xref: /openbmc/qemu/docs/devel/loads-stores.rst (revision b7bc6b18)
1..
2   Copyright (c) 2017 Linaro Limited
3   Written by Peter Maydell
4
5===================
6Load and Store APIs
7===================
8
9QEMU internally has multiple families of functions for performing
10loads and stores. This document attempts to enumerate them all
11and indicate when to use them. It does not provide detailed
12documentation of each API -- for that you should look at the
13documentation comments in the relevant header files.
14
15
16``ld*_p and st*_p``
17~~~~~~~~~~~~~~~~~~~
18
19These functions operate on a host pointer, and should be used
20when you already have a pointer into host memory (corresponding
21to guest ram or a local buffer). They deal with doing accesses
22with the desired endianness and with correctly handling
23potentially unaligned pointer values.
24
25Function names follow the pattern:
26
27load: ``ld{sign}{size}_{endian}_p(ptr)``
28
29store: ``st{size}_{endian}_p(ptr, val)``
30
31``sign``
32 - (empty) : for 32 or 64 bit sizes
33 - ``u`` : unsigned
34 - ``s`` : signed
35
36``size``
37 - ``b`` : 8 bits
38 - ``w`` : 16 bits
39 - ``l`` : 32 bits
40 - ``q`` : 64 bits
41
42``endian``
43 - ``he`` : host endian
44 - ``be`` : big endian
45 - ``le`` : little endian
46
47The ``_{endian}`` infix is omitted for target-endian accesses.
48
49The target endian accessors are only available to source
50files which are built per-target.
51
52There are also functions which take the size as an argument:
53
54load: ``ldn{endian}_p(ptr, sz)``
55
56which performs an unsigned load of ``sz`` bytes from ``ptr``
57as an ``{endian}`` order value and returns it in a uint64_t.
58
59store: ``stn{endian}_p(ptr, sz, val)``
60
61which stores ``val`` to ``ptr`` as an ``{endian}`` order value
62of size ``sz`` bytes.
63
64
65Regexes for git grep
66 - ``\<ld[us]\?[bwlq]\(_[hbl]e\)\?_p\>``
67 - ``\<st[bwlq]\(_[hbl]e\)\?_p\>``
68 - ``\<ldn_\([hbl]e\)?_p\>``
69 - ``\<stn_\([hbl]e\)?_p\>``
70
71``cpu_{ld,st}*_mmu``
72~~~~~~~~~~~~~~~~~~~~
73
74These functions operate on a guest virtual address, plus a context
75known as a "mmu index" which controls how that virtual address is
76translated, plus a ``MemOp`` which contains alignment requirements
77among other things.  The ``MemOp`` and mmu index are combined into
78a single argument of type ``MemOpIdx``.
79
80The meaning of the indexes are target specific, but specifying a
81particular index might be necessary if, for instance, the helper
82requires a "always as non-privileged" access rather than the
83default access for the current state of the guest CPU.
84
85These functions may cause a guest CPU exception to be taken
86(e.g. for an alignment fault or MMU fault) which will result in
87guest CPU state being updated and control longjmp'ing out of the
88function call.  They should therefore only be used in code that is
89implementing emulation of the guest CPU.
90
91The ``retaddr`` parameter is used to control unwinding of the
92guest CPU state in case of a guest CPU exception.  This is passed
93to ``cpu_restore_state()``.  Therefore the value should either be 0,
94to indicate that the guest CPU state is already synchronized, or
95the result of ``GETPC()`` from the top level ``HELPER(foo)``
96function, which is a return address into the generated code [#gpc]_.
97
98.. [#gpc] Note that ``GETPC()`` should be used with great care: calling
99          it in other functions that are *not* the top level
100          ``HELPER(foo)`` will cause unexpected behavior. Instead, the
101          value of ``GETPC()`` should be read from the helper and passed
102          if needed to the functions that the helper calls.
