, unsigned long pgoff) 526 unsigned long fd, unsigned long pgoff)
diff --git a/arch/powerpc/kernel/syscalls.c b/arch/powerpc/kernel/syscalls.c
index c04832c4a02e..f2496f2faecc 100644
--- a/arch/powerpc/kernel/syscalls.c
+++ b/arch/powerpc/kernel/syscalls.c
@@ -42,105 +42,10 @@
42#include <asm/time.h> 42#include <asm/time.h>
43#include <asm/unistd.h> 43#include <asm/unistd.h>
44 44
45/*
46 * sys_ipc() is the de-multiplexer for the SysV IPC calls..
47 *
48 * This is really horribly ugly.
49 */
50int sys_ipc(uint call, int first, unsigned long second, long third,
51 void __user *ptr, long fifth)
52{
53 int version, ret;
54
55 version = call >> 16; /* hack for backward compatibility */
56 call &= 0xffff;
57
58 ret = -ENOSYS;
59 switch (call) {
60 case SEMOP:
61 ret = sys_semtimedop(first, (struct sembuf __user *)ptr,
62 (unsigned)second, NULL);
63 break;
64 case SEMTIMEDOP:
65 ret = sys_semtimedop(first, (struct sembuf __user *)ptr,
66 (unsigned)second,
67 (const struct timespec __user *) fifth);
68 break;
69 case SEMGET:
70 ret = sys_semget (first, (int)second, third);
71 break;
72 case SEMCTL: {
73 union semun fourth;
74
75 ret = -EINVAL;
76 if (!ptr)
77 break;
78 if ((ret = get_user(fourth.__pad, (void __user * __user *)ptr)))
79 break;
80 ret = sys_semctl(first, (int)second, third, fourth);
81 break;
82 }
83 case MSGSND:
84 ret = sys_msgsnd(first, (struct msgbuf __user *)ptr,
85 (size_t)second, third);
86 break;
87 case MSGRCV:
88 switch (version) {
89 case 0: {
90 struct ipc_kludge tmp;
91
92 ret = -EINVAL;
93 if (!ptr)
94 break;
95 if ((ret = copy_from_user(&tmp,
96 (struct ipc_kludge __user *) ptr,
97 sizeof (tmp)) ? -EFAULT : 0))
98 break;
99 ret = sys_msgrcv(first, tmp.msgp, (size_t) second,
100 tmp.msgtyp, third);
101 break;
102 }
103 default:
104 ret = sys_msgrcv (first, (struct msgbuf __user *) ptr,
105 (size_t)second, fifth, third);
106 break;
107 }
108 break;
109 case MSGGET:
110 ret = sys_msgget((key_t)first, (int)second);
111 break;
112 case MSGCTL:
113 ret = sys_msgctl(first, (int)second,
114 (struct msqid_ds __user *)ptr);
115 break;
116 case SHMAT: {
117 ulong raddr;
118 ret = do_shmat(first, (char __user *)ptr, (int)second, &raddr);
119 if (ret)
120 break;
121 ret = put_user(raddr, (ulong __user *) third);
122 break;
123 }
124 case SHMDT:
125 ret = sys_shmdt((char __user *)ptr);
126 break;
127 case SHMGET:
128 ret = sys_shmget(first, (size_t)second, third);
129 break;
130 case SHMCTL:
131 ret = sys_shmctl(first, (int)second,
132 (struct shmid_ds __user *)ptr);
133 break;
134 }
135
136 return ret;
137}
138
139static inline unsigned long do_mmap2(unsigned long addr, size_t len, 45static inline unsigned long do_mmap2(unsigned long addr, size_t len,
140 unsigned long prot, unsigned long flags, 46 unsigned long prot, unsigned long flags,
141 unsigned long fd, unsigned long off, int shift) 47 unsigned long fd, unsigned long off, int shift)
142{ 48{
143 struct file * file = NULL;
144 unsigned long ret = -EINVAL; 49 unsigned long ret = -EINVAL;
145 50
146 if (!arch_validate_prot(prot)) 51 if (!arch_validate_prot(prot))
@@ -151,20 +56,8 @@ static inline unsigned long do_mmap2(unsigned long addr, size_t len,
151 goto out; 56 goto out;
152 off >>= shift; 57 off >>= shift;
153 } 58 }
154
155 ret = -EBADF;
156 if (!(flags & MAP_ANONYMOUS)) {
157 if (!(file = fget(fd)))
158 goto out;
159 }
160 59
161 flags &= ~(MAP_EXECUTABLE | MAP_DENYWRITE); 60 ret = sys_mmap_pgoff(addr, len, prot, flags, fd, off);
162
163 down_write(&current->mm->mmap_sem);
164 ret = do_mmap_pgoff(file, addr, len, prot, flags, off);
165 up_write(&current->mm->mmap_sem);
166 if (file)
167 fput(file);
168out: 61out:
169 return ret; 62 return ret;
170} 63}
@@ -223,76 +116,6 @@ long ppc64_personality(unsigned long personality)
223} 116}
224#endif 117#endif
225 118
226#ifdef CONFIG_PPC64
227#define OVERRIDE_MACHINE (personality(current->personality) == PER_LINUX32)
228#else
229#define OVERRIDE_MACHINE 0
230#endif
231
232static inline int override_machine(char __user *mach)
233{
234 if (OVERRIDE_MACHINE) {
235 /* change ppc64 to ppc */
236 if (__put_user(0, mach+3) || __put_user(0, mach+4))
237 return -EFAULT;
238 }
239 return 0;
240}
241
242long ppc_newuname(struct new_utsname __user * name)
243{
244 int err = 0;
245
246 down_read(&uts_sem);
247 if (copy_to_user(name, utsname(), sizeof(*name)))
248 err = -EFAULT;
249 up_read(&uts_sem);
250 if (!err)
251 err = override_machine(name->machine);
252 return err;
253}
254
255int sys_uname(struct old_utsname __user *name)
256{
257 int err = 0;
258
259 down_read(&uts_sem);
260 if (copy_to_user(name, utsname(), sizeof(*name)))
261 err = -EFAULT;
262 up_read(&uts_sem);
