OpenCores
URL https://opencores.org/ocsvn/neorv32/neorv32/trunk

Subversion Repositories neorv32

[/] [neorv32/] [trunk/] [docs/] [datasheet/] [cpu.adoc] - Blame information for rev 70

Go to most recent revision | Details | Compare with Previous | View Log

Line No. Rev Author Line
1 60 zero_gravi
:sectnums:
2
== NEORV32 Central Processing Unit (CPU)
3
 
4
image::riscv_logo.png[width=350,align=center]
5
 
6
**Key Features**
7
 
8 66 zero_gravi
* 32-bit multi-cycle in-order `rv32` RISC-V CPU
9 61 zero_gravi
* Optional RISC-V extensions:
10
** `A` - atomic memory access operations
11 66 zero_gravi
** `B` - bit-manipulation instructions
12 61 zero_gravi
** `C` - 16-bit compressed instructions
13
** `I` - integer base ISA (always enabled)
14
** `E` - embedded CPU version (reduced register file size)
15
** `M` - integer multiplication and division hardware
16
** `U` - less-privileged _user_ mode
17
** `Zfinx` - single-precision floating-point unit
18
** `Zicsr` - control and status register access (privileged architecture)
19 66 zero_gravi
** `Zicntr` - CPU base counters
20
** `Zihpm` - hardware performance monitors
21 61 zero_gravi
** `Zifencei` - instruction stream synchronization
22
** `Zmmul` - integer multiplication hardware
23
** `PMP` - physical memory protection
24 66 zero_gravi
** `Debug` - debug mode
25 65 zero_gravi
* Compatible to the RISC-V user specifications and a subset of the RISC-V privileged architecture specifications - passes the official RISC-V Architecture Tests (v2+)
26 60 zero_gravi
* Official RISC-V open-source architecture ID
27 65 zero_gravi
* Standard RISC-V interrupts (_external_, _timer_, _software_) plus 16 _fast_ interrupts
28 66 zero_gravi
* Supports _all_ of the machine-level traps from the RISC-V specifications (including bus access exceptions and all unimplemented/illegal/malformed instructions)
29
** This is a special aspect on _execution safety_ by <<_full_virtualization>>
30 60 zero_gravi
* Optional physical memory configuration (PMP), compatible to the RISC-V specifications
31
* Optional hardware performance monitors (HPM) for application benchmarking
32 66 zero_gravi
* Separated interfaces for instruction fetch and data access (merged into a single processor bus))
33 60 zero_gravi
* little-endian byte order
34
* Configurable hardware reset
35 65 zero_gravi
* No hardware support of unaligned data/instruction accesses - they will trigger an exception.
36 60 zero_gravi
 
37
[NOTE]
38
It is recommended to use the **NEORV32 Processor** as default top instance even if you only want to use the actual
39
CPU. Simply disable all the processor-internal modules via the generics and you will get a "CPU
40
wrapper" that provides a minimal CPU environment and an external bus interface (like AXI4). This
41
setup also allows to further use the default bootloader and software framework. From this base you
42 70 zero_gravi
can start building your own SoC. Of course you can also use the CPU in it's true stand-alone mode.
43 60 zero_gravi
 
44
[NOTE]
45
This documentation assumes the reader is familiar with the official RISC-V "User" and "Privileged Architecture" specifications.
46
 
47
<<<
48
// ####################################################################################################################
49
:sectnums:
50
=== Architecture
51
 
52
The NEORV32 CPU was designed from scratch based only on the official ISA / privileged architecture
53
specifications. The following figure shows the simplified architecture of the CPU.
54
 
55
image::neorv32_cpu.png[align=center]
56
 
57 66 zero_gravi
The CPU implements a _multi-cycle_ architecture. Hence, each instruction is executed as a series of consecutive
58
micro-operations. In order to increase performance, the CPU's **front-end** (instruction fetch) and **back-end**
59
(instruction execution) are de-couples via a FIFO (the "instruction prefetch buffer"). Therefore, the
60
front-end can already fetch new instructions while the back-end is still processing previously-fetched instructions.
61 60 zero_gravi
 
62 66 zero_gravi
The front-end is responsible for fetching 32-bit chunks of instruction words (one aligned 32-bit instruction,
63
two 16-bit instructions or a mixture if 32-bit instructions are not aligned to 32-bit boundaries). The instruction
64
data is stored to a FIFO queue - the instruction prefetch buffer.
65 60 zero_gravi
 
66 66 zero_gravi
The back-end is responsible for the actual execution of the instruction. It includes an "issue engine",
67
which takes data from the instruction prefetch buffer and assembles 32-bit instruction words (plain 32-bit
68
instruction or decompressed 16-bit instructions) for execution.
69
 
70
Front-end and back-end operate in parallel and with overlapping operations. Hence, the optimal CPI
71
(cycles per instructions) is 2, but it can be significantly higher: for instance when executing loads/stores
72
(accessing memory-mapped devices with high latency), executing multi-cycle ALU operations (like divisions) or
73
when the CPU front-end has to reload the prefetch buffer due to a taken branch.
74
 
75 60 zero_gravi
Basically, the NEORV32 CPU is somewhere between a classical pipelined architecture, where each stage
76
requires exactly one processing cycle (if not stalled) and a classical multi-cycle architecture, which executes
77 66 zero_gravi
every single instruction (_including_ fetch) in a series of consecutive micro-operations. The combination of
78
these two classical design paradigms allows an increased instruction execution in contrast to a pure multi-cycle
79
approach (due to overlapping operation of fetch and execute) at a reduced hardware footprint (due to the
80
multi-cycle concept).
81 60 zero_gravi
 
82 66 zero_gravi
As a Von-Neumann machine, the CPU provides independent interfaces for instruction fetch and data access.
83
These two bus interfaces are merged into a single processor-internal bus via a prioritizing bus switch (data accesses
84
have higher priority). Hence, ALL memory locations including peripheral devices are mapped to a single unified 32-bit
85
address space.
86 60 zero_gravi
 
87
 
88
// ####################################################################################################################
89
:sectnums:
90 66 zero_gravi
=== Full Virtualization
91
 
92
Just like the RISC-V ISA the NEORV32 aims to provide _maximum virtualization_ capabilities on CPU _and_ SoC level to
93
allow a high standard of **execution safety**. The CPU supports **all** traps specified by the official RISC-V specifications.
94
footnote:[If the `Zicsr` CPU extension is enabled (implementing the full set of the privileged architecture).]
95
Thus, the CPU provides defined hardware fall-backs via traps for any expected and unexpected situation (e.g. executing an
96
malformed instruction word or accessing a not-allocated memory address). For any kind of trap the core is always in a
97
defined and fully synchronized state throughout the whole architecture (i.e. there are no out-of-order operations that
98
might have to reverted). This allows predictable execution behavior at any time improving overall _execution safety_.
99
 
100
**Execution Safety - NEORV32 Virtualization Features**
101
 
102
* Due to the acknowledged memory accesses the CPU is _always_ sync with the memory system
103
(i.e. there is no speculative execution / no out-of-order states).
104
* The CPU supports _all_ RISC-V compatible bus exceptions including access exceptions, which are triggered if an
105
accessed address does not respond or encounters an internal error during access.
106
* Accessed memory addresses (plain memory, but also memory-mapped devices) need to respond within a fixed time
107
window. Otherwise a bus access exception is raised.
108
* The RISC-V specs. state that executing an malformed instruction results in unpredictable behavior. As an additional
109
execution safety feature the NEORV32 CPU ensures that _all_ unimplemented/malformed/illegal instructions do raise an
110
illegal instruction exceptions and do not commit any state-changing operation (like writing registers or triggering
111
memory operations).
112
* To be continued...
113
 
114
 
115
// ####################################################################################################################
116
:sectnums:
117 60 zero_gravi
=== RISC-V Compatibility
118
 
119
The NEORV32 CPU passes the rv32_m/I, rv32_m/M, rv32_m/C, rv32_m/privilege, and
120
rv32_m/Zifencei tests of the official RISC-V Architecture Tests (GitHub). The port files for the
121 62 zero_gravi
NEORV32 processor are located in the repository's `sw/isa-test` folder.
122
 
123
[NOTE]
124
See section https://stnolting.github.io/neorv32/ug/#_risc_v_architecture_test_framework[User Guide: RISC-V Architecture Test Framework]
125 60 zero_gravi
for information how to run the tests on the NEORV32.
126
 
127
.**RISC-V `rv32_m/C` Tests**
128
...................................
129
Check cadd-01           ... OK
130
Check caddi-01          ... OK
131
Check caddi16sp-01      ... OK
132
Check caddi4spn-01      ... OK
133
Check cand-01           ... OK
134
Check candi-01          ... OK
135
Check cbeqz-01          ... OK
136
Check cbnez-01          ... OK
137
Check cebreak-01        ... OK
138
Check cj-01             ... OK
139
Check cjal-01           ... OK
140
Check cjalr-01          ... OK
141
Check cjr-01            ... OK
142
Check cli-01            ... OK
143
Check clui-01           ... OK
144
Check clw-01            ... OK
145
Check clwsp-01          ... OK
146
Check cmv-01            ... OK
147
Check cnop-01           ... OK
148
Check cor-01            ... OK
149
Check cslli-01          ... OK
150
Check csrai-01          ... OK
151
Check csrli-01          ... OK
152
Check csub-01           ... OK
153
Check csw-01            ... OK
154
Check cswsp-01          ... OK
155
Check cxor-01           ... OK
156
--------------------------------
157
OK: 27/27 RISCV_TARGET=neorv32 RISCV_DEVICE=C XLEN=32
158
...................................
159
 
