slb.c 10.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4
/*
 * PowerPC64 SLB support.
 *
 * Copyright (C) 2004 David Gibson <dwg@au.ibm.com>, IBM
5
 * Based on earlier code written by:
L
Linus Torvalds 已提交
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
 * Dave Engebretsen and Mike Corrigan {engebret|mikejc}@us.ibm.com
 *    Copyright (c) 2001 Dave Engebretsen
 * Copyright (C) 2002 Anton Blanchard <anton@au.ibm.com>, IBM
 *
 *
 *      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; either version
 *      2 of the License, or (at your option) any later version.
 */

#include <asm/pgtable.h>
#include <asm/mmu.h>
#include <asm/mmu_context.h>
#include <asm/paca.h>
#include <asm/cputable.h>
22
#include <asm/cacheflush.h>
23 24
#include <asm/smp.h>
#include <linux/compiler.h>
25
#include <asm/udbg.h>
26
#include <asm/code-patching.h>
27

28 29 30 31 32
enum slb_index {
	LINEAR_INDEX	= 0, /* Kernel linear map  (0xc000000000000000) */
	VMALLOC_INDEX	= 1, /* Kernel virtual map (0xd000000000000000) */
	KSTACK_INDEX	= 2, /* Kernel stack map */
};
L
Linus Torvalds 已提交
33

34 35 36 37 38 39 40 41 42
extern void slb_allocate_realmode(unsigned long ea);

static void slb_allocate(unsigned long ea)
{
	/* Currently, we do real mode for all SLBs including user, but
	 * that will change if we bring back dynamic VSIDs
	 */
	slb_allocate_realmode(ea);
}
L
Linus Torvalds 已提交
43

44 45 46
#define slb_esid_mask(ssize)	\
	(((ssize) == MMU_SEGSIZE_256M)? ESID_MASK: ESID_MASK_1T)

P
Paul Mackerras 已提交
47
static inline unsigned long mk_esid_data(unsigned long ea, int ssize,
48
					 enum slb_index index)
L
Linus Torvalds 已提交
49
{
50
	return (ea & slb_esid_mask(ssize)) | SLB_ESID_V | index;
L
Linus Torvalds 已提交
51 52
}

P
Paul Mackerras 已提交
53 54
static inline unsigned long mk_vsid_data(unsigned long ea, int ssize,
					 unsigned long flags)
L
Linus Torvalds 已提交
55
{
P
Paul Mackerras 已提交
56 57
	return (get_kernel_vsid(ea, ssize) << slb_vsid_shift(ssize)) | flags |
		((unsigned long) ssize << SLB_VSID_SSIZE_SHIFT);
L
Linus Torvalds 已提交
58 59
}

P
Paul Mackerras 已提交
60
static inline void slb_shadow_update(unsigned long ea, int ssize,
61
				     unsigned long flags,
62
				     enum slb_index index)
L
Linus Torvalds 已提交
63
{
64 65
	struct slb_shadow *p = get_slb_shadow();

66 67
	/*
	 * Clear the ESID first so the entry is not valid while we are
68 69
	 * updating it.  No write barriers are needed here, provided
	 * we only update the current CPU's SLB shadow buffer.
70
	 */
71 72 73
	p->save_area[index].esid = 0;
	p->save_area[index].vsid = cpu_to_be64(mk_vsid_data(ea, ssize, flags));
	p->save_area[index].esid = cpu_to_be64(mk_esid_data(ea, ssize, index));
74 75
}

76
static inline void slb_shadow_clear(enum slb_index index)
77
{
78
	get_slb_shadow()->save_area[index].esid = 0;
L
Linus Torvalds 已提交
79 80
}

P
Paul Mackerras 已提交
81 82
static inline void create_shadowed_slbe(unsigned long ea, int ssize,
					unsigned long flags,
83
					enum slb_index index)
84 85 86 87 88 89
{
	/*
	 * Updating the shadow buffer before writing the SLB ensures
	 * we don't get a stale entry here if we get preempted by PHYP
	 * between these two statements.
	 */
90
	slb_shadow_update(ea, ssize, flags, index);
91 92

	asm volatile("slbmte  %0,%1" :
P
Paul Mackerras 已提交
93
		     : "r" (mk_vsid_data(ea, ssize, flags)),
94
		       "r" (mk_esid_data(ea, ssize, index))
95 96 97
		     : "memory" );
}