103
104Function names follow the pattern:
105
106load: ``cpu_ld{size}{end}_mmu(env, ptr, oi, retaddr)``
107
108store: ``cpu_st{size}{end}_mmu(env, ptr, val, oi, retaddr)``
109
110``size``
111 - ``b`` : 8 bits
112 - ``w`` : 16 bits
113 - ``l`` : 32 bits
114 - ``q`` : 64 bits
115
116``end``
117 - (empty) : for target endian, or 8 bit sizes
118 - ``_be`` : big endian
119 - ``_le`` : little endian
120
121Regexes for git grep:
122 - ``\<cpu_ld[bwlq](_[bl]e)\?_mmu\>``
123 - ``\<cpu_st[bwlq](_[bl]e)\?_mmu\>``
124
125
126``cpu_{ld,st}*_mmuidx_ra``
127~~~~~~~~~~~~~~~~~~~~~~~~~~
128
129These functions work like the ``cpu_{ld,st}_mmu`` functions except
130that the ``mmuidx`` parameter is not combined with a ``MemOp``,
131and therefore there is no required alignment supplied or enforced.
132
133Function names follow the pattern:
134
135load: ``cpu_ld{sign}{size}{end}_mmuidx_ra(env, ptr, mmuidx, retaddr)``
136
137store: ``cpu_st{size}{end}_mmuidx_ra(env, ptr, val, mmuidx, retaddr)``
138
139``sign``
140 - (empty) : for 32 or 64 bit sizes
141 - ``u`` : unsigned
142 - ``s`` : signed
143
144``size``
145 - ``b`` : 8 bits
146 - ``w`` : 16 bits
147 - ``l`` : 32 bits
148 - ``q`` : 64 bits
149
150``end``
151 - (empty) : for target endian, or 8 bit sizes
152 - ``_be`` : big endian
153 - ``_le`` : little endian
154
155Regexes for git grep:
156 - ``\<cpu_ld[us]\?[bwlq](_[bl]e)\?_mmuidx_ra\>``
157 - ``\<cpu_st[bwlq](_[bl]e)\?_mmuidx_ra\>``
158
159``cpu_{ld,st}*_data_ra``
160~~~~~~~~~~~~~~~~~~~~~~~~
161
162These functions work like the ``cpu_{ld,st}_mmuidx_ra`` functions
163except that the ``mmuidx`` parameter is taken from the current mode
164of the guest CPU, as determined by ``cpu_mmu_index(env, false)``.
165
166These are generally the preferred way to do accesses by guest
167virtual address from helper functions, unless the access should
168be performed with a context other than the default, or alignment
169should be enforced for the access.
170
171Function names follow the pattern:
172
173load: ``cpu_ld{sign}{size}{end}_data_ra(env, ptr, ra)``
174
175store: ``cpu_st{size}{end}_data_ra(env, ptr, val, ra)``
176
177``sign``
178 - (empty) : for 32 or 64 bit sizes
179 - ``u`` : unsigned
180 - ``s`` : signed
181
182``size``
183 - ``b`` : 8 bits
184 - ``w`` : 16 bits
185 - ``l`` : 32 bits
186 - ``q`` : 64 bits
187
188``end``
189 - (empty) : for target endian, or 8 bit sizes
190 - ``_be`` : big endian
191 - ``_le`` : little endian
192
193Regexes for git grep:
194 - ``\<cpu_ld[us]\?[bwlq](_[bl]e)\?_data_ra\>``
195 - ``\<cpu_st[bwlq](_[bl]e)\?_data_ra\>``
196
197``cpu_{ld,st}*_data``
198~~~~~~~~~~~~~~~~~~~~~
199
200These functions work like the ``cpu_{ld,st}_data_ra`` functions
201except that the ``retaddr`` parameter is 0, and thus does not
202unwind guest CPU state.
203
204This means they must only be used from helper functions where the
205translator has saved all necessary CPU state.  These functions are
206the right choice for calls made from hooks like the CPU ``do_interrupt``
207hook or when you know for certain that the translator had to save all
208the CPU state anyway.
