Subversion Repositories neorv32

## Project Change Log

The most recent version of the **NEORV32** project can be found at the top of this list.

"Stable releases" are linked and highlighted :rocket:.

The latest release is [![release](https://img.shields.io/github/v/release/stnolting/neorv32)](https://github.com/stnolting/neorv32/releases).

A list of all releases can be found [here](https://github.com/stnolting/neorv32/releases). The most recent version of the *NEORV32 data sheet*

can be found [here](https://raw.githubusercontent.com/stnolting/neorv32/master/docs/NEORV32.pdf) (pdf).

:information_source: To see a list of all commits between releases run `git log RELEASE_A..RELEASE_B` (example: `v1.4.7.0..v1.4.8.0`).

:information_source: The processor can determine it's version from the `mimpid` CSR (at CSR address 0xf13). A 8x4-bit BCD representation is used.

Leading zeros are optional. Example: `CSR(mimpid) = 0x01040312 => 01.04.03.12 = Version 01.04.03.12 = v1.4.3.12`. The version number is globally

defined by the `hw_version_c` constant in the main VHDL package file [`rtl/core/neorv32_package.vhd`](https://github.com/stnolting/neorv32/blob/master/rtl/core/neorv32_package.vhd).

### Version History

* :bug: = bug-fix

* :sparkles: = new feature

* :warning: = major change / important note

* :lock: = security issue

* :rocket: = release

| Date (*dd.mm.yyyy*) | Version | Comment |

|:----------:|:-------:|:--------|

| 10.05.2021 | [**:rocket:1.5.5.0**](https://github.com/stnolting/neorv32/releases/tag/v1.5.5.0) | **New release** |

| 10.05.2021 | 1.5.4.12 | :warning: `mip` CSR is now read-only (pending IRQs can be cleared by disabling (and re-enabling) the according `mie` bit), writing to `mip` will raise an illegal instruction exception; :sparkles: added non-maskable interrupt (NMI), top entity port `nm_irq_i`; added new NMI to NEORV32 runtime environment |

| 09.05.2021 | 1.5.4.11 | added new flags to `mzext` CSR: *CSR_MZEXT_PMP* (set if at least 1 PMP region is implemented at all), *CSR_MZEXT_HPM* (set if at least 1 HPM counter is implemented) |

| 03.05.2021 | 1.5.4.10 | minor code clean-ups; moved FIRQ synchronization registers to top, removed sync FFs for processor-internal sources; |

| 30.04.2021 | 1.5.4.9 | moved definitions of IO area from `crt0.S` to linker script; reworked CPU's CSR access system - highly reducing area overhead (removing decoding logic of not implemented CSRs by heavily using VHDL's `NULL` statement in `case` constructs) |

| 29.04.2021 | 1.5.4.8 | minor edits in CPU instruction fetch engine; reduced **processor-internal bus timeout** (`max_proc_int_response_time_c`) to 15 cycles; added flag to SYSINGO module (`SYSINFO_FEATURES_HW_RESET`) to check if a dedicated hardware reset of all core register is implemented (via package's `dedicated_reset_c` constant) |

| 28.04.2021 | 1.5.4.7 | :bug: fixed bug in instruction cache (iCACHE) when using two sets - `ICACHE_ASSOCIATIVITY` = 2: cache was corrupting the non-active set |

| 26.04.2021 | 1.5.4.6 | optimized CPU's instruction fetch unit: less overhead for branches, reduced unit's hardware complexity |

| 25.04.2021 | 1.5.4.5 | :sparkles: :warning: removed `cancel` signals from processor-internal bus system; removed CPU's internal bus access timeout counter; added new top generic: `MEM_EXT_TIMEOUT` - type `natural`, default = 255; used to configure optional auto-timeout of Wishbone interface (if an **external** device is not responding within `MEM_EXT_TIMEOUT` clock cycles); set to zero to disable auto-timeout (required to comply with AXI4-Lite specs. when using the top's AXI wrapper) |

| 25.04.2021 | 1.5.4.3 | :sparkles: converted NEORV32.pdf data sheet to [`asciidoc` using asciidoctor](https://asciidoctor.org/); added data sheet sources to [`docs/src_adoc`](https://github.com/stnolting/neorv32/blob/master/docs/src_adoc) |

| 21.04.2021 | 1.5.4.3 | :warning: :bug: reworked *atomic memory access* system due to conceptual design errors: new system will make atomic LR/SC combinations fail when there is a trap (like a context switch) between the two instructions; new system prohibits SC from writing to memory if exclusive access fails; removed top's `wb_tag_i` signal, pruned one bit of top's `wb_tag_o` signal (atomic access), added top's `wb_lock_o` signal; updated sections in NEORV32.pdf regarding atomic memory accesses |

| 19.04.2021 | 1.5.4.1 | added register stage to `MTIME.time` write access to improve timing closure |

| 17.04.2021 | [**:rocket:1.5.4.0**](https://github.com/stnolting/neorv32/releases/tag/v1.5.4.0) | **New release** |

| 16.04.2021 | 1.5.3.13 | :warning: added new top configuration generic `TINY_SHIFT_EN` (type = `boolean`, default = `false`) to configure a tiny single-bit (iterative) shifter for CPU ALU shift operations (for highly area-constrained setups) |