263 if (!err)
264 err = override_machine(name->machine);
265 return err;
266}
267
268int sys_olduname(struct oldold_utsname __user *name)
269{
270 int error;
271
272 if (!access_ok(VERIFY_WRITE, name, sizeof(struct oldold_utsname)))
273 return -EFAULT;
274
275 down_read(&uts_sem);
276 error = __copy_to_user(&name->sysname, &utsname()->sysname,
277 __OLD_UTS_LEN);
278 error |= __put_user(0, name->sysname + __OLD_UTS_LEN);
279 error |= __copy_to_user(&name->nodename, &utsname()->nodename,
280 __OLD_UTS_LEN);
281 error |= __put_user(0, name->nodename + __OLD_UTS_LEN);
282 error |= __copy_to_user(&name->release, &utsname()->release,
283 __OLD_UTS_LEN);
284 error |= __put_user(0, name->release + __OLD_UTS_LEN);
285 error |= __copy_to_user(&name->version, &utsname()->version,
286 __OLD_UTS_LEN);
287 error |= __put_user(0, name->version + __OLD_UTS_LEN);
288 error |= __copy_to_user(&name->machine, &utsname()->machine,
289 __OLD_UTS_LEN);
290 error |= override_machine(name->machine);
291 up_read(&uts_sem);
292
293 return error? -EFAULT: 0;
294}
295
296long ppc_fadvise64_64(int fd, int advice, u32 offset_high, u32 offset_low, 119long ppc_fadvise64_64(int fd, int advice, u32 offset_high, u32 offset_low,
297 u32 len_high, u32 len_low) 120 u32 len_high, u32 len_low)
298{ 121{
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 956ab33fd73f..e235e52dc4fe 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -461,6 +461,25 @@ static void unregister_cpu_online(unsigned int cpu)
461 461
462 cacheinfo_cpu_offline(cpu); 462 cacheinfo_cpu_offline(cpu);
463} 463}
464
465#ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
466ssize_t arch_cpu_probe(const char *buf, size_t count)
467{
468 if (ppc_md.cpu_probe)
469 return ppc_md.cpu_probe(buf, count);
470
471 return -EINVAL;
472}
473
474ssize_t arch_cpu_release(const char *buf, size_t count)
475{
476 if (ppc_md.cpu_release)
477 return ppc_md.cpu_release(buf, count);
478
479 return -EINVAL;
480}
481#endif /* CONFIG_ARCH_CPU_PROBE_RELEASE */
482
464#endif /* CONFIG_HOTPLUG_CPU */ 483#endif /* CONFIG_HOTPLUG_CPU */
465 484
466static int __cpuinit sysfs_cpu_notify(struct notifier_block *self, 485static int __cpuinit sysfs_cpu_notify(struct notifier_block *self,
diff --git a/arch/powerpc/kernel/tau_6xx.c b/arch/powerpc/kernel/tau_6xx.c
index c3a56d65c5a9..a753b72efbc0 100644
--- a/arch/powerpc/kernel/tau_6xx.c
+++ b/arch/powerpc/kernel/tau_6xx.c
@@ -59,7 +59,7 @@ void set_thresholds(unsigned long cpu)
59 mtspr(SPRN_THRM1, THRM1_THRES(tau[cpu].low) | THRM1_V | THRM1_TIE | THRM1_TID); 59 mtspr(SPRN_THRM1, THRM1_THRES(tau[cpu].low) | THRM1_V | THRM1_TIE | THRM1_TID);
60 60
61 /* setup THRM2, 61 /* setup THRM2,
62 * threshold, valid bit, enable interrupts, interrupt when above threshhold 62 * threshold, valid bit, enable interrupts, interrupt when above threshold
63 */ 63 */
64 mtspr (SPRN_THRM2, THRM1_THRES(tau[cpu].high) | THRM1_V | THRM1_TIE); 64 mtspr (SPRN_THRM2, THRM1_THRES(tau[cpu].high) | THRM1_V | THRM1_TIE);
65#else 65#else
diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
index a136a11c490d..0441bbdadbd1 100644
--- a/arch/powerpc/kernel/time.c
+++ b/arch/powerpc/kernel/time.c
@@ -54,6 +54,7 @@
54#include <linux/irq.h> 54#include <linux/irq.h>
55#include <linux/delay.h> 55#include <linux/delay.h>
56#include <linux/perf_event.h> 56#include <linux/perf_event.h>
57#include <asm/trace.h>
57 58
58#include <asm/io.h> 59#include <asm/io.h>
59#include <asm/processor.h> 60#include <asm/processor.h>
@@ -264,10 +265,11 @@ void account_system_vtime(struct task_struct *tsk)
264 account_system_time(tsk, 0, delta, deltascaled); 265 account_system_time(tsk, 0, delta, deltascaled);
265 else 266 else
266 account_idle_time(delta); 267 account_idle_time(delta);
267 per_cpu(cputime_last_delta, smp_processor_id()) = delta; 268 __get_cpu_var(cputime_last_delta) = delta;
268 per_cpu(cputime_scaled_last_delta, smp_processor_id()) = deltascaled; 269 __get_cpu_var(cputime_scaled_last_delta) = deltascaled;
269 local_irq_restore(flags); 270 local_irq_restore(flags);
270} 271}
272EXPORT_SYMBOL_GPL(account_system_vtime);
271 273
272/* 274/*
273 * Transfer the user and system times accumulated in the paca 275 * Transfer the user and system times accumulated in the paca
@@ -530,25 +532,60 @@ void __init iSeries_time_init_early(void)
530} 532}
531#endif /* CONFIG_PPC_ISERIES */ 533#endif /* CONFIG_PPC_ISERIES */
532 534
533#if defined(CONFIG_PERF_EVENTS) && defined(CONFIG_PPC32) 535#ifdef CONFIG_PERF_EVENTS
534DEFINE_PER_CPU(u8, perf_event_pending);
535 536
536void set_perf_event_pending(void) 537/*
538 * 64-bit uses a byte in the PACA, 32-bit uses a per-cpu variable...
