diff options
Diffstat (limited to 'arch/tile')
63 files changed, 1636 insertions, 585 deletions
diff --git a/arch/tile/Kconfig b/arch/tile/Kconfig index 932fa14de5f..4f3006b600e 100644 --- a/arch/tile/Kconfig +++ b/arch/tile/Kconfig @@ -3,15 +3,15 @@ config TILE def_bool y + select HAVE_PERF_EVENTS + select USE_PMC if PERF_EVENTS select HAVE_DMA_ATTRS select HAVE_DMA_API_DEBUG select HAVE_KVM if !TILEGX select GENERIC_FIND_FIRST_BIT select SYSCTL_EXCEPTION_TRACE - select USE_GENERIC_SMP_HELPERS select CC_OPTIMIZE_FOR_SIZE select HAVE_DEBUG_KMEMLEAK - select HAVE_GENERIC_HARDIRQS select GENERIC_IRQ_PROBE select GENERIC_PENDING_IRQ if SMP select GENERIC_IRQ_SHOW @@ -68,6 +68,10 @@ config HUGETLB_SUPER_PAGES config GENERIC_TIME_VSYSCALL def_bool y +# Enable PMC if PERF_EVENTS, OPROFILE, or WATCHPOINTS are enabled. +config USE_PMC + bool + # FIXME: tilegx can implement a more efficient rwsem. config RWSEM_GENERIC_SPINLOCK def_bool y @@ -121,6 +125,8 @@ config HVC_TILE config TILEGX bool "Building for TILE-Gx (64-bit) processor" + select SPARSE_IRQ + select GENERIC_IRQ_LEGACY_ALLOC_HWIRQ select HAVE_FUNCTION_TRACER select HAVE_FUNCTION_TRACE_MCOUNT_TEST select HAVE_FUNCTION_GRAPH_TRACER @@ -362,7 +368,7 @@ config CMDLINE_OVERRIDE config VMALLOC_RESERVE hex - default 0x1000000 + default 0x2000000 config HARDWALL bool "Hardwall support to allow access to user dynamic network" @@ -407,7 +413,7 @@ config PCI_DOMAINS config NO_IOMEM def_bool !PCI -config NO_IOPORT +config NO_IOPORT_MAP def_bool !PCI config TILE_PCI_IO diff --git a/arch/tile/gxio/iorpc_mpipe.c b/arch/tile/gxio/iorpc_mpipe.c index 4f8f3d619c4..e19325c4c43 100644 --- a/arch/tile/gxio/iorpc_mpipe.c +++ b/arch/tile/gxio/iorpc_mpipe.c @@ -21,7 +21,7 @@ struct alloc_buffer_stacks_param { unsigned int flags; }; -int gxio_mpipe_alloc_buffer_stacks(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_buffer_stacks(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { @@ -45,7 +45,7 @@ struct init_buffer_stack_aux_param { unsigned int buffer_size_enum; }; -int gxio_mpipe_init_buffer_stack_aux(gxio_mpipe_context_t * context, +int gxio_mpipe_init_buffer_stack_aux(gxio_mpipe_context_t *context, void *mem_va, size_t mem_size, unsigned int mem_flags, unsigned int stack, unsigned int buffer_size_enum) @@ -80,7 +80,7 @@ struct alloc_notif_rings_param { unsigned int flags; }; -int gxio_mpipe_alloc_notif_rings(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_notif_rings(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { @@ -102,7 +102,7 @@ struct init_notif_ring_aux_param { unsigned int ring; }; -int gxio_mpipe_init_notif_ring_aux(gxio_mpipe_context_t * context, void *mem_va, +int gxio_mpipe_init_notif_ring_aux(gxio_mpipe_context_t *context, void *mem_va, size_t mem_size, unsigned int mem_flags, unsigned int ring) { @@ -133,7 +133,7 @@ struct request_notif_ring_interrupt_param { unsigned int ring; }; -int gxio_mpipe_request_notif_ring_interrupt(gxio_mpipe_context_t * context, +int gxio_mpipe_request_notif_ring_interrupt(gxio_mpipe_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event, unsigned int ring) @@ -158,7 +158,7 @@ struct enable_notif_ring_interrupt_param { unsigned int ring; }; -int gxio_mpipe_enable_notif_ring_interrupt(gxio_mpipe_context_t * context, +int gxio_mpipe_enable_notif_ring_interrupt(gxio_mpipe_context_t *context, unsigned int ring) { struct enable_notif_ring_interrupt_param temp; @@ -179,7 +179,7 @@ struct alloc_notif_groups_param { unsigned int flags; }; -int gxio_mpipe_alloc_notif_groups(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_notif_groups(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { @@ -201,7 +201,7 @@ struct init_notif_group_param { gxio_mpipe_notif_group_bits_t bits; }; -int gxio_mpipe_init_notif_group(gxio_mpipe_context_t * context, +int gxio_mpipe_init_notif_group(gxio_mpipe_context_t *context, unsigned int group, gxio_mpipe_notif_group_bits_t bits) { @@ -223,7 +223,7 @@ struct alloc_buckets_param { unsigned int flags; }; -int gxio_mpipe_alloc_buckets(gxio_mpipe_context_t * context, unsigned int count, +int gxio_mpipe_alloc_buckets(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { struct alloc_buckets_param temp; @@ -244,7 +244,7 @@ struct init_bucket_param { MPIPE_LBL_INIT_DAT_BSTS_TBL_t bucket_info; }; -int gxio_mpipe_init_bucket(gxio_mpipe_context_t * context, unsigned int bucket, +int gxio_mpipe_init_bucket(gxio_mpipe_context_t *context, unsigned int bucket, MPIPE_LBL_INIT_DAT_BSTS_TBL_t bucket_info) { struct init_bucket_param temp; @@ -265,7 +265,7 @@ struct alloc_edma_rings_param { unsigned int flags; }; -int gxio_mpipe_alloc_edma_rings(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_edma_rings(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { @@ -288,7 +288,7 @@ struct init_edma_ring_aux_param { unsigned int channel; }; -int gxio_mpipe_init_edma_ring_aux(gxio_mpipe_context_t * context, void *mem_va, +int gxio_mpipe_init_edma_ring_aux(gxio_mpipe_context_t *context, void *mem_va, size_t mem_size, unsigned int mem_flags, unsigned int ring, unsigned int channel) { @@ -315,7 +315,7 @@ int gxio_mpipe_init_edma_ring_aux(gxio_mpipe_context_t * context, void *mem_va, EXPORT_SYMBOL(gxio_mpipe_init_edma_ring_aux); -int gxio_mpipe_commit_rules(gxio_mpipe_context_t * context, const void *blob, +int gxio_mpipe_commit_rules(gxio_mpipe_context_t *context, const void *blob, size_t blob_size) { const void *params = blob; @@ -332,7 +332,7 @@ struct register_client_memory_param { unsigned int flags; }; -int gxio_mpipe_register_client_memory(gxio_mpipe_context_t * context, +int gxio_mpipe_register_client_memory(gxio_mpipe_context_t *context, unsigned int iotlb, HV_PTE pte, unsigned int flags) { @@ -355,7 +355,7 @@ struct link_open_aux_param { unsigned int flags; }; -int gxio_mpipe_link_open_aux(gxio_mpipe_context_t * context, +int gxio_mpipe_link_open_aux(gxio_mpipe_context_t *context, _gxio_mpipe_link_name_t name, unsigned int flags) { struct link_open_aux_param temp; @@ -374,7 +374,7 @@ struct link_close_aux_param { int mac; }; -int gxio_mpipe_link_close_aux(gxio_mpipe_context_t * context, int mac) +int gxio_mpipe_link_close_aux(gxio_mpipe_context_t *context, int mac) { struct link_close_aux_param temp; struct link_close_aux_param *params = &temp; @@ -393,7 +393,7 @@ struct link_set_attr_aux_param { int64_t val; }; -int gxio_mpipe_link_set_attr_aux(gxio_mpipe_context_t * context, int mac, +int gxio_mpipe_link_set_attr_aux(gxio_mpipe_context_t *context, int mac, uint32_t attr, int64_t val) { struct link_set_attr_aux_param temp; @@ -415,8 +415,8 @@ struct get_timestamp_aux_param { uint64_t cycles; }; -int gxio_mpipe_get_timestamp_aux(gxio_mpipe_context_t * context, uint64_t * sec, - uint64_t * nsec, uint64_t * cycles) +int gxio_mpipe_get_timestamp_aux(gxio_mpipe_context_t *context, uint64_t *sec, + uint64_t *nsec, uint64_t *cycles) { int __result; struct get_timestamp_aux_param temp; @@ -440,7 +440,7 @@ struct set_timestamp_aux_param { uint64_t cycles; }; -int gxio_mpipe_set_timestamp_aux(gxio_mpipe_context_t * context, uint64_t sec, +int gxio_mpipe_set_timestamp_aux(gxio_mpipe_context_t *context, uint64_t sec, uint64_t nsec, uint64_t cycles) { struct set_timestamp_aux_param temp; @@ -460,8 +460,7 @@ struct adjust_timestamp_aux_param { int64_t nsec; }; -int gxio_mpipe_adjust_timestamp_aux(gxio_mpipe_context_t * context, - int64_t nsec) +int gxio_mpipe_adjust_timestamp_aux(gxio_mpipe_context_t *context, int64_t nsec) { struct adjust_timestamp_aux_param temp; struct adjust_timestamp_aux_param *params = &temp; @@ -475,25 +474,6 @@ int gxio_mpipe_adjust_timestamp_aux(gxio_mpipe_context_t * context, EXPORT_SYMBOL(gxio_mpipe_adjust_timestamp_aux); -struct adjust_timestamp_freq_param { - int32_t ppb; -}; - -int gxio_mpipe_adjust_timestamp_freq(gxio_mpipe_context_t * context, - int32_t ppb) -{ - struct adjust_timestamp_freq_param temp; - struct adjust_timestamp_freq_param *params = &temp; - - params->ppb = ppb; - - return hv_dev_pwrite(context->fd, 0, (HV_VirtAddr) params, - sizeof(*params), - GXIO_MPIPE_OP_ADJUST_TIMESTAMP_FREQ); -} - -EXPORT_SYMBOL(gxio_mpipe_adjust_timestamp_freq); - struct config_edma_ring_blks_param { unsigned int ering; unsigned int max_blks; @@ -501,7 +481,7 @@ struct config_edma_ring_blks_param { unsigned int db; }; -int gxio_mpipe_config_edma_ring_blks(gxio_mpipe_context_t * context, +int gxio_mpipe_config_edma_ring_blks(gxio_mpipe_context_t *context, unsigned int ering, unsigned int max_blks, unsigned int min_snf_blks, unsigned int db) { @@ -520,11 +500,29 @@ int gxio_mpipe_config_edma_ring_blks(gxio_mpipe_context_t * context, EXPORT_SYMBOL(gxio_mpipe_config_edma_ring_blks); +struct adjust_timestamp_freq_param { + int32_t ppb; +}; + +int gxio_mpipe_adjust_timestamp_freq(gxio_mpipe_context_t *context, int32_t ppb) +{ + struct adjust_timestamp_freq_param temp; + struct adjust_timestamp_freq_param *params = &temp; + + params->ppb = ppb; + + return hv_dev_pwrite(context->fd, 0, (HV_VirtAddr) params, + sizeof(*params), + GXIO_MPIPE_OP_ADJUST_TIMESTAMP_FREQ); +} + +EXPORT_SYMBOL(gxio_mpipe_adjust_timestamp_freq); + struct arm_pollfd_param { union iorpc_pollfd pollfd; }; -int gxio_mpipe_arm_pollfd(gxio_mpipe_context_t * context, int pollfd_cookie) +int gxio_mpipe_arm_pollfd(gxio_mpipe_context_t *context, int pollfd_cookie) { struct arm_pollfd_param temp; struct arm_pollfd_param *params = &temp; @@ -541,7 +539,7 @@ struct close_pollfd_param { union iorpc_pollfd pollfd; }; -int gxio_mpipe_close_pollfd(gxio_mpipe_context_t * context, int pollfd_cookie) +int gxio_mpipe_close_pollfd(gxio_mpipe_context_t *context, int pollfd_cookie) { struct close_pollfd_param temp; struct close_pollfd_param *params = &temp; @@ -558,7 +556,7 @@ struct get_mmio_base_param { HV_PTE base; }; -int gxio_mpipe_get_mmio_base(gxio_mpipe_context_t * context, HV_PTE *base) +int gxio_mpipe_get_mmio_base(gxio_mpipe_context_t *context, HV_PTE *base) { int __result; struct get_mmio_base_param temp; @@ -579,7 +577,7 @@ struct check_mmio_offset_param { unsigned long size; }; -int gxio_mpipe_check_mmio_offset(gxio_mpipe_context_t * context, +int gxio_mpipe_check_mmio_offset(gxio_mpipe_context_t *context, unsigned long offset, unsigned long size) { struct check_mmio_offset_param temp; diff --git a/arch/tile/gxio/iorpc_mpipe_info.c b/arch/tile/gxio/iorpc_mpipe_info.c index 64883aabeb9..77019c6e9b4 100644 --- a/arch/tile/gxio/iorpc_mpipe_info.c +++ b/arch/tile/gxio/iorpc_mpipe_info.c @@ -15,12 +15,11 @@ /* This file is machine-generated; DO NOT EDIT! */ #include "gxio/iorpc_mpipe_info.h" - struct instance_aux_param { _gxio_mpipe_link_name_t name; }; -int gxio_mpipe_info_instance_aux(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_instance_aux(gxio_mpipe_info_context_t *context, _gxio_mpipe_link_name_t name) { struct instance_aux_param temp; @@ -39,10 +38,10 @@ struct enumerate_aux_param { _gxio_mpipe_link_mac_t mac; }; -int gxio_mpipe_info_enumerate_aux(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_enumerate_aux(gxio_mpipe_info_context_t *context, unsigned int idx, - _gxio_mpipe_link_name_t * name, - _gxio_mpipe_link_mac_t * mac) + _gxio_mpipe_link_name_t *name, + _gxio_mpipe_link_mac_t *mac) { int __result; struct enumerate_aux_param temp; @@ -50,7 +49,7 @@ int gxio_mpipe_info_enumerate_aux(gxio_mpipe_info_context_t * context, __result = hv_dev_pread(context->fd, 0, (HV_VirtAddr) params, sizeof(*params), - (((uint64_t) idx << 32) | + (((uint64_t)idx << 32) | GXIO_MPIPE_INFO_OP_ENUMERATE_AUX)); *name = params->name; *mac = params->mac; @@ -64,7 +63,7 @@ struct get_mmio_base_param { HV_PTE base; }; -int gxio_mpipe_info_get_mmio_base(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_get_mmio_base(gxio_mpipe_info_context_t *context, HV_PTE *base) { int __result; @@ -86,7 +85,7 @@ struct check_mmio_offset_param { unsigned long size; }; -int gxio_mpipe_info_check_mmio_offset(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_check_mmio_offset(gxio_mpipe_info_context_t *context, unsigned long offset, unsigned long size) { struct check_mmio_offset_param temp; diff --git a/arch/tile/gxio/iorpc_trio.c b/arch/tile/gxio/iorpc_trio.c index da6e18e049c..1d3cedb9aeb 100644 --- a/arch/tile/gxio/iorpc_trio.c +++ b/arch/tile/gxio/iorpc_trio.c @@ -21,7 +21,7 @@ struct alloc_asids_param { unsigned int flags; }; -int gxio_trio_alloc_asids(gxio_trio_context_t * context, unsigned int count, +int gxio_trio_alloc_asids(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { struct alloc_asids_param temp; @@ -44,7 +44,7 @@ struct alloc_memory_maps_param { unsigned int flags; }; -int gxio_trio_alloc_memory_maps(gxio_trio_context_t * context, +int gxio_trio_alloc_memory_maps(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { @@ -67,7 +67,7 @@ struct alloc_scatter_queues_param { unsigned int flags; }; -int gxio_trio_alloc_scatter_queues(gxio_trio_context_t * context, +int gxio_trio_alloc_scatter_queues(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { @@ -91,7 +91,7 @@ struct alloc_pio_regions_param { unsigned int flags; }; -int gxio_trio_alloc_pio_regions(gxio_trio_context_t * context, +int gxio_trio_alloc_pio_regions(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags) { @@ -115,7 +115,7 @@ struct init_pio_region_aux_param { unsigned int flags; }; -int gxio_trio_init_pio_region_aux(gxio_trio_context_t * context, +int gxio_trio_init_pio_region_aux(gxio_trio_context_t *context, unsigned int pio_region, unsigned int mac, uint32_t bus_address_hi, unsigned int flags) { @@ -145,7 +145,7 @@ struct init_memory_map_mmu_aux_param { unsigned int order_mode; }; -int gxio_trio_init_memory_map_mmu_aux(gxio_trio_context_t * context, +int gxio_trio_init_memory_map_mmu_aux(gxio_trio_context_t *context, unsigned int map, unsigned long va, uint64_t size, unsigned int asid, unsigned int mac, uint64_t bus_address, @@ -175,7 +175,7 @@ struct get_port_property_param { struct pcie_trio_ports_property trio_ports; }; -int gxio_trio_get_port_property(gxio_trio_context_t * context, +int gxio_trio_get_port_property(gxio_trio_context_t *context, struct pcie_trio_ports_property *trio_ports) { int __result; @@ -198,7 +198,7 @@ struct config_legacy_intr_param { unsigned int intx; }; -int gxio_trio_config_legacy_intr(gxio_trio_context_t * context, int inter_x, +int gxio_trio_config_legacy_intr(gxio_trio_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event, unsigned int mac, unsigned int intx) { @@ -227,7 +227,7 @@ struct config_msi_intr_param { unsigned int asid; }; -int gxio_trio_config_msi_intr(gxio_trio_context_t * context, int inter_x, +int gxio_trio_config_msi_intr(gxio_trio_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event, unsigned int mac, unsigned int mem_map, uint64_t mem_map_base, uint64_t mem_map_limit, @@ -259,7 +259,7 @@ struct