uv_hub.h 12.3 KB
Newer Older
1 2 3 4 5 6 7
/*
 * 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
 *
8
 * Copyright (C) 2007-2008 Silicon Graphics, Inc. All rights reserved.
9 10
 */

11 12
#ifndef _ASM_X86_UV_UV_HUB_H
#define _ASM_X86_UV_UV_HUB_H
13

14
#ifdef CONFIG_X86_64
15 16
#include <linux/numa.h>
#include <linux/percpu.h>
17
#include <linux/timer.h>
18 19
#include <asm/types.h>
#include <asm/percpu.h>
20
#include <asm/uv/uv_mmrs.h>
21 22 23 24 25


/*
 * Addressing Terminology
 *
26 27 28 29
 *	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)..
30
 *
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
 * 	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)
54 55 56
 *
 *
 *  Memory/UV-HUB Processor Socket Address Format:
57 58 59 60
 *  +----------------+---------------+---------------------+
 *  |00..000000000000|   PNODE       |      NodeOffset     |
 *  +----------------+---------------+---------------------+
 *                   <--- N bits --->|<--------M bits ----->
61
 *
62 63
 *	M - number of node offset bits (35 .. 40)
 *	N - number of PNODE bits (0 .. 10)
64 65 66
 *
 *		Note: M + N cannot currently exceed 44 (x86_64) or 46 (IA64).
 *		The actual values are configuration dependent and are set at
67 68
 *		boot time. M & N values are set by the hardware/BIOS at boot.
 *
69 70 71 72 73 74 75 76
 *
 * APICID format
 * 	NOTE!!!!!! This is the current format of the APICID. However, code
 * 	should assume that this will change in the future. Use functions
 * 	in this file for all APICID bit manipulations and conversion.
 *
 * 		1111110000000000
 * 		5432109876543210
77
 *		pppppppppplc0cch
78 79
 *		sssssssssss
 *
80
 *			p  = pnode bits
81 82 83
 *			l =  socket number on board
 *			c  = core
 *			h  = hyperthread
84
 *			s  = bits that are in the SOCKET_ID CSR
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102
 *
 *	Note: Processor only supports 12 bits in the APICID register. The ACPI
 *	      tables hold all 16 bits. Software needs to be aware of this.
 *
 * 	      Unless otherwise specified, all references to APICID refer to
 * 	      the FULL value contained in ACPI tables, not the subset in the
 * 	      processor APICID register.
 */


/*
 * 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.
 *
103
 * NOTE: a brick may contain 1 or 2 OS nodes. Don't get these confused.
104 105 106 107 108 109 110 111 112 113 114 115 116 117
 */
#define UV_MAX_NUMALINK_BLADES	16384

/*
 * Maximum number of C/Mbricks within a software SSI (hardware may support
 * more).
 */
#define UV_MAX_SSI_BLADES	256

/*
 * The largest possible NASID of a C or M brick (+ 2)
 */
#define UV_MAX_NASID_VALUE	(UV_MAX_NUMALINK_NODES * 2)

118 119 120 121 122 123 124 125 126 127
struct uv_scir_s {
	struct timer_list timer;
	unsigned long	offset;
	unsigned long	last;
	unsigned long	idle_on;
	unsigned long	idle_off;
	unsigned char	state;
	unsigned char	enabled;
};

128 129 130 131 132 133
/*
 * 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 {
134 135
	unsigned long		global_mmr_base;
	unsigned long		gpa_mask;
136
	unsigned int		gnode_extra;
137 138 139 140 141 142 143 144 145 146 147
	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;
	struct uv_scir_s	scir;
148
};
149

150 151 152 153 154 155 156 157
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.
158 159 160
 * 		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)
161
 */
162
#define UV_NASID_TO_PNODE(n)		(((n) >> 1) & uv_hub_info->pnode_mask)
163 164
#define UV_PNODE_TO_GNODE(p)		((p) |uv_hub_info->gnode_extra)
#define UV_PNODE_TO_NASID(p)		(UV_PNODE_TO_GNODE(p) << 1)
165 166 167 168

#define UV_LOCAL_MMR_BASE		0xf4000000UL
#define UV_GLOBAL_MMR32_BASE		0xf8000000UL
#define UV_GLOBAL_MMR64_BASE		(uv_hub_info->global_mmr_base)
169 170
#define UV_LOCAL_MMR_SIZE		(64UL * 1024 * 1024)
#define UV_GLOBAL_MMR32_SIZE		(64UL * 1024 * 1024)
171