539 */
540#ifdef CONFIG_PPC64
541static inline unsigned long test_perf_event_pending(void)
537{ 542{
538 get_cpu_var(perf_event_pending) = 1; 543 unsigned long x;
539 set_dec(1); 544
540 put_cpu_var(perf_event_pending); 545 asm volatile("lbz %0,%1(13)"
546 : "=r" (x)
547 : "i" (offsetof(struct paca_struct, perf_event_pending)));
548 return x;
549}
550
551static inline void set_perf_event_pending_flag(void)
552{
553 asm volatile("stb %0,%1(13)" : :
554 "r" (1),
555 "i" (offsetof(struct paca_struct, perf_event_pending)));
541} 556}
542 557
558static inline void clear_perf_event_pending(void)
559{
560 asm volatile("stb %0,%1(13)" : :
561 "r" (0),
562 "i" (offsetof(struct paca_struct, perf_event_pending)));
563}
564
565#else /* 32-bit */
566
567DEFINE_PER_CPU(u8, perf_event_pending);
568
569#define set_perf_event_pending_flag() __get_cpu_var(perf_event_pending) = 1
543#define test_perf_event_pending() __get_cpu_var(perf_event_pending) 570#define test_perf_event_pending() __get_cpu_var(perf_event_pending)
544#define clear_perf_event_pending() __get_cpu_var(perf_event_pending) = 0 571#define clear_perf_event_pending() __get_cpu_var(perf_event_pending) = 0
545 572
546#else /* CONFIG_PERF_EVENTS && CONFIG_PPC32 */ 573#endif /* 32 vs 64 bit */
574
575void set_perf_event_pending(void)
576{
577 preempt_disable();
578 set_perf_event_pending_flag();
579 set_dec(1);
580 preempt_enable();
581}
582
583#else /* CONFIG_PERF_EVENTS */
547 584
548#define test_perf_event_pending() 0 585#define test_perf_event_pending() 0
549#define clear_perf_event_pending() 586#define clear_perf_event_pending()
550 587
551#endif /* CONFIG_PERF_EVENTS && CONFIG_PPC32 */ 588#endif /* CONFIG_PERF_EVENTS */
552 589
553/* 590/*
554 * For iSeries shared processors, we have to let the hypervisor 591 * For iSeries shared processors, we have to let the hypervisor
@@ -571,15 +608,15 @@ void timer_interrupt(struct pt_regs * regs)
571 struct clock_event_device *evt = &decrementer->event; 608 struct clock_event_device *evt = &decrementer->event;
572 u64 now; 609 u64 now;
573 610
611 trace_timer_interrupt_entry(regs);
612
613 __get_cpu_var(irq_stat).timer_irqs++;
614
574 /* Ensure a positive value is written to the decrementer, or else 615 /* Ensure a positive value is written to the decrementer, or else
575 * some CPUs will continuue to take decrementer exceptions */ 616 * some CPUs will continuue to take decrementer exceptions */
576 set_dec(DECREMENTER_MAX); 617 set_dec(DECREMENTER_MAX);
577 618
578#ifdef CONFIG_PPC32 619#ifdef CONFIG_PPC32
579 if (test_perf_event_pending()) {
580 clear_perf_event_pending();
581 perf_event_do_pending();
582 }
583 if (atomic_read(&ppc_n_lost_interrupts) != 0) 620 if (atomic_read(&ppc_n_lost_interrupts) != 0)
584 do_IRQ(regs); 621 do_IRQ(regs);
585#endif 622#endif
@@ -590,6 +627,7 @@ void timer_interrupt(struct pt_regs * regs)
590 now = decrementer->next_tb - now; 627 now = decrementer->next_tb - now;
591 if (now <= DECREMENTER_MAX) 628 if (now <= DECREMENTER_MAX)
592 set_dec((int)now); 629 set_dec((int)now);
630 trace_timer_interrupt_exit(regs);
593 return; 631 return;
594 } 632 }
595 old_regs = set_irq_regs(regs); 633 old_regs = set_irq_regs(regs);
@@ -597,6 +635,11 @@ void timer_interrupt(struct pt_regs * regs)
597 635
598 calculate_steal_time(); 636 calculate_steal_time();
599 637
638 if (test_perf_event_pending()) {
639 clear_perf_event_pending();
640 perf_event_do_pending();
641 }
642
600#ifdef CONFIG_PPC_ISERIES 643#ifdef CONFIG_PPC_ISERIES
601 if (firmware_has_feature(FW_FEATURE_ISERIES)) 644 if (firmware_has_feature(FW_FEATURE_ISERIES))
602 get_lppaca()->int_dword.fields.decr_int = 0; 645 get_lppaca()->int_dword.fields.decr_int = 0;
@@ -620,6 +663,8 @@ void timer_interrupt(struct pt_regs * regs)
620 663