set_mps_mrs_param { unsigned int mac; }; -int gxio_trio_set_mps_mrs(gxio_trio_context_t * context, uint16_t mps, +int gxio_trio_set_mps_mrs(gxio_trio_context_t *context, uint16_t mps, uint16_t mrs, unsigned int mac) { struct set_mps_mrs_param temp; @@ -279,7 +279,7 @@ struct force_rc_link_up_param { unsigned int mac; }; -int gxio_trio_force_rc_link_up(gxio_trio_context_t * context, unsigned int mac) +int gxio_trio_force_rc_link_up(gxio_trio_context_t *context, unsigned int mac) { struct force_rc_link_up_param temp; struct force_rc_link_up_param *params = &temp; @@ -296,7 +296,7 @@ struct force_ep_link_up_param { unsigned int mac; }; -int gxio_trio_force_ep_link_up(gxio_trio_context_t * context, unsigned int mac) +int gxio_trio_force_ep_link_up(gxio_trio_context_t *context, unsigned int mac) { struct force_ep_link_up_param temp; struct force_ep_link_up_param *params = &temp; @@ -313,7 +313,7 @@ struct get_mmio_base_param { HV_PTE base; }; -int gxio_trio_get_mmio_base(gxio_trio_context_t * context, HV_PTE *base) +int gxio_trio_get_mmio_base(gxio_trio_context_t *context, HV_PTE *base) { int __result; struct get_mmio_base_param temp; @@ -334,7 +334,7 @@ struct check_mmio_offset_param { unsigned long size; }; -int gxio_trio_check_mmio_offset(gxio_trio_context_t * context, +int gxio_trio_check_mmio_offset(gxio_trio_context_t *context, unsigned long offset, unsigned long size) { struct check_mmio_offset_param temp; diff --git a/arch/tile/gxio/iorpc_usb_host.c b/arch/tile/gxio/iorpc_usb_host.c index cf3c3cc1220..9c820073bfc 100644 --- a/arch/tile/gxio/iorpc_usb_host.c +++ b/arch/tile/gxio/iorpc_usb_host.c @@ -19,7 +19,7 @@ struct cfg_interrupt_param { union iorpc_interrupt interrupt; }; -int gxio_usb_host_cfg_interrupt(gxio_usb_host_context_t * context, int inter_x, +int gxio_usb_host_cfg_interrupt(gxio_usb_host_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event) { struct cfg_interrupt_param temp; @@ -41,7 +41,7 @@ struct register_client_memory_param { unsigned int flags; }; -int gxio_usb_host_register_client_memory(gxio_usb_host_context_t * context, +int gxio_usb_host_register_client_memory(gxio_usb_host_context_t *context, HV_PTE pte, unsigned int flags) { struct register_client_memory_param temp; @@ -61,7 +61,7 @@ struct get_mmio_base_param { HV_PTE base; }; -int gxio_usb_host_get_mmio_base(gxio_usb_host_context_t * context, HV_PTE *base) +int gxio_usb_host_get_mmio_base(gxio_usb_host_context_t *context, HV_PTE *base) { int __result; struct get_mmio_base_param temp; @@ -82,7 +82,7 @@ struct check_mmio_offset_param { unsigned long size; }; -int gxio_usb_host_check_mmio_offset(gxio_usb_host_context_t * context, +int gxio_usb_host_check_mmio_offset(gxio_usb_host_context_t *context, unsigned long offset, unsigned long size) { struct check_mmio_offset_param temp; diff --git a/arch/tile/gxio/usb_host.c b/arch/tile/gxio/usb_host.c index 66b002f54ec..785afad7922 100644 --- a/arch/tile/gxio/usb_host.c +++ b/arch/tile/gxio/usb_host.c @@ -26,7 +26,7 @@ #include <gxio/kiorpc.h> #include <gxio/usb_host.h> -int gxio_usb_host_init(gxio_usb_host_context_t * context, int usb_index, +int gxio_usb_host_init(gxio_usb_host_context_t *context, int usb_index, int is_ehci) { char file[32]; @@ -63,7 +63,7 @@ int gxio_usb_host_init(gxio_usb_host_context_t * context, int usb_index, EXPORT_SYMBOL_GPL(gxio_usb_host_init); -int gxio_usb_host_destroy(gxio_usb_host_context_t * context) +int gxio_usb_host_destroy(gxio_usb_host_context_t *context) { iounmap((void __force __iomem *)(context->mmio_base)); hv_dev_close(context->fd); @@ -76,14 +76,14 @@ int gxio_usb_host_destroy(gxio_usb_host_context_t * context) EXPORT_SYMBOL_GPL(gxio_usb_host_destroy); -void *gxio_usb_host_get_reg_start(gxio_usb_host_context_t * context) +void *gxio_usb_host_get_reg_start(gxio_usb_host_context_t *context) { return context->mmio_base; } EXPORT_SYMBOL_GPL(gxio_usb_host_get_reg_start); -size_t gxio_usb_host_get_reg_len(gxio_usb_host_context_t * context) +size_t gxio_usb_host_get_reg_len(gxio_usb_host_context_t *context) { return HV_USB_HOST_MMIO_SIZE; } diff --git a/arch/tile/include/arch/mpipe.h b/arch/tile/include/arch/mpipe.h index 8a33912fd6c..904538e754d 100644 --- a/arch/tile/include/arch/mpipe.h +++ b/arch/tile/include/arch/mpipe.h @@ -176,7 +176,18 @@ typedef union */ uint_reg_t stack_idx : 5; /* Reserved. */ - uint_reg_t __reserved_2 : 5; + uint_reg_t __reserved_2 : 3; + /* + * Instance ID. For devices that support automatic buffer return between + * mPIPE instances, this field indicates the buffer owner. If the INST + * field does not match the mPIPE's instance number when a packet is + * egressed, buffers with HWB set will be returned to the other mPIPE + * instance. Note that not all devices support multi-mPIPE buffer + * return. The MPIPE_EDMA_INFO.REMOTE_BUFF_RTN_SUPPORT bit indicates + * whether the INST field in the buffer descriptor is populated by iDMA + * hardware. This field is ignored on writes. + */ + uint_reg_t inst : 2; /* * Reads as one to indicate that this is a hardware managed buffer. * Ignored on writes since all buffers on a given stack are the same size. @@ -205,7 +216,8 @@ typedef union uint_reg_t c : 2; uint_reg_t size : 3; uint_reg_t hwb : 1; - uint_reg_t __reserved_2 : 5; + uint_reg_t inst : 2; + uint_reg_t __reserved_2 : 3; uint_reg_t stack_idx : 5; uint_reg_t __reserved_1 : 6; int_reg_t va : 35; @@ -231,9 +243,9 @@ typedef union /* Reserved. */ uint_reg_t __reserved_0 : 3; /* eDMA ring being accessed */ - uint_reg_t ring : 5; + uint_reg_t ring : 6; /* Reserved. */ - uint_reg_t __reserved_1 : 18; + uint_reg_t __reserved_1 : 17; /* * This field of the address selects the region (address space) to be * accessed. For the egress DMA post region, this field must be 5. @@ -250,8 +262,8 @@ typedef union uint_reg_t svc_dom : 5; uint_reg_t __reserved_2 : 6; uint_reg_t region : 3; - uint_reg_t __reserved_1 : 18; - uint_reg_t ring : 5; + uint_reg_t __reserved_1 : 17; + uint_reg_t ring : 6; uint_reg_t __reserved_0 : 3; #endif }; diff --git a/arch/tile/include/arch/mpipe_constants.h b/arch/tile/include/arch/mpipe_constants.h index 410a0400e05..84022ac5fe8 100644 --- a/arch/tile/include/arch/mpipe_constants.h +++ b/arch/tile/include/arch/mpipe_constants.h @@ -16,13 +16,13 @@ #ifndef __ARCH_MPIPE_CONSTANTS_H__ #define __ARCH_MPIPE_CONSTANTS_H__ -#define MPIPE_NUM_CLASSIFIERS 10 +#define MPIPE_NUM_CLASSIFIERS 16 #define MPIPE_CLS_MHZ 1200 -#define MPIPE_NUM_EDMA_RINGS 32 +#define MPIPE_NUM_EDMA_RINGS 64 #define MPIPE_NUM_SGMII_MACS 16 -#define MPIPE_NUM_XAUI_MACS 4 +#define MPIPE_NUM_XAUI_MACS 16 #define MPIPE_NUM_LOOPBACK_CHANNELS 4 #define MPIPE_NUM_NON_LB_CHANNELS 28 diff --git a/arch/tile/include/arch/mpipe_shm.h b/arch/tile/include/arch/mpipe_shm.h index f2e9e122818..13b3c4300e5 100644 --- a/arch/tile/include/arch/mpipe_shm.h +++ b/arch/tile/include/arch/mpipe_shm.h @@ -44,8 +44,14 @@ typedef union * descriptors toggles each time the ring tail pointer wraps. */ uint_reg_t gen : 1; + /** + * For devices with EDMA reorder support, this field allows the + * descriptor to select the egress FIFO. The associated DMA ring must + * have ALLOW_EFIFO_SEL enabled. + */ + uint_reg_t efifo_sel : 6; /** Reserved. Must be zero. */ - uint_reg_t r0 : 7; + uint_reg_t r0 : 1; /** Checksum generation enabled for this transfer. */ uint_reg_t csum : 1; /** @@ -110,7 +116,8 @@ typedef union uint_reg_t notif : 1; uint_reg_t ns : 1; uint_reg_t csum : 1; - uint_reg_t r0 : 7; + uint_reg_t r0 : 1; + uint_reg_t efifo_sel : 6; uint_reg_t gen : 1; #endif @@ -126,14 +133,16 @@ typedef union /** Reserved. */ uint_reg_t __reserved_1 : 3; /** - * Instance ID. For devices that support more than one mPIPE instance, - * this field indicates the buffer owner. If the INST field does not - * match the mPIPE's instance number when a packet is egressed, buffers - * with HWB set will be returned to the other mPIPE instance. + * Instance ID. For devices that support automatic buffer return between + * mPIPE instances, this field indicates the buffer owner. If the INST + * field does not match the mPIPE's instance number when a packet is + * egressed, buffers with HWB set will be returned to the other mPIPE + * instance. Note that not all devices support multi-mPIPE buffer + * return. The MPIPE_EDMA_INFO.REMOTE_BUFF_RTN_SUPPORT bit indicates + * whether the INST field in the buffer descriptor is populated by iDMA + * hardware. */ - uint_reg_t inst : 1; - /** Reserved. */ - uint_reg_t __reserved_2 : 1; + uint_reg_t inst : 2; /** * Always set to one by hardware in iDMA packet descriptors. For eDMA, * indicates whether the buffer will be released to the buffer stack @@ -166,8 +175,7 @@ typedef union uint_reg_t c : 2; uint_reg_t size : 3; uint_reg_t hwb : 1; - uint_reg_t __reserved_2 : 1; - uint_reg_t inst : 1; + uint_reg_t inst : 2; uint_reg_t __reserved_1 : 3; uint_reg_t stack_idx : 5; uint_reg_t __reserved_0 : 6; @@ -408,7 +416,10 @@ typedef union /** * Sequence number applied when packet is distributed. Classifier * selects which sequence number is to be applied by writing the 13-bit - * SQN-selector into this field. + * SQN-selector into this field. For devices that support EXT_SQN (as + * indicated in IDMA_INFO.EXT_SQN_SUPPORT), the GP_SQN can be extended to + * 32-bits via the IDMA_CTL.EXT_SQN register. In this case the + * PACKET_SQN will be reduced to 32 bits. */ uint_reg_t gp_sqn : 16; /** @@ -451,14 +462,16 @@ typedef union /** Reserved. */ uint_reg_t __reserved_5 : 3; /** - * Instance ID. For devices that support more than one mPIPE instance, - * this field indicates the buffer owner. If the INST field does not - * match the mPIPE's instance number when a packet is egressed, buffers - * with HWB set will be returned to the other mPIPE instance. + * Instance ID. For devices that support automatic buffer return between + * mPIPE instances, this field indicates the buffer owner. If the INST + * field does not match the mPIPE's instance number when a packet is + * egressed, buffers with HWB set will be returned to the other mPIPE + * instance. Note that not all devices support multi-mPIPE buffer + * return. The MPIPE_EDMA_INFO.REMOTE_BUFF_RTN_SUPPORT bit indicates + * whether the INST field in the buffer descriptor is populated by iDMA + * hardware. */ - uint_reg_t inst : 1; - /** Reserved. */ - uint_reg_t __reserved_6 : 1; + uint_reg_t inst : 2; /** * Always set to one by hardware in iDMA packet descriptors. For eDMA, * indicates whether the buffer will be released to the buffer stack @@ -491,8 +504,7 @@ typedef union uint_reg_t c : 2; uint_reg_t size : 3; uint_reg_t hwb : 1; - uint_reg_t __reserved_6 : 1; - uint_reg_t inst : 1; + uint_reg_t inst : 2; uint_reg_t __reserved_5 : 3; uint_reg_t stack_idx : 5; uint_reg_t __reserved_4 : 6; diff --git a/arch/tile/include/arch/trio_constants.h b/arch/tile/include/arch/trio_constants.h index 628b045436b..85647e91a45 100644 --- a/arch/tile/include/arch/trio_constants.h +++ b/arch/tile/include/arch/trio_constants.h @@ -16,21 +16,21 @@ #ifndef __ARCH_TRIO_CONSTANTS_H__ #define __ARCH_TRIO_CONSTANTS_H__ -#define TRIO_NUM_ASIDS 16 +#define TRIO_NUM_ASIDS 32 #define TRIO_NUM_TLBS_PER_ASID 16 #define TRIO_NUM_TPIO_REGIONS 8 #define TRIO_LOG2_NUM_TPIO_REGIONS 3 -#define TRIO_NUM_MAP_MEM_REGIONS 16 -#define TRIO_LOG2_NUM_MAP_MEM_REGIONS 4 +#define TRIO_NUM_MAP_MEM_REGIONS 32 +#define TRIO_LOG2_NUM_MAP_MEM_REGIONS 5 #define TRIO_NUM_MAP_SQ_REGIONS 8 #define TRIO_LOG2_NUM_MAP_SQ_REGIONS 3 #define TRIO_LOG2_NUM_SQ_FIFO_ENTRIES 6 -#define TRIO_NUM_PUSH_DMA_RINGS 32 +#define TRIO_NUM_PUSH_DMA_RINGS 64 -#define TRIO_NUM_PULL_DMA_RINGS 32 +#define TRIO_NUM_PULL_DMA_RINGS 64 #endif /* __ARCH_TRIO_CONSTANTS_H__ */ diff --git a/arch/tile/include/asm/Kbuild b/arch/tile/include/asm/Kbuild index 664d6ad23f8..0aa5675e702 100644 --- a/arch/tile/include/asm/Kbuild +++ b/arch/tile/include/asm/Kbuild @@ -11,6 +11,7 @@ generic-y += errno.h generic-y += exec.h generic-y += fb.h generic-y += fcntl.h +generic-y += hash.h generic-y += hw_irq.h generic-y += ioctl.h generic-y += ioctls.h @@ -18,12 +19,14 @@ generic-y += ipcbuf.h generic-y += irq_regs.h generic-y += local.h generic-y += local64.h +generic-y += mcs_spinlock.h generic-y += msgbuf.h generic-y += mutex.h generic-y += param.h generic-y += parport.h generic-y += poll.h generic-y += posix_types.h +generic-y += preempt.h generic-y += resource.h generic-y += scatterlist.h generic-y += sembuf.h diff --git a/arch/tile/include/asm/atomic.h b/arch/tile/include/asm/atomic.h index d385eaadece..70979846076 100644 --- a/arch/tile/include/asm/atomic.h +++ b/arch/tile/include/asm/atomic.h @@ -166,7 +166,7 @@ static inline int atomic_cmpxchg(atomic_t *v, int o, int n) * * Atomically sets @v to @i and returns old @v */ -static inline u64 atomic64_xchg(atomic64_t *v, u64 n) +static inline long long atomic64_xchg(atomic64_t *v, long long n) { return xchg64(&v->counter, n); } @@ -180,7 +180,8 @@ static inline u64 atomic64_xchg(atomic64_t *v, u64 n) * Atomically checks if @v holds @o and replaces it with @n if so. * Returns the old value at @v. */ -static inline u64 atomic64_cmpxchg(atomic64_t *v, u64 o, u64 n) +static inline long long atomic64_cmpxchg(atomic64_t *v, long long o, + long long n) { return cmpxchg64(&v->counter, o, n); } diff --git a/arch/tile/include/asm/atomic_32.h b/arch/tile/include/asm/atomic_32.h index 0d0395b1b15..1b109fad9ff 100644 --- a/arch/tile/include/asm/atomic_32.h +++ b/arch/tile/include/asm/atomic_32.h @@ -80,7 +80,7 @@ static inline void atomic_set(atomic_t *v, int n) /* A 64bit atomic type */ typedef struct { - u64 __aligned(8) counter; + long long counter; } atomic64_t; #define ATOMIC64_INIT(val) { (val) } @@ -91,14 +91,14 @@ typedef struct { * * Atomically reads the value of @v. */ -static inline u64 atomic64_read(const atomic64_t *v) +static inline long long atomic64_read(const atomic64_t *v) { /* * Requires an atomic op to read both 32-bit parts consistently. * Casting away const is safe since the atomic support routines * do not write to memory if the value has not been modified. */ - return _atomic64_xchg_add((u64 *)&v->counter, 0); + return _atomic64_xchg_add((long long *)&v->counter, 0); } /** @@ -108,7 +108,7 @@ static inline u64 atomic64_read(const atomic64_t *v) * * Atomically adds @i to @v. */ -static inline void atomic64_add(u64 i, atomic64_t *v) +static inline void atomic64_add(long long i, atomic64_t *v) { _atomic64_xchg_add(&v->counter, i); } @@ -120,7 +120,7 @@ static inline void atomic64_add(u64 i, atomic64_t *v) * * Atomically adds @i to @v and returns @i + @v */ -static inline u64 atomic64_add_return(u64 i, atomic64_t *v) +static inline long long atomic64_add_return(long long i, atomic64_t *v) { smp_mb(); /* barrier for proper semantics */ return _atomic64_xchg_add(&v->counter, i) + i; @@ -135,7 +135,8 @@ static inline u64 atomic64_add_return(u64 i, atomic64_t *v) * Atomically adds @a to @v, so long as @v was not already @u. * Returns non-zero if @v was not @u, and zero otherwise. */ -static inline u64 atomic64_add_unless(atomic64_t *v, u64 a, u64 u) +static inline long long atomic64_add_unless(atomic64_t *v, long long a, + long long u) { smp_mb(); /* barrier for proper semantics */ return _atomic64_xchg_add_unless(&v->counter, a, u) != u; @@ -151,7 +152,7 @@ static inline u64 atomic64_add_unless(atomic64_t *v, u64 a, u64 u) * atomic64_set() can't be just a raw store, since it would be lost if it * fell between the load and store of one of the other atomic ops. */ -static inline void atomic64_set(atomic64_t *v, u64 n) +static inline void atomic64_set(atomic64_t *v, long long n) { _atomic64_xchg(&v->counter, n); } @@ -168,16 +169,6 @@ static inline void atomic64_set(atomic64_t *v, u64 n) #define atomic64_dec_and_test(v) (atomic64_dec_return((v)) == 0) #define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1LL, 0LL) -/* - * We need to barrier before modifying the word, since the _atomic_xxx() - * routines just tns the lock and then read/modify/write of the word. - * But after the word is updated, the routine issues an "mf" before returning, - * and since it's a function call, we don't even need a compiler barrier. - */ -#define smp_mb__before_atomic_dec() smp_mb() -#define smp_mb__before_atomic_inc() smp_mb() -#define smp_mb__after_atomic_dec() do { } while (0) -#define smp_mb__after_atomic_inc() do { } while (0) #endif /* !