This repository has been archived by the owner on Jan 23, 2023. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 2.7k
/
JitHelpers_Slow.asm
422 lines (301 loc) · 13.6 KB
/
JitHelpers_Slow.asm
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
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
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
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
; Licensed to the .NET Foundation under one or more agreements.
; The .NET Foundation licenses this file to you under the MIT license.
; See the LICENSE file in the project root for more information.
; ==++==
;
;
; ==--==
; ***********************************************************************
; File: JitHelpers_Slow.asm, see history in jithelp.asm
;
; Notes: These are ASM routinues which we believe to be cold in normal
; AMD64 scenarios, mainly because they have other versions which
; have some more performant nature which will be used in the best
; cases.
; ***********************************************************************
include AsmMacros.inc
include asmconstants.inc
; Min amount of stack space that a nested function should allocate.
MIN_SIZE equ 28h
EXTERN g_ephemeral_low:QWORD
EXTERN g_ephemeral_high:QWORD
EXTERN g_lowest_address:QWORD
EXTERN g_highest_address:QWORD
EXTERN g_card_table:QWORD
ifdef FEATURE_USE_SOFTWARE_WRITE_WATCH_FOR_GC_HEAP
EXTERN g_sw_ww_table:QWORD
EXTERN g_sw_ww_enabled_for_gc_heap:BYTE
endif
ifdef WRITE_BARRIER_CHECK
; Those global variables are always defined, but should be 0 for Server GC
g_GCShadow TEXTEQU <?g_GCShadow@@3PEAEEA>
g_GCShadowEnd TEXTEQU <?g_GCShadowEnd@@3PEAEEA>
EXTERN g_GCShadow:QWORD
EXTERN g_GCShadowEnd:QWORD
endif
JIT_NEW equ ?JIT_New@@YAPEAVObject@@PEAUCORINFO_CLASS_STRUCT_@@@Z
CopyValueClassUnchecked equ ?CopyValueClassUnchecked@@YAXPEAX0PEAVMethodTable@@@Z
JIT_Box equ ?JIT_Box@@YAPEAVObject@@PEAUCORINFO_CLASS_STRUCT_@@PEAX@Z
g_pStringClass equ ?g_pStringClass@@3PEAVMethodTable@@EA
FramedAllocateString equ ?FramedAllocateString@@YAPEAVStringObject@@K@Z
JIT_NewArr1 equ ?JIT_NewArr1@@YAPEAVObject@@PEAUCORINFO_CLASS_STRUCT_@@_J@Z
INVALIDGCVALUE equ 0CCCCCCCDh
extern JIT_NEW:proc
extern CopyValueClassUnchecked:proc
extern JIT_Box:proc
extern g_pStringClass:QWORD
extern FramedAllocateString:proc
extern JIT_NewArr1:proc
extern JIT_GetSharedNonGCStaticBase_Helper:proc
extern JIT_GetSharedGCStaticBase_Helper:proc
extern JIT_InternalThrow:proc
ifdef _DEBUG
; Version for when we're sure to be in the GC, checks whether or not the card
; needs to be updated
;
; void JIT_WriteBarrier_Debug(Object** dst, Object* src)
LEAF_ENTRY JIT_WriteBarrier_Debug, _TEXT
ifdef WRITE_BARRIER_CHECK
; **ALSO update the shadow GC heap if that is enabled**
; Do not perform the work if g_GCShadow is 0
cmp g_GCShadow, 0
je NoShadow
; If we end up outside of the heap don't corrupt random memory
mov r10, rcx
sub r10, [g_lowest_address]
jb NoShadow
; Check that our adjusted destination is somewhere in the shadow gc
add r10, [g_GCShadow]
cmp r10, [g_GCShadowEnd]
ja NoShadow
; Write ref into real GC; see comment below about possibility of AV
mov [rcx], rdx
; Write ref into shadow GC
mov [r10], rdx
; Ensure that the write to the shadow heap occurs before the read from
; the GC heap so that race conditions are caught by INVALIDGCVALUE
mfence
; Check that GC/ShadowGC values match
mov r11, [rcx]
mov rax, [r10]
cmp rax, r11
je DoneShadow
mov r11, INVALIDGCVALUE
mov [r10], r11
jmp DoneShadow
; If we don't have a shadow GC we won't have done the write yet
NoShadow:
endif
mov rax, rdx
; Do the move. It is correct to possibly take an AV here, the EH code
; figures out that this came from a WriteBarrier and correctly maps it back
; to the managed method which called the WriteBarrier (see setup in
; InitializeExceptionHandling, vm\exceptionhandling.cpp).