621 irq_exit(); 664 irq_exit();
622 set_irq_regs(old_regs); 665 set_irq_regs(old_regs);
666
667 trace_timer_interrupt_exit(regs);
623} 668}
624 669
625void wakeup_decrementer(void) 670void wakeup_decrementer(void)
@@ -828,7 +873,8 @@ static cycle_t timebase_read(struct clocksource *cs)
828 return (cycle_t)get_tb(); 873 return (cycle_t)get_tb();
829} 874}
830 875
831void update_vsyscall(struct timespec *wall_time, struct clocksource *clock) 876void update_vsyscall(struct timespec *wall_time, struct clocksource *clock,
877 u32 mult)
832{ 878{
833 u64 t2x, stamp_xsec; 879 u64 t2x, stamp_xsec;
834 880
@@ -841,7 +887,7 @@ void update_vsyscall(struct timespec *wall_time, struct clocksource *clock)
841 887
842 /* XXX this assumes clock->shift == 22 */ 888 /* XXX this assumes clock->shift == 22 */
843 /* 4611686018 ~= 2^(20+64-22) / 1e9 */ 889 /* 4611686018 ~= 2^(20+64-22) / 1e9 */
844 t2x = (u64) clock->mult * 4611686018ULL; 890 t2x = (u64) mult * 4611686018ULL;
845 stamp_xsec = (u64) xtime.tv_nsec * XSEC_PER_SEC; 891 stamp_xsec = (u64) xtime.tv_nsec * XSEC_PER_SEC;
846 do_div(stamp_xsec, 1000000000); 892 do_div(stamp_xsec, 1000000000);
847 stamp_xsec += (u64) xtime.tv_sec * XSEC_PER_SEC; 893 stamp_xsec += (u64) xtime.tv_sec * XSEC_PER_SEC;
@@ -895,12 +941,21 @@ static void decrementer_set_mode(enum clock_event_mode mode,
895 decrementer_set_next_event(DECREMENTER_MAX, dev); 941 decrementer_set_next_event(DECREMENTER_MAX, dev);
896} 942}
897 943
944static inline uint64_t div_sc64(unsigned long ticks, unsigned long nsec,
945 int shift)
946{
947 uint64_t tmp = ((uint64_t)ticks) << shift;
948
949 do_div(tmp, nsec);
950 return tmp;
951}
952
898static void __init setup_clockevent_multiplier(unsigned long hz) 953static void __init setup_clockevent_multiplier(unsigned long hz)
899{ 954{
900 u64 mult, shift = 32; 955 u64 mult, shift = 32;
901 956
902 while (1) { 957 while (1) {
903 mult = div_sc(hz, NSEC_PER_SEC, shift); 958 mult = div_sc64(hz, NSEC_PER_SEC, shift);
904 if (mult && (mult >> 32UL) == 0UL) 959 if (mult && (mult >> 32UL) == 0UL)
905 break; 960 break;
906 961
@@ -918,8 +973,8 @@ static void register_decrementer_clockevent(int cpu)
918 *dec = decrementer_clockevent; 973 *dec = decrementer_clockevent;
919 dec->cpumask = cpumask_of(cpu); 974 dec->cpumask = cpumask_of(cpu);
920 975
921 printk(KERN_DEBUG "clockevent: %s mult[%lx] shift[%d] cpu[%d]\n", 976 printk_once(KERN_DEBUG "clockevent: %s mult[%x] shift[%d] cpu[%d]\n",
922 dec->name, dec->mult, dec->shift, cpu); 977 dec->name, dec->mult, dec->shift, cpu);
923 978
924 clockevents_register_device(dec); 979 clockevents_register_device(dec);
925} 980}
diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
index 6f0ae1a9bfae..29d128eb6c43 100644
--- a/arch/powerpc/kernel/traps.c
+++ b/arch/powerpc/kernel/traps.c
@@ -21,7 +21,6 @@
21#include <linux/stddef.h> 21#include <linux/stddef.h>
22#include <linux/unistd.h> 22#include <linux/unistd.h>
23#include <linux/ptrace.h> 23#include <linux/ptrace.h>
24#include <linux/slab.h>
25#include <linux/user.h> 24#include <linux/user.h>
26#include <linux/interrupt.h> 25#include <linux/interrupt.h>
27#include <linux/init.h> 26#include <linux/init.h>
@@ -60,13 +59,13 @@
60#endif 59#endif
61 60
62#if defined(CONFIG_DEBUGGER) || defined(CONFIG_KEXEC) 61#if defined(CONFIG_DEBUGGER) || defined(CONFIG_KEXEC)
63int (*__debugger)(struct pt_regs *regs); 62int (*__debugger)(struct pt_regs *regs) __read_mostly;
64int (*__debugger_ipi)(struct pt_regs *regs); 63int (*__debugger_ipi)(struct pt_regs *regs) __read_mostly;
65int (*__debugger_bpt)(struct pt_regs *regs); 64int (*__debugger_bpt)(struct pt_regs *regs) __read_mostly;
66int (*__debugger_sstep)(struct pt_regs *regs); 65int (*__debugger_sstep)(struct pt_regs *regs) __read_mostly;
