From 7868f1ed84696d3cca83558e8dd459f20a36d077 Mon Sep 17 00:00:00 2001 From: Jack Steiner Date: Tue, 6 May 2008 15:18:55 -0500 Subject: [PATCH 01/10] [IA64] Add header files for SGI UV platform Add new UV-specific header files. Signed-off-by: Jack Steiner Signed-off-by: Tony Luck --- include/asm-ia64/uv/uv_hub.h | 309 ++++++++++++++++++++++++++++++++++ include/asm-ia64/uv/uv_mmrs.h | 266 +++++++++++++++++++++++++++++ 2 files changed, 575 insertions(+) create mode 100644 include/asm-ia64/uv/uv_hub.h create mode 100644 include/asm-ia64/uv/uv_mmrs.h diff --git a/include/asm-ia64/uv/uv_hub.h b/include/asm-ia64/uv/uv_hub.h new file mode 100644 index 000000000000..f607018af4a1 --- /dev/null +++ b/include/asm-ia64/uv/uv_hub.h @@ -0,0 +1,309 @@ +/* + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + * + * SGI UV architectural definitions + * + * Copyright (C) 2008 Silicon Graphics, Inc. All rights reserved. + */ + +#ifndef __ASM_IA64_UV_HUB_H__ +#define __ASM_IA64_UV_HUB_H__ + +#include +#include +#include +#include + + +/* + * Addressing Terminology + * + * M - The low M bits of a physical address represent the offset + * into the blade local memory. RAM memory on a blade is physically + * contiguous (although various IO spaces may punch holes in + * it).. + * + * N - Number of bits in the node portion of a socket physical + * address. + * + * NASID - network ID of a router, Mbrick or Cbrick. Nasid values of + * routers always have low bit of 1, C/MBricks have low bit + * equal to 0. Most addressing macros that target UV hub chips + * right shift the NASID by 1 to exclude the always-zero bit. + * NASIDs contain up to 15 bits. + * + * GNODE - NASID right shifted by 1 bit. Most mmrs contain gnodes instead + * of nasids. + * + * PNODE - the low N bits of the GNODE. The PNODE is the most useful variant + * of the nasid for socket usage. + * + * + * NumaLink Global Physical Address Format: + * +--------------------------------+---------------------+ + * |00..000| GNODE | NodeOffset | + * +--------------------------------+---------------------+ + * |<-------53 - M bits --->|<--------M bits -----> + * + * M - number of node offset bits (35 .. 40) + * + * + * Memory/UV-HUB Processor Socket Address Format: + * +----------------+---------------+---------------------+ + * |00..000000000000| PNODE | NodeOffset | + * +----------------+---------------+---------------------+ + * <--- N bits --->|<--------M bits -----> + * + * M - number of node offset bits (35 .. 40) + * N - number of PNODE bits (0 .. 10) + * + * Note: M + N cannot currently exceed 44 (x86_64) or 46 (IA64). + * The actual values are configuration dependent and are set at + * boot time. M & N values are set by the hardware/BIOS at boot. + */ + + +/* + * Maximum number of bricks in all partitions and in all coherency domains. + * This is the total number of bricks accessible in the numalink fabric. It + * includes all C & M bricks. Routers are NOT included. + * + * This value is also the value of the maximum number of non-router NASIDs + * in the numalink fabric. + * + * NOTE: a brick may contain 1 or 2 OS nodes. Don't get these confused. + */ +#define UV_MAX_NUMALINK_BLADES 16384 + +/* + * Maximum number of C/Mbricks within a software SSI (hardware may support + * more). + */ +#define UV_MAX_SSI_BLADES 1 + +/* + * The largest possible NASID of a C or M brick (+ 2) + */ +#define UV_MAX_NASID_VALUE (UV_MAX_NUMALINK_NODES * 2) + +/* + * The following defines attributes of the HUB chip. These attributes are + * frequently referenced and are kept in the per-cpu data areas of each cpu. + * They are kept together in a struct to minimize cache misses. + */ +struct uv_hub_info_s { + unsigned long global_mmr_base; + unsigned long gpa_mask; + unsigned long gnode_upper; + unsigned long lowmem_remap_top; + unsigned long lowmem_remap_base; + unsigned short pnode; + unsigned short pnode_mask; + unsigned short coherency_domain_number; + unsigned short numa_blade_id; + unsigned char blade_processor_id; + unsigned char m_val; + unsigned char n_val; +}; +DECLARE_PER_CPU(struct uv_hub_info_s, __uv_hub_info); +#define uv_hub_info (&__get_cpu_var(__uv_hub_info)) +#define uv_cpu_hub_info(cpu) (&per_cpu(__uv_hub_info, cpu)) + +/* + * Local & Global MMR space macros. + * Note: macros are intended to be used ONLY by inline functions + * in this file - not by other kernel code. + * n - NASID (full 15-bit global nasid) + * g - GNODE (full 15-bit global nasid, right shifted 1) + * p - PNODE (local part of nsids, right shifted 1) + */ +#define UV_NASID_TO_PNODE(n) (((n) >> 1) & uv_hub_info->pnode_mask) +#define UV_PNODE_TO_NASID(p) (((p) << 1) | uv_hub_info->gnode_upper) + +#define UV_LOCAL_MMR_BASE 0xf4000000UL +#define UV_GLOBAL_MMR32_BASE 0xf8000000UL +#define UV_GLOBAL_MMR64_BASE (uv_hub_info->global_mmr_base) + +#define UV_GLOBAL_MMR32_PNODE_SHIFT 15 +#define UV_GLOBAL_MMR64_PNODE_SHIFT 26 + +#define UV_GLOBAL_MMR32_PNODE_BITS(p) ((p) << (UV_GLOBAL_MMR32_PNODE_SHIFT)) + +#define UV_GLOBAL_MMR64_PNODE_BITS(p) \ + ((unsigned long)(p) << UV_GLOBAL_MMR64_PNODE_SHIFT) + +/* + * Macros for converting between kernel virtual addresses, socket local physical + * addresses, and UV global physical addresses. + * Note: use the standard __pa() & __va() macros for converting + * between socket virtual and socket physical addresses. + */ + +/* socket phys RAM --> UV global physical address */ +static inline unsigned long uv_soc_phys_ram_to_gpa(unsigned long paddr) +{ + if (paddr < uv_hub_info->lowmem_remap_top) + paddr += uv_hub_info->lowmem_remap_base; + return paddr | uv_hub_info->gnode_upper; +} + + +/* socket virtual --> UV global physical address */ +static inline unsigned long uv_gpa(void *v) +{ + return __pa(v) | uv_hub_info->gnode_upper; +} + +/* socket virtual --> UV global physical address */ +static inline void *uv_vgpa(void *v) +{ + return (void *)uv_gpa(v); +} + +/* UV global physical address --> socket virtual */ +static inline void *uv_va(unsigned long