__ASSEMBLY__ */ @@ -236,11 +227,13 @@ extern struct __get_user __atomic_xchg_add_unless(volatile int *p, extern struct __get_user __atomic_or(volatile int *p, int *lock, int n); extern struct __get_user __atomic_andn(volatile int *p, int *lock, int n); extern struct __get_user __atomic_xor(volatile int *p, int *lock, int n); -extern u64 __atomic64_cmpxchg(volatile u64 *p, int *lock, u64 o, u64 n); -extern u64 __atomic64_xchg(volatile u64 *p, int *lock, u64 n); -extern u64 __atomic64_xchg_add(volatile u64 *p, int *lock, u64 n); -extern u64 __atomic64_xchg_add_unless(volatile u64 *p, - int *lock, u64 o, u64 n); +extern long long __atomic64_cmpxchg(volatile long long *p, int *lock, + long long o, long long n); +extern long long __atomic64_xchg(volatile long long *p, int *lock, long long n); +extern long long __atomic64_xchg_add(volatile long long *p, int *lock, + long long n); +extern long long __atomic64_xchg_add_unless(volatile long long *p, + int *lock, long long o, long long n); /* Return failure from the atomic wrappers. */ struct __get_user __atomic_bad_address(int __user *addr); diff --git a/arch/tile/include/asm/atomic_64.h b/arch/tile/include/asm/atomic_64.h index ad220eed05f..7b11c5fadd4 100644 --- a/arch/tile/include/asm/atomic_64.h +++ b/arch/tile/include/asm/atomic_64.h @@ -105,12 +105,6 @@ static inline long atomic64_add_unless(atomic64_t *v, long a, long u) #define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0) -/* Atomic dec and inc don't implement barrier, so provide them if needed. */ -#define smp_mb__before_atomic_dec() smp_mb() -#define smp_mb__after_atomic_dec() smp_mb() -#define smp_mb__before_atomic_inc() smp_mb() -#define smp_mb__after_atomic_inc() smp_mb() - /* Define this to indicate that cmpxchg is an efficient operation. */ #define __HAVE_ARCH_CMPXCHG diff --git a/arch/tile/include/asm/barrier.h b/arch/tile/include/asm/barrier.h index a9a73da5865..96a42ae79f4 100644 --- a/arch/tile/include/asm/barrier.h +++ b/arch/tile/include/asm/barrier.h @@ -22,59 +22,6 @@ #include <arch/spr_def.h> #include <asm/timex.h> -/* - * read_barrier_depends - Flush all pending reads that subsequents reads - * depend on. - * - * No data-dependent reads from memory-like regions are ever reordered - * over this barrier. All reads preceding this primitive are guaranteed - * to access memory (but not necessarily other CPUs' caches) before any - * reads following this primitive that depend on the data return by - * any of the preceding reads. This primitive is much lighter weight than - * rmb() on most CPUs, and is never heavier weight than is - * rmb(). - * - * These ordering constraints are respected by both the local CPU - * and the compiler. - * - * Ordering is not guaranteed by anything other than these primitives, - * not even by data dependencies. See the documentation for - * memory_barrier() for examples and URLs to more information. - * - * For example, the following code would force ordering (the initial - * value of "a" is zero, "b" is one, and "p" is "&a"): - * - * <programlisting> - * CPU 0 CPU 1 - * - * b = 2; - * memory_barrier(); - * p = &b; q = p; - * read_barrier_depends(); - * d = *q; - * </programlisting> - * - * because the read of "*q" depends on the read of "p" and these - * two reads are separated by a read_barrier_depends(). However, - * the following code, with the same initial values for "a" and "b": - * - * <programlisting> - * CPU 0 CPU 1 - * - * a = 2; - * memory_barrier(); - * b = 3; y = b; - * read_barrier_depends(); - * x = a; - * </programlisting> - * - * does not enforce ordering, since there is no data dependency between - * the read of "a" and the read of "b". Therefore, on some CPUs, such - * as Alpha, "y" could be set to 3 and "x" to 0. Use rmb() - * in cases like this where there are no data dependencies. - */ -#define read_barrier_depends() do { } while (0) - #define __sync() __insn_mf() #include <hv/syscall_public.h> @@ -125,20 +72,21 @@ mb_incoherent(void) #define mb() fast_mb() #define iob() fast_iob() -#ifdef CONFIG_SMP -#define smp_mb() mb() -#define smp_rmb() rmb() -#define smp_wmb() wmb() -#define smp_read_barrier_depends() read_barrier_depends() -#else -#define smp_mb() barrier() -#define smp_rmb() barrier() -#define smp_wmb() barrier() -#define smp_read_barrier_depends() do { } while (0) +#ifndef __tilegx__ /* 32 bit */ +/* + * We need to barrier before modifying the word, since the _atomic_xxx() + * routines just tns the lock and then read/modify/write of the word. + * But after the word is updated, the routine issues an "mf" before returning, + * and since it's a function call, we don't even need a compiler barrier. + */ +#define smp_mb__before_atomic() smp_mb() +#define smp_mb__after_atomic() do { } while (0) +#else /* 64 bit */ +#define smp_mb__before_atomic() smp_mb() +#define smp_mb__after_atomic() smp_mb() #endif -#define set_mb(var, value) \ - do { var = value; mb(); } while (0) +#include <asm-generic/barrier.h> #endif /* !__ASSEMBLY__ */ #endif /* _ASM_TILE_BARRIER_H */ diff --git a/arch/tile/include/asm/bitops.h b/arch/tile/include/asm/bitops.h index d5a20686503..20caa346ac0 100644 --- a/arch/tile/include/asm/bitops.h +++ b/arch/tile/include/asm/bitops.h @@ -17,6 +17,7 @@ #define _ASM_TILE_BITOPS_H #include <linux/types.h> +#include <asm/barrier.h> #ifndef _LINUX_BITOPS_H #error only <linux/bitops.h> can be included directly diff --git a/arch/tile/include/asm/bitops_32.h b/arch/tile/include/asm/bitops_32.h index 386865ad2f5..bbf7b666f21 100644 --- a/arch/tile/include/asm/bitops_32.h +++ b/arch/tile/include/asm/bitops_32.h @@ -49,8 +49,8 @@ static inline void set_bit(unsigned nr, volatile unsigned long *addr) * restricted to acting on a single-word quantity. * * clear_bit() may not contain a memory barrier, so if it is used for - * locking purposes, you should call smp_mb__before_clear_bit() and/or - * smp_mb__after_clear_bit() to ensure changes are visible on other cpus. + * locking purposes, you should call smp_mb__before_atomic() and/or + * smp_mb__after_atomic() to ensure changes are visible on other cpus. */ static inline void clear_bit(unsigned nr, volatile unsigned long *addr) { @@ -121,10 +121,6 @@ static inline int test_and_change_bit(unsigned nr, return (_atomic_xor(addr, mask) & mask) != 0; } -/* See discussion at smp_mb__before_atomic_dec() in <asm/atomic_32.h>. */ -#define smp_mb__before_clear_bit() smp_mb() -#define smp_mb__after_clear_bit() do {} while (0) - #include <asm-generic/bitops/ext2-atomic.h> #endif /* _ASM_TILE_BITOPS_32_H */ diff --git a/arch/tile/include/asm/bitops_64.h b/arch/tile/include/asm/bitops_64.h index ad34cd05608..bb1a29221fc 100644 --- a/arch/tile/include/asm/bitops_64.h +++ b/arch/tile/include/asm/bitops_64.h @@ -32,10 +32,6 @@ static inline void clear_bit(unsigned nr, volatile unsigned long *addr) __insn_fetchand((void *)(addr + nr / BITS_PER_LONG), ~mask); } -#define smp_mb__before_clear_bit() smp_mb() -#define smp_mb__after_clear_bit() smp_mb() - - static inline void change_bit(unsigned nr, volatile unsigned long *addr) { unsigned long mask = (1UL << (nr % BITS_PER_LONG)); diff --git a/arch/tile/include/asm/cmpxchg.h b/arch/tile/include/asm/cmpxchg.h index 4001d5eab4b..0ccda3c425b 100644 --- a/arch/tile/include/asm/cmpxchg.h +++ b/arch/tile/include/asm/cmpxchg.h @@ -35,10 +35,10 @@ int _atomic_xchg(int *ptr, int n); int _atomic_xchg_add(int *v, int i); int _atomic_xchg_add_unless(int *v, int a, int u); int _atomic_cmpxchg(int *ptr, int o, int n); -u64 _atomic64_xchg(u64 *v, u64 n); -u64 _atomic64_xchg_add(u64 *v, u64 i); -u64 _atomic64_xchg_add_unless(u64 *v, u64 a, u64 u); -u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n); +long long _atomic64_xchg(long long *v, long long n); +long long _atomic64_xchg_add(long long *v, long long i); +long long _atomic64_xchg_add_unless(long long *v, long long a, long long u); +long long _atomic64_cmpxchg(long long *v, long long o, long long n); #define xchg(ptr, n) \ ({ \ @@ -53,7 +53,8 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n); if (sizeof(*(ptr)) != 4) \ __cmpxchg_called_with_bad_pointer(); \ smp_mb(); \ - (typeof(*(ptr)))_atomic_cmpxchg((int *)ptr, (int)o, (int)n); \ + (typeof(*(ptr)))_atomic_cmpxchg((int *)ptr, (int)o, \ + (int)n); \ }) #define xchg64(ptr, n) \ @@ -61,7 +62,8 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n); if (sizeof(*(ptr)) != 8) \ __xchg_called_with_bad_pointer(); \ smp_mb(); \ - (typeof(*(ptr)))_atomic64_xchg((u64 *)(ptr), (u64)(n)); \ + (typeof(*(ptr)))_atomic64_xchg((long long *)(ptr), \ + (long long)(n)); \ }) #define cmpxchg64(ptr, o, n) \ @@ -69,7 +71,8 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n); if (sizeof(*(ptr)) != 8) \ __cmpxchg_called_with_bad_pointer(); \ smp_mb(); \ - (typeof(*(ptr)))_atomic64_cmpxchg((u64 *)ptr, (u64)o, (u64)n); \ + (typeof(*(ptr)))_atomic64_cmpxchg((long long *)ptr, \ + (long long)o, (long long)n); \ }) #else @@ -81,10 +84,11 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n); switch (sizeof(*(ptr))) { \ case 4: \ __x = (typeof(__x))(unsigned long) \ - __insn_exch4((ptr), (u32)(unsigned long)(n)); \ + __insn_exch4((ptr), \ + (u32)(unsigned long)(n)); \ break; \ case 8: \ - __x = (typeof(__x)) \ + __x = (typeof(__x)) \ __insn_exch((ptr), (unsigned long)(n)); \ break; \ default: \ @@ -103,10 +107,12 @@ u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n); switch (sizeof(*(ptr))) { \ case 4: \ __x = (typeof(__x))(unsigned long) \ - __insn_cmpexch4((ptr), (u32)(unsigned long)(n)); \ + __insn_cmpexch4((ptr), \ + (u32)(unsigned long)(n)); \ break; \ case 8: \ - __x = (typeof(__x))__insn_cmpexch((ptr), (u64)(n)); \ + __x = (typeof(__x))__insn_cmpexch((ptr), \ + (long long)(n)); \ break; \ default: \ __cmpxchg_called_with_bad_pointer(); \ diff --git a/arch/tile/include/asm/compat.h b/arch/tile/include/asm/compat.h index 78f1f2ded86..ffd4493efc7 100644 --- a/arch/tile/include/asm/compat.h +++ b/arch/tile/include/asm/compat.h @@ -281,7 +281,6 @@ long compat_sys_pread64(unsigned int fd, char __user *ubuf, size_t count, u32 dummy, u32 low, u32 high); long compat_sys_pwrite64(unsigned int fd, char __user *ubuf, size_t count, u32 dummy, u32 low, u32 high); -long compat_sys_lookup_dcookie(u32 low, u32 high, char __user *buf, size_t len); long compat_sys_sync_file_range2(int fd, unsigned int flags, u32 offset_lo, u32 offset_hi, u32 nbytes_lo, u32 nbytes_hi); diff --git a/arch/tile/include/asm/fixmap.h b/arch/tile/include/asm/fixmap.h index c6b9c1b38fd..ffe2637aeb3 100644 --- a/arch/tile/include/asm/fixmap.h +++ b/arch/tile/include/asm/fixmap.h @@ -25,9 +25,6 @@ #include <asm/kmap_types.h> #endif -#define __fix_to_virt(x) (FIXADDR_TOP - ((x) << PAGE_SHIFT)) -#define __virt_to_fix(x) ((FIXADDR_TOP - ((x)&PAGE_MASK)) >> PAGE_SHIFT) - /* * Here we define all the compile-time 'special' virtual * addresses. The point is to have a constant address at @@ -83,35 +80,7 @@ enum fixed_addresses { #define FIXADDR_START (FIXADDR_TOP + PAGE_SIZE - __FIXADDR_SIZE) #define FIXADDR_BOOT_START (FIXADDR_TOP + PAGE_SIZE - __FIXADDR_BOOT_SIZE) -extern void __this_fixmap_does_not_exist(void); - -/* - * 'index to address' translation. If anyone tries to use the idx - * directly without tranlation, we catch the bug with a NULL-deference - * kernel oops. Illegal ranges of incoming indices are caught too. - */ -static __always_inline unsigned long fix_to_virt(const unsigned int idx) -{ - /* - * this branch gets completely eliminated after inlining, - * except when someone tries to use fixaddr indices in an - * illegal way. (such as mixing up address types or using - * out-of-range indices). - * - * If it doesn't get removed, the linker will complain - * loudly with a reasonably clear error message.. - */ - if (idx >= __end_of_fixed_addresses) - __this_fixmap_does_not_exist(); - - return __fix_to_virt(idx); -} - -static inline unsigned long virt_to_fix(const unsigned long vaddr) -{ - BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START); - return __virt_to_fix(vaddr); -} +#include <asm-generic/fixmap.h> #endif /* !