67int (*__debugger_iabr_match)(struct pt_regs *regs); 66int (*__debugger_iabr_match)(struct pt_regs *regs) __read_mostly;
68int (*__debugger_dabr_match)(struct pt_regs *regs); 67int (*__debugger_dabr_match)(struct pt_regs *regs) __read_mostly;
69int (*__debugger_fault_handler)(struct pt_regs *regs); 68int (*__debugger_fault_handler)(struct pt_regs *regs) __read_mostly;
70 69
71EXPORT_SYMBOL(__debugger); 70EXPORT_SYMBOL(__debugger);
72EXPORT_SYMBOL(__debugger_ipi); 71EXPORT_SYMBOL(__debugger_ipi);
@@ -102,11 +101,11 @@ static inline void pmac_backlight_unblank(void) { }
102int die(const char *str, struct pt_regs *regs, long err) 101int die(const char *str, struct pt_regs *regs, long err)
103{ 102{
104 static struct { 103 static struct {
105 spinlock_t lock; 104 raw_spinlock_t lock;
106 u32 lock_owner; 105 u32 lock_owner;
107 int lock_owner_depth; 106 int lock_owner_depth;
108 } die = { 107 } die = {
109 .lock = __SPIN_LOCK_UNLOCKED(die.lock), 108 .lock = __RAW_SPIN_LOCK_UNLOCKED(die.lock),
110 .lock_owner = -1, 109 .lock_owner = -1,
111 .lock_owner_depth = 0 110 .lock_owner_depth = 0
112 }; 111 };
@@ -120,7 +119,7 @@ int die(const char *str, struct pt_regs *regs, long err)
120 119
121 if (die.lock_owner != raw_smp_processor_id()) { 120 if (die.lock_owner != raw_smp_processor_id()) {
122 console_verbose(); 121 console_verbose();
123 spin_lock_irqsave(&die.lock, flags); 122 raw_spin_lock_irqsave(&die.lock, flags);
124 die.lock_owner = smp_processor_id(); 123 die.lock_owner = smp_processor_id();
125 die.lock_owner_depth = 0; 124 die.lock_owner_depth = 0;
126 bust_spinlocks(1); 125 bust_spinlocks(1);
@@ -146,6 +145,11 @@ int die(const char *str, struct pt_regs *regs, long err)
146#endif 145#endif
147 printk("%s\n", ppc_md.name ? ppc_md.name : ""); 146 printk("%s\n", ppc_md.name ? ppc_md.name : "");
148 147
148 sysfs_printk_last_file();
149 if (notify_die(DIE_OOPS, str, regs, err, 255,
150 SIGSEGV) == NOTIFY_STOP)
151 return 1;
152
149 print_modules(); 153 print_modules();
150 show_regs(regs); 154 show_regs(regs);
151 } else { 155 } else {
@@ -155,7 +159,7 @@ int die(const char *str, struct pt_regs *regs, long err)
155 bust_spinlocks(0); 159 bust_spinlocks(0);
156 die.lock_owner = -1; 160 die.lock_owner = -1;
157 add_taint(TAINT_DIE); 161 add_taint(TAINT_DIE);
158 spin_unlock_irqrestore(&die.lock, flags); 162 raw_spin_unlock_irqrestore(&die.lock, flags);
159 163
160 if (kexec_should_crash(current) || 164 if (kexec_should_crash(current) ||
161 kexec_sr_activated(smp_processor_id())) 165 kexec_sr_activated(smp_processor_id()))
@@ -174,6 +178,15 @@ int die(const char *str, struct pt_regs *regs, long err)
174 return 0; 178 return 0;
175} 179}
176 180
181void user_single_step_siginfo(struct task_struct *tsk,
182 struct pt_regs *regs, siginfo_t *info)
183{
184 memset(info, 0, sizeof(*info));
185 info->si_signo = SIGTRAP;
186 info->si_code = TRAP_TRACE;
187 info->si_addr = (void __user *)regs->nip;
188}
189
177void _exception(int signr, struct pt_regs *regs, int code, unsigned long addr) 190void _exception(int signr, struct pt_regs *regs, int code, unsigned long addr)
178{ 191{
179 siginfo_t info; 192 siginfo_t info;
@@ -198,28 +211,6 @@ void _exception(int signr, struct pt_regs *regs, int code, unsigned long addr)
198 info.si_code = code; 211 info.si_code = code;
199 info.si_addr = (void __user *) addr; 212 info.si_addr = (void __user *) addr;
200 force_sig_info(signr, &info, current); 213 force_sig_info(signr, &info, current);
201
202 /*
203 * Init gets no signals that it doesn't have a handler for.
204 * That's all very well, but if it has caused a synchronous
205 * exception and we ignore the resulting signal, it will just
206 * generate the same exception over and over again and we get
207 * nowhere. Better to kill it and let the kernel panic.