mov [rcx], rax
ifdef WRITE_BARRIER_CHECK
; If we had a shadow GC then we already wrote to the real GC at the same time
; as the shadow GC so we want to jump over the real write immediately above
DoneShadow:
endif
ifdef FEATURE_USE_SOFTWARE_WRITE_WATCH_FOR_GC_HEAP
; Update the write watch table if necessary
cmp byte ptr [g_sw_ww_enabled_for_gc_heap], 0h
je CheckCardTable
mov r10, rcx
shr r10, 0Ch ; SoftwareWriteWatch::AddressToTableByteIndexShift
add r10, qword ptr [g_sw_ww_table]
cmp byte ptr [r10], 0h
jne CheckCardTable
mov byte ptr [r10], 0FFh
endif
CheckCardTable:
; See if we can just quick out
cmp rax, [g_ephemeral_low]
jb Exit
cmp rax, [g_ephemeral_high]
jnb Exit
; Check if we need to update the card table
; Calc pCardByte
shr rcx, 0Bh
add rcx, [g_card_table]
; Check if this card is dirty
cmp byte ptr [rcx], 0FFh
jne UpdateCardTable
REPRET
UpdateCardTable:
mov byte ptr [rcx], 0FFh
ret
align 16
Exit:
REPRET
LEAF_END_MARKED JIT_WriteBarrier_Debug, _TEXT
endif
extern g_global_alloc_lock:dword
extern g_global_alloc_context:qword
LEAF_ENTRY JIT_TrialAllocSFastSP, _TEXT
mov r8d, [rcx + OFFSET__MethodTable__m_BaseSize]
; m_BaseSize is guaranteed to be a multiple of 8.
inc [g_global_alloc_lock]
jnz JIT_NEW
mov rax, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr] ; alloc_ptr
mov r10, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_limit] ; limit_ptr
add r8, rax
cmp r8, r10
ja AllocFailed
mov qword ptr [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr], r8 ; update the alloc ptr
mov [rax], rcx
mov [g_global_alloc_lock], -1
ret
AllocFailed:
mov [g_global_alloc_lock], -1
jmp JIT_NEW
LEAF_END JIT_TrialAllocSFastSP, _TEXT
; HCIMPL2(Object*, JIT_Box, CORINFO_CLASS_HANDLE type, void* unboxedData)
NESTED_ENTRY JIT_BoxFastUP, _TEXT
mov rax, [rcx + OFFSETOF__MethodTable__m_pWriteableData]
; Check whether the class has not been initialized
test dword ptr [rax + OFFSETOF__MethodTableWriteableData__m_dwFlags], MethodTableWriteableData__enum_flag_Unrestored
jnz JIT_Box
mov r8d, [rcx + OFFSET__MethodTable__m_BaseSize]
; m_BaseSize is guaranteed to be a multiple of 8.
inc [g_global_alloc_lock]
jnz JIT_Box
mov rax, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr] ; alloc_ptr
mov r10, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_limit] ; limit_ptr
add r8, rax
cmp r8, r10
ja NoAlloc
mov qword ptr [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr], r8 ; update the alloc ptr
mov [rax], rcx
mov [g_global_alloc_lock], -1
; Check whether the object contains pointers
test dword ptr [rcx + OFFSETOF__MethodTable__m_dwFlags], MethodTable__enum_flag_ContainsPointers
jnz ContainsPointers
; We have no pointers - emit a simple inline copy loop
mov ecx, [rcx + OFFSET__MethodTable__m_BaseSize]
sub ecx, 18h ; sizeof(ObjHeader) + sizeof(Object) + last slot
CopyLoop:
mov r8, [rdx+rcx]
mov [rax+rcx+8], r8
sub ecx, 8
jge CopyLoop
REPRET
ContainsPointers:
; Do call to CopyValueClassUnchecked(object, data, pMT)
push_vol_reg rax
alloc_stack 20h
END_PROLOGUE
mov r8, rcx
lea rcx, [rax + 8]
call CopyValueClassUnchecked
add rsp, 20h
pop rax
ret
NoAlloc:
mov [g_global_alloc_lock], -1
jmp JIT_Box
NESTED_END JIT_BoxFastUP, _TEXT
LEAF_ENTRY AllocateStringFastUP, _TEXT
; We were passed the number of characters in ECX
; we need to load the method table for string from the global
mov r11, [g_pStringClass]
; Instead of doing elaborate overflow checks, we just limit the number of elements
; to (LARGE_OBJECT_SIZE - 256)/sizeof(WCHAR) or less.
; This will avoid all overflow problems, as well as making sure
; big string objects are correctly allocated in the big object heap.
cmp ecx, (ASM_LARGE_OBJECT_SIZE - 256)/2
jae FramedAllocateString
; Calculate the final size to allocate.