__ASSEMBLY__ */ diff --git a/arch/tile/include/asm/hardirq.h b/arch/tile/include/asm/hardirq.h index 822390f9a15..54110af2398 100644 --- a/arch/tile/include/asm/hardirq.h +++ b/arch/tile/include/asm/hardirq.h @@ -42,6 +42,4 @@ DECLARE_PER_CPU(irq_cpustat_t, irq_stat); #include <linux/irq_cpustat.h> /* Standard mappings for irq_cpustat_t above */ -#define HARDIRQ_BITS 8 - #endif /* _ASM_TILE_HARDIRQ_H */ diff --git a/arch/tile/include/asm/irq.h b/arch/tile/include/asm/irq.h index 33cff9a3058..1fe86911838 100644 --- a/arch/tile/include/asm/irq.h +++ b/arch/tile/include/asm/irq.h @@ -18,10 +18,12 @@ #include <linux/hardirq.h> /* The hypervisor interface provides 32 IRQs. */ -#define NR_IRQS 32 +#define NR_IRQS 32 /* IRQ numbers used for linux IPIs. */ -#define IRQ_RESCHEDULE 0 +#define IRQ_RESCHEDULE 0 +/* Interrupts for dynamic allocation start at 1. Let the core allocate irq0 */ +#define NR_IRQS_LEGACY 1 #define irq_canonicalize(irq) (irq) diff --git a/arch/tile/include/asm/page.h b/arch/tile/include/asm/page.h index 6346888f7bd..67276800861 100644 --- a/arch/tile/include/asm/page.h +++ b/arch/tile/include/asm/page.h @@ -182,10 +182,9 @@ static inline __attribute_const__ int get_order(unsigned long size) #define PAGE_OFFSET (-(_AC(1, UL) << (MAX_VA_WIDTH - 1))) #define KERNEL_HIGH_VADDR _AC(0xfffffff800000000, UL) /* high 32GB */ -#define FIXADDR_BASE (KERNEL_HIGH_VADDR - 0x400000000) /* 4 GB */ -#define FIXADDR_TOP (KERNEL_HIGH_VADDR - 0x300000000) /* 4 GB */ +#define FIXADDR_BASE (KERNEL_HIGH_VADDR - 0x300000000) /* 4 GB */ +#define FIXADDR_TOP (KERNEL_HIGH_VADDR - 0x200000000) /* 4 GB */ #define _VMALLOC_START FIXADDR_TOP -#define HUGE_VMAP_BASE (KERNEL_HIGH_VADDR - 0x200000000) /* 4 GB */ #define MEM_SV_START (KERNEL_HIGH_VADDR - 0x100000000) /* 256 MB */ #define MEM_MODULE_START (MEM_SV_START + (256*1024*1024)) /* 256 MB */ #define MEM_MODULE_END (MEM_MODULE_START + (256*1024*1024)) diff --git a/arch/tile/include/asm/percpu.h b/arch/tile/include/asm/percpu.h index 63294f5a8ef..4f7ae39fa20 100644 --- a/arch/tile/include/asm/percpu.h +++ b/arch/tile/include/asm/percpu.h @@ -15,9 +15,37 @@ #ifndef _ASM_TILE_PERCPU_H #define _ASM_TILE_PERCPU_H -register unsigned long __my_cpu_offset __asm__("tp"); -#define __my_cpu_offset __my_cpu_offset -#define set_my_cpu_offset(tp) (__my_cpu_offset = (tp)) +register unsigned long my_cpu_offset_reg asm("tp"); + +#ifdef CONFIG_PREEMPT +/* + * For full preemption, we can't just use the register variable + * directly, since we need barrier() to hazard against it, causing the + * compiler to reload anything computed from a previous "tp" value. + * But we also don't want to use volatile asm, since we'd like the + * compiler to be able to cache the value across multiple percpu reads. + * So we use a fake stack read as a hazard against barrier(). + * The 'U' constraint is like 'm' but disallows postincrement. + */ +static inline unsigned long __my_cpu_offset(void) +{ + unsigned long tp; + register unsigned long *sp asm("sp"); + asm("move %0, tp" : "=r" (tp) : "U" (*sp)); + return tp; +} +#define __my_cpu_offset __my_cpu_offset() +#else +/* + * We don't need to hazard against barrier() since "tp" doesn't ever + * change with PREEMPT_NONE, and with PREEMPT_VOLUNTARY it only + * changes at function call points, at which we are already re-reading + * the value of "tp" due to "my_cpu_offset_reg" being a global variable. + */ +#define __my_cpu_offset my_cpu_offset_reg +#endif + +#define set_my_cpu_offset(tp) (my_cpu_offset_reg = (tp)) #include <asm-generic/percpu.h> diff --git a/arch/tile/include/asm/perf_event.h b/arch/tile/include/asm/perf_event.h new file mode 100644 index 00000000000..59c5b164e5b --- /dev/null +++ b/arch/tile/include/asm/perf_event.h @@ -0,0 +1,22 @@ +/* + * Copyright 2014 Tilera Corporation. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation, version 2. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for + * more details. + */ + +#ifndef _ASM_TILE_PERF_EVENT_H +#define _ASM_TILE_PERF_EVENT_H + +#include <linux/percpu.h> +DECLARE_PER_CPU(u64, perf_irqs); + +unsigned long handle_syscall_link_address(void); +#endif /* _ASM_TILE_PERF_EVENT_H */ diff --git a/arch/tile/include/asm/pgtable_32.h b/arch/tile/include/asm/pgtable_32.h index 63142ab3b3d..d26a4227903 100644 --- a/arch/tile/include/asm/pgtable_32.h +++ b/arch/tile/include/asm/pgtable_32.h @@ -55,17 +55,9 @@ #define PKMAP_BASE ((FIXADDR_BOOT_START - PAGE_SIZE*LAST_PKMAP) & PGDIR_MASK) #ifdef CONFIG_HIGHMEM -# define __VMAPPING_END (PKMAP_BASE & ~(HPAGE_SIZE-1)) +# define _VMALLOC_END (PKMAP_BASE & ~(HPAGE_SIZE-1)) #else -# define __VMAPPING_END (FIXADDR_START & ~(HPAGE_SIZE-1)) -#endif - -#ifdef CONFIG_HUGEVMAP -#define HUGE_VMAP_END __VMAPPING_END -#define HUGE_VMAP_BASE (HUGE_VMAP_END - CONFIG_NR_HUGE_VMAPS * HPAGE_SIZE) -#define _VMALLOC_END HUGE_VMAP_BASE -#else -#define _VMALLOC_END __VMAPPING_END +# define _VMALLOC_END (FIXADDR_START & ~(HPAGE_SIZE-1)) #endif /* diff --git a/arch/tile/include/asm/pgtable_64.h b/arch/tile/include/asm/pgtable_64.h index 3421177f737..2c8a9cd102d 100644 --- a/arch/tile/include/asm/pgtable_64.h +++ b/arch/tile/include/asm/pgtable_64.h @@ -52,12 +52,10 @@ * memory allocation code). The vmalloc code puts in an internal * guard page between each allocation. */ -#define _VMALLOC_END HUGE_VMAP_BASE +#define _VMALLOC_END MEM_SV_START #define VMALLOC_END _VMALLOC_END #define VMALLOC_START _VMALLOC_START -#define HUGE_VMAP_END (HUGE_VMAP_BASE + PGDIR_SIZE) - #ifndef __ASSEMBLY__ /* We have no pud since we are a three-level page table. */ diff --git a/arch/tile/include/asm/pmc.h b/arch/tile/include/asm/pmc.h new file mode 100644 index 00000000000..7ae3956d900 --- /dev/null +++ b/arch/tile/include/asm/pmc.h @@ -0,0 +1,64 @@ +/* + * Copyright 2014 Tilera Corporation. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation, version 2. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for + * more details. + */ + +#ifndef _ASM_TILE_PMC_H +#define _ASM_TILE_PMC_H + +#include <linux/ptrace.h> + +#define TILE_BASE_COUNTERS 2 + +/* Bitfields below are derived from SPR PERF_COUNT_CTL*/ +#ifndef __tilegx__ +/* PERF_COUNT_CTL on TILEPro */ +#define TILE_CTL_EXCL_USER (1 << 7) /* exclude user level */ +#define TILE_CTL_EXCL_KERNEL (1 << 8) /* exclude kernel level */ +#define TILE_CTL_EXCL_HV (1 << 9) /* exclude hypervisor level */ + +#define TILE_SEL_MASK 0x7f /* 7 bits for event SEL, + COUNT_0_SEL */ +#define TILE_PLM_MASK 0x780 /* 4 bits priv level msks, + COUNT_0_MASK*/ +#define TILE_EVENT_MASK (TILE_SEL_MASK | TILE_PLM_MASK) + +#else /* __tilegx__*/ +/* PERF_COUNT_CTL on TILEGx*/ +#define TILE_CTL_EXCL_USER (1 << 10) /* exclude user level */ +#define TILE_CTL_EXCL_KERNEL (1 << 11) /* exclude kernel level */ +#define TILE_CTL_EXCL_HV (1 << 12) /* exclude hypervisor level */ + +#define TILE_SEL_MASK 0x3f /* 6 bits for event SEL, + COUNT_0_SEL*/ +#define TILE_BOX_MASK 0x1c0 /* 3 bits box msks, + COUNT_0_BOX */ +#define TILE_PLM_MASK 0x3c00 /* 4 bits priv level msks, + COUNT_0_MASK */ +#define TILE_EVENT_MASK (TILE_SEL_MASK | TILE_BOX_MASK | TILE_PLM_MASK) +#endif /* __tilegx__*/ + +/* Takes register and fault number. Returns error to disable the interrupt. */ +typedef int (*perf_irq_t)(struct pt_regs *, int); + +int userspace_perf_handler(struct pt_regs *regs, int fault); + +perf_irq_t reserve_pmc_hardware(perf_irq_t new_perf_irq); +void release_pmc_hardware(void); + +unsigned long pmc_get_overflow(void); +void pmc_ack_overflow(unsigned long status); + +void unmask_pmc_interrupts(void); +void mask_pmc_interrupts(void); + +#endif /* _ASM_TILE_PMC_H */ diff --git a/arch/tile/include/asm/thread_info.h b/arch/tile/include/asm/thread_info.h index b8aa6df3e10..48e4fd0f38e 100644 --- a/arch/tile/include/asm/thread_info.h +++ b/arch/tile/include/asm/thread_info.h @@ -94,7 +94,7 @@ register unsigned long stack_pointer __asm__("sp"); /* Sit on a nap instruction until interrupted. */ extern void smp_nap(void); -/* Enable interrupts racelessly and nap forever: helper for cpu_idle(). */ +/* Enable interrupts racelessly and nap forever: helper for arch_cpu_idle(). */ extern void _cpu_idle(void); #else /* __ASSEMBLY__ */ @@ -113,8 +113,6 @@ extern void _cpu_idle(void); #endif /* !__ASSEMBLY__ */ -#define PREEMPT_ACTIVE 0x10000000 - /* * Thread information flags that various assembly files may need to access. * Keep flags accessed frequently in low bits, particular since it makes @@ -131,6 +129,7 @@ extern void _cpu_idle(void); #define TIF_MEMDIE 7 /* OOM killer at work */ #define TIF_NOTIFY_RESUME 8 /* callback before returning to user */ #define TIF_SYSCALL_TRACEPOINT 9 /* syscall tracepoint instrumentation */ +#define TIF_POLLING_NRFLAG 10 /* idle is polling for TIF_NEED_RESCHED */ #define _TIF_SIGPENDING (1<<TIF_SIGPENDING) #define _TIF_NEED_RESCHED (1<<TIF_NEED_RESCHED) @@ -142,6 +141,7 @@ extern void _cpu_idle(void); #define _TIF_MEMDIE (1<<TIF_MEMDIE) #define _TIF_NOTIFY_RESUME (1<<TIF_NOTIFY_RESUME) #define _TIF_SYSCALL_TRACEPOINT (1<<TIF_SYSCALL_TRACEPOINT) +#define _TIF_POLLING_NRFLAG (1<<TIF_POLLING_NRFLAG) /* Work to do on any return to user space. */ #define _TIF_ALLWORK_MASK \ @@ -164,7 +164,6 @@ extern void _cpu_idle(void); #ifdef __tilegx__ #define TS_COMPAT 0x0001 /* 32-bit compatibility mode */ #endif -#define TS_POLLING 0x0004 /* in idle loop but not sleeping */ #define TS_RESTORE_SIGMASK 0x0008 /* restore signal mask in do_signal */ #ifndef __ASSEMBLY__ diff --git a/arch/tile/include/asm/topology.h b/arch/tile/include/asm/topology.h index d15c0d8d550..93831184423 100644 --- a/arch/tile/include/asm/topology.h +++ b/arch/tile/include/asm/topology.h @@ -44,39 +44,6 @@ static inline const struct cpumask *cpumask_of_node(int node) /* For now, use numa node -1 for global allocation. */ #define pcibus_to_node(bus) ((void)(bus), -1) -/* - * TILE architecture has many cores integrated in one processor, so we need - * setup bigger balance_interval for both CPU/NODE scheduling domains to - * reduce process scheduling costs. - */ - -/* sched_domains SD_CPU_INIT for TILE architecture */ -#define SD_CPU_INIT (struct sched_domain) { \ - .min_interval = 4, \ - .max_interval = 128, \ - .busy_factor = 64, \ - .imbalance_pct = 125, \ - .cache_nice_tries = 1, \ - .busy_idx = 2, \ - .idle_idx = 1, \ - .newidle_idx = 0, \ - .wake_idx = 0, \ - .forkexec_idx = 0, \ - \ - .flags = 1*SD_LOAD_BALANCE \ - | 1*SD_BALANCE_NEWIDLE \ - | 1*SD_BALANCE_EXEC \ - | 1*SD_BALANCE_FORK \ - | 0*SD_BALANCE_WAKE \ - | 0*SD_WAKE_AFFINE \ - | 0*SD_SHARE_CPUPOWER \ - | 0*SD_SHARE_PKG_RESOURCES \ - | 0*SD_SERIALIZE \ - , \ - .last_balance = jiffies, \ - .balance_interval = 32, \ -} - /* By definition, we create nodes based on online memory. */ #define node_has_online_mem(nid) 1 diff --git a/arch/tile/include/gxio/iorpc_mpipe.h b/arch/tile/include/gxio/iorpc_mpipe.h index fdd07f88cfd..4cda03de734 100644 --- a/arch/tile/include/gxio/iorpc_mpipe.h +++ b/arch/tile/include/gxio/iorpc_mpipe.h @@ -56,89 +56,89 @@ #define GXIO_MPIPE_OP_GET_MMIO_BASE IORPC_OPCODE(IORPC_FORMAT_NONE_NOUSER, 0x8000) #define GXIO_MPIPE_OP_CHECK_MMIO_OFFSET IORPC_OPCODE(IORPC_FORMAT_NONE_NOUSER, 0x8001) -int gxio_mpipe_alloc_buffer_stacks(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_buffer_stacks(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_mpipe_init_buffer_stack_aux(gxio_mpipe_context_t * context, +int gxio_mpipe_init_buffer_stack_aux(gxio_mpipe_context_t *context, void *mem_va, size_t mem_size, unsigned int mem_flags, unsigned int stack, unsigned int buffer_size_enum); -int gxio_mpipe_alloc_notif_rings(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_notif_rings(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_mpipe_init_notif_ring_aux(gxio_mpipe_context_t * context, void *mem_va, +int gxio_mpipe_init_notif_ring_aux(gxio_mpipe_context_t *context, void *mem_va, size_t mem_size, unsigned int mem_flags, unsigned int ring); -int gxio_mpipe_request_notif_ring_interrupt(gxio_mpipe_context_t * context, +int gxio_mpipe_request_notif_ring_interrupt(gxio_mpipe_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event, unsigned int ring); -int gxio_mpipe_enable_notif_ring_interrupt(gxio_mpipe_context_t * context, +int gxio_mpipe_enable_notif_ring_interrupt(gxio_mpipe_context_t *context, unsigned int ring); -int gxio_mpipe_alloc_notif_groups(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_notif_groups(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_mpipe_init_notif_group(gxio_mpipe_context_t * context, +int gxio_mpipe_init_notif_group(gxio_mpipe_context_t *context, unsigned int group, gxio_mpipe_notif_group_bits_t bits); -int gxio_mpipe_alloc_buckets(gxio_mpipe_context_t * context, unsigned int count, +int gxio_mpipe_alloc_buckets(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_mpipe_init_bucket(gxio_mpipe_context_t * context, unsigned int bucket, +int gxio_mpipe_init_bucket(gxio_mpipe_context_t *context, unsigned int bucket, MPIPE_LBL_INIT_DAT_BSTS_TBL_t bucket_info); -int gxio_mpipe_alloc_edma_rings(gxio_mpipe_context_t * context, +int gxio_mpipe_alloc_edma_rings(gxio_mpipe_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_mpipe_init_edma_ring_aux(gxio_mpipe_context_t * context, void *mem_va, +int gxio_mpipe_init_edma_ring_aux(gxio_mpipe_context_t *context, void *mem_va, size_t mem_size, unsigned int mem_flags, unsigned int ring, unsigned int channel); -int gxio_mpipe_commit_rules(gxio_mpipe_context_t * context, const void *blob, +int gxio_mpipe_commit_rules(gxio_mpipe_context_t *context, const void *blob, size_t blob_size); -int gxio_mpipe_register_client_memory(gxio_mpipe_context_t * context, +int gxio_mpipe_register_client_memory(gxio_mpipe_context_t *context, unsigned int iotlb, HV_PTE pte, unsigned int flags); -int gxio_mpipe_link_open_aux(gxio_mpipe_context_t * context, +int gxio_mpipe_link_open_aux(gxio_mpipe_context_t *context, _gxio_mpipe_link_name_t name, unsigned int flags); -int gxio_mpipe_link_close_aux(gxio_mpipe_context_t * context, int mac); +int gxio_mpipe_link_close_aux(gxio_mpipe_context_t *context, int mac); -int gxio_mpipe_link_set_attr_aux(gxio_mpipe_context_t * context, int mac, +int gxio_mpipe_link_set_attr_aux(gxio_mpipe_context_t *context, int mac, uint32_t attr, int64_t val); -int gxio_mpipe_get_timestamp_aux(gxio_mpipe_context_t * context, uint64_t * sec, - uint64_t * nsec, uint64_t * cycles); +int gxio_mpipe_get_timestamp_aux(gxio_mpipe_context_t *context, uint64_t *sec, + uint64_t *nsec, uint64_t *cycles); -int gxio_mpipe_set_timestamp_aux(gxio_mpipe_context_t * context, uint64_t sec, +int gxio_mpipe_set_timestamp_aux(gxio_mpipe_context_t *context, uint64_t sec, uint64_t nsec, uint64_t cycles); -int gxio_mpipe_adjust_timestamp_aux(gxio_mpipe_context_t * context, +int gxio_mpipe_adjust_timestamp_aux(gxio_mpipe_context_t *context, int64_t nsec); -int gxio_mpipe_adjust_timestamp_freq(gxio_mpipe_context_t * context, +int gxio_mpipe_adjust_timestamp_freq(gxio_mpipe_context_t *context, int32_t ppb); -int gxio_mpipe_arm_pollfd(gxio_mpipe_context_t * context, int pollfd_cookie); +int gxio_mpipe_arm_pollfd(gxio_mpipe_context_t *context, int pollfd_cookie); -int gxio_mpipe_close_pollfd(gxio_mpipe_context_t * context, int pollfd_cookie); +int gxio_mpipe_close_pollfd(gxio_mpipe_context_t *context, int pollfd_cookie); -int gxio_mpipe_get_mmio_base(gxio_mpipe_context_t * context, HV_PTE *base); +int gxio_mpipe_get_mmio_base(gxio_mpipe_context_t *context, HV_PTE *base); -int gxio_mpipe_check_mmio_offset(gxio_mpipe_context_t * context, +int gxio_mpipe_check_mmio_offset(gxio_mpipe_context_t *context, unsigned long offset, unsigned long size); #endif /* !