208 */
209 if (is_global_init(current)) {
210 __sighandler_t handler;
211
212 spin_lock_irq(&current->sighand->siglock);
213 handler = current->sighand->action[signr-1].sa.sa_handler;
214 spin_unlock_irq(&current->sighand->siglock);
215 if (handler == SIG_DFL) {
216 /* init has generated a synchronous exception
217 and it doesn't have a handler for the signal */
218 printk(KERN_CRIT "init has generated signal %d "
219 "but has no handler for it\n", signr);
220 do_exit(signr);
221 }
222 }
223} 214}
224 215
225#ifdef CONFIG_PPC64 216#ifdef CONFIG_PPC64
@@ -307,7 +298,7 @@ static inline int check_io_access(struct pt_regs *regs)
307 return 0; 298 return 0;
308} 299}
309 300
310#if defined(CONFIG_4xx) || defined(CONFIG_BOOKE) 301#ifdef CONFIG_PPC_ADV_DEBUG_REGS
311/* On 4xx, the reason for the machine check or program exception 302/* On 4xx, the reason for the machine check or program exception
312 is in the ESR. */ 303 is in the ESR. */
313#define get_reason(regs) ((regs)->dsisr) 304#define get_reason(regs) ((regs)->dsisr)
@@ -491,6 +482,8 @@ void machine_check_exception(struct pt_regs *regs)
491{ 482{
492 int recover = 0; 483 int recover = 0;
493 484
485 __get_cpu_var(irq_stat).mce_exceptions++;
486
494 /* See if any machine dependent calls. In theory, we would want 487 /* See if any machine dependent calls. In theory, we would want
495 * to call the CPU first, and call the ppc_md. one if the CPU 488 * to call the CPU first, and call the ppc_md. one if the CPU
496 * one returns a positive number. However there is existing code 489 * one returns a positive number. However there is existing code
@@ -759,7 +752,7 @@ static int emulate_instruction(struct pt_regs *regs)
759 752
760 /* Emulate the mfspr rD, PVR. */ 753 /* Emulate the mfspr rD, PVR. */
761 if ((instword & PPC_INST_MFSPR_PVR_MASK) == PPC_INST_MFSPR_PVR) { 754 if ((instword & PPC_INST_MFSPR_PVR_MASK) == PPC_INST_MFSPR_PVR) {
762 PPC_WARN_EMULATED(mfpvr); 755 PPC_WARN_EMULATED(mfpvr, regs);
763 rd = (instword >> 21) & 0x1f; 756 rd = (instword >> 21) & 0x1f;
764 regs->gpr[rd] = mfspr(SPRN_PVR); 757 regs->gpr[rd] = mfspr(SPRN_PVR);
765 return 0; 758 return 0;
@@ -767,7 +760,7 @@ static int emulate_instruction(struct pt_regs *regs)
767 760
768 /* Emulating the dcba insn is just a no-op. */ 761 /* Emulating the dcba insn is just a no-op. */
769 if ((instword & PPC_INST_DCBA_MASK) == PPC_INST_DCBA) { 762 if ((instword & PPC_INST_DCBA_MASK) == PPC_INST_DCBA) {
770 PPC_WARN_EMULATED(dcba); 763 PPC_WARN_EMULATED(dcba, regs);
771 return 0; 764 return 0;
772 } 765 }
773 766
@@ -776,7 +769,7 @@ static int emulate_instruction(struct pt_regs *regs)
776 int shift = (instword >> 21) & 0x1c; 769 int shift = (instword >> 21) & 0x1c;
777 unsigned long msk = 0xf0000000UL >> shift; 770 unsigned long msk = 0xf0000000UL >> shift;
778 771
779 PPC_WARN_EMULATED(mcrxr); 772 PPC_WARN_EMULATED(mcrxr, regs);
780 regs->ccr = (regs->ccr & ~msk) | ((regs->xer >> shift) & msk); 773 regs->ccr = (regs->ccr & ~msk) | ((regs->xer >> shift) & msk);
781 regs->xer &= ~0xf0000000UL; 774 regs->xer &= ~0xf0000000UL;
782 return 0; 775 return 0;
@@ -784,19 +777,19 @@ static int emulate_instruction(struct pt_regs *regs)
784 777
785 /* Emulate load/store string insn. */ 778 /* Emulate load/store string insn. */
786 if ((instword & PPC_INST_STRING_GEN_MASK) == PPC_INST_STRING) { 779 if ((instword & PPC_INST_STRING_GEN_MASK) == PPC_INST_STRING) {
787 PPC_WARN_EMULATED(string); 780 PPC_WARN_EMULATED(string, regs);
788 return emulate_string_inst(regs, instword); 781 return emulate_string_inst(regs, instword);
789 } 782 }
790 783
791 /* Emulate the popcntb (Population Count Bytes) instruction. */ 784 /* Emulate the popcntb (Population Count Bytes) instruction. */
792 if ((instword & PPC_INST_POPCNTB_MASK) == PPC_INST_POPCNTB) { 785 if ((instword & PPC_INST_POPCNTB_MASK) == PPC_INST_POPCNTB) {
793 PPC_WARN_EMULATED(popcntb); 786 PPC_WARN_EMULATED(popcntb, regs);
794 return emulate_popcntb_inst(regs, instword); 787 return emulate_popcntb_inst(regs, instword);
795 } 788 }
796 789
797 /* Emulate isel (Integer Select) instruction */ 790 /* Emulate isel (Integer Select) instruction */
798 if ((instword & PPC_INST_ISEL_MASK) == PPC_INST_ISEL) { 791 if ((instword & PPC_INST_ISEL_MASK) == PPC_INST_ISEL) {
799 PPC_WARN_EMULATED(isel); 792 PPC_WARN_EMULATED(isel, regs);
800 return emulate_isel(regs, instword); 793 return emulate_isel(regs, instword);