98
static void __slb_flush_and_rebolt(void)
L
Linus Torvalds 已提交
99 100
{
	/* If you change this make sure you change SLB_NUM_BOLTED
101
	 * and PR KVM appropriately too. */
102
	unsigned long linear_llp, vmalloc_llp, lflags, vflags;
P
Paul Mackerras 已提交
103
	unsigned long ksp_esid_data, ksp_vsid_data;
L
Linus Torvalds 已提交
104

105
	linear_llp = mmu_psize_defs[mmu_linear_psize].sllp;
106
	vmalloc_llp = mmu_psize_defs[mmu_vmalloc_psize].sllp;
107
	lflags = SLB_VSID_KERNEL | linear_llp;
108
	vflags = SLB_VSID_KERNEL | vmalloc_llp;
L
Linus Torvalds 已提交
109

110
	ksp_esid_data = mk_esid_data(get_paca()->kstack, mmu_kernel_ssize, KSTACK_INDEX);
P
Paul Mackerras 已提交
111
	if ((ksp_esid_data & ~0xfffffffUL) <= PAGE_OFFSET) {
L
Linus Torvalds 已提交
112
		ksp_esid_data &= ~SLB_ESID_V;
P
Paul Mackerras 已提交
113
		ksp_vsid_data = 0;
114
		slb_shadow_clear(KSTACK_INDEX);
115 116
	} else {
		/* Update stack entry; others don't change */
117
		slb_shadow_update(get_paca()->kstack, mmu_kernel_ssize, lflags, KSTACK_INDEX);
118
		ksp_vsid_data =
119
			be64_to_cpu(get_slb_shadow()->save_area[KSTACK_INDEX].vsid);
120
	}
121

L
Linus Torvalds 已提交
122 123 124 125 126 127 128 129 130
	/* We need to do this all in asm, so we're sure we don't touch
	 * the stack between the slbia and rebolting it. */
	asm volatile("isync\n"
		     "slbia\n"
		     /* Slot 1 - first VMALLOC segment */
		     "slbmte	%0,%1\n"
		     /* Slot 2 - kernel stack */
		     "slbmte	%2,%3\n"
		     "isync"
P
Paul Mackerras 已提交
131 132 133
		     :: "r"(mk_vsid_data(VMALLOC_START, mmu_kernel_ssize, vflags)),
		        "r"(mk_esid_data(VMALLOC_START, mmu_kernel_ssize, 1)),
		        "r"(ksp_vsid_data),
L
Linus Torvalds 已提交
134 135 136 137
		        "r"(ksp_esid_data)
		     : "memory");
}

138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
void slb_flush_and_rebolt(void)
{

	WARN_ON(!irqs_disabled());

	/*
	 * We can't take a PMU exception in the following code, so hard
	 * disable interrupts.
	 */
	hard_irq_disable();

	__slb_flush_and_rebolt();
	get_paca()->slb_cache_ptr = 0;
}

153 154 155 156 157
void slb_vmalloc_update(void)
{
	unsigned long vflags;

	vflags = SLB_VSID_KERNEL | mmu_psize_defs[mmu_vmalloc_psize].sllp;
158
	slb_shadow_update(VMALLOC_START, mmu_kernel_ssize, vflags, VMALLOC_INDEX);
159 160 161
	slb_flush_and_rebolt();
}

162 163 164 165 166 167 168 169 170 171 172
/* Helper function to compare esids.  There are four cases to handle.
 * 1. The system is not 1T segment size capable.  Use the GET_ESID compare.
 * 2. The system is 1T capable, both addresses are < 1T, use the GET_ESID compare.
 * 3. The system is 1T capable, only one of the two addresses is > 1T.  This is not a match.
 * 4. The system is 1T capable, both addresses are > 1T, use the GET_ESID_1T macro to compare.
 */
static inline int esids_match(unsigned long addr1, unsigned long addr2)
{
	int esid_1t_count;

	/* System is not 1T segment size capable. */
173
	if (!mmu_has_feature(MMU_FTR_1T_SEGMENT))
174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190
		return (GET_ESID(addr1) == GET_ESID(addr2));

	esid_1t_count = (((addr1 >> SID_SHIFT_1T) != 0) +
				((addr2 >> SID_SHIFT_1T) != 0));