gpa) +{ + return __va(gpa & uv_hub_info->gpa_mask); +} + +/* pnode, offset --> socket virtual */ +static inline void *uv_pnode_offset_to_vaddr(int pnode, unsigned long offset) +{ + return __va(((unsigned long)pnode << uv_hub_info->m_val) | offset); +} + + +/* + * Access global MMRs using the low memory MMR32 space. This region supports + * faster MMR access but not all MMRs are accessible in this space. + */ +static inline unsigned long *uv_global_mmr32_address(int pnode, + unsigned long offset) +{ + return __va(UV_GLOBAL_MMR32_BASE | + UV_GLOBAL_MMR32_PNODE_BITS(pnode) | offset); +} + +static inline void uv_write_global_mmr32(int pnode, unsigned long offset, + unsigned long val) +{ + *uv_global_mmr32_address(pnode, offset) = val; +} + +static inline unsigned long uv_read_global_mmr32(int pnode, + unsigned long offset) +{ + return *uv_global_mmr32_address(pnode, offset); +} + +/* + * Access Global MMR space using the MMR space located at the top of physical + * memory. + */ +static inline unsigned long *uv_global_mmr64_address(int pnode, + unsigned long offset) +{ + return __va(UV_GLOBAL_MMR64_BASE | + UV_GLOBAL_MMR64_PNODE_BITS(pnode) | offset); +} + +static inline void uv_write_global_mmr64(int pnode, unsigned long offset, + unsigned long val) +{ + *uv_global_mmr64_address(pnode, offset) = val; +} + +static inline unsigned long uv_read_global_mmr64(int pnode, + unsigned long offset) +{ + return *uv_global_mmr64_address(pnode, offset); +} + +/* + * Access hub local MMRs. Faster than using global space but only local MMRs + * are accessible. + */ +static inline unsigned long *uv_local_mmr_address(unsigned long offset) +{ + return __va(UV_LOCAL_MMR_BASE | offset); +} + +static inline unsigned long uv_read_local_mmr(unsigned long offset) +{ + return *uv_local_mmr_address(offset); +} + +static inline void uv_write_local_mmr(unsigned long offset, unsigned long val) +{ + *uv_local_mmr_address(offset) = val; +} + +/* + * Structures and definitions for converting between cpu, node, pnode, and blade + * numbers. + */ + +/* Blade-local cpu number of current cpu. Numbered 0 .. <# cpus on the blade> */ +static inline int uv_blade_processor_id(void) +{ + return smp_processor_id(); +} + +/* Blade number of current cpu. Numnbered 0 .. <#blades -1> */ +static inline int uv_numa_blade_id(void) +{ + return 0; +} + +/* Convert a cpu number to the the UV blade number */ +static inline int uv_cpu_to_blade_id(int cpu) +{ + return 0; +} + +/* Convert linux node number to the UV blade number */ +static inline int uv_node_to_blade_id(int nid) +{ + return 0; +} + +/* Convert a blade id to the PNODE of the blade */ +static inline int uv_blade_to_pnode(int bid) +{ + return 0; +} + +/* Determine the number of possible cpus on a blade */ +static inline int uv_blade_nr_possible_cpus(int bid) +{ + return num_possible_cpus(); +} + +/* Determine the number of online cpus on a blade */ +static inline int uv_blade_nr_online_cpus(int bid) +{ + return num_online_cpus(); +} + +/* Convert a cpu id to the PNODE of the blade containing the cpu */ +static inline int uv_cpu_to_pnode(int cpu) +{ + return 0; +} + +/* Convert a linux node number to the PNODE of the blade */ +static inline int uv_node_to_pnode(int nid) +{ + return 0; +} + +/* Maximum possible number of blades */ +static inline int uv_num_possible_blades(void) +{ + return 1; +} + +#endif /* __ASM_IA64_UV_HUB__ */ + diff --git a/include/asm-ia64/uv/uv_mmrs.h b/include/asm-ia64/uv/uv_mmrs.h new file mode 100644 index 000000000000..1cc1dbb0182f --- /dev/null +++ b/include/asm-ia64/uv/uv_mmrs.h @@ -0,0 +1,266 @@ +/* + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + * + * SGI UV MMR definitions + * + * Copyright (C) 2007-2008 Silicon Graphics, Inc. All rights reserved. + */ + +#ifndef __ASM_IA64_UV_MMRS__ +#define __ASM_IA64_UV_MMRS__ + +/* + * AUTO GENERATED - Do not edit + */ + + #define UV_MMR_ENABLE (1UL << 63) + +/* ========================================================================= */ +/* UVH_NODE_ID */ +/* ========================================================================= */ +#define UVH_NODE_ID 0x0UL + +#define UVH_NODE_ID_FORCE1_SHFT 0 +#define UVH_NODE_ID_FORCE1_MASK 0x0000000000000001UL +#define UVH_NODE_ID_MANUFACTURER_SHFT 1 +#define UVH_NODE_ID_MANUFACTURER_MASK 0x0000000000000ffeUL +#define UVH_NODE_ID_PART_NUMBER_SHFT 12 +#define UVH_NODE_ID_PART_NUMBER_MASK 0x000000000ffff000UL +#define UVH_NODE_ID_REVISION_SHFT 28 +#define UVH_NODE_ID_REVISION_MASK 0x00000000f0000000UL +#define UVH_NODE_ID_NODE_ID_SHFT 32 +#define UVH_NODE_ID_NODE_ID_MASK 0x00007fff00000000UL +#define UVH_NODE_ID_NODES_PER_BIT_SHFT 48 +#define UVH_NODE_ID_NODES_PER_BIT_MASK 0x007f000000000000UL +#define UVH_NODE_ID_NI_PORT_SHFT 56 +#define UVH_NODE_ID_NI_PORT_MASK 0x0f00000000000000UL + +union uvh_node_id_u { + unsigned long v; + struct uvh_node_id_s { + unsigned long force1 : 1; /* RO */ + unsigned long manufacturer : 11; /* RO */ + unsigned long part_number : 16; /* RO */ + unsigned long revision : 4; /* RO */ + unsigned long node_id : 15; /* RW */ + unsigned long rsvd_47 : 1; /* */ + unsigned long nodes_per_bit : 7; /* RW */ + unsigned long rsvd_55 : 1; /* */ + unsigned long ni_port : 4; /* RO */ + unsigned long rsvd_60_63 : 4; /* */ + } s; +}; + +/* ========================================================================= */ +/* UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR */ +/* ========================================================================= */ +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR 0x16000d0UL + +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR_DEST_BASE_SHFT 24 +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR_DEST_BASE_MASK 0x00003fffff000000UL + +union uvh_rh_gam_alias210_redirect_config_0_mmr_u { + unsigned long v; + struct uvh_rh_gam_alias210_redirect_config_0_mmr_s { + unsigned long rsvd_0_23 : 24; /* */ + unsigned long dest_base : 22; /* RW */ + unsigned long rsvd_46_63: 18; /* */ + } s; +}; + +/* ========================================================================= */ +/* UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_1_MMR */ +/* ========================================================================= */ +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_1_MMR 0x16000e0UL + +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_1_MMR_DEST_BASE_SHFT 24 +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_1_MMR_DEST_BASE_MASK 0x00003fffff000000UL + +union uvh_rh_gam_alias210_redirect_config_1_mmr_u { + unsigned long v; + struct uvh_rh_gam_alias210_redirect_config_1_mmr_s { + unsigned long rsvd_0_23 : 24; /* */ + unsigned long dest_base : 22; /* RW */ + unsigned long rsvd_46_63: 18; /* */ + } s; +}; + +/* ========================================================================= */ +/* UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_2_MMR */ +/* ========================================================================= */ +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_2_MMR 0x16000f0UL + +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_2_MMR_DEST_BASE_SHFT 24 +#define UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_2_MMR_DEST_BASE_MASK 0x00003fffff000000UL + +union uvh_rh_gam_alias210_redirect_config_2_mmr_u { + unsigned long v; + struct uvh_rh_gam_alias210_redirect_config_2_mmr_s { + unsigned long rsvd_0_23 : 24; /* */ + unsigned long dest_base : 22; /* RW */ + unsigned long rsvd_46_63: 18; /* */ + } s; +}; + +/* ========================================================================= */ +/* UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR */ +/* ========================================================================= */ +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR 0x1600010UL + +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_BASE_SHFT 28 +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_BASE_MASK 0x00003ffff0000000UL +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_GR4_SHFT 46 +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_GR4_MASK 0x0000400000000000UL +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_N_GRU_SHFT 52 +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_N_GRU_MASK 0x00f0000000000000UL +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_ENABLE_SHFT 63 +#define UVH_RH_GAM_GRU_OVERLAY_CONFIG_MMR_ENABLE_MASK 0x8000000000000000UL + +union uvh_rh_gam_gru_overlay_config_mmr_u { + unsigned long v; + struct uvh_rh_gam_gru_overlay_config_mmr_s { + unsigned long rsvd_0_27: 28; /* */ + unsigned long base : 18; /* RW */ + unsigned long gr4 : 1; /* RW */ + unsigned long rsvd_47_51: 5; /* */ + unsigned long n_gru : 4; /* RW */ + unsigned long rsvd_56_62: 7; /* */ + unsigned long enable : 1; /* RW */ + } s; +}; + +/* ========================================================================= */ +/* UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR */ +/* ========================================================================= */ +#define UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR 0x1600028UL + +#define UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR_BASE_SHFT 26 +#define UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR_BASE_MASK 0x00003ffffc000000UL +#define UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR_DUAL_HUB_SHFT 46 +#define UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR_DUAL_HUB_MASK 0x0000400000000000UL +#define UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR_ENABLE_SHFT 63 +#define UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR_ENABLE_MASK 0x8000000000000000UL + +union uvh_rh_gam_mmr_overlay_config_mmr_u { + unsigned long v; + struct uvh_rh_gam_mmr_overlay_config_mmr_s { + unsigned long rsvd_0_25: 26; /* */ + unsigned long base : 20; /* RW */ + unsigned long dual_hub : 1; /* RW */ + unsigned long rsvd_47_62: 16; /* */ + unsigned long enable : 1; /* RW */ + } s; +}; + +/* ========================================================================= */ +/* UVH_RTC */ +/* ========================================================================= */ +#define UVH_RTC 0x28000UL + +#define UVH_RTC_REAL_TIME_CLOCK_SHFT 0 +#define UVH_RTC_REAL_TIME_CLOCK_MASK 0x00ffffffffffffffUL + +union uvh_rtc_u { + unsigned long v; + struct uvh_rtc_s { + unsigned long real_time_clock : 56; /* RW */ + unsigned long rsvd_56_63 : 8; /* */ + } s; +}; + +/* ========================================================================= */ +/* UVH_SI_ADDR_MAP_CONFIG */ +/* ========================================================================= */ +#define UVH_SI_ADDR_MAP_CONFIG 0xc80000UL + +#define UVH_SI_ADDR_MAP_CONFIG_M_SKT_SHFT 0 +#define UVH_SI_ADDR_MAP_CONFIG_M_SKT_MASK 0x000000000000003fUL +#define UVH_SI_ADDR_MAP_CONFIG_N_SKT_SHFT 8 +#define UVH_SI_ADDR_MAP_CONFIG_N_SKT_MASK 0x0000000000000f00UL + +union uvh_si_addr_map_config_u { + unsigned long v; + struct uvh_si_addr_map_config_s { + unsigned long m_skt : 6; /* RW */ + unsigned long rsvd_6_7: 2; /* */ + unsigned long n_skt : 4; /* RW */ + unsigned long rsvd_12_63: 52; /* */ + } s; +}; + +/* ========================================================================= */ +/* UVH_SI_ALIAS0_OVERLAY_CONFIG */ +/* ========================================================================= */ +#define UVH_SI_ALIAS0_OVERLAY_CONFIG 0xc80008UL + +#define UVH_SI_ALIAS0_OVERLAY_CONFIG_BASE_SHFT 24 +#define UVH_SI_ALIAS0_OVERLAY_CONFIG_BASE_MASK 0x00000000ff000000UL +#define UVH_SI_ALIAS0_OVERLAY_CONFIG_M_ALIAS_SHFT 48 +#define UVH_SI_ALIAS0_OVERLAY_CONFIG_M_ALIAS_MASK 0x001f000000000000UL +#define UVH_SI_ALIAS0_OVERLAY_CONFIG_ENABLE_SHFT 63 +#define UVH_SI_ALIAS0_OVERLAY_CONFIG_ENABLE_MASK 0x8000000000000000UL + +union uvh_si_alias0_overlay_config_u { + unsigned long v; + struct uvh_si_alias0_overlay_config_s { + unsigned long rsvd_0_23: 24; /* */ + unsigned long base : 8; /* RW */ + unsigned long rsvd_32_47: 16; /* */ + unsigned long m_alias : 5; /* RW */ + unsigned long rsvd_53_62: 10; /* */ + unsigned long enable : 1; /* RW */ + } s; +}; + +/* ========================================================================= */ +/* UVH_SI_ALIAS1_OVERLAY_CONFIG */ +/* ========================================================================= */ +#define UVH_SI_ALIAS1_OVERLAY_CONFIG 0xc80010UL + +#define UVH_SI_ALIAS1_OVERLAY_CONFIG_BASE_SHFT 24 +#define UVH_SI_ALIAS1_OVERLAY_CONFIG_BASE_MASK 0x00000000ff000000UL +#define UVH_SI_ALIAS1_OVERLAY_CONFIG_M_ALIAS_SHFT 48 +#define UVH_SI_ALIAS1_OVERLAY_CONFIG_M_ALIAS_MASK 0x001f000000000000UL +#define UVH_SI_ALIAS1_OVERLAY_CONFIG_ENABLE_SHFT 63 +#define UVH_SI_ALIAS1_OVERLAY_CONFIG_ENABLE_MASK 0x8000000000000000UL + +union uvh_si_alias1_overlay_config_u { + unsigned long v; + struct uvh_si_alias1_overlay_config_s { + unsigned long rsvd_0_23: 24; /* */ + unsigned long base : 8; /* RW */ + unsigned long rsvd_32_47: 16; /* */ + unsigned long m_alias : 5; /* RW */ + unsigned long rsvd_53_62: 10; /* */ + unsigned long enable : 1; /* RW */ + } s; +}; + +/* ========================================================================= */ +/* UVH_SI_ALIAS2_OVERLAY_CONFIG */ +/* ========================================================================= */ +#define UVH_SI_ALIAS2_OVERLAY_CONFIG 0xc80018UL + +#define UVH_SI_ALIAS2_OVERLAY_CONFIG_BASE_SHFT 24 +#define UVH_SI_ALIAS2_OVERLAY_CONFIG_BASE_MASK 0x00000000ff000000UL +#define UVH_SI_ALIAS2_OVERLAY_CONFIG_M_ALIAS_SHFT 48 +#define UVH_SI_ALIAS2_OVERLAY_CONFIG_M_ALIAS_MASK 0x001f000000000000UL +#define UVH_SI_ALIAS2_OVERLAY_CONFIG_ENABLE_SHFT 63 +#define UVH_SI_ALIAS2_OVERLAY_CONFIG_ENABLE_MASK 0x8000000000000000UL + +union uvh_si_alias2_overlay_config_u { + unsigned long v; + struct uvh_si_alias2_overlay_config_s { + unsigned long rsvd_0_23: 24; /* */ + unsigned long base : 8; /* RW */ + unsigned long rsvd_32_47: 16; /* */ + unsigned long m_alias : 5; /* RW */ + unsigned long rsvd_53_62: 10; /* */ + unsigned long enable : 1; /* RW */ + } s; +}; + + +#endif /* __ASM_IA64_UV_MMRS__ */ From 2224661494278bfc1c35b392cf6ee6f58e1d5e64 Mon Sep 17 00:00:00 2001 From: Jack Steiner Date: Tue, 6 May 2008 15:18:57 -0500 Subject: [PATCH 02/10] [IA64] machvec support for SGI UV platform This patch adds the basic IA64 machvec infrastructure to support the SGI "UV" platform. Signed-off-by: Jack Steiner Signed-off-by: Tony Luck --- arch/ia64/Kconfig | 13 +++++ arch/ia64/Makefile | 2 +- arch/ia64/kernel/acpi.c | 7 ++- arch/ia64/uv/Makefile | 12 +++++ arch/ia64/uv/kernel/Makefile | 13 +++++ arch/ia64/uv/kernel/machvec.c | 11 ++++ arch/ia64/uv/kernel/setup.c | 98 +++++++++++++++++++++++++++++++++++ include/asm-ia64/machvec.h | 2 + include/asm-ia64/machvec_uv.h | 26 ++++++++++ 9 files changed, 182 insertions(+), 2 deletions(-) create mode 100644 arch/ia64/uv/Makefile create mode 100644 arch/ia64/uv/kernel/Makefile create mode 100644 arch/ia64/uv/kernel/machvec.c create mode 100644 arch/ia64/uv/kernel/setup.c create mode 100644 include/asm-ia64/machvec_uv.h diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig index 0df5f6f75edf..16be41446b5b 100644 --- a/arch/ia64/Kconfig +++ b/arch/ia64/Kconfig @@ -135,6 +135,7 @@ config IA64_GENERIC HP-zx1/sx1000 For HP systems HP-zx1/sx1000+swiotlb For HP systems with (broken) DMA-constrained devices. SGI-SN2 For SGI Altix systems + SGI-UV For SGI UV systems Ski-simulator For the HP simulator If you don't know what to do, choose "generic". @@ -170,6 +171,18 @@ config IA64_SGI_SN2 to select this option. If in doubt, select ia64 generic support instead. +config IA64_SGI_UV` + bool "SGI-UV`" + select NUMA + select ACPI_NUMA + select SWIOTLB + help + Selecting this option will optimize the kernel for use on UV based + systems, but the resulting kernel binary will not run on other + types of ia64 systems. If you have an SGI UV system, it's safe + to select this option. If in doubt, select ia64 generic support + instead. + config IA64_HP_SIM bool "Ski-simulator" select SWIOTLB diff --git a/arch/ia64/Makefile b/arch/ia64/Makefile index ec4cca477f49..88f1a55c6c94 100644 --- a/arch/ia64/Makefile +++ b/arch/ia64/Makefile @@ -63,7 +63,7 @@ drivers-$(CONFIG_PCI) += arch/ia64/pci/ drivers-$(CONFIG_IA64_HP_SIM) += arch/ia64/hp/sim/ drivers-$(CONFIG_IA64_HP_ZX1) += arch/ia64/hp/common/ arch/ia64/hp/zx1/ drivers-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += arch/ia64/hp/common/ arch/ia64/hp/zx1/ -drivers-$(CONFIG_IA64_GENERIC) += arch/ia64/hp/common/ arch/ia64/hp/zx1/ arch/ia64/hp/sim/ arch/ia64/sn/ +drivers-$(CONFIG_IA64_GENERIC) += arch/ia64/hp/common/ arch/ia64/hp/zx1/ arch/ia64/hp/sim/ arch/ia64/sn/ arch/ia64/uv/ drivers-$(CONFIG_OPROFILE) += arch/ia64/oprofile/ boot := arch/ia64/hp/sim/boot diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c index 19709a079635..6ff6815d34ef 100644 --- a/arch/ia64/kernel/acpi.c +++ b/arch/ia64/kernel/acpi.c @@ -117,7 +117,10 @@ acpi_get_sysname(void) if (!strcmp(hdr->oem_id, "HP")) { return "hpzx1"; } else if (!strcmp(hdr->oem_id, "SGI")) { - return "sn2"; + if (!strcmp(hdr->oem_table_id + 4, "UV")) + return "uv"; + else + return "sn2"; } return "dig"; @@ -130,6 +133,8 @@ acpi_get_sysname(void) return "hpzx1_swiotlb"; # elif defined (CONFIG_IA64_SGI_SN2) return "sn2"; +# elif defined (CONFIG_IA64_SGI_UV) + return "uv"; # elif defined (CONFIG_IA64_DIG) return "dig"; # else diff --git a/arch/ia64/uv/Makefile b/arch/ia64/uv/Makefile new file mode 100644 index 000000000000..aa9f91947c49 --- /dev/null +++ b/arch/ia64/uv/Makefile @@ -0,0 +1,12 @@ +# arch/ia64/uv/Makefile +# +# This file is subject to the terms and conditions of the GNU General Public +# License. See the file "COPYING" in the main directory of this archive +# for more details. +# +# Copyright (C) 2008 Silicon Graphics, Inc. All Rights Reserved. +# +# Makefile for the sn uv subplatform +# + +obj-y += kernel/ diff --git a/arch/ia64/uv/kernel/Makefile b/arch/ia64/uv/kernel/Makefile new file mode 100644 index 000000000000..8d92b4684d8e --- /dev/null +++ b/arch/ia64/uv/kernel/Makefile @@ -0,0 +1,13 @@ +# arch/ia64/uv/kernel/Makefile +# +# This file is subject to the terms and conditions of the GNU General Public +# License. See the file "COPYING" in the main directory of this archive +# for more details. +# +# Copyright (C) 2008 Silicon Graphics, Inc. All Rights Reserved. +# + +EXTRA_CFLAGS += -Iarch/ia64/sn/include + +obj-y += setup.o +obj-$(CONFIG_IA64_GENERIC) += machvec.o diff --git a/arch/ia64/uv/kernel/machvec.c b/arch/ia64/uv/kernel/machvec.c new file mode 100644 index 000000000000..50737a9dca74 --- /dev/null +++ b/arch/ia64/uv/kernel/machvec.c @@ -0,0 +1,11 @@ +/* + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + * + * Copyright (c) 2008 Silicon Graphics, Inc. All Rights Reserved. + */ + +#define MACHVEC_PLATFORM_NAME uv +#define MACHVEC_PLATFORM_HEADER +#include diff --git a/arch/ia64/uv/kernel/setup.c b/arch/ia64/uv/kernel/setup.c new file mode 100644 index 000000000000..9aa743203c3c --- /dev/null +++ b/arch/ia64/uv/kernel/setup.c @@ -0,0 +1,98 @@ +/* + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + * + * SGI UV Core Functions + * + * Copyright (C) 2008 Silicon Graphics, Inc. All rights reserved. + */ + +#include +#include +#include +#include +#include + +DEFINE_PER_CPU(struct uv_hub_info_s, __uv_hub_info); +EXPORT_PER_CPU_SYMBOL_GPL(__uv_hub_info); + + +struct redir_addr { + unsigned long redirect; + unsigned long alias; +}; + +#define DEST_SHIFT UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR_DEST_BASE_SHFT + +static __initdata struct redir_addr redir_addrs[] = { + {UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR, UVH_SI_ALIAS0_OVERLAY_CONFIG}, + {UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_1_MMR, UVH_SI_ALIAS1_OVERLAY_CONFIG}, + {UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_2_MMR, UVH_SI_ALIAS2_OVERLAY_CONFIG}, +}; + +static __init void get_lowmem_redirect(unsigned long *base, unsigned long *size) +{ + union uvh_si_alias0_overlay_config_u alias; + union uvh_rh_gam_alias210_redirect_config_2_mmr_u redirect; + int i; + + for (i = 0; i < ARRAY_SIZE(redir_addrs); i++) { + alias.v = uv_read_local_mmr(redir_addrs[i].alias); + if (alias.s.base == 0) { + *size = (1UL << alias.s.m_alias); + redirect.v = uv_read_local_mmr(redir_addrs[i].redirect); + *base = (unsigned long)redirect.s.dest_base << DEST_SHIFT; + return; + } + } + BUG(); +} + +void __init uv_setup(char **cmdline_p) +{ + union uvh_si_addr_map_config_u m_n_config; + union uvh_node_id_u node_id; + unsigned long gnode_upper; + int nid, cpu, m_val, n_val; + unsigned long mmr_base, lowmem_redir_base, lowmem_redir_size; + + if (IS_MEDUSA()) { + lowmem_redir_base = 0; + lowmem_redir_size = 0; + node_id.v = 0; + m_n_config.s.m_skt = 37; + m_n_config.s.n_skt = 0; + mmr_base = 0; + } else { + get_lowmem_redirect(&lowmem_redir_base, &lowmem_redir_size); + node_id.v = uv_read_local_mmr(UVH_NODE_ID); + m_n_config.v = uv_read_local_mmr(UVH_SI_ADDR_MAP_CONFIG); + mmr_base = + uv_read_local_mmr(UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR) & + ~UV_MMR_ENABLE; + } + + m_val = m_n_config.s.m_skt; + n_val = m_n_config.s.n_skt; + printk(KERN_DEBUG "UV: global MMR base 0x%lx\n", mmr_base); + + gnode_upper = (((unsigned long)node_id.s.node_id) & + ~((1 << n_val) - 1)) << m_val; + + for_each_present_cpu(cpu) { + nid = cpu_to_node(cpu); + uv_cpu_hub_info(cpu)->lowmem_remap_base = lowmem_redir_base; + uv_cpu_hub_info(cpu)->lowmem_remap_top = + lowmem_redir_base + lowmem_redir_size; + uv_cpu_hub_info(cpu)->m_val = m_val; + uv_cpu_hub_info(cpu)->n_val = m_val; + uv_cpu_hub_info(cpu)->pnode_mask = (1 << n_val) -1; + uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1; + uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper; + uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base; + uv_cpu_hub_info(cpu)->coherency_domain_number = 0;/* ZZZ */ + printk(KERN_DEBUG "UV cpu %d, nid %d\n", cpu, nid); + } +} + diff --git a/include/asm-ia64/machvec.h b/include/asm-ia64/machvec.h index 9f020eb825c5..0721a5e8271e 100644 --- a/include/asm-ia64/machvec.h +++ b/include/asm-ia64/machvec.h @@ -126,6 +126,8 @@ extern void machvec_tlb_migrate_finish (struct mm_struct *); # include # elif defined (CONFIG_IA64_SGI_SN2) # include +# elif defined (CONFIG_IA64_SGI_UV) +# include # elif defined (CONFIG_IA64_GENERIC) # ifdef MACHVEC_PLATFORM_HEADER diff --git a/include/asm-ia64/machvec_uv.h b/include/asm-ia64/machvec_uv.h new file mode 100644 index 000000000000..2931447f3813 --- /dev/null +++ b/include/asm-ia64/machvec_uv.h @@ -0,0 +1,26 @@ +/* + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + * + * SGI UV Core Functions + * + * Copyright (C) 2008 Silicon Graphics, Inc. All rights reserved. + */ + +#ifndef _ASM_IA64_MACHVEC_UV_H +#define _ASM_IA64_MACHVEC_UV_H + +extern ia64_mv_setup_t uv_setup; + +/* + * This stuff has dual use! + * + * For a generic kernel, the macros are used to initialize the + * platform's machvec structure. When compiling a non-generic kernel, + * the macros are used directly. + */ +#define platform_name "uv" +#define platform_setup uv_setup + +#endif /* _ASM_IA64_MACHVEC_UV_H */ From 8a3360f06cd4d3b1d57f73e029f2f8b6057fdcba Mon Sep 17 00:00:00 2001 From: Bernhard Walle Date: Wed, 14 May 2008 15:40:40 -0700 Subject: [PATCH 03/10] [IA64] Don't reserve crashkernel memory > 4 GB Some IA64 machines map all cell-local memory above 4 GB (32 bit limit). However, in most cases, the kernel needs some memory below that limit that is DMA-capable. So in this machine configuration, the crashkernel will be reserved above 4 GB. For machines that use SWIOTLB implementation because they lack an I/O MMU the low memory is required by the SWIOTLB implementation. In that case, it doesn't make sense to reserve the crashkernel at all because it's unusable for kdump. A special case is the "hpzx1" machine vector. In theory, it has a I/O MMU, so it can be booted above 4 GB. However, in the kdump case that is not possible because of changeset 51b58e3e26ebfb8cd56825c4b396ed251f51dec9: On HP zx1 machines, the 'machvec=dig' parameter is needed for the kdump kernel to avoid problems with the HP sba iommu. The problem is that during the boot of the kdump kernel, the iommu is re-initialized, so in-flight DMA from improperly shutdown drivers causes an IOTLB miss which leads to an MCA. With kdump, the idea is to get into the kdump kernel with as little code as we can, so shutting down drivers properly is not an option. The workaround is to add 'machvec=dig' to the kdump kernel boot parameters. This makes the kdump kernel avoid using the sba iommu altogether, leaving the IOTLB intact. Any ongoing DMA falls harmlessly outside the kdump kernel. After the kdump kernel reboots, all devices will have been shutdown properly and DMA stopped. This patch pushes that functionality into the sba iommu initialization code, so that users won't have to find the obscure documentation telling them about 'machvec=dig'. This means that also for hpzx1 it's not possible to boot when all memory is above the 4 GB limit. So the only machine vectors that can handle this case are "sn2" and "uv". Signed-off-by: Bernhard Walle Signed-off-by: Tony Luck --- arch/ia64/kernel/setup.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) diff --git a/arch/ia64/kernel/setup.c b/arch/ia64/kernel/setup.c index 5015ca1275ca..e9596cd0cdab 100644 --- a/arch/ia64/kernel/setup.c +++ b/arch/ia64/kernel/setup.c @@ -239,6 +239,25 @@ __initcall(register_memory); #ifdef CONFIG_KEXEC + +/* + * This function checks if the reserved crashkernel is allowed on the specific + * IA64 machine flavour. Machines without an IO TLB use swiotlb and require + * some memory below 4 GB (i.e. in 32 bit area), see the implementation of + * lib/swiotlb.c. The hpzx1 architecture has an IO TLB but cannot use that + * in kdump case. See the comment in sba_init() in sba_iommu.c. + * + * So, the only machvec that really supports loading the kdump kernel + * over 4 GB is "sn2". + */ +static int __init check_crashkernel_memory(unsigned long pbase, size_t size) +{ + if (ia64_platform_is("sn2") || ia64_platform_is("uv")) + return 1; + else + return pbase < (1UL << 32); +} + static void __init setup_crashkernel(unsigned long total, int *n) { unsigned long long base = 0, size = 0; @@ -252,6 +271,16 @@ static void __init setup_crashkernel(unsigned long total, int *n) base = kdump_find_rsvd_region(size, rsvd_region, *n); } + + if (!check_crashkernel_memory(base, size)) { + pr_warning("crashkernel: There would be kdump memory " + "at %ld GB but this is unusable because it " + "must\nbe below 4 GB. Change the memory " + "configuration of the machine.\n", + (unsigned long)(base >> 30)); + return; + } + if (base != ~0UL) { printk(KERN_INFO "Reserving %ldMB of memory at %ldMB " "for crashkernel (System RAM: %ldMB)\n", From f13ae30e1397e3bfb38feb3b6e889af5d021f13d Mon Sep 17 00:00:00 2001 From: Alex Chiang Date: Thu, 8 May 2008 14:03:23 -0600 Subject: [PATCH 04/10] [IA64] allow user to force_pal_cache_flush The sequence executed in check_sal_cache_flush: - pend a timer interrupt - call SAL_CACHE_FLUSH - see if interrupt is still pending can hang HP machines with buggy SAL_CACHE_FLUSH implementations. Provide a kernel command-line argument to allow users skip this check if desired. Using this parameter will force ia64_sal_cache_flush to call ia64_pal_cache_flush() instead of SAL_CACHE_FLUSH. Signed-off-by: Alex Chiang Signed-off-by: Tony Luck --- Documentation/kernel-parameters.txt | 6 ++++++ arch/ia64/kernel/sal.c | 11 +++++++++++ 2 files changed, 17 insertions(+) diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index cdd5b934f43e..654d630f6751 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -689,6 +689,12 @@ and is between 256 and 4096 characters. It is defined in the file floppy= [HW] See Documentation/floppy.txt. + force_pal_cache_flush + [IA-64] Avoid check_sal_cache_flush which may hang on + buggy SAL_CACHE_FLUSH implementations. Using this + parameter will force ia64_sal_cache_flush to call + ia64_pal_cache_flush instead of SAL_CACHE_FLUSH. + gamecon.map[2|3]= [HW,JOY] Multisystem joystick and NES/SNES/PSX pad support via parallel port (up to 5 devices per port) diff --git a/arch/ia64/kernel/sal.c b/arch/ia64/kernel/sal.c index a3022dc48ef8..7e0259709c04 100644 --- a/arch/ia64/kernel/sal.c +++ b/arch/ia64/kernel/sal.c @@ -229,6 +229,14 @@ static void __init sal_desc_ap_wakeup(void *p) { } */ static int sal_cache_flush_drops_interrupts; +static int __init +force_pal_cache_flush(char *str) +{ + sal_cache_flush_drops_interrupts = 1; + return 0; +} +early_param("force_pal_cache_flush", force_pal_cache_flush); + void __init check_sal_cache_flush (void) { @@ -237,6 +245,9 @@ check_sal_cache_flush (void) u64 vector, cache_type = 3; struct ia64_sal_retval isrv; + if (sal_cache_flush_drops_interrupts) + return; + cpu = get_cpu(); local_irq_save(flags); From 3633c7308005e8c1dab594f69ef904424f8b639a Mon Sep 17 00:00:00 2001 From: Hidetoshi Seto Date: Fri, 9 May 2008 15:26:35 +0900 Subject: [PATCH 05/10] [IA64] fix interrupt masking for pending works on kernel leave [Bug-fix for "[BUG?][2.6.25-mm1] sleeping during IRQ disabled"] This patch does: - enable interrupts before calling schedule() as same as others, ex. x86 - enable interrupts during ia64_do_signal() and ia64_sync_krbs() - do_notify_resume_user() is still called with interrupts disabled, since we can take short path of fsys_mode if-statement quickly. - pfm_handle_work() is also called with interrupts disabled, since it can deal interrupt mask within itself. - fix/add some comments/notes Reported-by: KOSAKI Motohiro Signed-off-by: Hidetoshi Seto Signed-off-by: Tony Luck --- arch/ia64/kernel/entry.S | 14 ++++++++++---- arch/ia64/kernel/process.c | 25 +++++++++++++++++++++---- 2 files changed, 31 insertions(+), 8 deletions(-) diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S index e49ad8c5dc69..ef6b0313c857 100644 --- a/arch/ia64/kernel/entry.S +++ b/arch/ia64/kernel/entry.S @@ -1156,6 +1156,9 @@ skip_rbs_switch: * r31 = current->thread_info->flags * On exit: * p6 = TRUE if work-pending-check needs to be redone + * + * Interrupts are disabled on entry, reenabled depend on work, and + * disabled on exit. */ .work_pending_syscall: add r2=-8,r2 @@ -1170,8 +1173,8 @@ skip_rbs_switch: (pKStk) dep r21=-1,r0,PREEMPT_ACTIVE_BIT,1 ;; (pKStk) st4 [r20]=r21 - ssm psr.i // enable interrupts #endif + ssm psr.i // enable interrupts br.call.spnt.many rp=schedule .ret9: cmp.eq p6,p0=r0,r0 // p6 <- 1 rsm psr.i // disable interrupts @@ -1234,9 +1237,12 @@ GLOBAL_ENTRY(ia64_invoke_schedule_tail) END(ia64_invoke_schedule_tail) /* - * Setup stack and call do_notify_resume_user(). Note that pSys and pNonSys need to - * be set up by the caller. We declare 8 input registers so the system call - * args get preserved, in case we need to restart a system call. + * Setup stack and call do_notify_resume_user(), keeping interrupts + * disabled. + * + * Note that pSys and pNonSys need to be set up by the caller. + * We declare 8 input registers so the system call args get preserved, + * in case we need to restart a system call. */ ENTRY(notify_resume_user) .prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8) diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c index 58dcfac5ea88..a3a34b4eb038 100644 --- a/arch/ia64/kernel/process.c +++ b/arch/ia64/kernel/process.c @@ -167,11 +167,18 @@ void tsk_clear_notify_resume(struct task_struct *tsk) clear_ti_thread_flag(task_thread_info(tsk), TIF_NOTIFY_RESUME); } +/* + * do_notify_resume_user(): + * Called from notify_resume_user at entry.S, with interrupts disabled. + */ void -do_notify_resume_user (sigset_t *unused, struct sigscratch *scr, long in_syscall) +do_notify_resume_user(sigset_t *unused, struct sigscratch *scr, long in_syscall) { if (fsys_mode(current, &scr->pt)) { - /* defer signal-handling etc. until we return to privilege-level 0. */ + /* + * defer signal-handling etc. until we return to + * privilege-level 0. + */ if (!ia64_psr(&scr->pt)->lp) ia64_psr(&scr->pt)->lp = 1; return; @@ -179,16 +186,26 @@ do_notify_resume_user (sigset_t *unused, struct sigscratch *scr, long in_syscall #ifdef CONFIG_PERFMON if (current->thread.pfm_needs_checking) + /* + * Note: pfm_handle_work() allow us to call it with interrupts + * disabled, and may enable interrupts within the function. + */ pfm_handle_work(); #endif /* deal with pending signal delivery */ - if (test_thread_flag(TIF_SIGPENDING)) + if (test_thread_flag(TIF_SIGPENDING)) { + local_irq_enable(); /* force interrupt enable */ ia64_do_signal(scr, in_syscall); + } /* copy user rbs to kernel rbs */ - if (unlikely(test_thread_flag(TIF_RESTORE_RSE))) + if (unlikely(test_thread_flag(TIF_RESTORE_RSE))) { + local_irq_enable(); /* force interrupt enable */ ia64_sync_krbs(); + } + + local_irq_disable(); /* force interrupt disable */ } static int pal_halt = 1; From 2e513fe4903c62450a9f8c3759f75bc4cd7e4dfd Mon Sep 17 00:00:00 2001 From: Hidetoshi Seto Date: Fri, 9 May 2008 15:26:51 +0900 Subject: [PATCH 06/10] [IA64] trivial cleanup for entry.S This patch does: - make comment at next to resched check more robust - move "re-check" comments to next to where change predicate regs Signed-off-by: Hidetoshi Seto Signed-off-by: Tony Luck --- arch/ia64/kernel/entry.S | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S index ef6b0313c857..ca2bb95726de 100644 --- a/arch/ia64/kernel/entry.S +++ b/arch/ia64/kernel/entry.S @@ -1167,7 +1167,7 @@ skip_rbs_switch: st8 [r2]=r8 st8 [r3]=r10 .work_pending: - tbit.z p6,p0=r31,TIF_NEED_RESCHED // current_thread_info()->need_resched==0? + tbit.z p6,p0=r31,TIF_NEED_RESCHED // is resched not needed? (p6) br.cond.sptk.few .notify #ifdef CONFIG_PREEMPT (pKStk) dep r21=-1,r0,PREEMPT_ACTIVE_BIT,1 @@ -1176,7 +1176,7 @@ skip_rbs_switch: #endif ssm psr.i // enable interrupts br.call.spnt.many rp=schedule -.ret9: cmp.eq p6,p0=r0,r0 // p6 <- 1 +.ret9: cmp.eq p6,p0=r0,r0 // p6 <- 1 (re-check) rsm psr.i // disable interrupts ;; #ifdef CONFIG_PREEMPT @@ -1185,13 +1185,13 @@ skip_rbs_switch: (pKStk) st4 [r20]=r0 // preempt_count() <- 0 #endif (pLvSys)br.cond.sptk.few .work_pending_syscall_end - br.cond.sptk.many .work_processed_kernel // re-check + br.cond.sptk.many .work_processed_kernel .notify: (pUStk) br.call.spnt.many rp=notify_resume_user -.ret10: cmp.ne p6,p0=r0,r0 // p6 <- 0 +.ret10: cmp.ne p6,p0=r0,r0 // p6 <- 0 (don't re-check) (pLvSys)br.cond.sptk.few .work_pending_syscall_end - br.cond.sptk.many .work_processed_kernel // don't re-check + br.cond.sptk.many .work_processed_kernel .work_pending_syscall_end: adds r2=PT(R8)+16,r12 @@ -1199,7 +1199,7 @@ skip_rbs_switch: ;; ld8 r8=[r2] ld8 r10=[r3] - br.cond.sptk.many .work_processed_syscall // re-check + br.cond.sptk.many .work_processed_syscall END(ia64_leave_kernel) From 0fb232fdb2a2674003ef4b874034e872b7256aa9 Mon Sep 17 00:00:00 2001 From: Hidetoshi Seto Date: Fri, 9 May 2008 15:27:09 +0900 Subject: [PATCH 07/10] [IA64] trivial cleanup for perfmon.c Fix a typo, and coding style cleanups for pfm_handle_work(). Signed-off-by: Hidetoshi Seto Signed-off-by: Tony Luck --- arch/ia64/kernel/perfmon.