; We need to calculate baseSize + cnt*2, then round that up by adding 7 and anding ~7.
lea r8d, [STRING_BASE_SIZE + ecx*2 + 7]
and r8d, -8
inc [g_global_alloc_lock]
jnz FramedAllocateString
mov rax, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr] ; alloc_ptr
mov r10, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_limit] ; limit_ptr
add r8, rax
cmp r8, r10
ja AllocFailed
mov qword ptr [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr], r8 ; update the alloc ptr
mov [rax], r11
mov [g_global_alloc_lock], -1
mov [rax + OFFSETOF__StringObject__m_StringLength], ecx
ret
AllocFailed:
mov [g_global_alloc_lock], -1
jmp FramedAllocateString
LEAF_END AllocateStringFastUP, _TEXT
; HCIMPL2(Object*, JIT_NewArr1VC_UP, CORINFO_CLASS_HANDLE arrayMT, INT_PTR size)
LEAF_ENTRY JIT_NewArr1VC_UP, _TEXT
; We were passed a (shared) method table in RCX, which contains the element type.
; The element count is in RDX
; NOTE: if this code is ported for CORINFO_HELP_NEWSFAST_ALIGN8, it will need
; to emulate the double-specific behavior of JIT_TrialAlloc::GenAllocArray.
; Do a conservative check here. This is to avoid overflow while doing the calculations. We don't
; have to worry about "large" objects, since the allocation quantum is never big enough for
; LARGE_OBJECT_SIZE.
; For Value Classes, this needs to be 2^16 - slack (2^32 / max component size),
; The slack includes the size for the array header and round-up ; for alignment. Use 256 for the
; slack value out of laziness.
; In both cases we do a final overflow check after adding to the alloc_ptr.
cmp rdx, (65535 - 256)
jae JIT_NewArr1
movzx r8d, word ptr [rcx + OFFSETOF__MethodTable__m_dwFlags] ; component size is low 16 bits
imul r8d, edx ; signed mul, but won't overflow due to length restriction above
add r8d, dword ptr [rcx + OFFSET__MethodTable__m_BaseSize]
; round the size to a multiple of 8
add r8d, 7
and r8d, -8
inc [g_global_alloc_lock]
jnz JIT_NewArr1
mov rax, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr] ; alloc_ptr
mov r10, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_limit] ; limit_ptr
add r8, rax
jc AllocFailed
cmp r8, r10
ja AllocFailed
mov qword ptr [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr], r8 ; update the alloc ptr
mov [rax], rcx
mov [g_global_alloc_lock], -1
mov dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
ret
AllocFailed:
mov [g_global_alloc_lock], -1
jmp JIT_NewArr1
LEAF_END JIT_NewArr1VC_UP, _TEXT
; HCIMPL2(Object*, JIT_NewArr1OBJ_UP, CORINFO_CLASS_HANDLE arrayMT, INT_PTR size)
LEAF_ENTRY JIT_NewArr1OBJ_UP, _TEXT
; We were passed a (shared) method table in RCX, which contains the element type.
; The element count is in RDX
; NOTE: if this code is ported for CORINFO_HELP_NEWSFAST_ALIGN8, it will need
; to emulate the double-specific behavior of JIT_TrialAlloc::GenAllocArray.
; Verifies that LARGE_OBJECT_SIZE fits in 32-bit. This allows us to do array size
; arithmetic using 32-bit registers.
.erre ASM_LARGE_OBJECT_SIZE lt 100000000h
cmp rdx, (ASM_LARGE_OBJECT_SIZE - 256)/8 ; sizeof(void*)
jae OversizedArray
; In this case we know the element size is sizeof(void *), or 8 for x64
; This helps us in two ways - we can shift instead of multiplying, and
; there's no need to align the size either
mov r8d, dword ptr [rcx + OFFSET__MethodTable__m_BaseSize]
lea r8d, [r8d + edx * 8]
; No need for rounding in this case - element size is 8, and m_BaseSize is guaranteed
; to be a multiple of 8.
inc [g_global_alloc_lock]
jnz JIT_NewArr1
mov rax, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr] ; alloc_ptr
mov r10, [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_limit] ; limit_ptr
add r8, rax
cmp r8, r10
ja AllocFailed
mov qword ptr [g_global_alloc_context + OFFSETOF__gc_alloc_context__alloc_ptr], r8 ; update the alloc ptr
mov [rax], rcx
mov [g_global_alloc_lock], -1
mov dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
ret
AllocFailed:
mov [g_global_alloc_lock], -1
OversizedArray:
jmp JIT_NewArr1
LEAF_END JIT_NewArr1OBJ_UP, _TEXT
end