	/* both addresses are < 1T */
	if (esid_1t_count == 0)
		return (GET_ESID(addr1) == GET_ESID(addr2));

	/* One address < 1T, the other > 1T.  Not a match */
	if (esid_1t_count == 1)
		return 0;

	/* Both addresses are > 1T. */
	return (GET_ESID_1T(addr1) == GET_ESID_1T(addr2));
}

L
Linus Torvalds 已提交
191 192 193
/* Flush all user entries from the segment table of the current processor. */
void switch_slb(struct task_struct *tsk, struct mm_struct *mm)
{
194
	unsigned long offset;
P
Paul Mackerras 已提交
195
	unsigned long slbie_data = 0;
L
Linus Torvalds 已提交
196 197
	unsigned long pc = KSTK_EIP(tsk);
	unsigned long stack = KSTK_ESP(tsk);
198
	unsigned long exec_base;
L
Linus Torvalds 已提交
199

200 201 202 203 204 205 206 207
	/*
	 * We need interrupts hard-disabled here, not just soft-disabled,
	 * so that a PMU interrupt can't occur, which might try to access
	 * user memory (to get a stack trace) and possible cause an SLB miss
	 * which would update the slb_cache/slb_cache_ptr fields in the PACA.
	 */
	hard_irq_disable();
	offset = get_paca()->slb_cache_ptr;
208
	if (!mmu_has_feature(MMU_FTR_NO_SLBIE_B) &&
209
	    offset <= SLB_CACHE_ENTRIES) {
L
Linus Torvalds 已提交
210 211 212
		int i;
		asm volatile("isync" : : : "memory");
		for (i = 0; i < offset; i++) {
P
Paul Mackerras 已提交
213 214 215 216 217 218
			slbie_data = (unsigned long)get_paca()->slb_cache[i]
				<< SID_SHIFT; /* EA */
			slbie_data |= user_segment_size(slbie_data)
				<< SLBIE_SSIZE_SHIFT;
			slbie_data |= SLBIE_C; /* C set for user addresses */
			asm volatile("slbie %0" : : "r" (slbie_data));
L
Linus Torvalds 已提交
219 220 221
		}
		asm volatile("isync" : : : "memory");
	} else {
222
		__slb_flush_and_rebolt();
L
Linus Torvalds 已提交
223 224 225 226
	}

	/* Workaround POWER5 < DD2.1 issue */
	if (offset == 1 || offset > SLB_CACHE_ENTRIES)
P
Paul Mackerras 已提交
227
		asm volatile("slbie %0" : : "r" (slbie_data));
L
Linus Torvalds 已提交
228 229

	get_paca()->slb_cache_ptr = 0;
230
	copy_mm_to_paca(&mm->context);
L
Linus Torvalds 已提交
231 232 233

	/*
	 * preload some userspace segments into the SLB.
234 235
	 * Almost all 32 and 64bit PowerPC executables are linked at
	 * 0x10000000 so it makes sense to preload this segment.
L
Linus Torvalds 已提交
236
	 */
237
	exec_base = 0x10000000;
L
Linus Torvalds 已提交
238

239
	if (is_kernel_addr(pc) || is_kernel_addr(stack) ||
240
	    is_kernel_addr(exec_base))
L
Linus Torvalds 已提交
241 242
		return;

243
	slb_allocate(pc);
L
Linus Torvalds 已提交
244

245 246
	if (!esids_match(pc, stack))
		slb_allocate(stack);
L
Linus Torvalds 已提交
247

248 249 250
	if (!esids_match(pc, exec_base) &&
	    !esids_match(stack, exec_base))
		slb_allocate(exec_base);
L
Linus Torvalds 已提交
251 252
}

253 254 255
static inline void patch_slb_encoding(unsigned int *insn_addr,
				      unsigned int immed)
{
256 257 258 259 260 261 262 263 264 265 266 267 268 269 270

	/*
	 * This function patches either an li or a cmpldi instruction with
	 * a new immediate value. This relies on the fact that both li
	 * (which is actually addi) and cmpldi both take a 16-bit immediate
	 * value, and it is situated in the same location in the instruction,
	 * ie. bits 16-31 (Big endian bit order) or the lower 16 bits.
	 * The signedness of the immediate operand differs between the two
	 * instructions however this code is only ever patching a small value,
	 * much less than 1 << 15, so we can get away with it.
	 * To patch the value we read the existing instruction, clear the
	 * immediate value, and or in our new value, then write the instruction
	 * back.
	 */
	unsigned int insn = (*insn_addr & 0xffff0000) | immed;
271
	patch_instruction(insn_addr, insn);
272 273
}