209
210Function names follow the pattern:
211
212load: ``cpu_ld{sign}{size}{end}_data(env, ptr)``
213
214store: ``cpu_st{size}{end}_data(env, ptr, val)``
215
216``sign``
217 - (empty) : for 32 or 64 bit sizes
218 - ``u`` : unsigned
219 - ``s`` : signed
220
221``size``
222 - ``b`` : 8 bits
223 - ``w`` : 16 bits
224 - ``l`` : 32 bits
225 - ``q`` : 64 bits
226
227``end``
228 - (empty) : for target endian, or 8 bit sizes
229 - ``_be`` : big endian
230 - ``_le`` : little endian
231
232Regexes for git grep
233 - ``\<cpu_ld[us]\?[bwlq](_[bl]e)\?_data\>``
234 - ``\<cpu_st[bwlq](_[bl]e)\?_data\+\>``
235
236``cpu_ld*_code``
237~~~~~~~~~~~~~~~~
238
239These functions perform a read for instruction execution.  The ``mmuidx``
240parameter is taken from the current mode of the guest CPU, as determined
241by ``cpu_mmu_index(env, true)``.  The ``retaddr`` parameter is 0, and
242thus does not unwind guest CPU state, because CPU state is always
243synchronized while translating instructions.  Any guest CPU exception
244that is raised will indicate an instruction execution fault rather than
245a data read fault.
246
247In general these functions should not be used directly during translation.
248There are wrapper functions that are to be used which also take care of
249plugins for tracing.
250
251Function names follow the pattern:
252
253load: ``cpu_ld{sign}{size}_code(env, ptr)``
254
255``sign``
256 - (empty) : for 32 or 64 bit sizes
257 - ``u`` : unsigned
258 - ``s`` : signed
259
260``size``
261 - ``b`` : 8 bits
262 - ``w`` : 16 bits
263 - ``l`` : 32 bits
264 - ``q`` : 64 bits
265
266Regexes for git grep:
267 - ``\<cpu_ld[us]\?[bwlq]_code\>``
268
269``translator_ld*``
270~~~~~~~~~~~~~~~~~~
271
272These functions are a wrapper for ``cpu_ld*_code`` which also perform
273any actions required by any tracing plugins.  They are only to be
274called during the translator callback ``translate_insn``.
275
276There is a set of functions ending in ``_swap`` which, if the parameter
277is true, returns the value in the endianness that is the reverse of
278the guest native endianness, as determined by ``TARGET_WORDS_BIGENDIAN``.
279
280Function names follow the pattern:
281
282load: ``translator_ld{sign}{size}(env, ptr)``
283
284swap: ``translator_ld{sign}{size}_swap(env, ptr, swap)``
285
286``sign``
287 - (empty) : for 32 or 64 bit sizes
288 - ``u`` : unsigned
289 - ``s`` : signed
290
291``size``
292 - ``b`` : 8 bits
293 - ``w`` : 16 bits
294 - ``l`` : 32 bits
295 - ``q`` : 64 bits
296
297Regexes for git grep
298 - ``\<translator_ld[us]\?[bwlq]\(_swap\)\?\>``
299
300``helper_*_{ld,st}*_mmu``
301~~~~~~~~~~~~~~~~~~~~~~~~~
302
303These functions are intended primarily to be called by the code
304generated by the TCG backend. They may also be called by target
305CPU helper function code. Like the ``cpu_{ld,st}_mmuidx_ra`` functions
306they perform accesses by guest virtual address, with a given ``mmuidx``.
307
308These functions specify an ``opindex`` parameter which encodes
309(among other things) the mmu index to use for the access.  This parameter
310should be created by calling ``make_memop_idx()``.
311
312The ``retaddr`` parameter should be the result of GETPC() called directly
313from the top level HELPER(foo) function (or 0 if no guest CPU state
314unwinding is required).
315
316**TODO** The names of these functions are a bit odd for historical
317reasons because they were originally expected to be called only from
318within generated code. We should rename them to bring them more in
319line with the other memory access functions. The explicit endianness
320is the only feature they have beyond ``*_mmuidx_ra``.