__GXIO_MPIPE_LINUX_RPC_H__ */ diff --git a/arch/tile/include/gxio/iorpc_mpipe_info.h b/arch/tile/include/gxio/iorpc_mpipe_info.h index 476c5e5ca22..f0b04284468 100644 --- a/arch/tile/include/gxio/iorpc_mpipe_info.h +++ b/arch/tile/include/gxio/iorpc_mpipe_info.h @@ -33,18 +33,18 @@ #define GXIO_MPIPE_INFO_OP_CHECK_MMIO_OFFSET IORPC_OPCODE(IORPC_FORMAT_NONE_NOUSER, 0x8001) -int gxio_mpipe_info_instance_aux(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_instance_aux(gxio_mpipe_info_context_t *context, _gxio_mpipe_link_name_t name); -int gxio_mpipe_info_enumerate_aux(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_enumerate_aux(gxio_mpipe_info_context_t *context, unsigned int idx, - _gxio_mpipe_link_name_t * name, - _gxio_mpipe_link_mac_t * mac); + _gxio_mpipe_link_name_t *name, + _gxio_mpipe_link_mac_t *mac); -int gxio_mpipe_info_get_mmio_base(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_get_mmio_base(gxio_mpipe_info_context_t *context, HV_PTE *base); -int gxio_mpipe_info_check_mmio_offset(gxio_mpipe_info_context_t * context, +int gxio_mpipe_info_check_mmio_offset(gxio_mpipe_info_context_t *context, unsigned long offset, unsigned long size); #endif /* !__GXIO_MPIPE_INFO_LINUX_RPC_H__ */ diff --git a/arch/tile/include/gxio/iorpc_trio.h b/arch/tile/include/gxio/iorpc_trio.h index d95b96fd6c9..376a4f77116 100644 --- a/arch/tile/include/gxio/iorpc_trio.h +++ b/arch/tile/include/gxio/iorpc_trio.h @@ -46,59 +46,59 @@ #define GXIO_TRIO_OP_GET_MMIO_BASE IORPC_OPCODE(IORPC_FORMAT_NONE_NOUSER, 0x8000) #define GXIO_TRIO_OP_CHECK_MMIO_OFFSET IORPC_OPCODE(IORPC_FORMAT_NONE_NOUSER, 0x8001) -int gxio_trio_alloc_asids(gxio_trio_context_t * context, unsigned int count, +int gxio_trio_alloc_asids(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_trio_alloc_memory_maps(gxio_trio_context_t * context, +int gxio_trio_alloc_memory_maps(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_trio_alloc_scatter_queues(gxio_trio_context_t * context, +int gxio_trio_alloc_scatter_queues(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_trio_alloc_pio_regions(gxio_trio_context_t * context, +int gxio_trio_alloc_pio_regions(gxio_trio_context_t *context, unsigned int count, unsigned int first, unsigned int flags); -int gxio_trio_init_pio_region_aux(gxio_trio_context_t * context, +int gxio_trio_init_pio_region_aux(gxio_trio_context_t *context, unsigned int pio_region, unsigned int mac, uint32_t bus_address_hi, unsigned int flags); -int gxio_trio_init_memory_map_mmu_aux(gxio_trio_context_t * context, +int gxio_trio_init_memory_map_mmu_aux(gxio_trio_context_t *context, unsigned int map, unsigned long va, uint64_t size, unsigned int asid, unsigned int mac, uint64_t bus_address, unsigned int node, unsigned int order_mode); -int gxio_trio_get_port_property(gxio_trio_context_t * context, +int gxio_trio_get_port_property(gxio_trio_context_t *context, struct pcie_trio_ports_property *trio_ports); -int gxio_trio_config_legacy_intr(gxio_trio_context_t * context, int inter_x, +int gxio_trio_config_legacy_intr(gxio_trio_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event, unsigned int mac, unsigned int intx); -int gxio_trio_config_msi_intr(gxio_trio_context_t * context, int inter_x, +int gxio_trio_config_msi_intr(gxio_trio_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event, unsigned int mac, unsigned int mem_map, uint64_t mem_map_base, uint64_t mem_map_limit, unsigned int asid); -int gxio_trio_set_mps_mrs(gxio_trio_context_t * context, uint16_t mps, +int gxio_trio_set_mps_mrs(gxio_trio_context_t *context, uint16_t mps, uint16_t mrs, unsigned int mac); -int gxio_trio_force_rc_link_up(gxio_trio_context_t * context, unsigned int mac); +int gxio_trio_force_rc_link_up(gxio_trio_context_t *context, unsigned int mac); -int gxio_trio_force_ep_link_up(gxio_trio_context_t * context, unsigned int mac); +int gxio_trio_force_ep_link_up(gxio_trio_context_t *context, unsigned int mac); -int gxio_trio_get_mmio_base(gxio_trio_context_t * context, HV_PTE *base); +int gxio_trio_get_mmio_base(gxio_trio_context_t *context, HV_PTE *base); -int gxio_trio_check_mmio_offset(gxio_trio_context_t * context, +int gxio_trio_check_mmio_offset(gxio_trio_context_t *context, unsigned long offset, unsigned long size); #endif /* !__GXIO_TRIO_LINUX_RPC_H__ */ diff --git a/arch/tile/include/gxio/iorpc_usb_host.h b/arch/tile/include/gxio/iorpc_usb_host.h index 8622e7d126a..79962a97de8 100644 --- a/arch/tile/include/gxio/iorpc_usb_host.h +++ b/arch/tile/include/gxio/iorpc_usb_host.h @@ -31,16 +31,16 @@ #define GXIO_USB_HOST_OP_GET_MMIO_BASE IORPC_OPCODE(IORPC_FORMAT_NONE_NOUSER, 0x8000) #define GXIO_USB_HOST_OP_CHECK_MMIO_OFFSET IORPC_OPCODE(IORPC_FORMAT_NONE_NOUSER, 0x8001) -int gxio_usb_host_cfg_interrupt(gxio_usb_host_context_t * context, int inter_x, +int gxio_usb_host_cfg_interrupt(gxio_usb_host_context_t *context, int inter_x, int inter_y, int inter_ipi, int inter_event); -int gxio_usb_host_register_client_memory(gxio_usb_host_context_t * context, +int gxio_usb_host_register_client_memory(gxio_usb_host_context_t *context, HV_PTE pte, unsigned int flags); -int gxio_usb_host_get_mmio_base(gxio_usb_host_context_t * context, +int gxio_usb_host_get_mmio_base(gxio_usb_host_context_t *context, HV_PTE *base); -int gxio_usb_host_check_mmio_offset(gxio_usb_host_context_t * context, +int gxio_usb_host_check_mmio_offset(gxio_usb_host_context_t *context, unsigned long offset, unsigned long size); #endif /* !__GXIO_USB_HOST_LINUX_RPC_H__ */ diff --git a/arch/tile/include/gxio/usb_host.h b/arch/tile/include/gxio/usb_host.h index 5eedec0e988..93c9636d2dd 100644 --- a/arch/tile/include/gxio/usb_host.h +++ b/arch/tile/include/gxio/usb_host.h @@ -53,7 +53,7 @@ typedef struct { * @return Zero if the context was successfully initialized, else a * GXIO_ERR_xxx error code. */ -extern int gxio_usb_host_init(gxio_usb_host_context_t * context, int usb_index, +extern int gxio_usb_host_init(gxio_usb_host_context_t *context, int usb_index, int is_ehci); /* Destroy a USB context. @@ -68,20 +68,20 @@ extern int gxio_usb_host_init(gxio_usb_host_context_t * context, int usb_index, * @return Zero if the context was successfully destroyed, else a * GXIO_ERR_xxx error code. */ -extern int gxio_usb_host_destroy(gxio_usb_host_context_t * context); +extern int gxio_usb_host_destroy(gxio_usb_host_context_t *context); /* Retrieve the address of the shim's MMIO registers. * * @param context Pointer to a properly initialized gxio_usb_host_context_t. * @return The address of the shim's MMIO registers. */ -extern void *gxio_usb_host_get_reg_start(gxio_usb_host_context_t * context); +extern void *gxio_usb_host_get_reg_start(gxio_usb_host_context_t *context); /* Retrieve the length of the shim's MMIO registers. * * @param context Pointer to a properly initialized gxio_usb_host_context_t. * @return The length of the shim's MMIO registers. */ -extern size_t gxio_usb_host_get_reg_len(gxio_usb_host_context_t * context); +extern size_t gxio_usb_host_get_reg_len(gxio_usb_host_context_t *context); #endif /* _GXIO_USB_H_ */ diff --git a/arch/tile/kernel/Makefile b/arch/tile/kernel/Makefile index 27a2bf39dae..21f77bf68c6 100644 --- a/arch/tile/kernel/Makefile +++ b/arch/tile/kernel/Makefile @@ -25,6 +25,8 @@ obj-$(CONFIG_PCI) += pci_gx.o else obj-$(CONFIG_PCI) += pci.o endif +obj-$(CONFIG_PERF_EVENTS) += perf_event.o +obj-$(CONFIG_USE_PMC) += pmc.o obj-$(CONFIG_TILE_USB) += usb.o obj-$(CONFIG_TILE_HVGLUE_TRACE) += hvglue_trace.o obj-$(CONFIG_FUNCTION_TRACER) += ftrace.o mcount_64.o diff --git a/arch/tile/kernel/compat.c b/arch/tile/kernel/compat.c index ed378416b86..49120843ff9 100644 --- a/arch/tile/kernel/compat.c +++ b/arch/tile/kernel/compat.c @@ -84,7 +84,7 @@ COMPAT_SYSCALL_DEFINE5(llseek, unsigned int, fd, unsigned int, offset_high, { return sys_llseek(fd, offset_high, offset_low, result, origin); } - + /* Provide the compat syscall number to call mapping. */ #undef __SYSCALL #define __SYSCALL(nr, call) [nr] = (call), diff --git a/arch/tile/kernel/compat_signal.c b/arch/tile/kernel/compat_signal.c index 85e00b2f39b..19c04b5ce40 100644 --- a/arch/tile/kernel/compat_signal.c +++ b/arch/tile/kernel/compat_signal.c @@ -49,7 +49,7 @@ struct compat_rt_sigframe { struct compat_ucontext uc; }; -int copy_siginfo_to_user32(struct compat_siginfo __user *to, siginfo_t *from) +int copy_siginfo_to_user32(struct compat_siginfo __user *to, const siginfo_t *from) { int err; diff --git a/arch/tile/kernel/ftrace.c b/arch/tile/kernel/ftrace.c index f1c452092ee..8d52d83cc51 100644 --- a/arch/tile/kernel/ftrace.c +++ b/arch/tile/kernel/ftrace.c @@ -167,10 +167,8 @@ int ftrace_make_nop(struct module *mod, return ret; } -int __init ftrace_dyn_arch_init(void *data) +int __init ftrace_dyn_arch_init(void) { - *(unsigned long *)data = 0; - return 0; } #endif /* CONFIG_DYNAMIC_FTRACE */ diff --git a/arch/tile/kernel/futex_64.S b/arch/tile/kernel/futex_64.S deleted file mode 100644 index f465d1eda20..00000000000 --- a/arch/tile/kernel/futex_64.S +++ /dev/null @@ -1,55 +0,0 @@ -/* - * Copyright 2011 Tilera Corporation. All Rights Reserved. - * - * This program is free software; you can redistribute it and/or - * modify it under the terms of the GNU General Public License - * as published by the Free Software Foundation, version 2. - * - * This program is distributed in the hope that it will be useful, but - * WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or - * NON INFRINGEMENT. See the GNU General Public License for - * more details. - * - * Atomically access user memory, but use MMU to avoid propagating - * kernel exceptions. - */ - -#include <linux/linkage.h> -#include <asm/errno.h> -#include <asm/futex.h> -#include <asm/page.h> -#include <asm/processor.h> - -/* - * Provide a set of atomic memory operations supporting <asm/futex.h>. - * - * r0: user address to manipulate - * r1: new value to write, or for cmpxchg, old value to compare against - * r2: (cmpxchg only) new value to write - * - * Return __get_user struct, r0 with value, r1 with error. - */ -#define FUTEX_OP(name, ...) \ -STD_ENTRY(futex_##name) \ - __VA_ARGS__; \ - { \ - move r1, zero; \ - jrp lr \ - }; \ - STD_ENDPROC(futex_##name); \ - .pushsection __ex_table,"a"; \ - .quad 1b, get_user_fault; \ - .popsection - - .pushsection .fixup,"ax" -get_user_fault: - { movei r1, -EFAULT; jrp lr } - ENDPROC(get_user_fault) - .popsection - -FUTEX_OP(cmpxchg, mtspr CMPEXCH_VALUE, r1; 1: cmpexch4 r0, r0, r2) -FUTEX_OP(set, 1: exch4 r0, r0, r1) -FUTEX_OP(add, 1: fetchadd4 r0, r0, r1) -FUTEX_OP(or, 1: fetchor4 r0, r0, r1) -FUTEX_OP(andn, nor r1, r1, zero; 1: fetchand4 r0, r0, r1) diff --git a/arch/tile/kernel/hardwall.c b/arch/tile/kernel/hardwall.c index df27a1fd94a..531f4c36535 100644 --- a/arch/tile/kernel/hardwall.c +++ b/arch/tile/kernel/hardwall.c @@ -66,7 +66,7 @@ static struct hardwall_type hardwall_types[] = { 0, "udn", LIST_HEAD_INIT(hardwall_types[HARDWALL_UDN].list), - __SPIN_LOCK_INITIALIZER(hardwall_types[HARDWALL_UDN].lock), + __SPIN_LOCK_UNLOCKED(hardwall_types[HARDWALL_UDN].lock), NULL }, #ifndef __tilepro__ @@ -77,7 +77,7 @@ static struct hardwall_type hardwall_types[] = { 1, /* disabled pending hypervisor support */ "idn", LIST_HEAD_INIT(hardwall_types[HARDWALL_IDN].list), - __SPIN_LOCK_INITIALIZER(hardwall_types[HARDWALL_IDN].lock), + __SPIN_LOCK_UNLOCKED(hardwall_types[HARDWALL_IDN].lock), NULL }, { /* access to user-space IPI */ @@ -87,7 +87,7 @@ static struct hardwall_type hardwall_types[] = { 0, "ipi", LIST_HEAD_INIT(hardwall_types[HARDWALL_IPI].list), - __SPIN_LOCK_INITIALIZER(hardwall_types[HARDWALL_IPI].lock), + __SPIN_LOCK_UNLOCKED(hardwall_types[HARDWALL_IPI].lock), NULL }, #endif diff --git a/arch/tile/kernel/intvec_32.S b/arch/tile/kernel/intvec_32.S index 088d5c141e6..cdbda45a4e4 100644 --- a/arch/tile/kernel/intvec_32.S +++ b/arch/tile/kernel/intvec_32.S @@ -313,13 +313,13 @@ intvec_\vecname: movei r3, 0 } .else - .ifc \c_routine, op_handle_perf_interrupt + .ifc \c_routine, handle_perf_interrupt { mfspr r2, PERF_COUNT_STS movei r3, -1 /* not used, but set for consistency */ } .else - .ifc \c_routine, op_handle_aux_perf_interrupt + .ifc \c_routine, handle_perf_interrupt { mfspr r2, AUX_PERF_COUNT_STS movei r3, -1 /* not used, but set for consistency */ @@ -815,6 +815,9 @@ STD_ENTRY(interrupt_return) } bzt r28, 1f bnz r29, 1f + /* Disable interrupts explicitly for preemption. */ + IRQ_DISABLE(r20,r21) + TRACE_IRQS_OFF jal preempt_schedule_irq FEEDBACK_REENTER(interrupt_return) 1: @@ -943,6 +946,13 @@ STD_ENTRY(interrupt_return) bzt r30, .Lrestore_regs 3: + /* We are relying on INT_PERF_COUNT at 33, and AUX_PERF_COUNT at 48 */ + { + moveli r0, lo16(1 << (INT_PERF_COUNT - 32)) + bz r31, .Lrestore_regs + } + auli r0, r0, ha16(1 << (INT_AUX_PERF_COUNT - 32)) + mtspr SPR_INTERRUPT_MASK_RESET_K_1, r0 /* * We now commit to returning from this interrupt, since we will be @@ -1168,6 +1178,10 @@ handle_nmi: PTREGS_PTR(r0, PTREGS_OFFSET_BASE) } FEEDBACK_REENTER(handle_nmi) + { + movei r30, 1 + seq r31, r0, zero + } j interrupt_return STD_ENDPROC(handle_nmi) @@ -1832,8 +1846,9 @@ int_unalign: /* Include .intrpt array of interrupt vectors */ .section ".intrpt", "ax" -#define op_handle_perf_interrupt bad_intr -#define op_handle_aux_perf_interrupt bad_intr +#ifndef CONFIG_USE_PMC +#define handle_perf_interrupt bad_intr +#endif #ifndef CONFIG_HARDWALL #define do_hardwall_trap bad_intr @@ -1874,7 +1889,7 @@ int_unalign: int_hand INT_IDN_AVAIL, IDN_AVAIL, bad_intr int_hand INT_UDN_AVAIL, UDN_AVAIL, bad_intr int_hand INT_PERF_COUNT, PERF_COUNT, \ - op_handle_perf_interrupt, handle_nmi + handle_perf_interrupt, handle_nmi int_hand INT_INTCTRL_3, INTCTRL_3, bad_intr #if CONFIG_KERNEL_PL == 2 dc_dispatch INT_INTCTRL_2, INTCTRL_2 @@ -1899,7 +1914,7 @@ int_unalign: int_hand INT_SN_CPL, SN_CPL, bad_intr int_hand INT_DOUBLE_FAULT, DOUBLE_FAULT, do_trap int_hand INT_AUX_PERF_COUNT, AUX_PERF_COUNT, \ - op_handle_aux_perf_interrupt, handle_nmi + handle_perf_interrupt, handle_nmi /* Synthetic interrupt delivered only by the simulator */ int_hand INT_BREAKPOINT, BREAKPOINT, do_breakpoint diff --git a/arch/tile/kernel/intvec_64.S b/arch/tile/kernel/intvec_64.S index ec755d3f373..5b67efcecab 100644 --- a/arch/tile/kernel/intvec_64.S +++ b/arch/tile/kernel/intvec_64.S @@ -509,10 +509,10 @@ intvec_\vecname: .ifc \c_routine, do_trap mfspr r2, GPV_REASON .else - .ifc \c_routine, op_handle_perf_interrupt + .ifc \c_routine, handle_perf_interrupt mfspr r2, PERF_COUNT_STS .else - .ifc \c_routine, op_handle_aux_perf_interrupt + .ifc \c_routine, handle_perf_interrupt mfspr r2, AUX_PERF_COUNT_STS .endif .endif @@ -841,6 +841,9 @@ STD_ENTRY(interrupt_return) } beqzt r28, 1f bnez r29, 1f + /* Disable interrupts explicitly for preemption. */ + IRQ_DISABLE(r20,r21) + TRACE_IRQS_OFF jal preempt_schedule_irq FEEDBACK_REENTER(interrupt_return) 1: @@ -968,6 +971,15 @@ STD_ENTRY(interrupt_return) beqzt r30, .Lrestore_regs 3: +#if INT_PERF_COUNT + 1 != INT_AUX_PERF_COUNT +# error Bad interrupt assumption +#endif + { + movei r0, 3 /* two adjacent bits for the PERF_COUNT mask */ + beqz r31, .Lrestore_regs + } + shli r0, r0, INT_PERF_COUNT + mtspr SPR_INTERRUPT_MASK_RESET_K, r0 /* * We now commit to returning from this interrupt, since we will be @@ -1184,7 +1196,7 @@ handle_nmi: FEEDBACK_REENTER(handle_nmi) { movei r30, 1 - move r31, r0 + cmpeq r31, r0, zero } j interrupt_return STD_ENDPROC(handle_nmi) @@ -1488,8 +1500,9 @@ STD_ENTRY(fill_ra_stack) .global intrpt_start intrpt_start: -#define op_handle_perf_interrupt bad_intr -#define op_handle_aux_perf_interrupt bad_intr +#ifndef CONFIG_USE_PMC +#define handle_perf_interrupt bad_intr +#endif #ifndef CONFIG_HARDWALL #define do_hardwall_trap bad_intr @@ -1537,9 +1550,9 @@ intrpt_start: #endif int_hand INT_IPI_0, IPI_0, bad_intr int_hand INT_PERF_COUNT, PERF_COUNT, \ - op_handle_perf_interrupt, handle_nmi + handle_perf_interrupt, handle_nmi int_hand INT_AUX_PERF_COUNT, AUX_PERF_COUNT, \ - op_handle_perf_interrupt, handle_nmi + handle_perf_interrupt, handle_nmi int_hand INT_INTCTRL_3, INTCTRL_3, bad_intr #if CONFIG_KERNEL_PL == 2 dc_dispatch INT_INTCTRL_2, INTCTRL_2 diff --git a/arch/tile/kernel/irq.c b/arch/tile/kernel/irq.c index 0586fdb9352..637f2ffaa5f 100644 --- a/arch/tile/kernel/irq.c +++ b/arch/tile/kernel/irq.c @@ -21,6 +21,7 @@ #include <hv/drv_pcie_rc_intf.h> #include <arch/spr_def.h> #include <asm/traps.h> +#include <linux/perf_event.h> /* Bit-flag stored in irq_desc->chip_data to indicate HW-cleared irqs. */ #define IS_HW_CLEARED 1 @@ -53,13 +54,6 @@ static DEFINE_PER_CPU(unsigned long, irq_disable_mask) */ static DEFINE_PER_CPU(int, irq_depth); -/* State for allocating IRQs on Gx. */ -#if CHIP_HAS_IPI() -static unsigned long available_irqs = ((1UL << NR_IRQS) - 1) & - (~(1UL << IRQ_RESCHEDULE)); -static DEFINE_SPINLOCK(available_irqs_lock); -#endif - #if CHIP_HAS_IPI() /* Use SPRs to manipulate device interrupts. */ #define mask_irqs(irq_mask) __insn_mtspr(SPR_IPI_MASK_SET_K, irq_mask) @@ -261,37 +255,27 @@ void ack_bad_irq(unsigned int irq) } /* - * Generic, controller-independent functions: + * /proc/interrupts printing: */ - -#if CHIP_HAS_IPI() -int create_irq(void) +int arch_show_interrupts(struct seq_file *p, int prec) { - unsigned long flags; - int result; - - spin_lock_irqsave(&available_irqs_lock, flags); - if (available_irqs == 0) - result = -ENOMEM; - else { - result = __ffs(available_irqs); - available_irqs &= ~(1UL << result); - dynamic_irq_init(result); - } - spin_unlock_irqrestore(&available_irqs_lock, flags); +#ifdef CONFIG_PERF_EVENTS + int i; - return result; + seq_printf(p, "%*s: ", prec, "PMI"); + + for_each_online_cpu(i) + seq_printf(p, "%10llu ", per_cpu(perf_irqs, i)); + seq_puts(p, " perf_events\n"); +#endif + return 0; } -EXPORT_SYMBOL(create_irq); -void destroy_irq(unsigned int irq) +#if CHIP_HAS_IPI() +int arch_setup_hwirq(unsigned int irq, int node) { - unsigned long flags; - - spin_lock_irqsave(&available_irqs_lock, flags); - available_irqs |= (1UL << irq); - dynamic_irq_cleanup(irq); - spin_unlock_irqrestore(&available_irqs_lock, flags); + return irq >= NR_IRQS ? -EINVAL : 0; } -EXPORT_SYMBOL(destroy_irq); + +void arch_teardown_hwirq(unsigned int irq) { } #endif diff --git a/arch/tile/kernel/messaging.c b/arch/tile/kernel/messaging.c index 00331af9525..7867266f971 100644 --- a/arch/tile/kernel/messaging.c +++ b/arch/tile/kernel/messaging.c @@ -68,8 +68,8 @@ void hv_message_intr(struct pt_regs *regs, int intnum) #endif while (1) { - rmi = hv_receive_message(__get_cpu_var(msg_state), - (HV_VirtAddr) message, + HV_MsgState *state = this_cpu_ptr(&msg_state); + rmi = hv_receive_message(*state, (HV_VirtAddr) message, sizeof(message)); if (rmi.msglen == 0) break; diff --git a/arch/tile/kernel/pci.c b/arch/tile/kernel/pci.c index b7180e6e900..1f80a88c75a 100644 --- a/arch/tile/kernel/pci.c +++ b/arch/tile/kernel/pci.c @@ -250,16 +250,11 @@ static void fixup_read_and_payload_sizes(void) /* Scan for the smallest maximum payload size. */ for_each_pci_dev(dev) { - u32 devcap; - int max_payload; - if (!pci_is_pcie(dev)) continue; - pcie_capability_read_dword(dev, PCI_EXP_DEVCAP, &devcap); - max_payload = devcap & PCI_EXP_DEVCAP_PAYLOAD; - if (max_payload < smallest_max_payload) - smallest_max_payload = max_payload; + if (dev->pcie_mpss < smallest_max_payload) + smallest_max_payload = dev->pcie_mpss; } /* Now, set the max_payload_size for all devices to that value. */ diff --git a/arch/tile/kernel/pci_gx.c b/arch/tile/kernel/pci_gx.c index a97a6452b81..e39f9c54280 100644 --- a/arch/tile/kernel/pci_gx.c +++ b/arch/tile/kernel/pci_gx.c @@ -350,10 +350,9 @@ static int tile_init_irqs(struct pci_controller *controller) int cpu; /* Ask the kernel to allocate an IRQ. */ - irq = create_irq(); - if (irq < 0) { + irq = irq_alloc_hwirq(-1); + if (!irq) { pr_err("PCI: no free irq vectors, failed for %d\n", i); - goto free_irqs; } controller->irq_intx_table[i] = irq; @@ -382,7 +381,7 @@ static int tile_init_irqs(struct pci_controller *controller) free_irqs: for (j = 0; j < i; j++) - destroy_irq(controller->irq_intx_table[j]); + irq_free_hwirq(controller->irq_intx_table[j]); return -1; } @@ -1065,18 +1064,6 @@ char *__init pcibios_setup(char *str) } /* - * Enable memory address decoding, as appropriate, for the - * device described by the 'dev' struct. - * - * This is called from the generic PCI layer, and can be called - * for bridges or endpoints. - */ -int pcibios_enable_device(struct pci_dev *dev, int mask) -{ - return pci_enable_resources(dev, mask); -} - -/* * Called for each device after PCI setup is done. * We initialize the PCI device capabilities conservatively, assuming that * all devices can only address the 32-bit DMA space. The exception here is @@ -1512,9 +1499,9 @@ int arch_setup_msi_irq(struct pci_dev *pdev, struct msi_desc *desc) int irq; int ret; - irq = create_irq(); - if (irq < 0) - return irq; + irq = irq_alloc_hwirq(-1); + if (!irq) + return -ENOSPC; /* * Since we use a 64-bit Mem-Map to accept the MSI write, we fail @@ -1613,11 +1600,11 @@ hv_msi_config_failure: /* Free mem-map */ msi_mem_map_alloc_failure: is_64_failure: - destroy_irq(irq); + irq_free_hwirq(irq); return ret; } void arch_teardown_msi_irq(unsigned int irq) { - destroy_irq(irq); + irq_free_hwirq(irq); } diff --git a/arch/tile/kernel/perf_event.c b/arch/tile/kernel/perf_event.c new file mode 100644 index 00000000000..2bf6c9c135c --- /dev/null +++ b/arch/tile/kernel/perf_event.c @@ -0,0 +1,1005 @@ +/* + * Copyright 2014 Tilera Corporation. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation, version 2. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for + * more details. + * + * + * Perf_events support for Tile processor. + * + * This code is based upon the x86 perf event + * code, which is: + * + * Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de> + * Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar + * Copyright (C) 2009 Jaswinder Singh Rajput + * Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter + * Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com> + * Copyright (C) 2009 Intel Corporation, <markus.t.metzger@intel.com> + * Copyright (C) 2009 Google, Inc., Stephane Eranian + */ + +#include <linux/kprobes.h> +#include <linux/kernel.h> +#include <linux/kdebug.h> +#include <linux/mutex.h> +#include <linux/bitmap.h> +#include <linux/irq.h> +#include <linux/interrupt.h> +#include <linux/perf_event.h> +#include <linux/atomic.h> +#include <asm/traps.h> +#include <asm/stack.h> +#include <asm/pmc.h> +#include <hv/hypervisor.h> + +#define TILE_MAX_COUNTERS 4 + +#define PERF_COUNT_0_IDX 0 +#define PERF_COUNT_1_IDX 1 +#define AUX_PERF_COUNT_0_IDX 2 +#define AUX_PERF_COUNT_1_IDX 3 + +struct cpu_hw_events { + int n_events; + struct perf_event *events[TILE_MAX_COUNTERS]; /* counter order */ + struct perf_event *event_list[TILE_MAX_COUNTERS]; /* enabled + order */ + int assign[TILE_MAX_COUNTERS]; + unsigned long active_mask[BITS_TO_LONGS(TILE_MAX_COUNTERS)]; + unsigned long used_mask; +}; + +/* TILE arch specific performance monitor unit */ +struct tile_pmu { + const char *name; + int version; + const int *hw_events; /* generic hw events table */ + /* generic hw cache events table */ + const int (*cache_events)[PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX]; + int (*map_hw_event)(u64); /*method used to map + hw events */ + int (*map_cache_event)(u64); /*method used to map + cache events */ + + u64 max_period; /* max sampling period */ + u64 cntval_mask; /* counter width mask */ + int cntval_bits; /* counter width */ + int max_events; /* max generic hw events + in map */ + int num_counters; /* number base + aux counters */ + int num_base_counters; /* number base counters */ +}; + +DEFINE_PER_CPU(u64, perf_irqs); +static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events); + +#define TILE_OP_UNSUPP (-1) + +#ifndef __tilegx__ +/* TILEPro hardware events map */ +static const int tile_hw_event_map[] = { + [PERF_COUNT_HW_CPU_CYCLES] = 0x01, /* ONE */ + [PERF_COUNT_HW_INSTRUCTIONS] = 0x06, /* MP_BUNDLE_RETIRED */ + [PERF_COUNT_HW_CACHE_REFERENCES] = TILE_OP_UNSUPP, + [PERF_COUNT_HW_CACHE_MISSES] = TILE_OP_UNSUPP, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x16, /* + MP_CONDITIONAL_BRANCH_ISSUED */ + [PERF_COUNT_HW_BRANCH_MISSES] = 0x14, /* + MP_CONDITIONAL_BRANCH_MISSPREDICT */ + [PERF_COUNT_HW_BUS_CYCLES] = TILE_OP_UNSUPP, +}; +#else +/* TILEGx hardware events map */ +static const int tile_hw_event_map[] = { + [PERF_COUNT_HW_CPU_CYCLES] = 0x181, /* ONE */ + [PERF_COUNT_HW_INSTRUCTIONS] = 0xdb, /* INSTRUCTION_BUNDLE */ + [PERF_COUNT_HW_CACHE_REFERENCES] = TILE_OP_UNSUPP, + [PERF_COUNT_HW_CACHE_MISSES] = TILE_OP_UNSUPP, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0xd9, /* + COND_BRANCH_PRED_CORRECT */ + [PERF_COUNT_HW_BRANCH_MISSES] = 0xda, /* + COND_BRANCH_PRED_INCORRECT */ + [PERF_COUNT_HW_BUS_CYCLES] = TILE_OP_UNSUPP, +}; +#endif + +#define C(x) PERF_COUNT_HW_CACHE_##x + +/* + * Generalized hw caching related hw_event table, filled + * in on a per model basis. A value of -1 means + * 'not supported', any other value means the + * raw hw_event ID. + */ +#ifndef __tilegx__ +/* TILEPro hardware cache event map */ +static const int tile_cache_event_map[PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = { +[C(L1D)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = 0x21, /* RD_MISS */ + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = 0x22, /* WR_MISS */ + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(L1I)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = 0x12, /* MP_ICACHE_HIT_ISSUED */ + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(LL)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(DTLB)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = 0x1d, /* TLB_CNT */ + [C(RESULT_MISS)] = 0x20, /* TLB_EXCEPTION */ + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(ITLB)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = 0x13, /* MP_ITLB_HIT_ISSUED */ + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(BPU)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +}; +#else +/* TILEGx hardware events map */ +static const int tile_cache_event_map[PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = { +[C(L1D)] = { + /* + * Like some other architectures (e.g. ARM), the performance + * counters don't differentiate between read and write + * accesses/misses, so this isn't strictly correct, but it's the + * best we can do. Writes and reads get combined. + */ + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = 0x44, /* RD_MISS */ + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = 0x45, /* WR_MISS */ + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(L1I)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(LL)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(DTLB)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = 0x40, /* TLB_CNT */ + [C(RESULT_MISS)] = 0x43, /* TLB_EXCEPTION */ + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = 0x40, /* TLB_CNT */ + [C(RESULT_MISS)] = 0x43, /* TLB_EXCEPTION */ + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(ITLB)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = 0xd4, /* ITLB_MISS_INT */ + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = 0xd4, /* ITLB_MISS_INT */ + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +[C(BPU)] = { + [C(OP_READ)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_WRITE)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, + [C(OP_PREFETCH)] = { + [C(RESULT_ACCESS)] = TILE_OP_UNSUPP, + [C(RESULT_MISS)] = TILE_OP_UNSUPP, + }, +}, +}; +#endif + +static atomic_t tile_active_events; +static DEFINE_MUTEX(perf_intr_reserve_mutex); + +static int tile_map_hw_event(u64 config); +static int tile_map_cache_event(u64 config); + +static int tile_pmu_handle_irq(struct pt_regs *regs, int fault); + +/* + * To avoid new_raw_count getting larger then pre_raw_count + * in tile_perf_event_update(), we limit the value of max_period to 2^31 - 1. + */ +static const struct tile_pmu tilepmu = { +#ifndef __tilegx__ + .name = "tilepro", +#else + .name = "tilegx", +#endif + .max_events = ARRAY_SIZE(tile_hw_event_map), + .map_hw_event = tile_map_hw_event, + .hw_events = tile_hw_event_map, + .map_cache_event = tile_map_cache_event, + .cache_events = &tile_cache_event_map, + .cntval_bits = 32, + .cntval_mask = (1ULL << 32) - 1, + .max_period = (1ULL << 31) - 1, + .num_counters = TILE_MAX_COUNTERS, + .num_base_counters = TILE_BASE_COUNTERS, +}; + +static const struct tile_pmu *tile_pmu __read_mostly; + +/* + * Check whether perf event is enabled. + */ +int tile_perf_enabled(void) +{ + return atomic_read(&tile_active_events) != 0; +} + +/* + * Read Performance Counters. + */ +static inline u64 read_counter(int idx) +{ + u64 val = 0; + + /* __insn_mfspr() only takes an immediate argument */ + switch (idx) { + case PERF_COUNT_0_IDX: + val = __insn_mfspr(SPR_PERF_COUNT_0); + break; + case PERF_COUNT_1_IDX: + val = __insn_mfspr(SPR_PERF_COUNT_1); + break; + case AUX_PERF_COUNT_0_IDX: + val = __insn_mfspr(SPR_AUX_PERF_COUNT_0); + break; + case AUX_PERF_COUNT_1_IDX: + val = __insn_mfspr(SPR_AUX_PERF_COUNT_1); + break; + default: + WARN_ON_ONCE(idx > AUX_PERF_COUNT_1_IDX || + idx < PERF_COUNT_0_IDX); + } + + return val; +} + +/* + * Write Performance Counters. + */ +static inline void write_counter(int idx, u64 value) +{ + /* __insn_mtspr() only takes an immediate argument */ + switch (idx) { + case PERF_COUNT_0_IDX: + __insn_mtspr(SPR_PERF_COUNT_0, value); + break; + case PERF_COUNT_1_IDX: + __insn_mtspr(SPR_PERF_COUNT_1, value); + break; + case AUX_PERF_COUNT_0_IDX: + __insn_mtspr(SPR_AUX_PERF_COUNT_0, value); + break; + case AUX_PERF_COUNT_1_IDX: + __insn_mtspr(SPR_AUX_PERF_COUNT_1, value); + break; + default: + WARN_ON_ONCE(idx > AUX_PERF_COUNT_1_IDX || + idx < PERF_COUNT_0_IDX); + } +} + +/* + * Enable performance