160
.**RISC-V `rv32_m/I` Tests**
161
...................................
162
Check add-01            ... OK
163
Check addi-01           ... OK
164
Check and-01            ... OK
165
Check andi-01           ... OK
166
Check auipc-01          ... OK
167
Check beq-01            ... OK
168
Check bge-01            ... OK
169
Check bgeu-01           ... OK
170
Check blt-01            ... OK
171
Check bltu-01           ... OK
172
Check bne-01            ... OK
173
Check fence-01          ... OK
174
Check jal-01            ... OK
175
Check jalr-01           ... OK
176
Check lb-align-01       ... OK
177
Check lbu-align-01      ... OK
178
Check lh-align-01       ... OK
179
Check lhu-align-01      ... OK
180
Check lui-01            ... OK
181
Check lw-align-01       ... OK
182
Check or-01             ... OK
183
Check ori-01            ... OK
184
Check sb-align-01       ... OK
185
Check sh-align-01       ... OK
186
Check sll-01            ... OK
187
Check slli-01           ... OK
188
Check slt-01            ... OK
189
Check slti-01           ... OK
190
Check sltiu-01          ... OK
191
Check sltu-01           ... OK
192
Check sra-01            ... OK
193
Check srai-01           ... OK
194
Check srl-01            ... OK
195
Check srli-01           ... OK
196
Check sub-01            ... OK
197
Check sw-align-01       ... OK
198
Check xor-01            ... OK
199
Check xori-01           ... OK
200 70 zero_gravi
Check fence-01          ... OK
201 60 zero_gravi
--------------------------------
202 70 zero_gravi
OK: 39/39 RISCV_TARGET=neorv32 RISCV_DEVICE=I XLEN=32
203 60 zero_gravi
...................................
204
 
205
.**RISC-V `rv32_m/M` Tests**
206
...................................
207
Check div-01            ... OK
208
Check divu-01           ... OK
209
Check mul-01            ... OK
210
Check mulh-01           ... OK
211
Check mulhsu-01         ... OK
212
Check mulhu-01          ... OK
213
Check rem-01            ... OK
214
Check remu-01           ... OK
215
--------------------------------
216
OK: 8/8 RISCV_TARGET=neorv32 RISCV_DEVICE=M XLEN=32
217
...................................
218
 
219
.**RISC-V `rv32_m/privilege` Tests**
220
...................................
221
Check ebreak            ... OK
222
Check ecall             ... OK
223
Check misalign-beq-01   ... OK
224
Check misalign-bge-01   ... OK
225
Check misalign-bgeu-01  ... OK
226
Check misalign-blt-01   ... OK
227
Check misalign-bltu-01  ... OK
228
Check misalign-bne-01   ... OK
229
Check misalign-jal-01   ... OK
230
Check misalign-lh-01    ... OK
231
Check misalign-lhu-01   ... OK
232
Check misalign-lw-01    ... OK
233
Check misalign-sh-01    ... OK
234
Check misalign-sw-01    ... OK
235
Check misalign1-jalr-01 ... OK
236
Check misalign2-jalr-01 ... OK
237
--------------------------------
238
OK: 16/16 RISCV_TARGET=neorv32 RISCV_DEVICE=privilege XLEN=32
239
...................................
240
 
241
.**RISC-V `rv32_m/Zifencei` Tests**
242
...................................
243
Check Fencei            ... OK
244
--------------------------------
245
OK: 1/1 RISCV_TARGET=neorv32 RISCV_DEVICE=Zifencei XLEN=32
246
...................................
247
 
248
 
249
<<<
250
:sectnums:
251
==== RISC-V Incompatibility Issues and Limitations
252
 
253 64 zero_gravi
This list shows the currently identified issues regarding full RISC-V-compatibility. More specific information
254 60 zero_gravi
can be found in section <<_instruction_sets_and_extensions>>.
255
 
256 69 zero_gravi
.Read-Only "Read-Write" CSRs
257 60 zero_gravi
[IMPORTANT]
258 69 zero_gravi
The `misa` and `mtval` CSRs in the NEORV32 are _read-only_.
259
Any machine-mode write access to them is ignored and will _not_ cause any exceptions or side-effects to maintain
260
RISC-V compatibility.
261 60 zero_gravi
 
262 69 zero_gravi
.Physical Memory Protection
263 60 zero_gravi
[IMPORTANT]
264 70 zero_gravi
The physical memory protection (see section <<_machine_physical_memory_protection_csrs>>)
265 60 zero_gravi
only supports the modes _OFF_ and _NAPOT_ yet and a minimal granularity of 8 bytes per region.
266
 
267 69 zero_gravi
.Atomic Memory Operations
268 60 zero_gravi
[IMPORTANT]
269 64 zero_gravi
The `A` CPU extension only implements the `lr.w` and `sc.w` instructions yet.
270
However, these instructions are sufficient to emulate all further atomic memory operations.
271 60 zero_gravi
 
272 69 zero_gravi
.Bit-Manipulation ISA Extension
273 66 zero_gravi
[IMPORTANT]
274
The NEORV32 `B` extension only implements the _basic bit-manipulation instructions_ (`Zbb`) subset
275
and the _address generation instructions_ (`Zba`) subset yet.
276
 
277 60 zero_gravi
 
278 69 zero_gravi
 
279 60 zero_gravi
<<<
280
// ####################################################################################################################
281
:sectnums:
282
=== CPU Top Entity - Signals
283
 
284
The following table shows all interface signals of the CPU top entity `rtl/core/neorv32_cpu.vhd`. The
285
type of all signals is _std_ulogic_ or _std_ulogic_vector_, respectively. The "Dir." column shows the signal
286
direction seen from the CPU.
287
 
288
.NEORV32 CPU top entity signals
289
[cols="<2,^1,^1,<6"]
290
[options="header", grid="rows"]
291
|=======================
292
| Signal           | Width | Dir.   | Function
293
4+^| **Global Signals**
294
| `clk_i`          |     1 | in  | global clock line, all registers triggering on rising edge
295
| `rstn_i`         |     1 | in  | global reset, low-active
296
| `sleep_o`        |     1 | out | CPU is in sleep mode when set
297 69 zero_gravi
| `debug_o`        |     1 | out | CPU is in debug mode when set
298 60 zero_gravi
4+^| **Instruction Bus Interface (<<_bus_interface>>)**
299
| `i_bus_addr_o`   |    32 | out | destination address
300
| `i_bus_rdata_i`  |    32 | in  | read data
301
| `i_bus_wdata_o`  |    32 | out | write data (always zero)
302
| `i_bus_ben_o`    |     4 | out | byte enable
303
| `i_bus_we_o`     |     1 | out | write transaction (always zero)
304
| `i_bus_re_o`     |     1 | out | read transaction
305
| `i_bus_lock_o`   |     1 | out | exclusive access request (always zero)
306
| `i_bus_ack_i`    |     1 | in  | bus transfer acknowledge from accessed peripheral
307
| `i_bus_err_i`    |     1 | in  | bus transfer terminate from accessed peripheral
308
| `i_bus_fence_o`  |     1 | out | indicates an executed _fence.i_ instruction
309
| `i_bus_priv_o`   |     2 | out | current CPU privilege level
310
4+^| **Data Bus Interface (<<_bus_interface>>)**
311
| `d_bus_addr_o`   |    32 | out | destination address
312
| `d_bus_rdata_i`  |    32 | in  | read data
313
| `d_bus_wdata_o`  |    32 | out | write data
314
| `d_bus_ben_o`    |     4 | out | byte enable
315
| `d_bus_we_o`     |     1 | out | write transaction
316
| `d_bus_re_o`     |     1 | out | read transaction
317
| `d_bus_lock_o`   |     1 | out | exclusive access request
318
| `d_bus_ack_i`    |     1 | in  | bus transfer acknowledge from accessed peripheral
319
| `d_bus_err_i`    |     1 | in  | bus transfer terminate from accessed peripheral
320
| `d_bus_fence_o`  |     1 | out | indicates an executed _fence_ instruction
321
| `d_bus_priv_o`   |     2 | out | current CPU privilege level
322
4+^| **System Time (see <<_timeh>> CSR)**
323
| `time_i`         |    64 | in  | system time input (from MTIME)
324
4+^| **Interrupts, RISC-V-compatible (<<_traps_exceptions_and_interrupts>>)**
325
| `msw_irq_i`      |     1 | in  | RISC-V machine software interrupt
326
| `mext_irq_i`     |     1 | in  | RISC-V machine external interrupt
327
| `mtime_irq_i`    |     1 | in  | RISC-V machine timer interrupt
328
4+^| **Fast Interrupts, NEORV32-specific (<<_traps_exceptions_and_interrupts>>)**
329
| `firq_i`         |    16 | in  | fast interrupt request signals
330
4+^| **Enter Debug Mode Request (<<_on_chip_debugger_ocd>>)**
331
| `db_halt_req_i`  |     1 | in  | request CPU to halt and enter debug mode
332
|=======================
333
 
334
<<<
335
// ####################################################################################################################
336
:sectnums:
337
=== CPU Top Entity - Generics
338
 