321
322load: ``helper_{endian}_ld{sign}{size}_mmu(env, addr, opindex, retaddr)``
323
324store: ``helper_{endian}_st{size}_mmu(env, addr, val, opindex, retaddr)``
325
326``sign``
327 - (empty) : for 32 or 64 bit sizes
328 - ``u`` : unsigned
329 - ``s`` : signed
330
331``size``
332 - ``b`` : 8 bits
333 - ``w`` : 16 bits
334 - ``l`` : 32 bits
335 - ``q`` : 64 bits
336
337``endian``
338 - ``le`` : little endian
339 - ``be`` : big endian
340 - ``ret`` : target endianness
341
342Regexes for git grep
343 - ``\<helper_\(le\|be\|ret\)_ld[us]\?[bwlq]_mmu\>``
344 - ``\<helper_\(le\|be\|ret\)_st[bwlq]_mmu\>``
345
346``address_space_*``
347~~~~~~~~~~~~~~~~~~~
348
349These functions are the primary ones to use when emulating CPU
350or device memory accesses. They take an AddressSpace, which is the
351way QEMU defines the view of memory that a device or CPU has.
352(They generally correspond to being the "master" end of a hardware bus
353or bus fabric.)
354
355Each CPU has an AddressSpace. Some kinds of CPU have more than
356one AddressSpace (for instance Arm guest CPUs have an AddressSpace
357for the Secure world and one for NonSecure if they implement TrustZone).
358Devices which can do DMA-type operations should generally have an
359AddressSpace. There is also a "system address space" which typically
360has all the devices and memory that all CPUs can see. (Some older
361device models use the "system address space" rather than properly
362modelling that they have an AddressSpace of their own.)
363
364Functions are provided for doing byte-buffer reads and writes,
365and also for doing one-data-item loads and stores.
366
367In all cases the caller provides a MemTxAttrs to specify bus
368transaction attributes, and can check whether the memory transaction
369succeeded using a MemTxResult return code.
370
371``address_space_read(address_space, addr, attrs, buf, len)``
372
373``address_space_write(address_space, addr, attrs, buf, len)``
374
375``address_space_rw(address_space, addr, attrs, buf, len, is_write)``
376
377``address_space_ld{sign}{size}_{endian}(address_space, addr, attrs, txresult)``
378
379``address_space_st{size}_{endian}(address_space, addr, val, attrs, txresult)``
380
381``sign``
382 - (empty) : for 32 or 64 bit sizes
383 - ``u`` : unsigned
384
385(No signed load operations are provided.)
386
387``size``
388 - ``b`` : 8 bits
389 - ``w`` : 16 bits
390 - ``l`` : 32 bits
391 - ``q`` : 64 bits
392
393``endian``
394 - ``le`` : little endian
395 - ``be`` : big endian
396
397The ``_{endian}`` suffix is omitted for byte accesses.
398
399Regexes for git grep
400 - ``\<address_space_\(read\|write\|rw\)\>``
401 - ``\<address_space_ldu\?[bwql]\(_[lb]e\)\?\>``
402 - ``\<address_space_st[bwql]\(_[lb]e\)\?\>``
403
404``address_space_write_rom``
405~~~~~~~~~~~~~~~~~~~~~~~~~~~
406
407This function performs a write by physical address like
408``address_space_write``, except that if the write is to a ROM then
409the ROM contents will be modified, even though a write by the guest
410CPU to the ROM would be ignored. This is used for non-guest writes
411like writes from the gdb debug stub or initial loading of ROM contents.
412
413Note that portions of the write which attempt to write data to a
414device will be silently ignored -- only real RAM and ROM will
415be written to.
416
417Regexes for git grep
418 - ``address_space_write_rom``
419
420``{ld,st}*_phys``
421~~~~~~~~~~~~~~~~~
422
423These are functions which are identical to
424``address_space_{ld,st}*``, except that they always pass
425``MEMTXATTRS_UNSPECIFIED`` for the transaction attributes, and ignore
426whether the transaction succeeded or failed.
427
428The fact that they ignore whether the transaction succeeded means
429they should not be used in new code, unless you know for certain
430that your code will only be used in a context where the CPU or
431device doing the access has no way to report such an error.
432
433``load: ld{sign}{size}_{endian}_phys``
434
435``store: st{size}_{endian}_phys``
436
437``sign``
438 - (empty) : for 32 or 64 bit sizes
439 - ``u`` : unsigned
440
441(No signed load operations are provided.)
