1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
|
#ifndef __ALPHA_MMU_CONTEXT_H
#define __ALPHA_MMU_CONTEXT_H
/*
* get a new mmu context..
*
* Copyright (C) 1996, Linus Torvalds
*/
#include <asm/system.h>
#include <asm/machvec.h>
#include <asm/compiler.h>
#include <asm-generic/mm_hooks.h>
/*
* Force a context reload. This is needed when we change the page
* table pointer or when we update the ASN of the current process.
*/
/* Don't get into trouble with dueling __EXTERN_INLINEs. */
#ifndef __EXTERN_INLINE
#include <asm/io.h>
#endif
static inline unsigned long
__reload_thread(struct pcb_struct *pcb)
{
register unsigned long a0 __asm__("$16");
register unsigned long v0 __asm__("$0");
a0 = virt_to_phys(pcb);
__asm__ __volatile__(
"call_pal %2 #__reload_thread"
: "=r"(v0), "=r"(a0)
: "i"(PAL_swpctx), "r"(a0)
: "$1", "$22", "$23", "$24", "$25");
return v0;
}
/*
* The maximum ASN's the processor supports. On the EV4 this is 63
* but the PAL-code doesn't actually use this information. On the
* EV5 this is 127, and EV6 has 255.
*
* On the EV4, the ASNs are more-or-less useless anyway, as they are
* only used as an icache tag, not for TB entries. On the EV5 and EV6,
* ASN's also validate the TB entries, and thus make a lot more sense.
*
* The EV4 ASN's don't even match the architecture manual, ugh. And
* I quote: "If a processor implements address space numbers (ASNs),
* and the old PTE has the Address Space Match (ASM) bit clear (ASNs
* in use) and the Valid bit set, then entries can also effectively be
* made coherent by assigning a new, unused ASN to the currently
* running process and not reusing the previous ASN before calling the
* appropriate PALcode routine to invalidate the translation buffer (TB)".
*
* In short, the EV4 has a "kind of" ASN capability, but it doesn't actually
* work correctly and can thus not be used (explaining the lack of PAL-code
* support).
*/
#define EV4_MAX_ASN 63
#define EV5_MAX_ASN 127
#define EV6_MAX_ASN 255
#ifdef CONFIG_ALPHA_GENERIC
# define MAX_ASN (alpha_mv.max_asn)
#else
# ifdef CONFIG_ALPHA_EV4
# define MAX_ASN EV4_MAX_ASN
# elif defined(CONFIG_ALPHA_EV5)
# define MAX_ASN EV5_MAX_ASN
# else
# define MAX_ASN EV6_MAX_ASN
# endif
#endif
/*
* cpu_last_asn(processor):
* 63 0
* +-------------+----------------+--------------+
* | asn version | this processor | hardware asn |
* +-------------+----------------+--------------+
*/
#include <asm/smp.h>
#ifdef CONFIG_SMP
#define cpu_last_asn(cpuid) (cpu_data[cpuid].last_asn)
#else
extern unsigned long last_asn;
#define cpu_last_asn(cpuid) last_asn
#endif /* CONFIG_SMP */
#define WIDTH_HARDWARE_ASN 8
#define ASN_FIRST_VERSION (1UL << WIDTH_HARDWARE_ASN)
#define HARDWARE_ASN_MASK ((1UL << WIDTH_HARDWARE_ASN) - 1)
/*
* NOTE! The way this is set up, the high bits of the "asn_cache" (and
* the "mm->context") are the ASN _version_ code. A version of 0 is
* always considered invalid, so to invalidate another process you only
* need to do "p->mm->context = 0".
*
* If we need more ASN's than the processor has, we invalidate the old
* user TLB's (tbiap()) and start a new ASN version. That will automatically
* force a new asn for any other processes the next time they want to
* run.