339
Most of the CPU configuration generics are a subset of the actual Processor configuration generics (see section <<_processor_top_entity_generics>>).
340
and are not listed here. However, the CPU provides some _specific_ generics that are used to configure the CPU for the
341
NEORV32 processor setup. These generics are assigned by the processor setup only and are not available for user defined configuration.
342
The _specific_ generics are listed below.
343
 
344
[cols="4,4,2"]
345
[frame="all",grid="none"]
346
|======
347
| **CPU_BOOT_ADDR** | _std_ulogic_vector(31 downto 0)_ | 0x00000000
348
3+| This address defines the reset address at which the CPU starts fetching instructions after reset. In terms of the NEORV32 processor, this
349
generic is configured with the base address of the bootloader ROM (default) or with the base address of the processor-internal instruction
350 61 zero_gravi
memory (IMEM) if the bootloader is disabled (_INT_BOOTLOADER_EN_ = _false_). See section <<_address_space>> for more information.
351 60 zero_gravi
|======
352
 
353
[cols="4,4,2"]
354
[frame="all",grid="none"]
355
|======
356
| **CPU_DEBUG_ADDR** | _std_ulogic_vector(31 downto 0)_ | 0x00000000
357
3+| This address defines the entry address for the "execution based" on-chip debugger. By default, this generic is configured with the base address
358
of the debugger memory. See section <<_on_chip_debugger_ocd>> for more information.
359
|======
360
 
361
[cols="4,4,2"]
362
[frame="all",grid="none"]
363
|======
364
| **CPU_EXTENSION_RISCV_DEBUG** | _boolean_ | false
365
3+| Implement RISC-V-compatible "debug" CPU operation mode. See section <<_cpu_debug_mode>> for more information.
366
|======
367
 
368
 
369
<<<
370
// ####################################################################################################################
371
:sectnums:
372
=== Instruction Sets and Extensions
373
 
374 65 zero_gravi
The basic NEORV32 is a RISC-V `rv32i` architecture that provides several _optional_ RISC-V CPU and ISA
375 60 zero_gravi
(instruction set architecture) extensions. For more information regarding the RISC-V ISA extensions please
376 65 zero_gravi
see the the _RISC-V Instruction Set Manual - Volume I: Unprivileged ISA_ and _The RISC-V Instruction Set Manual
377 60 zero_gravi
Volume II: Privileged Architecture_, which are available in the projects `docs/references` folder.
378
 
379
[TIP]
380 63 zero_gravi
The CPU can discover available ISA extensions via the <<_misa>> CSR and the
381 64 zero_gravi
`CPU` <<_system_configuration_information_memory_sysinfo, SYSINFO>> register
382 63 zero_gravi
or by executing an instruction and checking for an _illegal instruction exception_.
383 60 zero_gravi
 
384 63 zero_gravi
[NOTE]
385 65 zero_gravi
Executing an instruction from an extension that is not supported yet or that is currently not enabled
386
(via the according top entity generic) will raise an _illegal instruction_ exception.
387 60 zero_gravi
 
388 63 zero_gravi
 
389 60 zero_gravi
==== **`A`** - Atomic Memory Access
390
 
391 65 zero_gravi
Atomic memory access instructions allow more sophisticated memory operations like implementing semaphores and mutexes.
392
The RICS-C specs. defines a specific _atomic_ extension that provides instructions for atomic memory accesses. The `A`
393
ISA extension is enabled if the `CPU_EXTENSION_RISCV_A` configuration generic is _true_.
394
In this case the following additional instructions are available:
395 60 zero_gravi
 
396
* `lr.w`: load-reservate
397
* `sc.w`: store-conditional
398
 
399
[NOTE]
400
Even though only `lr.w` and `sc.w` instructions are implemented yet, all further atomic operations
401
(load-modify-write instruction) can be emulated using these two instruction. Furthermore, the
402 65 zero_gravi
instruction's ordering flags (`aq` and `lr`) are ignored by the CPU hardware. Using any other (not yet
403
implemented) AMO (atomic memory operation) will raise an illegal instruction exception.
404 60 zero_gravi
 
405 65 zero_gravi
The *load-reservate* instruction behaves as a "normal" load-word instruction (`lw`) but will also set a CPU-internal
406
_data memory access lock_. Executing a *store-conditional* behaves as "normal" store-word instruction (`sw`) that will
407
only conduct an actual memory write operations if the lock is still intact. Additionally, the store-conditional instruction
408
will also return the lock state (returns zero if the lock is still intact or non-zero if the lock has been broken).
409
After the execution of the `sc` instruction, the lock is automatically removed.
410
 
411
The lock is broken if at least one of the following conditions occur:
412
. executing any data memory access instruction other than `lr.w`
413
. raising _any_ t (for example an interrupt or a memory access exception)
414
 
415 60 zero_gravi
[NOTE]
416
The atomic instructions have special requirements for memory system / bus interconnect. More
417
information can be found in sections <<_bus_interface>> and <<_processor_external_memory_interface_wishbone_axi4_lite>>, respectively.
418
 
419
 
420 66 zero_gravi
==== **`B`** - Bit-Manipulation Operations
421
 
422
The `B` ISA extension adds instructions for bit-manipulation operations. This extension is enabled if the
423
`CPU_EXTENSION_RISCV_B` configuration generic is _true_.
424
The official RISC-V specifications can be found here: https://github.com/riscv/riscv-bitmanip
425
 
426
[IMPORTANT]
427
The NEORV32 `B` extension only implements the _basic bit-manipulation instructions_ (`Zbb`) subset
428
and the _address generation instructions_ (`Zba`) subset yet.
429
 
430 70 zero_gravi
The `Zbb` sub-extension adds the following instructions:
431 66 zero_gravi
 
432 70 zero_gravi
* `andn` `orn` `xnor`
433
* `clz` `ctz` `cpop`
434
* `max` `maxu` `min` `minu`
435
* `sext.b` `sext.h` `zext.h`
436
* `rol` `ror` `rori`
437
* `orc.b` `rev8`
438 66 zero_gravi
 
439 70 zero_gravi
The `Zba` sub-extension adds the following instructions:
440 66 zero_gravi
 
441 70 zero_gravi
* `sh1add` `sh2add` `sh3add`
442 66 zero_gravi
 
443
[TIP]
444
By default, the bit-manipulation unit uses an _iterative_ approach to compute shift-related operations
445
like `clz` and `rol`. To increase performance (at the cost of additional hardware resources) the
446
<<_fast_shift_en>> generic can be enabled to implement full-parallel logic (like barrel shifters) for all
447
shift-related `B` instructions.
448
 
449
[WARNING]
450
The `B` extension is frozen but not officially ratified yet. There is no
451
software support for this extension in the upstream GCC RISC-V port yet. However, an
452
intrinsic library is provided to utilize the provided `B` extension features from C-language
453
code (see `sw/example/bitmanip_test`).
454
 
455
 
456 60 zero_gravi
==== **`C`** - Compressed Instructions
457
 
458 65 zero_gravi
The _compressed_ ISA extension provides 16-bit encodings of commonly used instructions to reduce code space size.
459
The `C` extension is available when the `CPU_EXTENSION_RISCV_C` configuration generic is _true_.
460
In this case the following instructions are available:
461 60 zero_gravi
 
462 70 zero_gravi
* `c.addi4spn` `c.lw` `c.sw` `c.nop` `c.addi` `c.jal` `c.li` `c.addi16sp` `c.lui` `c.srli` `c.srai` `c.andi` `c.sub`
463
`c.xor` `c.or` `c.and` `c.j` `c.beqz` `c.bnez` `c.slli` `c.lwsp` `c.jr` `c.mv` `c.ebreak` `c.jalr` `c.add` `c.swsp`
464 60 zero_gravi
 
465
[NOTE]
466 65 zero_gravi
When the compressed instructions extension is enabled, branches to an _unaligned_ and _uncompressed_ instruction require
467
an additional instruction fetch to load the according second half-word of that instruction. The performance can be increased
468 60 zero_gravi
again by forcing a 32-bit alignment of branch target addresses. By default, this is enforced via the GCC `-falign-functions=4`,
469
`-falign-labels=4`, `-falign-loops=4` and `-falign-jumps=4` compile flags (via the makefile).
470
 
471
 
472
==== **`E`** - Embedded CPU
473
 
474 65 zero_gravi
The embedded CPU extensions reduces the size of the general purpose register file from 32 entries to 16 entries to
475
decrease physical hardware requirements (for example block RAM). This extensions is enabled when the `CPU_EXTENSION_RISCV_E`
476
configuration generic is _true_. Accesses to registers beyond `x15` will raise and _illegal instruction exception_.
477
This extension does not add any additional instructions or features.
478 60 zero_gravi
 
479 70 zero_gravi
[NOTE]
480 63 zero_gravi
Due to the reduced register file size an alternate toolchain ABI (**`ilp32e`**) is required.
481 60 zero_gravi
 
482
 
483
==== **`I`** - Base Integer ISA
484 65 zero_gravi
 
485 60 zero_gravi
The CPU always supports the complete `rv32i` base integer instruction set. This base set is always enabled
486
regardless of the setting of the remaining exceptions. The base instruction set includes the following
487
instructions:
488
 