274 275 276 277 278
extern u32 slb_miss_kernel_load_linear[];
extern u32 slb_miss_kernel_load_io[];
extern u32 slb_compare_rr_to_size[];
extern u32 slb_miss_kernel_load_vmemmap[];

279 280 281 282 283 284 285 286 287
void slb_set_size(u16 size)
{
	if (mmu_slb_size == size)
		return;

	mmu_slb_size = size;
	patch_slb_encoding(slb_compare_rr_to_size, mmu_slb_size);
}

L
Linus Torvalds 已提交
288 289
void slb_initialize(void)
{
290
	unsigned long linear_llp, vmalloc_llp, io_llp;
291
	unsigned long lflags, vflags;
292
	static int slb_encoding_inited;
293 294 295
#ifdef CONFIG_SPARSEMEM_VMEMMAP
	unsigned long vmemmap_llp;
#endif
296 297 298

	/* Prepare our SLB miss handler based on our page size */
	linear_llp = mmu_psize_defs[mmu_linear_psize].sllp;
299 300 301
	io_llp = mmu_psize_defs[mmu_io_psize].sllp;
	vmalloc_llp = mmu_psize_defs[mmu_vmalloc_psize].sllp;
	get_paca()->vmalloc_sllp = SLB_VSID_KERNEL | vmalloc_llp;
302 303 304
#ifdef CONFIG_SPARSEMEM_VMEMMAP
	vmemmap_llp = mmu_psize_defs[mmu_vmemmap_psize].sllp;
#endif
305 306 307 308
	if (!slb_encoding_inited) {
		slb_encoding_inited = 1;
		patch_slb_encoding(slb_miss_kernel_load_linear,
				   SLB_VSID_KERNEL | linear_llp);
309 310
		patch_slb_encoding(slb_miss_kernel_load_io,
				   SLB_VSID_KERNEL | io_llp);
311 312
		patch_slb_encoding(slb_compare_rr_to_size,
				   mmu_slb_size);
313

314 315
		pr_devel("SLB: linear  LLP = %04lx\n", linear_llp);
		pr_devel("SLB: io      LLP = %04lx\n", io_llp);
316 317 318 319

#ifdef CONFIG_SPARSEMEM_VMEMMAP
		patch_slb_encoding(slb_miss_kernel_load_vmemmap,
				   SLB_VSID_KERNEL | vmemmap_llp);
320
		pr_devel("SLB: vmemmap LLP = %04lx\n", vmemmap_llp);
321
#endif
322 323
	}

324 325
	get_paca()->stab_rr = SLB_NUM_BOLTED;

326
	lflags = SLB_VSID_KERNEL | linear_llp;
327
	vflags = SLB_VSID_KERNEL | vmalloc_llp;
L
Linus Torvalds 已提交
328

329
	/* Invalidate the entire SLB (even entry 0) & all the ERATS */
330 331 332
	asm volatile("isync":::"memory");
	asm volatile("slbmte  %0,%0"::"r" (0) : "memory");
	asm volatile("isync; slbia; isync":::"memory");
333 334
	create_shadowed_slbe(PAGE_OFFSET, mmu_kernel_ssize, lflags, LINEAR_INDEX);
	create_shadowed_slbe(VMALLOC_START, mmu_kernel_ssize, vflags, VMALLOC_INDEX);
335

336 337 338 339 340
	/* For the boot cpu, we're running on the stack in init_thread_union,
	 * which is in the first segment of the linear mapping, and also
	 * get_paca()->kstack hasn't been initialized yet.
	 * For secondary cpus, we need to bolt the kernel stack entry now.
	 */
341
	slb_shadow_clear(KSTACK_INDEX);
342 343 344
	if (raw_smp_processor_id() != boot_cpuid &&
	    (get_paca()->kstack & slb_esid_mask(mmu_kernel_ssize)) > PAGE_OFFSET)
		create_shadowed_slbe(get_paca()->kstack,
345
				     mmu_kernel_ssize, lflags, KSTACK_INDEX);
346

347
	asm volatile("isync":::"memory");
L
Linus Torvalds 已提交
348
}