*/
#ifndef __EXTERN_INLINE
#define __EXTERN_INLINE extern inline
#define __MMU_EXTERN_INLINE
#endif
extern inline unsigned long
__get_new_mm_context(struct mm_struct *mm, long cpu)
{
unsigned long asn = cpu_last_asn(cpu);
unsigned long next = asn + 1;
if ((asn & HARDWARE_ASN_MASK) >= MAX_ASN) {
tbiap();
imb();
next = (asn & ~HARDWARE_ASN_MASK) + ASN_FIRST_VERSION;
}
cpu_last_asn(cpu) = next;
return next;
}
__EXTERN_INLINE void
ev5_switch_mm(struct mm_struct *prev_mm, struct mm_struct *next_mm,
struct task_struct *next)
{
/* Check if our ASN is of an older version, and thus invalid. */
unsigned long asn;
unsigned long mmc;
long cpu = smp_processor_id();
#ifdef CONFIG_SMP
cpu_data[cpu].asn_lock = 1;
barrier();
#endif
asn = cpu_last_asn(cpu);
mmc = next_mm->context[cpu];
if ((mmc ^ asn) & ~HARDWARE_ASN_MASK) {
mmc = __get_new_mm_context(next_mm, cpu);
next_mm->context[cpu] = mmc;
}
#ifdef CONFIG_SMP
else
cpu_data[cpu].need_new_asn = 1;
#endif
/* Always update the PCB ASN. Another thread may have allocated
a new mm->context (via flush_tlb_mm) without the ASN serial
number wrapping. We have no way to detect when this is needed. */
task_thread_info(next)->pcb.asn = mmc & HARDWARE_ASN_MASK;
}
__EXTERN_INLINE void
ev4_switch_mm(struct mm_struct *prev_mm, struct mm_struct *next_mm,
struct task_struct *next)
{
/* As described, ASN's are broken for TLB usage. But we can
optimize for switching between threads -- if the mm is
unchanged from current we needn't flush. */
/* ??? May not be needed because EV4 PALcode recognizes that
ASN's are broken and does a tbiap itself on swpctx, under
the "Must set ASN or flush" rule. At least this is true
for a 1992 SRM, reports Joseph Martin (jmartin@hlo.dec.com).
I'm going to leave this here anyway, just to Be Sure. -- r~ */
if (prev_mm != next_mm)
tbiap();
/* Do continue to allocate ASNs, because we can still use them
to avoid flushing the icache. */
ev5_switch_mm(prev_mm, next_mm, next);
}
extern void __load_new_mm_context(struct mm_struct *);
#ifdef CONFIG_SMP
#define check_mmu_context() \
do { \
int cpu = smp_processor_id(); \
cpu_data[cpu].asn_lock = 0; \
barrier(); \
if (cpu_data[cpu].need_new_asn) { \
struct mm_struct * mm = current->active_mm; \
cpu_data[cpu].need_new_asn = 0; \
if (!mm->context[cpu]) \
__load_new_mm_context(mm); \
} \
} while(0)
#else
#define check_mmu_context() do { } while(0)
#endif
__EXTERN_INLINE void
ev5_activate_mm(struct mm_struct *prev_mm, struct mm_struct *next_mm)
{
__load_new_mm_context(next_mm);
}
__EXTERN_INLINE void
ev4_activate_mm(struct mm_struct *prev_mm, struct mm_struct *next_mm)
{
__load_new_mm_context(next_mm);
tbiap();
}
#define deactivate_mm(tsk,mm) do { } while (0)
#ifdef CONFIG_ALPHA_GENERIC
# define switch_mm(a,b,c) alpha_mv.mv_switch_mm((a),(b),(c))
# define activate_mm(x,y) alpha_mv.mv_activate_mm((x),(y))
#else
# ifdef CONFIG_ALPHA_EV4
# define switch_mm(a,b,c) ev4_switch_mm((a),(b),(c))
# define activate_mm(x,y) ev4_activate_mm((x),(y))
# else
# define switch_mm(a,b,c) ev5_switch_mm((a),(b),(c))
# define activate_mm(x,y) ev5_activate_mm((x),(y))
# endif
#endif
static inline int
init_new_context(struct task_struct *tsk, struct mm_struct *mm)
{
int i;
for_each_online_cpu(i)
mm->context[i] = 0;
if (tsk != current)
task_thread_info(tsk)->pcb.ptbr
= ((unsigned long)mm->pgd - IDENT_ADDR) >> PAGE_SHIFT;
return 0;
}
extern inline void
destroy_context(struct mm_struct *mm)
{
/* Nothing to do. */
}
static inline void
enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk)
{
task_thread_info(tsk)->pcb.ptbr
= ((unsigned long)mm->pgd - IDENT_ADDR) >> PAGE_SHIFT;
}
#ifdef __MMU_EXTERN_INLINE
#undef __EXTERN_INLINE
#undef __MMU_EXTERN_INLINE
#endif
#endif /* __ALPHA_MMU_CONTEXT_H */
|