489 70 zero_gravi
* immediate: `lui` `auipc`
490
* jumps: `jal` `jalr`
491
* branches: `beq` `bne` `blt` `bge` `bltu` `bgeu`
492
* memory: `lb` `lh` `lw` `lbu` `lhu` `sb` `sh` `sw`
493
* alu: `addi` `slti` `sltiu` `xori` `ori` `andi` `slli` `srli` `srai` `add` `sub` `sll` `slt` `sltu` `xor` `srl` `sra` `or` `and`
494
* environment: `ecall` `ebreak` `fence`
495 60 zero_gravi
 
496
[NOTE]
497 70 zero_gravi
In order to keep the hardware footprint low, the CPU's shift unit uses a bit-serial approach. Hence, shift operations
498 61 zero_gravi
take up to 32 cycles (plus overhead) depending on the actual shift amount. Alternatively, the shift operations can be processed
499 70 zero_gravi
completely in parallel by a fast (but large) barrel shifter if the `FAST_SHIFT_EN` generic is _true_. In that case, shift operations
500 62 zero_gravi
complete within 2 cycles (plus overhead) regardless of the actual shift amount.
501 60 zero_gravi
 
502
[NOTE]
503
Internally, the `fence` instruction does not perform any operation inside the CPU. It only sets the
504 70 zero_gravi
top's `d_bus_fence_o` signal high for one cycle to inform the memory system a `fence` instruction has been
505 60 zero_gravi
executed. Any flags within the `fence` instruction word are ignore by the hardware.
506
 
507
 
508
==== **`M`** - Integer Multiplication and Division
509
 
510 65 zero_gravi
Hardware-accelerated integer multiplication and division operations are available when the
511 60 zero_gravi
`CPU_EXTENSION_RISCV_M` configuration generic is _true_. In this case the following instructions are
512
available:
513
 
514 70 zero_gravi
* multiplication: `mul` `mulh` `mulhsu` `mulhu`
515
* division: `div` `divu` `rem` `remu`
516 60 zero_gravi
 
517
[NOTE]
518
By default, multiplication and division operations are executed in a bit-serial approach.
519
Alternatively, the multiplier core can be implemented using DSP blocks if the `FAST_MUL_EN`
520
generic is _true_ allowing faster execution. Multiplications and divisions
521
always require a fixed amount of cycles to complete - regardless of the input operands.
522
 
523
 
524 61 zero_gravi
==== **`Zmmul`** - Integer Multiplication
525
 
526
This is a _sub-extension_ of the `M` ISA extension. It implements the multiplication-only operations
527 65 zero_gravi
of the `M` extensions and is intended for size-constrained setups that require hardware-based
528 61 zero_gravi
integer multiplications but not hardware-based divisions, which will be computed entirely in software.
529 65 zero_gravi
This extension requires only ~50% of the hardware utilization of the "full" `M` extension.
530 61 zero_gravi
 
531 70 zero_gravi
* multiplication: `mul` `mulh` `mulhsu` `mulhu`
532 61 zero_gravi
 
533 63 zero_gravi
If `Zmmul` is enabled, executing any division instruction from the `M` ISA extension (`div`, `divu`, `rem`, `remu`)
534
will raise an _illegal instruction exception_.
535 61 zero_gravi
 
536 63 zero_gravi
Note that `M` and `Zmmul` extensions _cannot_ be enabled at the same time.
537 61 zero_gravi
 
538
[TIP]
539
If your RISC-V GCC toolchain does not (yet) support the `_Zmmul` ISA extensions, it can be "emulated"
540
using a `rv32im` machine architecture and setting the `-mno-div` compiler flag
541 65 zero_gravi
(example `$ make MARCH=rv32im USER_FLAGS+=-mno-div clean_all exe`).
542 61 zero_gravi
 
543
 
544 60 zero_gravi
==== **`U`** - Less-Privileged User Mode
545
 
546 65 zero_gravi
In addition to the basic (and highest-privileged) machine-mode, the _user-mode_ ISA extensions adds a second less-privileged
547
operation mode. It is implemented if the `CPU_EXTENSION_RISCV_U` configuration generic is _true_.
548
Code executed in user-mode cannot access machine-mode CSRs. Furthermore, user-mode access to the address space (like
549
peripheral/IO devices) can be constrained via the physical memory protection (_PMP_).
550
Any kind of privilege rights violation will raise an exception to allow full virtualization.
551 60 zero_gravi
 
552
 
553
==== **`X`** - NEORV32-Specific (Custom) Extensions
554
 
555
The NEORV32-specific extensions are always enabled and are indicated by the set `X` bit in the `misa` CSR.
556
 
557 63 zero_gravi
The most important points of the NEORV32-specific extensions are:
558
* The CPU provides 16 _fast interrupt_ interrupts (`FIRQ)`, which are controlled via custom bits in the `mie`
559 69 zero_gravi
and `mip` CSR. This extension is mapped to CSR bits, that are available for custom use (according to the
560 60 zero_gravi
RISC-V specs). Also, custom trap codes for `mcause` are implemented.
561 63 zero_gravi
* All undefined/unimplemented/malformed/illegal instructions do raise an illegal instruction exception (see <<_full_virtualization>>).
562 60 zero_gravi
 
563
 
564 63 zero_gravi
==== **`Zfinx`** Single-Precision Floating-Point Operations
565 60 zero_gravi
 
566 65 zero_gravi
The `Zfinx` floating-point extension is an _alternative_ of the standard `F` floating-point ISA extension.
567
The `Zfinx` extensions also uses the integer register file `x` to store and operate on floating-point data
568
instead of a dedicated floating-point register file (hence, `F-in-x`). Thus, the `Zfinx` extension requires
569
less hardware resources and features faster context changes. This also implies that there are NO dedicated `f`
570
register file-related load/store or move instructions.
571
The official RISC-V specifications can be found here: https://github.com/riscv/riscv-zfinx
572 60 zero_gravi
 
573 70 zero_gravi
[NOTE]
574 60 zero_gravi
The NEORV32 floating-point unit used by the `Zfinx` extension is compatible to the _IEEE-754_ specifications.
575
 
576 65 zero_gravi
The `Zfinx` extensions only supports single-precision (`.s` instruction suffix), so it is a direct alternative
577
to the `F` extension. The `Zfinx` extension is implemented when the `CPU_EXTENSION_RISCV_Zfinx` configuration
578 60 zero_gravi
generic is _true_. In this case the following instructions and CSRs are available:
579
 
580 70 zero_gravi
* conversion: `fcvt.s.w` `fcvt.s.wu` `fcvt.w.s` `fcvt.wu.s`
581
* comparison: `fmin.s` `fmax.s` `feq.s` `flt.s` `fle.s`
582
* computational: `fadd.s` `fsub.s` `fmul.s`
583
* sign-injection: `fsgnj.s` `fsgnjn.s` `fsgnjx.s`
584 60 zero_gravi
* number classification: `fclass.s`
585
 
586 70 zero_gravi
* additional CSRs: `fcsr` `frm` `fflags`
587 60 zero_gravi
 
588
[WARNING]
589
Fused multiply-add instructions `f[n]m[add/sub].s` are not supported!
590
Division `fdiv.s` and square root `fsqrt.s` instructions are not supported yet!
591
 
592
[WARNING]
593 65 zero_gravi
Subnormal numbers ("de-normalized" numbers) are not supported by the NEORV32 FPU.
594
Subnormal numbers (exponent = 0) are _flushed to zero_ setting them to +/- 0 before entering the
595 60 zero_gravi
FPU's processing core. If a computational instruction (like `fmul.s`) generates a subnormal result, the
596
result is also flushed to zero during normalization.
597
 
598
[WARNING]
599
The `Zfinx` extension is not yet officially ratified, but is expected to stay unchanged. There is no
600
software support for the `Zfinx` extension in the upstream GCC RISC-V port yet. However, an
601
intrinsic library is provided to utilize the provided `Zfinx` floating-point extension from C-language
602
code (see `sw/example/floating_point_test`).
603
 
604 63 zero_gravi
 
605 60 zero_gravi
==== **`Zicsr`** Control and Status Register Access / Privileged Architecture
606
 
607 65 zero_gravi
The CSR access instructions as well as the exception and interrupt system (= the privileged architecture)
608
is implemented when the `CPU_EXTENSION_RISCV_Zicsr` configuration generic is _true_.
609 68 zero_gravi
 
610
[IMPORTANT]
611
If the `Zicsr` extension is disabled the CPU does not provide any _privileged architecture_ features at all!
612
In order to provide the full set of privileged functions that are required to run more complex tasks like
613 70 zero_gravi
operating system and to allow a secure execution environment the `Zicsr` extension should be always enabled.
614 68 zero_gravi
 
615 65 zero_gravi
In this case the following instructions are available:
616 60 zero_gravi
 
617 70 zero_gravi
* CSR access: `csrrw` `csrrs` `csrrc` `csrrwi` `csrrsi` `csrrci`
618
* environment: `mret` `wfi`
619 60 zero_gravi
 
620 68 zero_gravi
[NOTE]
621
If `rd=x0` for the `csrrw[i]` instructions there will be no actual read access to the according CSR.
622
However, access privileges are still enforced so these instruction variants _do_ cause side-effects
623
(the RISC-V spec. state that these combinations "_shall_ not cause any side-effects").
624 60 zero_gravi
 