172 173
#define UV_GLOBAL_MMR32_PNODE_SHIFT	15
#define UV_GLOBAL_MMR64_PNODE_SHIFT	26
174

175
#define UV_GLOBAL_MMR32_PNODE_BITS(p)	((p) << (UV_GLOBAL_MMR32_PNODE_SHIFT))
176

177
#define UV_GLOBAL_MMR64_PNODE_BITS(p)					\
178
	((unsigned long)(UV_PNODE_TO_GNODE(p)) << UV_GLOBAL_MMR64_PNODE_SHIFT)
179 180 181

#define UV_APIC_PNODE_SHIFT	6

182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205
/* Local Bus from cpu's perspective */
#define LOCAL_BUS_BASE		0x1c00000
#define LOCAL_BUS_SIZE		(4 * 1024 * 1024)

/*
 * System Controller Interface Reg
 *
 * Note there are NO leds on a UV system.  This register is only
 * used by the system controller to monitor system-wide operation.
 * There are 64 regs per node.  With Nahelem cpus (2 cores per node,
 * 8 cpus per core, 2 threads per cpu) there are 32 cpu threads on
 * a node.
 *
 * The window is located at top of ACPI MMR space
 */
#define SCIR_WINDOW_COUNT	64
#define SCIR_LOCAL_MMR_BASE	(LOCAL_BUS_BASE + \
				 LOCAL_BUS_SIZE - \
				 SCIR_WINDOW_COUNT)

#define SCIR_CPU_HEARTBEAT	0x01	/* timer interrupt */
#define SCIR_CPU_ACTIVITY	0x02	/* not idle */
#define SCIR_CPU_HB_INTERVAL	(HZ)	/* once per second */

206 207 208 209
/* Loop through all installed blades */
#define for_each_possible_blade(bid)		\
	for ((bid) = 0; (bid) < uv_num_possible_blades(); (bid)++)

210 211 212 213 214 215 216 217 218 219 220
/*
 * 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)
221
		paddr |= uv_hub_info->lowmem_remap_base;
222 223 224 225 226 227 228
	return paddr | uv_hub_info->gnode_upper;
}


/* socket virtual --> UV global physical address */
static inline unsigned long uv_gpa(void *v)
{
229
	return uv_soc_phys_ram_to_gpa(__pa(v));
230 231 232 233 234 235 236
}

/* 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);
}
237 238 239


/*
240
 * Extract a PNODE from an APICID (full apicid, not processor subset)
241
 */
242
static inline int uv_apicid_to_pnode(int apicid)
243
{
244
	return (apicid >> UV_APIC_PNODE_SHIFT);
245 246 247 248 249 250
}

/*
 * Access global MMRs using the low memory MMR32 space. This region supports
 * faster MMR access but not all MMRs are accessible in this space.
 */
251
static inline unsigned long *uv_global_mmr32_address(int pnode,
252 253 254
				unsigned long offset)
{
	return __va(UV_GLOBAL_MMR32_BASE |
255
		       UV_GLOBAL_MMR32_PNODE_BITS(pnode) | offset);
256 257
}

258
static inline void uv_write_global_mmr32(int pnode, unsigned long offset,
259 260
				 unsigned long val)
{
261
	*uv_global_mmr32_address(pnode, offset) = val;
262 263
}

264
static inline unsigned long uv_read_global_mmr32(int pnode,
265 266
						 unsigned long offset)
{
267
	return *uv_global_mmr32_address(pnode, offset);
268 269 270 271 272 273
}

/*
 * Access Global MMR space using the MMR space located at the top of physical
 * memory.
 */
274
static inline unsigned long *uv_global_mmr64_address(int pnode,
275 276 277
				unsigned long offset)
{
	return __va(UV_GLOBAL_MMR64_BASE |
278
		    UV_GLOBAL_MMR64_PNODE_BITS(pnode) | offset);
279 280
}

281
static inline void uv_write_global_mmr64(int pnode, unsigned long offset,
282 283
				unsigned long val)
{
284
	*uv_global_mmr64_address(pnode, offset) = val;
285 286
}

287
static inline unsigned long uv_read_global_mmr64(int pnode,
288 289
						 unsigned long offset)
{
290
	return *uv_global_mmr64_address(pnode, offset);
291 292 293
}