event by setting + * Performance Counter Control registers. + */ +static inline void tile_pmu_enable_event(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + unsigned long cfg, mask; + int shift, idx = hwc->idx; + + /* + * prevent early activation from tile_pmu_start() in hw_perf_enable + */ + + if (WARN_ON_ONCE(idx == -1)) + return; + + if (idx < tile_pmu->num_base_counters) + cfg = __insn_mfspr(SPR_PERF_COUNT_CTL); + else + cfg = __insn_mfspr(SPR_AUX_PERF_COUNT_CTL); + + switch (idx) { + case PERF_COUNT_0_IDX: + case AUX_PERF_COUNT_0_IDX: + mask = TILE_EVENT_MASK; + shift = 0; + break; + case PERF_COUNT_1_IDX: + case AUX_PERF_COUNT_1_IDX: + mask = TILE_EVENT_MASK << 16; + shift = 16; + break; + default: + WARN_ON_ONCE(idx < PERF_COUNT_0_IDX || + idx > AUX_PERF_COUNT_1_IDX); + return; + } + + /* Clear mask bits to enable the event. */ + cfg &= ~mask; + cfg |= hwc->config << shift; + + if (idx < tile_pmu->num_base_counters) + __insn_mtspr(SPR_PERF_COUNT_CTL, cfg); + else + __insn_mtspr(SPR_AUX_PERF_COUNT_CTL, cfg); +} + +/* + * Disable performance event by clearing + * Performance Counter Control registers. + */ +static inline void tile_pmu_disable_event(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + unsigned long cfg, mask; + int idx = hwc->idx; + + if (idx == -1) + return; + + if (idx < tile_pmu->num_base_counters) + cfg = __insn_mfspr(SPR_PERF_COUNT_CTL); + else + cfg = __insn_mfspr(SPR_AUX_PERF_COUNT_CTL); + + switch (idx) { + case PERF_COUNT_0_IDX: + case AUX_PERF_COUNT_0_IDX: + mask = TILE_PLM_MASK; + break; + case PERF_COUNT_1_IDX: + case AUX_PERF_COUNT_1_IDX: + mask = TILE_PLM_MASK << 16; + break; + default: + WARN_ON_ONCE(idx < PERF_COUNT_0_IDX || + idx > AUX_PERF_COUNT_1_IDX); + return; + } + + /* Set mask bits to disable the event. */ + cfg |= mask; + + if (idx < tile_pmu->num_base_counters) + __insn_mtspr(SPR_PERF_COUNT_CTL, cfg); + else + __insn_mtspr(SPR_AUX_PERF_COUNT_CTL, cfg); +} + +/* + * Propagate event elapsed time into the generic event. + * Can only be executed on the CPU where the event is active. + * Returns the delta events processed. + */ +static u64 tile_perf_event_update(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + int shift = 64 - tile_pmu->cntval_bits; + u64 prev_raw_count, new_raw_count; + u64 oldval; + int idx = hwc->idx; + u64 delta; + + /* + * Careful: an NMI might modify the previous event value. + * + * Our tactic to handle this is to first atomically read and + * exchange a new raw count - then add that new-prev delta + * count to the generic event atomically: + */ +again: + prev_raw_count = local64_read(&hwc->prev_count); + new_raw_count = read_counter(idx); + + oldval = local64_cmpxchg(&hwc->prev_count, prev_raw_count, + new_raw_count); + if (oldval != prev_raw_count) + goto again; + + /* + * Now we have the new raw value and have updated the prev + * timestamp already. We can now calculate the elapsed delta + * (event-)time and add that to the generic event. + * + * Careful, not all hw sign-extends above the physical width + * of the count. + */ + delta = (new_raw_count << shift) - (prev_raw_count << shift); + delta >>= shift; + + local64_add(delta, &event->count); + local64_sub(delta, &hwc->period_left); + + return new_raw_count; +} + +/* + * Set the next IRQ period, based on the hwc->period_left value. + * To be called with the event disabled in hw: + */ +static int tile_event_set_period(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; + s64 left = local64_read(&hwc->period_left); + s64 period = hwc->sample_period; + int ret = 0; + + /* + * If we are way outside a reasonable range then just skip forward: + */ + if (unlikely(left <= -period)) { + left = period; + local64_set(&hwc->period_left, left); + hwc->last_period = period; + ret = 1; + } + + if (unlikely(left <= 0)) { + left += period; + local64_set(&hwc->period_left, left); + hwc->last_period = period; + ret = 1; + } + if (left > tile_pmu->max_period) + left = tile_pmu->max_period; + + /* + * The hw event starts counting from this event offset, + * mark it to be able to extra future deltas: + */ + local64_set(&hwc->prev_count, (u64)-left); + + write_counter(idx, (u64)(-left) & tile_pmu->cntval_mask); + + perf_event_update_userpage(event); + + return ret; +} + +/* + * Stop the event but do not release the PMU counter + */ +static void tile_pmu_stop(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; + + if (__test_and_clear_bit(idx, cpuc->active_mask)) { + tile_pmu_disable_event(event); + cpuc->events[hwc->idx] = NULL; + WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED); + hwc->state |= PERF_HES_STOPPED; + } + + if ((flags & PERF_EF_UPDATE) && !(hwc->state & PERF_HES_UPTODATE)) { + /* + * Drain the remaining delta count out of a event + * that we are disabling: + */ + tile_perf_event_update(event); + hwc->state |= PERF_HES_UPTODATE; + } +} + +/* + * Start an event (without re-assigning counter) + */ +static void tile_pmu_start(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int idx = event->hw.idx; + + if (WARN_ON_ONCE(!(event->hw.state & PERF_HES_STOPPED))) + return; + + if (WARN_ON_ONCE(idx == -1)) + return; + + if (flags & PERF_EF_RELOAD) { + WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE)); + tile_event_set_period(event); + } + + event->hw.state = 0; + + cpuc->events[idx] = event; + __set_bit(idx, cpuc->active_mask); + + unmask_pmc_interrupts(); + + tile_pmu_enable_event(event); + + perf_event_update_userpage(event); +} + +/* + * Add a single event to the PMU. + * + * The event is added to the group of enabled events + * but only if it can be scehduled with existing events. + */ +static int tile_pmu_add(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc; + unsigned long mask; + int b, max_cnt; + + hwc = &event->hw; + + /* + * We are full. + */ + if (cpuc->n_events == tile_pmu->num_counters) + return -ENOSPC; + + cpuc->event_list[cpuc->n_events] = event; + cpuc->n_events++; + + hwc->state = PERF_HES_UPTODATE | PERF_HES_STOPPED; + if (!(flags & PERF_EF_START)) + hwc->state |= PERF_HES_ARCH; + + /* + * Find first empty counter. + */ + max_cnt = tile_pmu->num_counters; + mask = ~cpuc->used_mask; + + /* Find next free counter. */ + b = find_next_bit(&mask, max_cnt, 0); + + /* Should not happen. */ + if (WARN_ON_ONCE(b == max_cnt)) + return -ENOSPC; + + /* + * Assign counter to event. + */ + event->hw.idx = b; + __set_bit(b, &cpuc->used_mask); + + /* + * Start if requested. + */ + if (flags & PERF_EF_START) + tile_pmu_start(event, PERF_EF_RELOAD); + + return 0; +} + +/* + * Delete a single event from the PMU. + * + * The event is deleted from the group of enabled events. + * If it is the last event, disable PMU interrupt. + */ +static void tile_pmu_del(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int i; + + /* + * Remove event from list, compact list if necessary. + */ + for (i = 0; i < cpuc->n_events; i++) { + if (cpuc->event_list[i] == event) { + while (++i < cpuc->n_events) + cpuc->event_list[i-1] = cpuc->event_list[i]; + --cpuc->n_events; + cpuc->events[event->hw.idx] = NULL; + __clear_bit(event->hw.idx, &cpuc->used_mask); + tile_pmu_stop(event, PERF_EF_UPDATE); + break; + } + } + /* + * If there are no events left, then mask PMU interrupt. + */ + if (cpuc->n_events == 0) + mask_pmc_interrupts(); + perf_event_update_userpage(event); +} + +/* + * Propagate event elapsed time into the event. + */ +static inline void tile_pmu_read(struct perf_event *event) +{ + tile_perf_event_update(event); +} + +/* + * Map generic events to Tile PMU. + */ +static int tile_map_hw_event(u64 config) +{ + if (config >= tile_pmu->max_events) + return -EINVAL; + return tile_pmu->hw_events[config]; +} + +/* + * Map generic hardware cache events to Tile PMU. + */ +static int tile_map_cache_event(u64 config) +{ + unsigned int cache_type, cache_op, cache_result; + int code; + + if (!tile_pmu->cache_events) + return -ENOENT; + + cache_type = (config >> 0) & 0xff; + if (cache_type >= PERF_COUNT_HW_CACHE_MAX) + return -EINVAL; + + cache_op = (config >> 8) & 0xff; + if (cache_op >= PERF_COUNT_HW_CACHE_OP_MAX) + return -EINVAL; + + cache_result = (config >> 16) & 0xff; + if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX) + return -EINVAL; + + code = (*tile_pmu->cache_events)[cache_type][cache_op][cache_result]; + if (code == TILE_OP_UNSUPP) + return -EINVAL; + + return code; +} + +static void tile_event_destroy(struct perf_event *event) +{ + if (atomic_dec_return(&tile_active_events) == 0) + release_pmc_hardware(); +} + +static int __tile_event_init(struct perf_event *event) +{ + struct perf_event_attr *attr = &event->attr; + struct hw_perf_event *hwc = &event->hw; + int code; + + switch (attr->type) { + case PERF_TYPE_HARDWARE: + code = tile_pmu->map_hw_event(attr->config); + break; + case PERF_TYPE_HW_CACHE: + code = tile_pmu->map_cache_event(attr->config); + break; + case PERF_TYPE_RAW: + code = attr->config & TILE_EVENT_MASK; + break; + default: + /* Should not happen. */ + return -EOPNOTSUPP; + } + + if (code < 0) + return code; + + hwc->config = code; + hwc->idx = -1; + + if (attr->exclude_user) + hwc->config |= TILE_CTL_EXCL_USER; + + if (attr->exclude_kernel) + hwc->config |= TILE_CTL_EXCL_KERNEL; + + if (attr->exclude_hv) + hwc->config |= TILE_CTL_EXCL_HV; + + if (!hwc->sample_period) { + hwc->sample_period = tile_pmu->max_period; + hwc->last_period = hwc->sample_period; + local64_set(&hwc->period_left, hwc->sample_period); + } + event->destroy = tile_event_destroy; + return 0; +} + +static int tile_event_init(struct perf_event *event) +{ + int err = 0; + perf_irq_t old_irq_handler = NULL; + + if (atomic_inc_return(&tile_active_events) == 1) + old_irq_handler = reserve_pmc_hardware(tile_pmu_handle_irq); + + if (old_irq_handler) { + pr_warn("PMC hardware busy (reserved by oprofile)\n"); + + atomic_dec(&tile_active_events); + return -EBUSY; + } + + switch (event->attr.type) { + case PERF_TYPE_RAW: + case PERF_TYPE_HARDWARE: + case PERF_TYPE_HW_CACHE: + break; + + default: + return -ENOENT; + } + + err = __tile_event_init(event); + if (err) { + if (event->destroy) + event->destroy(event); + } + return err; +} + +static struct pmu tilera_pmu = { + .event_init = tile_event_init, + .add = tile_pmu_add, + .del = tile_pmu_del, + + .start = tile_pmu_start, + .stop = tile_pmu_stop, + + .read = tile_pmu_read, +}; + +/* + * PMU's IRQ handler, PMU has 2 interrupts, they share the same handler. + */ +int tile_pmu_handle_irq(struct pt_regs *regs, int fault) +{ + struct perf_sample_data data; + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct perf_event *event; + struct hw_perf_event *hwc; + u64 val; + unsigned long status; + int bit; + + __get_cpu_var(perf_irqs)++; + + if (!atomic_read(&tile_active_events)) + return 0; + + status = pmc_get_overflow(); + pmc_ack_overflow(status); + + for_each_set_bit(bit, &status, tile_pmu->num_counters) { + + event = cpuc->events[bit]; + + if (!event) + continue; + + if (!test_bit(bit, cpuc->active_mask)) + continue; + + hwc = &event->hw; + + val = tile_perf_event_update(event); + if (val & (1ULL << (tile_pmu->cntval_bits - 1))) + continue; + + perf_sample_data_init(&data, 0, event->hw.last_period); + if (!tile_event_set_period(event)) + continue; + + if (perf_event_overflow(event, &data, regs)) + tile_pmu_stop(event, 0); + } + + return 0; +} + +static bool __init supported_pmu(void) +{ + tile_pmu = &tilepmu; + return true; +} + +int __init init_hw_perf_events(void) +{ + supported_pmu(); + perf_pmu_register(&tilera_pmu, "cpu", PERF_TYPE_RAW); + return 0; +} +arch_initcall(init_hw_perf_events); + +/* Callchain handling code. */ + +/* + * Tile specific backtracing code for perf_events. + */ +static inline void perf_callchain(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ + struct KBacktraceIterator kbt; + unsigned int i; + + /* + * Get the address just after the "jalr" instruction that + * jumps to the handler for a syscall. When we find this + * address in a backtrace, we silently ignore it, which gives + * us a one-step backtrace connection from the sys_xxx() + * function in the kernel to the xxx() function in libc. + * Otherwise, we lose the ability to properly attribute time + * from the libc calls to the kernel implementations, since + * oprofile only considers PCs from backtraces a pair at a time. + */ + unsigned long handle_syscall_pc = handle_syscall_link_address(); + + KBacktraceIterator_init(&kbt, NULL, regs); + kbt.profile = 1; + + /* + * The sample for the pc is already recorded. Now we are adding the + * address of the callsites on the stack. Our iterator starts + * with the frame of the (already sampled) call site. If our + * iterator contained a "return address" field, we could have just + * used it and wouldn't have needed to skip the first + * frame. That's in effect what the arm and x86 versions do. + * Instead we peel off the first iteration to get the equivalent + * behavior. + */ + + if (KBacktraceIterator_end(&kbt)) + return; + KBacktraceIterator_next(&kbt); + + /* + * Set stack depth to 16 for user and kernel space respectively, that + * is, total 32 stack frames. + */ + for (i = 0; i < 16; ++i) { + unsigned long pc; + if (KBacktraceIterator_end(&kbt)) + break; + pc = kbt.it.pc; + if (pc != handle_syscall_pc) + perf_callchain_store(entry, pc); + KBacktraceIterator_next(&kbt); + } +} + +void perf_callchain_user(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ + perf_callchain(entry, regs); +} + +void perf_callchain_kernel(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ + perf_callchain(entry, regs); +} diff --git a/arch/tile/kernel/pmc.c b/arch/tile/kernel/pmc.c new file mode 100644 index 00000000000..db62cc34b95 --- /dev/null +++ b/arch/tile/kernel/pmc.c @@ -0,0 +1,121 @@ +/* + * Copyright 2014 Tilera Corporation. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation, version 2. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for + * more details. + */ + +#include <linux/errno.h> +#include <linux/spinlock.h> +#include <linux/module.h> +#include <linux/atomic.h> +#include <linux/interrupt.h> + +#include <asm/processor.h> +#include <asm/pmc.h> + +perf_irq_t perf_irq = NULL; +int handle_perf_interrupt(struct pt_regs *regs, int fault) +{ + int retval; + + if (!perf_irq) + panic("Unexpected PERF_COUNT interrupt %d\n", fault); + + nmi_enter(); + retval = perf_irq(regs, fault); + nmi_exit(); + return retval; +} + +/* Reserve PMC hardware if it is available. */ +perf_irq_t reserve_pmc_hardware(perf_irq_t new_perf_irq) +{ + return cmpxchg(&perf_irq, NULL, new_perf_irq); +} +EXPORT_SYMBOL(reserve_pmc_hardware); + +/* Release PMC hardware. */ +void release_pmc_hardware(void) +{ + perf_irq = NULL; +} +EXPORT_SYMBOL(release_pmc_hardware); + + +/* + * Get current overflow status of each performance counter, + * and auxiliary performance counter. + */ +unsigned long +pmc_get_overflow(void) +{ + unsigned long status; + + /* + * merge base+aux into a single vector + */ + status = __insn_mfspr(SPR_PERF_COUNT_STS); + status |= __insn_mfspr(SPR_AUX_PERF_COUNT_STS) << TILE_BASE_COUNTERS; + return status; +} + +/* + * Clear the status bit for the corresponding counter, if written + * with a one. + */ +void +pmc_ack_overflow(unsigned long status) +{ + /* + * clear overflow status by writing ones + */ + __insn_mtspr(SPR_PERF_COUNT_STS, status); + __insn_mtspr(SPR_AUX_PERF_COUNT_STS, status >> TILE_BASE_COUNTERS); +} + +/* + * The perf count interrupts are masked and unmasked explicitly, + * and only here. The normal irq_enable() does not enable them, + * and irq_disable() does not disable them. That lets these + * routines drive the perf count interrupts orthogonally. + * + * We also mask the perf count interrupts on entry to the perf count + * interrupt handler in assembly code, and by default unmask them + * again (with interrupt critical section protection) just before + * returning from the interrupt. If the perf count handler returns + * a non-zero error code, then we don't re-enable them before returning. + * + * For Pro, we rely on both interrupts being in the same word to update + * them atomically so we never have one enabled and one disabled. + */ + +#if CHIP_HAS_SPLIT_INTR_MASK() +# if INT_PERF_COUNT < 32 || INT_AUX_PERF_COUNT < 32 +# error Fix assumptions about which word PERF_COUNT interrupts are in +# endif +#endif + +static inline unsigned long long pmc_mask(void) +{ + unsigned long long mask = 1ULL << INT_PERF_COUNT; + mask |= 1ULL << INT_AUX_PERF_COUNT; + return mask; +} + +void unmask_pmc_interrupts(void) +{ + interrupt_mask_reset_mask(pmc_mask()); +} + +void mask_pmc_interrupts(void) +{ + interrupt_mask_set_mask(pmc_mask()); +} diff --git a/arch/tile/kernel/proc.c b/arch/tile/kernel/proc.c index 681100c59fd..6829a950864 100644 --- a/arch/tile/kernel/proc.c +++ b/arch/tile/kernel/proc.c @@ -113,7 +113,7 @@ arch_initcall(proc_tile_init); * Support /proc/sys/tile directory */ -static ctl_table unaligned_subtable[] = { +static struct ctl_table unaligned_subtable[] = { { .procname = "enabled", .data = &unaligned_fixup, @@ -138,7 +138,7 @@ static ctl_table unaligned_subtable[] = { {} }; -static ctl_table unaligned_table[] = { +static struct ctl_table unaligned_table[] = { { .procname = "unaligned_fixup", .mode = 0555, diff --git a/arch/tile/kernel/setup.c b/arch/tile/kernel/setup.c index 4c34caea9dd..112ababa9e5 100644 --- a/arch/tile/kernel/setup.c +++ b/arch/tile/kernel/setup.c @@ -228,13 +228,10 @@ early_param("isolnodes", setup_isolnodes); #if defined(CONFIG_PCI) && !defined(__tilegx__) static int __init setup_pci_reserve(char* str) { - unsigned long mb; - - if (str == NULL || strict_strtoul(str, 0, &mb) != 0 || - mb > 3 * 1024) + if (str == NULL || kstrtouint(str, 0, &pci_reserve_mb) != 0 || + pci_reserve_mb > 3 * 1024) return -EINVAL; - pci_reserve_mb = mb; pr_info("Reserving %dMB for PCIE root complex mappings\n", pci_reserve_mb); return 0; @@ -691,7 +688,7 @@ static void __init setup_bootmem_allocator(void) /* Reserve any memory excluded by "memmap" arguments. */ for (i = 0; i < memmap_nr; ++i) { struct memmap_entry *m = &memmap_map[i]; - reserve_bootmem(m->addr, m->size, 0); + reserve_bootmem(m->addr, m->size, BOOTMEM_DEFAULT); } #ifdef CONFIG_BLK_DEV_INITRD @@ -715,7 +712,8 @@ static void __init setup_bootmem_allocator(void) #ifdef CONFIG_KEXEC if (crashk_res.start != crashk_res.end) - reserve_bootmem(crashk_res.start, resource_size(&crashk_res), 0); + reserve_bootmem(crashk_res.start, resource_size(&crashk_res), + BOOTMEM_DEFAULT); #endif } @@ -1268,8 +1266,7 @@ static void __init validate_va(void) if ((long)VMALLOC_START >= 0) early_panic( "Linux VMALLOC region below the 2GB line (%#lx)!\n" - "Reconfigure the kernel with fewer NR_HUGE_VMAPS\n" - "or smaller VMALLOC_RESERVE.\n", + "Reconfigure the kernel with smaller VMALLOC_RESERVE.\n", VMALLOC_START); #endif } diff --git a/arch/tile/kernel/signal.c b/arch/tile/kernel/signal.c index 2d1dbf38a9a..d1d026f0126 100644 --- a/arch/tile/kernel/signal.c +++ b/arch/tile/kernel/signal.c @@ -321,14 +321,13 @@ int show_unhandled_signals = 1; static int __init crashinfo(char *str) { - unsigned long val; const char *word; if (*str == '\0') - val = 2; - else if (*str != '=' || strict_strtoul(++str, 0, &val) != 0) + show_unhandled_signals = 2; + else if (*str != '=' || kstrtoint(++str, 0, &show_unhandled_signals) != 0) return 0; - show_unhandled_signals = val; + switch (show_unhandled_signals) { case 0: word = "No"; diff --git a/arch/tile/kernel/stack.c b/arch/tile/kernel/stack.c index 362284af3af..c93977a6211 100644 --- a/arch/tile/kernel/stack.c +++ b/arch/tile/kernel/stack.c @@ -23,6 +23,7 @@ #include <linux/mmzone.h> #include <linux/dcache.h> #include <linux/fs.h> +#include <linux/string.h> #include <asm/backtrace.h> #include <asm/page.h> #include <asm/ucontext.h> @@ -332,21 +333,18 @@ static void describe_addr(struct KBacktraceIterator *kbt, } if (vma->vm_file) { - char *s; p = d_path(&vma->vm_file->f_path, buf, bufsize); if (IS_ERR(p)) p = "?"; - s = strrchr(p, '/'); - if (s) - p = s+1; + name = kbasename(p); } else { - p = "anon"; + name = "anon"; } /* Generate a string description of the vma info. */ - namelen = strlen(p); + namelen = strlen(name); remaining = (bufsize - 1) - namelen; - memmove(buf, p, namelen); + memmove(buf, name, namelen); snprintf(buf + namelen, remaining, "[%lx+%lx] ", vma->vm_start, vma->vm_end - vma->vm_start); } diff --git a/arch/tile/kernel/time.c b/arch/tile/kernel/time.c index 5d10642db63..462dcd0c170 100644 --- a/arch/tile/kernel/time.c +++ b/arch/tile/kernel/time.c @@ -236,7 +236,15 @@ cycles_t ns2cycles(unsigned long nsecs) * clock frequency. */ struct clock_event_device *dev = &__raw_get_cpu_var(tile_timer); - return ((u64)nsecs * dev->mult) >> dev->shift; + + /* + * as in clocksource.h and x86's timer.h, we split the calculation + * into 2 parts to avoid unecessary overflow of the intermediate + * value. This will not lead to any loss of precision. + */ + u64 quot = (u64)nsecs >> dev->shift; + u64 rem = (u64)nsecs & ((1ULL << dev->shift) - 1); + return quot * dev->mult + ((rem * dev->mult) >> dev->shift); } void update_vsyscall_tz(void) diff --git a/arch/tile/kernel/traps.c b/arch/tile/kernel/traps.c index 6b603d556ca..f3ceb6308e4 100644 --- a/arch/tile/kernel/traps.c +++ b/arch/tile/kernel/traps.c @@ -42,10 +42,9 @@ static int __init setup_unaligned_fixup(char *str) * will still parse the instruction, then fire a SIGBUS with * the correct address from inside the single_step code. */ - long val; - if (strict_strtol(str, 0, &val) != 0) + if (kstrtoint(str, 0, &unaligned_fixup) != 0) return 0; - unaligned_fixup = val; + pr_info("Fixups for unaligned data accesses are %s\n", unaligned_fixup >= 0 ? (unaligned_fixup ? "enabled" : "disabled") : diff --git a/arch/tile/kernel/unaligned.c b/arch/tile/kernel/unaligned.c index b425fb6a480..c02ea2a45f6 100644 --- a/arch/tile/kernel/unaligned.c +++ b/arch/tile/kernel/unaligned.c @@ -182,18 +182,7 @@ static void find_regs(tilegx_bundle_bits bundle, uint64_t *rd, uint64_t *ra, int i; uint64_t reg; uint64_t reg_map = 0, alias_reg_map = 0, map; - bool alias; - - *ra = -1; - *rb = -1; - - if (rd) - *rd = -1; - - *clob1 = -1; - *clob2 = -1; - *clob3 = -1; - alias = false; + bool alias = false; /* * Parse fault bundle, find potential used registers and mark @@ -551,8 +540,8 @@ static tilegx_bundle_bits jit_x1_bnezt(int ra, int broff) /* * This function generates unalign fixup JIT. * - * We fist find unalign load/store instruction's destination, source - * reguisters: ra, rb and rd. and 3 scratch registers by calling + * We first find unalign load/store instruction's destination, source + * registers: ra, rb and rd. and 3 scratch registers by calling * find_regs(...). 3 scratch clobbers should not alias with any register * used in the fault bundle. Then analyze the fault bundle to determine * if it's a load or store, operand width, branch or address increment etc. @@ -569,7 +558,7 @@ void jit_bundle_gen(struct pt_regs *regs, tilegx_bundle_bits bundle, tilegx_bundle_bits bundle_2 = 0; /* If bundle_2_enable = false, bundle_2 is fnop/nop operation. */ bool bundle_2_enable = true; - uint64_t ra, rb, rd = -1, clob1, clob2, clob3; + uint64_t ra = -1, rb = -1, rd = -1, clob1 = -1, clob2 = -1, clob3 = -1; /* * Indicate if the unalign access * instruction's registers hit with diff --git a/arch/tile/kernel/vdso/Makefile b/arch/tile/kernel/vdso/Makefile index e2b7a2f4ee4..a025f63d54c 100644 --- a/arch/tile/kernel/vdso/Makefile +++ b/arch/tile/kernel/vdso/Makefile @@ -104,7 +104,7 @@ $(obj-vdso32:%=%): KBUILD_AFLAGS = $(KBUILD_AFLAGS_32) $(obj-vdso32:%=%): KBUILD_CFLAGS = $(KBUILD_CFLAGS_32) $(obj)/vgettimeofday32.o: $(obj)/vgettimeofday.c - $(call if_changed,cc_o_c) + $(call if_changed_rule,cc_o_c) $(obj)/vrt_sigreturn32.o: $(obj)/vrt_sigreturn.S $(call if_changed,as_o_S) diff --git a/arch/tile/lib/atomic_32.c b/arch/tile/lib/atomic_32.c index 759efa337be..c89b211fd9e 100644 --- a/arch/tile/lib/atomic_32.c +++ b/arch/tile/lib/atomic_32.c @@ -107,19 +107,19 @@ unsigned long _atomic_xor(volatile unsigned long *p, unsigned long mask) EXPORT_SYMBOL(_atomic_xor); -u64 _atomic64_xchg(u64 *v, u64 n) +long long _atomic64_xchg(long long *v, long long n) { return __atomic64_xchg(v, __atomic_setup(v), n); } EXPORT_SYMBOL(_atomic64_xchg); -u64 _atomic64_xchg_add(u64 *v, u64 i) +long long _atomic64_xchg_add(long long *v, long long i) { return __atomic64_xchg_add(v, __atomic_setup(v), i); } EXPORT_SYMBOL(_atomic64_xchg_add); -u64 _atomic64_xchg_add_unless(u64 *v, u64 a, u64 u) +long long _atomic64_xchg_add_unless(long long *v, long long a, long long u) { /* * Note: argument order is switched here since it is easier @@ -130,7 +130,7 @@ u64 _atomic64_xchg_add_unless(u64 *v, u64 a, u64 u) } EXPORT_SYMBOL(_atomic64_xchg_add_unless); -u64 _atomic64_cmpxchg(u64 *v, u64 o, u64 n) +long long _atomic64_cmpxchg(long long *v, long long o, long long n) { return __atomic64_cmpxchg(v, __atomic_setup(v), o, n); } diff --git a/arch/tile/mm/fault.c b/arch/tile/mm/fault.c index 111d5a9b76f..6c0571216a9 100644 --- a/arch/tile/mm/fault.c +++ b/arch/tile/mm/fault.c @@ -149,8 +149,6 @@ static inline int vmalloc_fault(pgd_t *pgd, unsigned long address) pmd_k = vmalloc_sync_one(pgd, address); if (!pmd_k) return -1; - if (pmd_huge(*pmd_k)) - return 0; /* support TILE huge_vmap() API */ pte_k = pte_offset_kernel(pmd_k, address); if (!pte_present(*pte_k)) return -1; @@ -280,8 +278,7 @@ static int handle_page_fault(struct pt_regs *regs, if (!is_page_fault) write = 1; - flags = (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE | - (write ? FAULT_FLAG_WRITE : 0)); + flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE; is_kernel_mode = !user_mode(regs); @@ -365,6 +362,9 @@ static int handle_page_fault(struct pt_regs *regs, goto bad_area_nosemaphore; } + if (!is_kernel_mode) + flags |= FAULT_FLAG_USER; + /* * When running in the kernel we expect faults to occur only to * addresses in user space. All other faults represent errors in the @@ -425,12 +425,12 @@ good_area: #endif if (!(vma->vm_flags & VM_WRITE)) goto bad_area; + flags |= FAULT_FLAG_WRITE; } else { if (!is_page_fault || !(vma->vm_flags & VM_READ)) goto bad_area; } - survive: /* * If for any reason at all we couldn't handle the fault, * make sure we exit gracefully rather than endlessly redo @@ -555,11 +555,6 @@ no_context: */ out_of_memory: up_read(&mm->mmap_sem); - if (is_global_init(tsk)) { - yield(); - down_read(&mm->mmap_sem); - goto survive; - } if (is_kernel_mode) goto no_context; pagefault_out_of_memory(); diff --git a/arch/tile/mm/homecache.c b/arch/tile/mm/homecache.c index 004ba568d93..33294fdc402 100644 --- a/arch/tile/mm/homecache.c +++ b/arch/tile/mm/homecache.c @@ -417,7 +417,7 @@ void __homecache_free_pages(struct page *page, unsigned int order) if (put_page_testzero(page)) { homecache_change_page_home(page, order, PAGE_HOME_HASH); if (order == 0) { - free_hot_cold_page(page, 0); + free_hot_cold_page(page, false); } else { init_page_count(page); __free_pages(page, order); diff --git a/arch/tile/mm/init.c b/arch/tile/mm/init.c index 4e316deb92f..bfb3127b4df 100644 --- a/arch/tile/mm/init.c +++ b/arch/tile/mm/init.c @@ -273,9 +273,9 @@ static pgprot_t __init init_pgprot(ulong address) /* * Otherwise we just hand out consecutive cpus. To avoid * requiring this function to hold state, we just walk forward from - * _sdata by PAGE_SIZE, skipping the readonly and init data, to reach - * the requested address, while walking cpu home around kdata_mask. - * This is typically no more than a dozen or so iterations. + * __end_rodata by PAGE_SIZE, skipping the readonly and init data, to + * reach the requested address, while walking cpu home around + * kdata_mask. This is typically no more than a dozen or so iterations. */ page = (((ulong)__end_rodata) + PAGE_SIZE - 1) & PAGE_MASK; BUG_ON(address < page || address >= (ulong)_end); @@ -828,10 +828,6 @@ void __init mem_init(void) printk(KERN_DEBUG " PKMAP %#lx - %#lx\n", PKMAP_BASE, PKMAP_ADDR(LAST_PKMAP) - 1); #endif -#ifdef CONFIG_HUGEVMAP - printk(KERN_DEBUG " HUGEMAP %#lx - %#lx\n", - HUGE_VMAP_BASE, HUGE_VMAP_END - 1); -#endif printk(KERN_DEBUG " VMALLOC %#lx - %#lx\n", _VMALLOC_START, _VMALLOC_END - 1); #ifdef __tilegx__ @@ -916,7 +912,7 @@ static long __write_once initfree = 1; static int __init set_initfree(char *str) { long val; - if (strict_strtol(str, 0, &val) == 0) { + if (kstrtol(str, 0, &val) == 0) { initfree = val; pr_info("initfree: %s free init pages\n", initfree ? "will" : "won't"); diff --git a/arch/tile/mm/pgtable.c b/arch/tile/mm/pgtable.c index 2deaddf3e01..5e86eac4bfa 100644 --- a/arch/tile/mm/pgtable.c +++ b/arch/tile/mm/pgtable.c @@ -127,8 +127,7 @@ void shatter_huge_page(unsigned long addr) } /* Shatter the huge page into the preallocated L2 page table. */ - pmd_populate_kernel(&init_mm, pmd, - get_prealloc_pte(pte_pfn(*(pte_t *)pmd))); + pmd_populate_kernel(&init_mm, pmd, get_prealloc_pte(pmd_pfn(*pmd))); #ifdef __PAGETABLE_PMD_FOLDED /* Walk every pgd on the system and update the pmd there. */ @@ -242,6 +241,11 @@ struct page *pgtable_alloc_one(struct mm_struct *mm, unsigned long address, if (p == NULL) return NULL; + if (!pgtable_page_ctor(p)) { + __free_pages(p, L2_USER_PGTABLE_ORDER); + return NULL; + } + /* * Make every page have a page_count() of one, not just the first. * We don't use __GFP_COMP since it doesn't look like it works @@ -252,7 +256,6 @@ struct page *pgtable_alloc_one(struct mm_struct *mm, unsigned long address, inc_zone_page_state(p+i, NR_PAGETABLE); } - pgtable_page_ctor(p); return p; } |