625
[NOTE]
626 68 zero_gravi
The "wait for interrupt instruction" `wfi` acts like a sleep command. When executed, the CPU is
627 60 zero_gravi
halted until a valid interrupt request occurs. To wake up again, the according interrupt source has to
628
be enabled via the `mie` CSR and the global interrupt enable flag in `mstatus` has to be set.
629 65 zero_gravi
The `wfi` instruction may also be executed in user-mode without causing an exception as <<_mstatus>> bit
630 68 zero_gravi
`TW` (timeout wait) is _hardwired_ to zero.
631 60 zero_gravi
 
632 62 zero_gravi
 
633 66 zero_gravi
 
634
==== **`Zicntr`** CPU Base Counters
635
 
636
The `Zicntr` ISA extension adds the basic cycle `[m]cycle[h]`), instruction-retired (`[m]instret[h]`) and time (`time[h]`)
637
counters. This extensions is stated is _mandatory_ by the RISC-V spec. However, size-constrained setups may remove support for
638
these counters. Section <<_machine_counter_and_timer_csrs>> shows a list of all `Zicntr`-related CSRs.
639
These are available if the `Zicntr` ISA extensions is enabled via the <<_cpu_extension_riscv_zicntr>> generic.
640
 
641
[NOTE]
642
Disabling the `Zicntr` extension does not remove the `time[h]`-driving MTIME unit.
643
 
644
If `Zicntr` is disabled, all accesses to the according counter CSRs will raise an illegal instruction exception.
645
 
646
 
647
 
648
==== **`Zihpm`** Hardware Performance Monitors
649
 
650
In additions to the base cycle, instructions-retired and time counters the NEORV32 CPU provides
651
up to 29 hardware performance monitors (HPM 3..31), which can be used to benchmark applications. Each HPM consists of an
652
N-bit wide counter (split in a high-word 32-bit CSR and a low-word 32-bit CSR), where N is defined via the top's
653
`HPM_CNT_WIDTH` generic (0..64-bit) and a corresponding event configuration CSR. The event configuration
654
CSR defines the architectural events that lead to an increment of the associated HPM counter.
655
 
656
The HPM counters are available if the `Zihpm` ISA extensions is enabled via the <<_cpu_extension_riscv_zihpm>> generic.
657
 
658
Depending on the configuration the following additional CSR are available:
659
 
660
* counters: `mhpmcounter*[h]` (3..31, depending on `HPM_NUM_CNTS`)
661
* event configuration: `mhpmevent*` (3..31, depending on `HPM_NUM_CNTS`)
662
 
663
[IMPORTANT]
664
The HPM counter CSR can only be accessed in machine-mode. Hence, the according `mcounteren` CSR bits
665
are always zero and read-only. Any access from less-privileged modes will raise an illegal instruction
666
exception.
667
 
668
[TIP]
669
Auto-increment of the HPMs can be individually deactivated via the `mcountinhibit` CSR.
670
 
671
[TIP]
672
For a list of all HPM-related CSRs and all provided event configurations
673
see section <<_hardware_performance_monitors_hpm>>.
674
 
675
 
676 60 zero_gravi
==== **`Zifencei`** Instruction Stream Synchronization
677
 
678
The `Zifencei` CPU extension is implemented if the `CPU_EXTENSION_RISCV_Zifencei` configuration
679
generic is _true_. It allows manual synchronization of the instruction stream via the following instruction:
680
 
681
* `fence.i`
682
 
683 66 zero_gravi
The `fence.i` instruction resets the CPU's front-end (instruction fetch) and flushes the prefetch buffer.
684 64 zero_gravi
This allows a clean re-fetch of modified instructions from memory. Also, the top's `i_bus_fencei_o` signal is set
685
high for one cycle to inform the memory system (like the i-cache to perform a flush/reload.
686
Any additional flags within the `fence.i` instruction word are ignore by the hardware.
687 60 zero_gravi
 
688
 
689
==== **`PMP`** Physical Memory Protection
690
 
691 65 zero_gravi
The NEORV32 physical memory protection (PMP) is compatible to the RISC-V PMP specifications. It can be used
692
to constrain memory read/write/execute rights for each available privilege level.
693 60 zero_gravi
 
694 65 zero_gravi
The NEORV32 PMP only supports _NAPOT_ mode yet and a minimal region size (granularity) of 8 bytes. Larger
695
minimal sizes can be configured via the top `PMP_MIN_GRANULARITY` generic to reduce hardware requirements.
696
The physical memory protection system is implemented when the `PMP_NUM_REGIONS` configuration generic is >0.
697
In this case the following additional CSRs are available:
698
 
699 60 zero_gravi
* `pmpcfg*` (0..15, depending on configuration): PMP configuration registers
700
* `pmpaddr*` (0..63, depending on configuration): PMP address registers
701
 
702 65 zero_gravi
[TIP]
703 70 zero_gravi
See section <<_machine_physical_memory_protection_csrs>> for more information regarding the PMP CSRs.
704 60 zero_gravi
 
705
The actual number of regions and the minimal region granularity are defined via the top entity
706
`PMP_MIN_GRANULARITY` and `PMP_NUM_REGIONS` generics. `PMP_MIN_GRANULARITY` defines the minimal available
707
granularity of each region in bytes. `PMP_NUM_REGIONS` defines the total number of implemented regions and thus, the
708
number of available `pmpcfg*` and `pmpaddr*` CSRs.
709
 
710
When implementing more PMP regions that a _certain critical limit_ *an additional register stage
711
is automatically inserted* into the CPU's memory interfaces to reduce critical path length. Unfortunately, this will also
712
increase the latency of instruction fetches and data access by +1 cycle.
713
 
714
The critical limit can be adapted for custom use by a constant from the main VHDL package file
715
(`rtl/core/neorv32_package.vhd`). The default value is 8:
716
 
717
[source,vhdl]
718
----
719
-- "critical" number of PMP regions --
720
constant pmp_num_regions_critical_c : natural := 8;
721
----
722
 
723
**Operation**
724
 
725 65 zero_gravi
Any CPU memory access address (from the instruction fetch or data access interface) is tested if it is accessing _any_
726
of the specified  PMP regions(configured via `pmpaddr*` and enabled via `pmpcfg*`). If an
727
address matches one of these regions, the configured access rights (attributes in `pmpcfg*`) are enforced:
728 60 zero_gravi
 
729
* a write access (store) will fail if no write attribute is set
730
* a read access (load) will fail if no read attribute is set
731
* an instruction fetch access will fail if no execute attribute is set
732
 
733 65 zero_gravi
If an access to a protected region does not have the according access rights it will raise the according
734
instruction/load/store _access fault_ exception.
735 60 zero_gravi
 
736
By default, all PMP checks are enforced for user-level programs only. If you wish to enforce the physical
737 65 zero_gravi
memory protection also for machine-level programs you need to set the _locked bit_ in the according
738
`pmpcfg*` configuration CSR.
739 60 zero_gravi
 
740
[IMPORTANT]
741
After updating the address configuration registers `pmpaddr*` the system requires up to 33 cycles for
742
internal (iterative) computations before the configuration becomes valid.
743
 
744
[NOTE]
745
For more information regarding RISC-V physical memory protection see the official _The RISC-V
746 65 zero_gravi
Instruction Set Manual - Volume II: Privileged Architecture_ specifications.
747 60 zero_gravi
 
748
 
749
 
750
<<<
751
// ####################################################################################################################
752
:sectnums:
753
=== Instruction Timing
754
 
755
The instruction timing listed in the table below shows the required clock cycles for executing a certain
756
instruction. These instruction cycles assume a bus access without additional wait states and a filled
757
pipeline.
758
 
759
Average CPI (cycles per instructions) values for "real applications" like for executing the CoreMark benchmark for different CPU
760
configurations are presented in <<_cpu_performance>>.
761
 