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c index c1ad27de2dd2..71d05133f556 100644 --- a/arch/ia64/kernel/perfmon.c +++ b/arch/ia64/kernel/perfmon.c @@ -5013,12 +5013,13 @@ pfm_context_force_terminate(pfm_context_t *ctx, struct pt_regs *regs) } static int pfm_ovfl_notify_user(pfm_context_t *ctx, unsigned long ovfl_pmds); + /* * pfm_handle_work() can be called with interrupts enabled * (TIF_NEED_RESCHED) or disabled. The down_interruptible * call may sleep, therefore we must re-enable interrupts * to avoid deadlocks. It is safe to do so because this function - * is called ONLY when returning to user level (PUStk=1), in which case + * is called ONLY when returning to user level (pUStk=1), in which case * there is no risk of kernel stack overflow due to deep * interrupt nesting. */ @@ -5034,7 +5035,8 @@ pfm_handle_work(void) ctx = PFM_GET_CTX(current); if (ctx == NULL) { - printk(KERN_ERR "perfmon: [%d] has no PFM context\n", task_pid_nr(current)); + printk(KERN_ERR "perfmon: [%d] has no PFM context\n", + task_pid_nr(current)); return; } @@ -5058,11 +5060,12 @@ pfm_handle_work(void) /* * must be done before we check for simple-reset mode */ - if (ctx->ctx_fl_going_zombie || ctx->ctx_state == PFM_CTX_ZOMBIE) goto do_zombie; - + if (ctx->ctx_fl_going_zombie || ctx->ctx_state == PFM_CTX_ZOMBIE) + goto do_zombie; //if (CTX_OVFL_NOBLOCK(ctx)) goto skip_blocking; - if (reason == PFM_TRAP_REASON_RESET) goto skip_blocking; + if (reason == PFM_TRAP_REASON_RESET) + goto skip_blocking; /* * restore interrupt mask to what it was on entry. @@ -5110,7 +5113,8 @@ do_zombie: /* * in case of interruption of down() we don't restart anything */ - if (ret < 0) goto nothing_to_do; + if (ret < 0) + goto nothing_to_do; skip_blocking: pfm_resume_after_ovfl(ctx, ovfl_regs, regs); From 7af1d7532b6cf905230c72c67ad85a480b122374 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Simon=20Holm=20Th=C3=B8gersen?= Date: Mon, 12 May 2008 17:40:50 +0200 Subject: [PATCH 08/10] [IA64] Remove NULL pointer check for argument never passed as NULL. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit There is only palinfo_handle_smp as (indirect) user of palinfo_smp_call (by way of smp_call_function_single) and surely palinfo_handle_smp never pass NULL as parameter for info. Signed-off-by: Simon Holm Thøgersen Signed-off-by: Tony Luck --- arch/ia64/kernel/palinfo.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/arch/ia64/kernel/palinfo.c b/arch/ia64/kernel/palinfo.c index 4547a2092af9..9dc00f7fe10e 100644 --- a/arch/ia64/kernel/palinfo.c +++ b/arch/ia64/kernel/palinfo.c @@ -900,12 +900,6 @@ static void palinfo_smp_call(void *info) { palinfo_smp_data_t *data = (palinfo_smp_data_t *)info; - if (data == NULL) { - printk(KERN_ERR "palinfo: data pointer is NULL\n"); - data->ret = 0; /* no output */ - return; - } - /* does this actual call */ data->ret = (*data->func)(data->page); } From 3fb2c74ee20b77affd494c6b8ce7928d0ebbb62e Mon Sep 17 00:00:00 2001 From: Prarit Bhargava Date: Wed, 14 May 2008 12:00:24 -0400 Subject: [PATCH 09/10] [IA64] Properly unregister legacy interrupts acpi_unregister_gsi() should "undo" what acpi_register_gsi() does. On systems that have legacy interrupts, acpi_unregister_gsi erroneously calls iosapci_unregister_intr() which is wrong to do and causes a loud warning. acpi_unregister_gsi() should just return in these cases. Signed-off-by: Prarit Bhargava Signed-off-by: Tony Luck --- arch/ia64/kernel/acpi.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c index 6ff6815d34ef..853d1f11be00 100644 --- a/arch/ia64/kernel/acpi.c +++ b/arch/ia64/kernel/acpi.c @@ -627,6 +627,9 @@ void acpi_unregister_gsi(u32 gsi) if (acpi_irq_model == ACPI_IRQ_MODEL_PLATFORM) return; + if (has_8259 && gsi < 16) + return; + iosapic_unregister_intr(gsi); } From 839052d27e8db0c1545256fe5827abcd00fb51c5 Mon Sep 17 00:00:00 2001 From: "Huang, Xiaolan" Date: Thu, 15 May 2008 10:18:41 +0800 Subject: [PATCH 10/10] [IA64] fix personality(PER_LINUX32) performance issue The patch aims to fix a performance issue for the syscall personality(PER_LINUX32). On IA-64 box, the syscall personality (PER_LINUX32) has poor performance because it failed to find the Linux/x86 execution domain. Then it tried to load the kernel module however it failed always and it used the default execution domain PER_LINUX instead. Requesting kernel modules is very expensive. It caused the performance issue. (see the function lookup_exec_domain in kernel/exec_domain.c). To resolve the issue, execution domain Linux/x86 is always registered in initialization time for IA-64 architecture. Signed-off-by: Xiaolan Huang Signed-off-by: Tony Luck --- arch/ia64/ia32/ia32_support.c | 10 ---------- arch/ia64/mm/init.c | 25 +++++++++++++++++++++++++ 2 files changed, 25 insertions(+), 10 deletions(-) diff --git a/arch/ia64/ia32/ia32_support.c b/arch/ia64/ia32/ia32_support.c index 896b1ebbfb26..a6965ddafc46 100644 --- a/arch/ia64/ia32/ia32_support.c +++ b/arch/ia64/ia32/ia32_support.c @@ -15,7 +15,6 @@ #include #include #include -#include #include #include @@ -29,7 +28,6 @@ extern int die_if_kernel (char *str, struct pt_regs *regs, long err); -struct exec_domain ia32_exec_domain; struct page *ia32_shared_page[NR_CPUS]; unsigned long *ia32_boot_gdt; unsigned long *cpu_gdt_table[NR_CPUS]; @@ -240,14 +238,6 @@ ia32_cpu_init (void) static int __init ia32_init (void) { - ia32_exec_domain.name = "Linux/x86"; - ia32_exec_domain.handler = NULL; - ia32_exec_domain.pers_low = PER_LINUX32; - ia32_exec_domain.pers_high = PER_LINUX32; - ia32_exec_domain.signal_map = default_exec_domain.signal_map; - ia32_exec_domain.signal_invmap = default_exec_domain.signal_invmap; - register_exec_domain(&ia32_exec_domain); - #if PAGE_SHIFT > IA32_PAGE_SHIFT { extern struct kmem_cache *ia64_partial_page_cachep; diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c index fc6c6636ffda..200100ea7610 100644 --- a/arch/ia64/mm/init.c +++ b/arch/ia64/mm/init.c @@ -719,3 +719,28 @@ out: EXPORT_SYMBOL_GPL(remove_memory); #endif /* CONFIG_MEMORY_HOTREMOVE */ #endif + +/* + * Even when CONFIG_IA32_SUPPORT is not enabled it is + * useful to have the Linux/x86 domain registered to + * avoid an attempted module load when emulators call + * personality(PER_LINUX32). This saves several milliseconds + * on each such call. + */ +static struct exec_domain ia32_exec_domain; + +static int __init +per_linux32_init(void) +{ + ia32_exec_domain.name = "Linux/x86"; + ia32_exec_domain.handler = NULL; + ia32_exec_domain.pers_low = PER_LINUX32; + ia32_exec_domain.pers_high = PER_LINUX32; + ia32_exec_domain.signal_map = default_exec_domain.signal_map; + ia32_exec_domain.signal_invmap = default_exec_domain.signal_invmap; + register_exec_domain(&ia32_exec_domain); + + return 0; +} + +__initcall(per_linux32_init);