442
443``size``
444 - ``b`` : 8 bits
445 - ``w`` : 16 bits
446 - ``l`` : 32 bits
447 - ``q`` : 64 bits
448
449``endian``
450 - ``le`` : little endian
451 - ``be`` : big endian
452
453The ``_{endian}_`` infix is omitted for byte accesses.
454
455Regexes for git grep
456 - ``\<ldu\?[bwlq]\(_[bl]e\)\?_phys\>``
457 - ``\<st[bwlq]\(_[bl]e\)\?_phys\>``
458
459``cpu_physical_memory_*``
460~~~~~~~~~~~~~~~~~~~~~~~~~
461
462These are convenience functions which are identical to
463``address_space_*`` but operate specifically on the system address space,
464always pass a ``MEMTXATTRS_UNSPECIFIED`` set of memory attributes and
465ignore whether the memory transaction succeeded or failed.
466For new code they are better avoided:
467
468* there is likely to be behaviour you need to model correctly for a
469  failed read or write operation
470* a device should usually perform operations on its own AddressSpace
471  rather than using the system address space
472
473``cpu_physical_memory_read``
474
475``cpu_physical_memory_write``
476
477``cpu_physical_memory_rw``
478
479Regexes for git grep
480 - ``\<cpu_physical_memory_\(read\|write\|rw\)\>``
481
482``cpu_memory_rw_debug``
483~~~~~~~~~~~~~~~~~~~~~~~
484
485Access CPU memory by virtual address for debug purposes.
486
487This function is intended for use by the GDB stub and similar code.
488It takes a virtual address, converts it to a physical address via
489an MMU lookup using the current settings of the specified CPU,
490and then performs the access (using ``address_space_rw`` for
491reads or ``cpu_physical_memory_write_rom`` for writes).
492This means that if the access is a write to a ROM then this
493function will modify the contents (whereas a normal guest CPU access
494would ignore the write attempt).
495
496``cpu_memory_rw_debug``
497
498``dma_memory_*``
499~~~~~~~~~~~~~~~~
500
501These behave like ``address_space_*``, except that they perform a DMA
502barrier operation first.
503
504**TODO**: We should provide guidance on when you need the DMA
505barrier operation and when it's OK to use ``address_space_*``, and
506make sure our existing code is doing things correctly.
507
508``dma_memory_read``
509
510``dma_memory_write``
511
512``dma_memory_rw``
513
514Regexes for git grep
515 - ``\<dma_memory_\(read\|write\|rw\)\>``
516 - ``\<ldu\?[bwlq]\(_[bl]e\)\?_dma\>``
517 - ``\<st[bwlq]\(_[bl]e\)\?_dma\>``
518
519``pci_dma_*`` and ``{ld,st}*_pci_dma``
520~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
521
522These functions are specifically for PCI device models which need to
523perform accesses where the PCI device is a bus master. You pass them a
524``PCIDevice *`` and they will do ``dma_memory_*`` operations on the
525correct address space for that device.
526
527``pci_dma_read``
528
529``pci_dma_write``
530
531``pci_dma_rw``
532
533``load: ld{sign}{size}_{endian}_pci_dma``
534
535``store: st{size}_{endian}_pci_dma``
536
537``sign``
538 - (empty) : for 32 or 64 bit sizes
539 - ``u`` : unsigned
540
541(No signed load operations are provided.)
542
543``size``
544 - ``b`` : 8 bits
545 - ``w`` : 16 bits
546 - ``l`` : 32 bits
547 - ``q`` : 64 bits
548
549``endian``
550 - ``le`` : little endian
551 - ``be`` : big endian
552
553The ``_{endian}_`` infix is omitted for byte accesses.
554
555Regexes for git grep
556 - ``\<pci_dma_\(read\|write\|rw\)\>``
557 - ``\<ldu\?[bwlq]\(_[bl]e\)\?_pci_dma\>``
558 - ``\<st[bwlq]\(_[bl]e\)\?_pci_dma\>``
559