/*
294
 * Access hub local MMRs. Faster than using global space but only local MMRs
295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
 * 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;
}

312 313 314 315 316 317 318 319 320 321
static inline unsigned char uv_read_local_mmr8(unsigned long offset)
{
	return *((unsigned char *)uv_local_mmr_address(offset));
}

static inline void uv_write_local_mmr8(unsigned long offset, unsigned char val)
{
	*((unsigned char *)uv_local_mmr_address(offset)) = val;
}

322
/*
323
 * Structures and definitions for converting between cpu, node, pnode, and blade
324 325 326
 * numbers.
 */
struct uv_blade_info {
327
	unsigned short	nr_possible_cpus;
328
	unsigned short	nr_online_cpus;
329
	unsigned short	pnode;
330
	short		memory_nid;
331
};
332
extern struct uv_blade_info *uv_blade_info;
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
extern short *uv_node_to_blade;
extern short *uv_cpu_to_blade;
extern short uv_possible_blades;

/* Blade-local cpu number of current cpu. Numbered 0 .. <# cpus on the blade> */
static inline int uv_blade_processor_id(void)
{
	return uv_hub_info->blade_processor_id;
}

/* Blade number of current cpu. Numnbered 0 .. <#blades -1> */
static inline int uv_numa_blade_id(void)
{
	return uv_hub_info->numa_blade_id;
}

/* Convert a cpu number to the the UV blade number */
static inline int uv_cpu_to_blade_id(int cpu)
{
	return uv_cpu_to_blade[cpu];
}

/* Convert linux node number to the UV blade number */
static inline int uv_node_to_blade_id(int nid)
{
	return uv_node_to_blade[nid];
}

361 362
/* Convert a blade id to the PNODE of the blade */
static inline int uv_blade_to_pnode(int bid)
363
{
364
	return uv_blade_info[bid].pnode;
365 366
}

367 368 369 370 371 372
/* Nid of memory node on blade. -1 if no blade-local memory */
static inline int uv_blade_to_memory_nid(int bid)
{
	return uv_blade_info[bid].memory_nid;
}

373 374 375
/* Determine the number of possible cpus on a blade */
static inline int uv_blade_nr_possible_cpus(int bid)
{
376
	return uv_blade_info[bid].nr_possible_cpus;
377 378 379 380 381 382 383 384
}

/* Determine the number of online cpus on a blade */
static inline int uv_blade_nr_online_cpus(int bid)
{
	return uv_blade_info[bid].nr_online_cpus;
}

385 386
/* Convert a cpu id to the PNODE of the blade containing the cpu */
static inline int uv_cpu_to_pnode(int cpu)
387
{
388
	return uv_blade_info[uv_cpu_to_blade_id(cpu)].pnode;
389 390
}

391 392
/* Convert a linux node number to the PNODE of the blade */
static inline int uv_node_to_pnode(int nid)
393
{
394
	return uv_blade_info[uv_node_to_blade_id(nid)].pnode;
395 396 397 398 399 400 401 402
}

/* Maximum possible number of blades */
static inline int uv_num_possible_blades(void)
{
	return uv_possible_blades;
}

403 404 405 406 407 408 409 410
/* Update SCIR state */
static inline void uv_set_scir_bits(unsigned char value)
{
	if (uv_hub_info->scir.state != value) {
		uv_hub_info->scir.state = value;
		uv_write_local_mmr8(uv_hub_info->scir.offset, value);
	}
}
411

412 413 414 415 416 417 418
static inline void uv_set_cpu_scir_bits(int cpu, unsigned char value)
{
	if (uv_cpu_hub_info(cpu)->scir.state != value) {
		uv_cpu_hub_info(cpu)->scir.state = value;
		uv_write_local_mmr8(uv_cpu_hub_info(cpu)->scir.offset, value);
	}
}
419

420 421 422 423 424 425 426 427 428 429
static inline void uv_hub_send_ipi(int pnode, int apicid, int vector)
{
	unsigned long val;

	val = (1UL << UVH_IPI_INT_SEND_SHFT) |
			((apicid & 0x3f) << UVH_IPI_INT_APIC_ID_SHFT) |
			(vector << UVH_IPI_INT_VECTOR_SHFT);
	uv_write_global_mmr64(pnode, UVH_IPI_INT, val);
}

430
#endif /* CONFIG_X86_64 */
431
#endif /* _ASM_X86_UV_UV_HUB_H */