762
.Clock cycles per instruction
763
[cols="<2,^1,^4,<3"]
764
[options="header", grid="rows"]
765
|=======================
766
| Class | ISA | Instruction(s) | Execution cycles
767
| ALU           | `I/E` | `addi` `slti` `sltiu` `xori` `ori` `andi` `add` `sub` `slt` `sltu` `xor` `or` `and` `lui` `auipc` | 2
768
| ALU           | `C`   | `c.addi4spn` `c.nop` `c.addi` `c.li` `c.addi16sp` `c.lui` `c.andi` `c.sub` `c.xor` `c.or` `c.and` `c.add` `c.mv` | 2
769
| ALU           | `I/E` | `slli` `srli` `srai` `sll` `srl` `sra` | 3 + SAfootnote:[Shift amount.]/4 + SA%4; FAST_SHIFTfootnote:[Barrel shift when `FAST_SHIFT_EN` is enabled.]: 4; TINY_SHIFTfootnote:[Serial shift when `TINY_SHIFT_EN` is enabled.]: 2..32
770 61 zero_gravi
| ALU           | `C`   | `c.srli` `c.srai` `c.slli` | 3 + SAfootnote:[Shift amount (0..31).]; FAST_SHIFTfootnote:[Barrel shifter when `FAST_SHIFT_EN` is enabled.]:
771 60 zero_gravi
| Branches      | `I/E` | `beq` `bne` `blt` `bge` `bltu` `bgeu` | Taken: 5 + MLfootnote:[Memory latency.]; Not taken: 3
772
| Branches      | `C`   | `c.beqz` `c.bnez`                     | Taken: 5 + MLfootnote:[Memory latency.]; Not taken: 3
773
| Jumps / Calls | `I/E` | `jal` `jalr`                  | 4 + ML
774
| Jumps / Calls | `C`   | `c.jal` `c.j` `c.jr` `c.jalr` | 4 + ML
775
| Memory access | `I/E` | `lb` `lh` `lw` `lbu` `lhu` `sb` `sh` `sw` | 4 + ML
776
| Memory access | `C`   | `c.lw` `c.sw` `c.lwsp` `c.swsp`           | 4 + ML
777
| Memory access | `A`   | `lr.w` `sc.w`                             | 4 + ML
778 69 zero_gravi
| Multiplication | `M`  | `mul` `mulh` `mulhsu` `mulhu` | 2+32+2; FAST_MULfootnote:[DSP-based multiplication; enabled via `FAST_MUL_EN`.]: 4
779
| Division       | `M`  | `div` `divu` `rem` `remu`     | 2+32+2
780 60 zero_gravi
| CSR access | `Zicsr` | `csrrw` `csrrs` `csrrc` `csrrwi` `csrrsi` `csrrci` | 4
781
| System | `I/E`+`Zicsr` | `ecall` `ebreak` | 4
782
| System | `I/E` | `fence` | 3
783
| System | `C`+`Zicsr` | `c.break` | 4
784
| System | `Zicsr` | `mret` `wfi` | 5
785 66 zero_gravi
| System | `Zifencei` | `fence.i` | 3 + ML
786 60 zero_gravi
| Floating-point - artihmetic | `Zfinx` | `fadd.s` | 110
787
| Floating-point - artihmetic | `Zfinx` | `fsub.s` | 112
788
| Floating-point - artihmetic | `Zfinx` | `fmul.s` | 22
789
| Floating-point - compare | `Zfinx` | `fmin.s` `fmax.s` `feq.s` `flt.s` `fle.s` | 13
790
| Floating-point - misc | `Zfinx` | `fsgnj.s` `fsgnjn.s` `fsgnjx.s` `fclass.s` | 12
791
| Floating-point - conversion | `Zfinx` | `fcvt.w.s` `fcvt.wu.s` | 47
792
| Floating-point - conversion | `Zfinx` | `fcvt.s.w` `fcvt.s.wu` | 48
793 66 zero_gravi
| Bit-manipulation - arithmetic/logic | `B(Zbb)` | `sext.b` `sext.h` `min` `minu` `max` `maxu` `andn` `orn` `xnor` `zext`(pack) `rev8`(grevi) `orc.b`(gorci) | 3
794
| Bit-manipulation - arithmetic/logic | `B(Zba)` | `sh1add` `sh2add` `sh3add` | 3
795
| Bit-manipulation - shifts | `B(Zbb)` | `clz` `ctz` | 3 + 0..32
796
| Bit-manipulation - shifts | `B(Zbb)` | `cpop` | 3 + 32
797
| Bit-manipulation - shifts | `B(Zbb)` | `rol` `ror` `rori` | 3 + SA
798
| Bit-manipulation - single-bit | `B(Zbs)` | `sbset[i]` `sbclr[i]` `sbinv[i]` `sbext[i]` | 3
799
| Bit-manipulation - shifted-add | `B(Zba)` | `sh1add` `sh2add` `sh3add` | 3
800 60 zero_gravi
|=======================
801
 
802
[NOTE]
803 65 zero_gravi
The presented values of the *floating-point execution cycles* are average values - obtained from
804 60 zero_gravi
4096 instruction executions using pseudo-random input values. The execution time for emulating the
805
instructions (using pure-software libraries) is ~17..140 times higher.
806
 
807
 
808 66 zero_gravi
<<<
809 60 zero_gravi
// ####################################################################################################################
810
include::cpu_csr.adoc[]
811
 
812
 
813
<<<
814
// ####################################################################################################################
815
:sectnums:
816
==== Traps, Exceptions and Interrupts
817
 
818 61 zero_gravi
In this document the following nomenclature regarding traps is used:
819 60 zero_gravi
 
820 64 zero_gravi
* _interrupts_ = asynchronous exceptions
821 60 zero_gravi
* _exceptions_ = synchronous exceptions
822
* _traps_ = exceptions + interrupts (synchronous or asynchronous exceptions)
823
 
824 61 zero_gravi
Whenever an exception or interrupt is triggered, the CPU transfers control to the address stored in `mtvec`
825
CSR. The cause of the according interrupt or exception can be determined via the content of `mcause`
826
CSR. The address that reflects the current program counter when a trap was taken is stored to `mepc` CSR.
827 70 zero_gravi
Additional information regarding the cause of the trap can be retrieved from `mtval` CSR and the processor's
828
<<_internal_bus_monitor_buskeeper>> (for memory access exceptions)
829 60 zero_gravi
 
830 70 zero_gravi
The traps are prioritized. If several _synchronous exceptions_ occur at once only the one with highest priority is triggered
831
while all remaining exceptions are ignored. If several _asynchronous exceptions_ (interrupts) trigger at once, the one with highest priority
832 64 zero_gravi
is serviced first while the remaining ones stay _pending_. After completing the interrupt handler the interrupt with
833 70 zero_gravi
the second highest priority will get serviced and so on until no further interrupts are pending.
834 60 zero_gravi
 
835 69 zero_gravi
.Interrupt Signal Requirements - Standard RISC-V Interrupts
836 61 zero_gravi
[IMPORTANT]
837 69 zero_gravi
All standard RISC-V interrupts request signals are **high-active**. A request has to stay at high-level (=asserted)
838 65 zero_gravi
until it is explicitly acknowledged by the CPU software (for example by writing to a specific memory-mapped register).
839 60 zero_gravi
 
840 69 zero_gravi
.Interrupt Signal Requirements - Fast Interrupt Requests
841
[IMPORTANT]
842 70 zero_gravi
The NEORV32-specific FIRQ request lines are triggered by a one-shot high-level (i.e. rising edge). Each request is buffered in the CPU control
843 69 zero_gravi
unit until the channel is either disabled (by clearing the according `mie` CSR bit) or the request is explicitly cleared (by setting
844
the according `mip` CSR bit).
845
 
846 61 zero_gravi
.Instruction Atomicity
847
[NOTE]
848 70 zero_gravi
All instructions execute as atomic operations - interrupts can only trigger _between_ two instructions.
849
So even if there is a permanent interrupt request, exactly one instruction from the interrupt program will be executed before
850
another interrupt handler can start. This allows program progress even if there are permanent interrupt requests.
851 60 zero_gravi
 
852
 
853 61 zero_gravi
:sectnums:
854 70 zero_gravi
==== Memory Access Exceptions
855 60 zero_gravi
 
856 61 zero_gravi
If a load operation causes any exception, the instruction's destination register is
857
_not written_ at all. Load exceptions caused by a misalignment or a physical memory protection fault do not
858 70 zero_gravi
trigger a bus/memory read-operation at all. Vice versa, exceptions caused by a store address misalignment or a store physical
859
memory protection fault do not trigger a bus/memory write-operation at all.
860 60 zero_gravi
 
861
 
862 61 zero_gravi
:sectnums:
863
==== Custom Fast Interrupt Request Lines
864 60 zero_gravi
 
865 61 zero_gravi
As a custom extension, the NEORV32 CPU features 16 fast interrupt request (FIRQ) lines via the `firq_i` CPU top
866 60 zero_gravi
entity signals. These interrupts have custom configuration and status flags in the `mie` and `mip` CSRs and also
867 65 zero_gravi
provide custom trap codes in `mcause`. These FIRQs are reserved for NEORV32 processor-internal usage only.
868 60 zero_gravi
 
869
 
870
 
871
<<<
872
// ####################################################################################################################
873 69 zero_gravi
:sectnums:
874
==== NEORV32 Trap Listing
875 60 zero_gravi
 
876 69 zero_gravi
The following table shows all traps that are currently supported by the NEORV32 CPU. It also shows the prioritization
877
and the CSR side-effects. A more detailed description of the actual trap triggering events is provided in a further table.
878
 
879
[NOTE]
880
_Asynchronous exceptions_ (= interrupts) set the MSB of `mcause` while _synchronous exception_ (= "software exception")
881
clear the MSB.
882
 
883
**Table Annotations**
884
 
885
The "Prio." column shows the priority of each trap. The highest priority is 1. The "`mcause`" column shows the
886
cause ID of the according trap that is written to `mcause` CSR. The "[RISC-V]" columns show the interrupt/exception code value from the
887 70 zero_gravi
official RISC-V privileged architecture manual. The "[C]" names are defined by the NEORV32 core library (the runtime environment _RTE_) and can
888 69 zero_gravi
be used in plain C code. The "`mepc`" and "`mtval`" columns show the value written to
889
`mepc` and `mtval` CSRs when a trap is triggered:
890
 
891
* _I-PC_ - address of interrupted instruction (instruction has not been execute/completed yet)
892
* _B-ADR_- bad memory access address that cause the trap
893
* _PC_ - address of instruction that caused the trap
894
* _0_ - zero
895
* _Inst_ - the faulting instruction itself
896
 