801 } 794 }
802 795
@@ -973,6 +966,8 @@ void vsx_unavailable_exception(struct pt_regs *regs)
973 966
974void performance_monitor_exception(struct pt_regs *regs) 967void performance_monitor_exception(struct pt_regs *regs)
975{ 968{
969 __get_cpu_var(irq_stat).pmu_irqs++;
970
976 perf_irq(regs); 971 perf_irq(regs);
977} 972}
978 973
@@ -995,7 +990,7 @@ void SoftwareEmulation(struct pt_regs *regs)
995#ifdef CONFIG_MATH_EMULATION 990#ifdef CONFIG_MATH_EMULATION
996 errcode = do_mathemu(regs); 991 errcode = do_mathemu(regs);
997 if (errcode >= 0) 992 if (errcode >= 0)
998 PPC_WARN_EMULATED(math); 993 PPC_WARN_EMULATED(math, regs);
999 994
1000 switch (errcode) { 995 switch (errcode) {
1001 case 0: 996 case 0:
@@ -1018,7 +1013,7 @@ void SoftwareEmulation(struct pt_regs *regs)
1018#elif defined(CONFIG_8XX_MINIMAL_FPEMU) 1013#elif defined(CONFIG_8XX_MINIMAL_FPEMU)
1019 errcode = Soft_emulate_8xx(regs); 1014 errcode = Soft_emulate_8xx(regs);
1020 if (errcode >= 0) 1015 if (errcode >= 0)
1021 PPC_WARN_EMULATED(8xx); 1016 PPC_WARN_EMULATED(8xx, regs);
1022 1017
1023 switch (errcode) { 1018 switch (errcode) {
1024 case 0: 1019 case 0:
@@ -1037,10 +1032,69 @@ void SoftwareEmulation(struct pt_regs *regs)
1037} 1032}
1038#endif /* CONFIG_8xx */ 1033#endif /* CONFIG_8xx */
1039 1034
1040#if defined(CONFIG_40x) || defined(CONFIG_BOOKE) 1035#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1036static void handle_debug(struct pt_regs *regs, unsigned long debug_status)
1037{
1038 int changed = 0;
1039 /*
1040 * Determine the cause of the debug event, clear the
1041 * event flags and send a trap to the handler. Torez
1042 */
1043 if (debug_status & (DBSR_DAC1R | DBSR_DAC1W)) {
1044 dbcr_dac(current) &= ~(DBCR_DAC1R | DBCR_DAC1W);
1045#ifdef CONFIG_PPC_ADV_DEBUG_DAC_RANGE
1046 current->thread.dbcr2 &= ~DBCR2_DAC12MODE;
1047#endif
1048 do_send_trap(regs, mfspr(SPRN_DAC1), debug_status, TRAP_HWBKPT,
1049 5);
1050 changed |= 0x01;
1051 } else if (debug_status & (DBSR_DAC2R | DBSR_DAC2W)) {
1052 dbcr_dac(current) &= ~(DBCR_DAC2R | DBCR_DAC2W);
1053 do_send_trap(regs, mfspr(SPRN_DAC2), debug_status, TRAP_HWBKPT,
1054 6);
1055 changed |= 0x01;
1056 } else if (debug_status & DBSR_IAC1) {
1057 current->thread.dbcr0 &= ~DBCR0_IAC1;
1058 dbcr_iac_range(current) &= ~DBCR_IAC12MODE;
1059 do_send_trap(regs, mfspr(SPRN_IAC1), debug_status, TRAP_HWBKPT,
1060 1);
1061 changed |= 0x01;
1062 } else if (debug_status & DBSR_IAC2) {
1063 current->thread.dbcr0 &= ~DBCR0_IAC2;
1064 do_send_trap(regs, mfspr(SPRN_IAC2), debug_status, TRAP_HWBKPT,
1065 2);
1066 changed |= 0x01;
1067 } else if (debug_status & DBSR_IAC3) {
1068 current->thread.dbcr0 &= ~DBCR0_IAC3;
1069 dbcr_iac_range(current) &= ~DBCR_IAC34MODE;
1070 do_send_trap(regs, mfspr(SPRN_IAC3), debug_status, TRAP_HWBKPT,
1071 3);
1072 changed |= 0x01;
1073 } else if (debug_status & DBSR_IAC4) {
1074 current->thread.dbcr0 &= ~DBCR0_IAC4;
1075 do_send_trap(regs, mfspr(SPRN_IAC4), debug_status, TRAP_HWBKPT,
1076 4);
1077 changed |= 0x01;
1078 }
1079 /*
1080 * At the point this routine was called, the MSR(DE) was turned off.
1081 * Check all other debug flags and see if that bit needs to be turned
1082 * back on or not.
1083 */
1084 if (DBCR_ACTIVE_EVENTS(current->thread.dbcr0, current->thread.dbcr1))
1085 regs->msr |= MSR_DE;
1086 else
1087 /* Make sure the IDM flag is off */
1088 current->thread.dbcr0 &= ~DBCR0_IDM;
1089
1090 if (changed & 0x01)
1091 mtspr(SPRN_DBCR0, current->thread.dbcr0);
1092}
1041 1093
1042void __kprobes DebugException(struct pt_regs *regs, unsigned long debug_status) 1094void __kprobes DebugException(struct pt_regs *regs, unsigned long debug_status)
1043{ 1095{
1096 current->thread.dbsr = debug_status;
1097
1044 /* Hack alert: On BookE, Branch Taken stops on the branch itself, while 1098 /* Hack alert: On BookE, Branch Taken stops on the branch itself, while
1045 * on server, it stops on the target of the branch. In order to simulate 1099 * on server, it stops on the target of the branch. In order to simulate
1046 * the server behaviour, we thus restart right away with a single step 1100 * the server behaviour, we thus restart right away with a single step
@@ -1084,29 +1138,23 @@ void __kprobes DebugException(struct pt_regs *regs, unsigned long debug_status)
1084 if (debugger_sstep(regs)) 1138 if (debugger_sstep(regs))
1085 return; 1139 return;
1086 1140
1087 if (user_mode(regs))
1088 current->thread.dbcr0 &= ~(DBCR0_IC);