897
.NEORV32 Trap Listing
898 60 zero_gravi
[cols="3,6,5,14,11,4,4"]
899
[options="header",grid="rows"]
900
|=======================
901 64 zero_gravi
| Prio. | `mcause` | [RISC-V] | ID [C] | Cause | `mepc` | `mtval`
902
| 1  | `0x00000000` | 0.0  | _TRAP_CODE_I_MISALIGNED_ | instruction address misaligned | _B-ADR_ | _PC_
903
| 2  | `0x00000001` | 0.1  | _TRAP_CODE_I_ACCESS_     | instruction access fault | _B-ADR_ | _PC_
904
| 3  | `0x00000002` | 0.2  | _TRAP_CODE_I_ILLEGAL_    | illegal instruction | _PC_ | _Inst_
905
| 4  | `0x0000000B` | 0.11 | _TRAP_CODE_MENV_CALL_    | environment call from M-mode (`ecall` in machine-mode) | _PC_ | _PC_
906
| 5  | `0x00000008` | 0.8  | _TRAP_CODE_UENV_CALL_    | environment call from U-mode (`ecall` in user-mode) | _PC_ | _PC_
907 69 zero_gravi
| 6  | `0x00000003` | 0.3  | _TRAP_CODE_BREAKPOINT_   | breakpoint (`ebreak`) | _PC_ | _PC_
908 64 zero_gravi
| 7  | `0x00000006` | 0.6  | _TRAP_CODE_S_MISALIGNED_ | store address misaligned | _B-ADR_ | _B-ADR_
909
| 8  | `0x00000004` | 0.4  | _TRAP_CODE_L_MISALIGNED_ | load address misaligned | _B-ADR_ | _B-ADR_
910
| 9  | `0x00000007` | 0.7  | _TRAP_CODE_S_ACCESS_     | store access fault | _B-ADR_ | _B-ADR_
911
| 10 | `0x00000005` | 0.5  | _TRAP_CODE_L_ACCESS_     | load access fault | _B-ADR_ | _B-ADR_
912
| 11 | `0x80000010` | 1.16 | _TRAP_CODE_FIRQ_0_       | fast interrupt request channel 0 | _I-PC_ | _0_
913
| 12 | `0x80000011` | 1.17 | _TRAP_CODE_FIRQ_1_       | fast interrupt request channel 1 | _I-PC_ | _0_
914
| 13 | `0x80000012` | 1.18 | _TRAP_CODE_FIRQ_2_       | fast interrupt request channel 2 | _I-PC_ | _0_
915
| 14 | `0x80000013` | 1.19 | _TRAP_CODE_FIRQ_3_       | fast interrupt request channel 3 | _I-PC_ | _0_
916
| 15 | `0x80000014` | 1.20 | _TRAP_CODE_FIRQ_4_       | fast interrupt request channel 4 | _I-PC_ | _0_
917
| 16 | `0x80000015` | 1.21 | _TRAP_CODE_FIRQ_5_       | fast interrupt request channel 5 | _I-PC_ | _0_
918
| 17 | `0x80000016` | 1.22 | _TRAP_CODE_FIRQ_6_       | fast interrupt request channel 6 | _I-PC_ | _0_
919
| 18 | `0x80000017` | 1.23 | _TRAP_CODE_FIRQ_7_       | fast interrupt request channel 7 | _I-PC_ | _0_
920
| 19 | `0x80000018` | 1.24 | _TRAP_CODE_FIRQ_8_       | fast interrupt request channel 8 | _I-PC_ | _0_
921
| 20 | `0x80000019` | 1.25 | _TRAP_CODE_FIRQ_9_       | fast interrupt request channel 9 | _I-PC_ | _0_
922
| 21 | `0x8000001a` | 1.26 | _TRAP_CODE_FIRQ_10_      | fast interrupt request channel 10 | _I-PC_ | _0_
923
| 22 | `0x8000001b` | 1.27 | _TRAP_CODE_FIRQ_11_      | fast interrupt request channel 11 | _I-PC_ | _0_
924
| 23 | `0x8000001c` | 1.28 | _TRAP_CODE_FIRQ_12_      | fast interrupt request channel 12 | _I-PC_ | _0_
925
| 24 | `0x8000001d` | 1.29 | _TRAP_CODE_FIRQ_13_      | fast interrupt request channel 13 | _I-PC_ | _0_
926
| 25 | `0x8000001e` | 1.30 | _TRAP_CODE_FIRQ_14_      | fast interrupt request channel 14 | _I-PC_ | _0_
927
| 26 | `0x8000001f` | 1.31 | _TRAP_CODE_FIRQ_15_      | fast interrupt request channel 15 | _I-PC_ | _0_
928
| 27 | `0x8000000B` | 1.11 | _TRAP_CODE_MEI_          | machine external interrupt | _I-PC_ | _0_
929
| 28 | `0x80000003` | 1.3  | _TRAP_CODE_MSI_          | machine software interrupt | _I-PC_ | _0_
930
| 29 | `0x80000007` | 1.7  | _TRAP_CODE_MTI_          | machine timer interrupt | _I-PC_ | _0_
931 60 zero_gravi
|=======================
932
 
933
 
934 69 zero_gravi
The following table provides a summarized description of the actual events for triggering a specific trap.
935 60 zero_gravi
 
936 69 zero_gravi
.NEORV32 Trap Description
937
[cols="<3,<7"]
938
[options="header",grid="rows"]
939
|=======================
940 70 zero_gravi
| Trap ID [C] | Triggered when ...
941 69 zero_gravi
| _TRAP_CODE_I_MISALIGNED_ | fetching an 32-bit instruction word that is not 32-bit-aligned (_see note below!_)
942
| _TRAP_CODE_I_ACCESS_     | bus timeout or bus error during instruction word fetch
943
| _TRAP_CODE_I_ILLEGAL_    | trying to execute an invalid instruction word (malformed or not supported) or on a privilege violation
944
| _TRAP_CODE_MENV_CALL_    | executing `ecall` instruction in machine-mode
945
| _TRAP_CODE_UENV_CALL_    | executing `ecall` instruction in user-mode
946
| _TRAP_CODE_BREAKPOINT_   | executing `ebreak` instruction (or triggered by on-chip debugger)
947
| _TRAP_CODE_S_MISALIGNED_ | storing data to an address that is not naturally aligned to the data size (byte, half, word) being stored
948
| _TRAP_CODE_L_MISALIGNED_ | loading data from an address that is not naturally aligned to the data size  (byte, half, word) being loaded
949
| _TRAP_CODE_S_ACCESS_     | bus timeout or bus error during load data operation
950
| _TRAP_CODE_L_ACCESS_     | bus timeout or bus error during store data operation
951
| _TRAP_CODE_FIRQ_0_ ... _TRAP_CODE_FIRQ_15_| caused by interrupt-condition of processor-internal modules, see <<_neorv32_specific_fast_interrupt_requests>>
952
| _TRAP_CODE_MEI_          | user-defined processor-external source (via dedicated top-entity signal)
953
| _TRAP_CODE_MSI_          | user-defined processor-external source (via dedicated top-entity signal)
954
| _TRAP_CODE_MTI_          | processor-internal machine timer overflow OR user-defined processor-external source (via dedicated top-entity signal)
955
|=======================
956 60 zero_gravi
 
957 69 zero_gravi
.Instruction Address Misaligned Exception
958
[NOTE]
959
For 32-bit-only instructions (= no `C` extension) the misaligned instruction exception
960
is raised if bit 1 of the fetch address is set (i.e. not on a 32-bit boundary). If the `C` extension is implemented
961
there will never be a misaligned instruction exception _at all_.
962
In both cases bit 0 of the program counter (and all related registers) is hardwired to zero.
963 60 zero_gravi
 
964
 
965
<<<
966
// ####################################################################################################################
967
:sectnums:
968
==== Bus Interface
969
 
970
The CPU provides two independent bus interfaces: One for fetching instructions (`i_bus_*`) and one for
971
accessing data (`d_bus_*`) via load and store operations. Both interfaces use the same interface protocol.
972
 
973
:sectnums:
974
===== Address Space
975
 
976
The CPU is a 32-bit architecture with separated instruction and data interfaces making it a Harvard
977
Architecture. Each of this interfaces can access an address space of up to 2^32^ bytes (4GB). The memory
978
system is based on 32-bit words with a minimal granularity of 1 byte. Please note, that the NEORV32 CPU
979 65 zero_gravi
does not support unaligned memory accesses _in hardware_ - however, a software-based handling can be
980 60 zero_gravi
implemented as any unaligned memory access will trigger an according exception.
981
 
982
:sectnums:
983
===== Interface Signals
984
 
985
The following table shows the signals of the data and instruction interfaces seen from the CPU
986
(`*_o` signals are driven by the CPU / outputs, `*_i` signals are read by the CPU / inputs).
987
 
988
.CPU bus interface
989
[cols="<2,^1,<7"]
990
[options="header",grid="rows"]
991
|=======================
992
| Signal | Size | Function
993
| `bus_addr_o`   | 32 | access address
994
| `bus_rdata_i`  | 32 | data input for read operations
995
| `bus_wdata_o`  | 32 | data output for write operations
996
| `bus_ben_o`    | 4  | byte enable signal for write operations
997
| `bus_we_o`     | 1  | bus write access
998
| `bus_re_o`     | 1  | bus read access
999
| `bus_lock_o`   | 1  | exclusive access request
1000
| `bus_ack_i`    | 1  | accessed peripheral indicates a successful completion of the bus transaction
1001
| `bus_err_i`    | 1  | accessed peripheral indicates an error during the bus transaction
1002
| `bus_fence_o`  | 1  | this signal is set for one cycle when the CPU executes a data/instruction fence operation
1003
| `bus_priv_o`   | 2  | current CPU privilege level
1004
|=======================
1005
 
1006
[NOTE]
1007
Currently, there a no pipelined or overlapping operations implemented within the same bus interface.
1008 70 zero_gravi
So only a single transfer request can be "on the fly" (pending) at once.
1009 60 zero_gravi
 
1010
:sectnums:
1011
===== Protocol
1012
 
1013
A bus request is triggered either by the `bus_re_o` signal (for reading data) or by the `bus_we_o` signal (for
1014
writing data). These signals are active for exactly one cycle and initiate either a read or a write transaction. The transaction is
1015
completed when the accessed peripheral either sets the `bus_ack_i` signal (-> successful completion) or the
1016
`bus_err_i` signal is set (-> failed completion). All these control signals are only active (= high) for one
1017
single cycle. An error indicated via the `bus_err_i` signal during a transfer will trigger the according instruction bus
1018
access fault or load/store bus access fault exception.
1019
 
1020
[NOTE]
1021
The transfer can be completed directly in the same cycle as it was initiated (via the `bus_re_o` or `bus_we_o`
1022
signal) if the peripheral sets `bus_ack_i` or `bus_err_i` high for one cycle. However, in order to shorten the critical path such "asynchronous"
1023
completion should be avoided. The default processor-internal module provide exactly **one cycle delay** between initiation and completion of transfers.
1024
 
1025
.Bus Keeper: Processor-internal memories and memory-mapped devices with variable / high latency
1026
[IMPORTANT]
1027
Processor-internal peripherals or memories do not have to respond within one cycle after the transfer initiation (= latency > 1 cycle).
1028
However, the bus transaction has to be completed (= acknowledged) within a certain **response time window**. This time window is defined
1029
by the global `max_proc_int_response_time_c` constant (default = 15 cycles) from the processor's VHDL package file (`rtl/neorv32_package.vhd`).
1030
It defines the maximum number of cycles after which an _unacknowledged_ processor-internal bus transfer will timeout and raise a **bus fault exception**.
1031 66 zero_gravi
The _BUSKEEPER_ hardware module (see section <<_internal_bus_monitor_buskeeper>>) keeps track of all _internal_ bus transactions. If any bus operations times out
1032 60 zero_gravi
(for example when accessing "address space holes") this unit will issue a bus error to the CPU that will raise the according instruction fetch or data access bus exception.
1033
Note that **the bus keeper does not track external accesses via the external memory bus interface**. However, the external memory bus interface also provides
1034
an _optional_ bus timeout (see section <<_processor_external_memory_interface_wishbone_axi4_lite>>).
1035
 
1036
**Exemplary Bus Accesses**
1037
 
1038
.Example bus accesses: see read/write access description below
1039
[cols="^2,^2"]
1040
[grid="none"]
1041
|=======================
1042
a| image::cpu_interface_read_long.png[read,300,150]
1043
a| image::cpu_interface_write_long.png[write,300,150]
1044
| Read access | Write access
1045
|=======================
1046
 
1047
**Write Access**
1048
 
1049
For a write access, the accessed address (`bus_addr_o`), the data to be written (`bus_wdata_o`) and the byte
1050
enable signals (`bus_ben_o`) are set when bus_we_o goes high. These three signals are kept stable until the
1051
transaction is completed. In the example the accessed peripheral cannot answer directly in the next
1052
cycle after issuing. Here, the transaction is successful and the peripheral sets the `bus_ack_i` signal several
1053
cycles after issuing.
1054
 
1055
**Read Access**
1056
 
1057
For a read access, the accessed address (`bus_addr_o`) is set when `bus_re_o` goes high. The address is kept
1058
stable until the transaction is completed. In the example the accessed peripheral cannot answer
1059
directly in the next cycle after issuing. The peripheral hast to apply the read data right in the same cycle as
1060
the bus transaction is completed (here, the transaction is successful and the peripheral sets the `bus_ack_i`
1061
signal).
1062
 
1063
**Access Boundaries**
1064
 
1065
The instruction interface will always access memory on word (= 32-bit) boundaries even if fetching
1066
compressed (16-bit) instructions. The data interface can access memory on byte (= 8-bit), half-word (= 16-
1067
bit) and word (= 32-bit) boundaries.
1068
 
1069
**Exclusive (Atomic) Access**
1070
 
1071
The CPU can access memory in an exclusive manner by generating a load-reservate and store-conditional
1072
combination. Normally, these combinations should target the same memory address.
1073
 
1074
The CPU starts an exclusive access to memory via the _load-reservate instruction_ (`lr.w`). This instruction
1075
will set the CPU-internal _exclusive access lock_, which directly drives the `d_bus_lock_o`. It is the task of
1076
the memory system to manage this exclusive access reservation by storing the according access address and
1077
the source of the access itself (for example via the CPU ID in a multi-core system).
1078
 
1079
When the CPU executes a _store-conditional instruction_ (`sc.w`) the _CPU-internal exclusive access lock_ is
1080
evaluated to check if the exclusive access was successful. If the lock is still OK, the instruction will write-back
1081
zero and will allow the according store operation to the memory system. If the lock is broken, the
1082
instruction will write-back non-zero and will not generate an actual memory store operation.
1083
 
1084
The CPU-internal exclusive access lock is broken if at least one of the situations appear.
1085
 
1086
* when executing any other memory-access operation than `lr.w`
1087
* when any trap (sync. or async.) is triggered (for example to force a context switch)
1088
* when the memory system signals a bus error (via the `bus_err_i` signal)
1089
 
1090
[TIP]
1091
For more information regarding the SoC-level behavior and requirements of atomic operations see
1092
section <<_processor_external_memory_interface_wishbone_axi4_lite>>.
1093
 
1094
**Memory Barriers**
1095
 
1096
Whenever the CPU executes a fence instruction, the according interface signal is set high for one cycle
1097
(`d_bus_fence_o` for a _fence_ instruction; `i_bus_fence_o` for a _fencei_ instruction). It is the task of the
1098
memory system to perform the necessary operations (like a cache flush and refill).
1099
 
1100
 
1101
 
1102
<<<
1103
// ####################################################################################################################
1104
:sectnums:
1105
==== CPU Hardware Reset
1106
 
1107
In order to reduce routing constraints (and by this the actual hardware requirements), most uncritical
1108
registers of the NEORV32 CPU as well as most register of the whole NEORV32 Processor do not use **a
1109
dedicated hardware reset**. "Uncritical registers" in this context means that the initial value of these registers
1110
after power-up is not relevant for a defined CPU boot process.
1111
 
1112 70 zero_gravi
**Rationale**
1113 60 zero_gravi
 
1114
A good example to illustrate the concept of uncritical registers is a pipelined processing engine. Each stage
1115
of the engine features an N-bit _data register_ and a 1-bit _status register_. The status register is set when the
1116 66 zero_gravi
data in the according data register is valid. At the end of the pipeline the status register might trigger a write-back
1117 60 zero_gravi
of the processing result to some kind of memory. The initial status of the data registers after power-up is
1118
irrelevant as long as the status registers are all reset to a defined value that indicates there is no valid data in
1119 70 zero_gravi
the pipeline's data register. Therefore, the pipeline data register do no require a dedicated reset as they do not
1120 60 zero_gravi
control the actual operation (in contrast to the status register). This makes the pipeline data registers from
1121
this example "uncritical registers".
1122
 
1123
**NEORV32 CPU Reset**
1124
 
1125
In terms of the NEORV32 CPU, there are several pipeline registers, state machine registers and even status
1126
and control registers (CSRs) that do not require a defined initial state to ensure a correct boot process. The
1127 70 zero_gravi
pipeline register will get initialized by the CPU's internal state machines, which are initialized from the main
1128 60 zero_gravi
control engine that actually features a defined reset. The initialization of most of the CPU's core CSRs (like
1129
interrupt control) is done by the software (to be more specific, this is done by the `crt0.S` start-up code).
1130
 
1131
During the very early boot process (where `crt0.S` is running) there is no chance for undefined behavior due to
1132
the lack of dedicated hardware resets of certain CSRs. For example the machine interrupt-enable CSR (`mie`)
1133
does not provide a dedicated reset. The value after reset of this register is uncritical as interrupts cannot fire
1134 70 zero_gravi
because the global interrupt enabled flag in the status register (`mstatsus(mie)`) _do_ provide a dedicated
1135
hardware reset setting this bit to low (globally disabling interrupts).
1136 60 zero_gravi
 
1137
**Reset Configuration**
1138
 
1139 70 zero_gravi
Most CPU-internal register do provide an asynchronous reset in the VHDL code, but the "don't care" value
1140
(VHDL `'-'`) is used for initialization of all uncritical registers, effectively generating a flip-flop without a
1141 60 zero_gravi
reset. However, certain applications or situations (like advanced gate-level / timing simulations) might
1142 70 zero_gravi
require a more deterministic reset state. For this case, a defined reset level (reset-to-low) of all CPU registers can
1143
be enabled ba enabling a constant in the main VHDL package file (`rtl/core/neorv32_package.vhd`):
1144 60 zero_gravi
 
1145
[source,vhdl]
1146
----
1147
-- "critical" number of PMP regions --
1148 70 zero_gravi
constant dedicated_reset_c : boolean := false; -- use dedicated hardware reset value for UNCRITICAL registers (FALSE=reset value is irrelevant (might simplify HW), default; TRUE=defined LOW reset value)
1149 60 zero_gravi
----

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.