1089
1090 _exception(SIGTRAP, regs, TRAP_TRACE, regs->nip);
1091 } else if (debug_status & (DBSR_DAC1R | DBSR_DAC1W)) {
1092 regs->msr &= ~MSR_DE;
1093
1094 if (user_mode(regs)) { 1141 if (user_mode(regs)) {
1095 current->thread.dbcr0 &= ~(DBSR_DAC1R | DBSR_DAC1W | 1142 current->thread.dbcr0 &= ~DBCR0_IC;
1096 DBCR0_IDM); 1143#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1097 } else { 1144 if (DBCR_ACTIVE_EVENTS(current->thread.dbcr0,
1098 /* Disable DAC interupts */ 1145 current->thread.dbcr1))
1099 mtspr(SPRN_DBCR0, mfspr(SPRN_DBCR0) & ~(DBSR_DAC1R | 1146 regs->msr |= MSR_DE;
1100 DBSR_DAC1W | DBCR0_IDM)); 1147 else
1101 1148 /* Make sure the IDM bit is off */
1102 /* Clear the DAC event */ 1149 current->thread.dbcr0 &= ~DBCR0_IDM;
1103 mtspr(SPRN_DBSR, (DBSR_DAC1R | DBSR_DAC1W)); 1150#endif
1104 } 1151 }
1105 /* Setup and send the trap to the handler */ 1152
1106 do_dabr(regs, mfspr(SPRN_DAC1), debug_status); 1153 _exception(SIGTRAP, regs, TRAP_TRACE, regs->nip);
1107 } 1154 } else
1155 handle_debug(regs, debug_status);
1108} 1156}
1109#endif /* CONFIG_4xx || CONFIG_BOOKE */ 1157#endif /* CONFIG_PPC_ADV_DEBUG_REGS */
1110 1158
1111#if !defined(CONFIG_TAU_INT) 1159#if !defined(CONFIG_TAU_INT)
1112void TAUException(struct pt_regs *regs) 1160void TAUException(struct pt_regs *regs)
@@ -1129,7 +1177,7 @@ void altivec_assist_exception(struct pt_regs *regs)
1129 1177
1130 flush_altivec_to_thread(current); 1178 flush_altivec_to_thread(current);
1131 1179
1132 PPC_WARN_EMULATED(altivec); 1180 PPC_WARN_EMULATED(altivec, regs);
1133 err = emulate_altivec(regs); 1181 err = emulate_altivec(regs);
1134 if (err == 0) { 1182 if (err == 0) {
1135 regs->nip += 4; /* skip emulated instruction */ 1183 regs->nip += 4; /* skip emulated instruction */
diff --git a/arch/powerpc/kernel/udbg.c b/arch/powerpc/kernel/udbg.c
index fc9af47e2128..e39cad83c884 100644
--- a/arch/powerpc/kernel/udbg.c
+++ b/arch/powerpc/kernel/udbg.c
@@ -60,6 +60,8 @@ void __init udbg_early_init(void)
60 udbg_init_40x_realmode(); 60 udbg_init_40x_realmode();
61#elif defined(CONFIG_PPC_EARLY_DEBUG_CPM) 61#elif defined(CONFIG_PPC_EARLY_DEBUG_CPM)
62 udbg_init_cpm(); 62 udbg_init_cpm();
63#elif defined(CONFIG_PPC_EARLY_DEBUG_USBGECKO)
64 udbg_init_usbgecko();
63#endif 65#endif
64 66
65#ifdef CONFIG_PPC_EARLY_DEBUG 67#ifdef CONFIG_PPC_EARLY_DEBUG
diff --git a/arch/powerpc/kernel/vdso.c b/arch/powerpc/kernel/vdso.c
index 137dc22afa42..d84d19224a95 100644
--- a/arch/powerpc/kernel/vdso.c
+++ b/arch/powerpc/kernel/vdso.c
@@ -721,7 +721,7 @@ static int __init vdso_init(void)
721 721
722#ifdef CONFIG_PPC64 722#ifdef CONFIG_PPC64
723 /* 723 /*
724 * Fill up the "systemcfg" stuff for backward compatiblity 724 * Fill up the "systemcfg" stuff for backward compatibility
725 */ 725 */
726 strcpy((char *)vdso_data->eye_catcher, "SYSTEMCFG:PPC64"); 726 strcpy((char *)vdso_data->eye_catcher, "SYSTEMCFG:PPC64");
727 vdso_data->version.major = SYSTEMCFG_MAJOR; 727 vdso_data->version.major = SYSTEMCFG_MAJOR;
diff --git a/arch/powerpc/kernel/vector.S b/arch/powerpc/kernel/vector.S
index 67b6916f0e94..fe460482fa68 100644
--- a/arch/powerpc/kernel/vector.S
+++ b/arch/powerpc/kernel/vector.S
@@ -58,7 +58,7 @@ _GLOBAL(load_up_altivec)
58 * all 1's 58 * all 1's
59 */ 59 */
60 mfspr r4,SPRN_VRSAVE 60 mfspr r4,SPRN_VRSAVE
61 cmpdi 0,r4,0 61 cmpwi 0,r4,0
62 bne+ 1f 62 bne+ 1f
63 li r4,-1 63 li r4,-1
64 mtspr SPRN_VRSAVE,r4 64 mtspr SPRN_VRSAVE,r4
diff --git a/arch/powerpc/kernel/vio.c b/arch/powerpc/kernel/vio.c
index 77f64218abf3..82237176a2a3 100644
--- a/arch/powerpc/kernel/vio.c
+++ b/arch/powerpc/kernel/vio.c
@@ -17,6 +17,7 @@
17#include <linux/types.h> 17#include <linux/types.h>
18#include <linux/device.h> 18#include <linux/device.h>
19#include <linux/init.h> 19#include <linux/init.h>
20#include <linux/slab.h>
20#include <linux/console.h> 21#include <linux/console.h>
21#include <linux/module.h> 22#include <linux/module.h>
22#include <linux/mm.h> 23#include <linux/mm.h>
diff --git a/arch/powerpc/kernel/vmlinux.lds.S b/arch/powerpc/kernel/vmlinux.lds.S
index 27735a7ac12b..dcd01c82e701 100644
--- a/arch/powerpc/kernel/vmlinux.lds.S
+++ b/arch/powerpc/kernel/vmlinux.lds.S
@@ -38,6 +38,9 @@ jiffies = jiffies_64 + 4;
38#endif 38#endif
39SECTIONS 39SECTIONS
40{ 40{
41 . = 0;
42 reloc_start = .;
43
41 . = KERNELBASE; 44 . = KERNELBASE;
42 45
43/* 46/*
generated by cgit v1.2.2 (git 2.25.0) at 2025-04-18 17:26:16 -0400