| 16.04.2021 | 1.5.3.12 | :sparkles: reworked reset system of the complete CPU: by default most registers (= "uncritical registers") **do not** provide an initialization via hardware reset; a **defined reset value** can be enabled by setting a constant from the main VHDL package (`rtl/core/neorv32_package.vhd`): `constant dedicated_reset_c : boolean := false;` (set `true` to enable CPU-wide dedicated register reset); see new section "2.11. CPU Hardware Reset" of NEORV32.pdf for more information |

| 14.04.2021 | 1.5.3.11 | minor rtl edits to allow synthesis using [`ghdl-yosys-plugin`](https://github.com/ghdl/ghdl-yosys-plugin) (:construction: work in progress :construction:)

| 13.04.2021 | 1.5.3.10 | :bug: fixed bug when configuring `HPM_CNT_WIDTH` less than 32; :warning: added new generic `CPU_CNT_WIDTH` to configure total size of CPU's `cycle` and `instret` CSRs (default = 64-bit); added `Zxnocnt` (no counters) and `Zxscnt` (small counters) flags to `mzexr` CSR to check if `CPU_CNT_WIDTH` is zero or less than 64, respectively; :bug: fixed bug in `crt0.S` start-up code: stack pointer has to be initialized before an exception can occur; updated `cpu_test` example program |

| 11.04.2021 | 1.5.3.9 | :warning: reworked CPU reset system (in addition to modifications in v1.5.3.7): default reset: most register are "initialized" with '-' (don't care) since no real reset is required; however, a "real" reset can be configured using the packages 'def_rst_val_c' constant that defines the reset value for all "uncritical regsiter" (see new NEORV32.pdf section 2.11 "CPU Hardware Reset") |

| 09.04.2021 | 1.5.3.8 | optimized CPU control: register write back during multi-cycle ALU operation only when result is really available (reducing switching activity; avoids possible source operand corruption); optimized `M` extension's co-processor: multiplications and divisions are 2 cycles faster |

| 08.04.2021 | 1.5.3.7 | :bug: fixed bug in HPM event configuration via `mhpmevent*` CSRs - there was a CSR address decoding overlap between the HPM event CSRs and the machine trap setup CSRs (introduced in version 1.5.3.6); :warning: reworked CPU core CSRs: most CSRs are not reset by hardware and need explicit initialization (done by crt0.S start-up code) |

| 02.04.2021 | 1.5.3.6 | :bug: fixed bug in external memory interface (`neorv32_wishbone.vhd`) that caused bus exceptions when using external memories with very high access latencies (race condition in bus timeouts); VHDL code clean-up |

| 30.03.2021 | 1.5.3.5 | added new top's generic `HPM_CNT_WIDTH` (type `natural`, default=40) to configure the total bit width of the hardware performance monitors (HPM) counter (min 1, max 64); mofified `crt0.S`: stops all counters (incl. HPMs), no user-level access to ANY counter; `neorv32.h`: added missing `mcounteren` and `mcountinhibit` CSR bit definitions |

| 28.03.2021 | 1.5.3.4 | default "test setup" `rtl/top_templetes/neorv32_test_setup.vhd`: disabled PMP, implementing 4 HPM counters; :sparkles: added [`boards`](https://github.com/stnolting/neorv32/tree/master/boards) folder for exemplary FPGA setups |

| 27.03.2021 | 1.5.3.3 | minor optimization in CPU control engine; FPU comparator now uses comparatos results from main ALU (reduces FPU hardware footprint) |

| 26.03.2021 | 1.5.3.2 | :sparkles: **added single-precision floating-point unit (FPU)** `rtl/core/neorv32_cpu_cp_fpu.vhd` implementing the `Zfinx` CPU extension; added/updated `Zfinx` **intrinsic library** and verification framework: [`sw/example/floating_point_test`](https://github.com/stnolting/neorv32/tree/master/sw/example/floating_point_test); added co-processor timeout counter to CPU to auto-terminate co-processor operations (for debugging only; defaullt=deactivated) |

| 25.03.2021 | 1.5.3.1 | :bug: fixed bug in invalid floating-point instruction detection (caused CPU to stall if executing an invalid floating-point operation); intrinsic core library (mainly used for not-yet-supported CPU extensions like `B` and `Zfinx`): clean-up, added R3 instruction type |

| 24.03.2021 | [**:rocket:1.5.3.0**](https://github.com/stnolting/neorv32/releases/tag/v1.5.3.0) | **New release** |

| 23.03.2021 | 1.5.2.9 | :sparkles: added new top generic to enable single-precision floating-point extensions `Zfinx`: `CPU_EXTENSION_RISCV_Zfinx` - type `boolean`, default = `false`, :warning: **extension is not yet operational!**; bootloader now shows available `Z*` extensions (from `mzext` CSR) like `Zifencei` |

| 21.03.2021 | 1.5.2.8 | :bug: fixed problem with linking `math.h` library in makefile; added floating-point-related global definitions to main VHDL package; added intrinsic core library file `sw/lib/include/neorv32_intrinsics.h` - intrinsic library support for CPU extensions, which are not yet supported by the upstream GCC, are based on this |

| 18.03.2021 | 1.5.2.7 | :bug: fixed bug in `sw/common/crt0.S` dummy exception handler (wrong order of register push/pop); changed upcoming floating-point extension (orignally `F` extension) to `Zfinx` extension (-> [RISC-V `Zfinx` spec](https://github.com/riscv/riscv-zfinx)) - updated CPU infrastructure |

| 16.03.2021 | 1.5.2.6 | reworked atomic/exclusive memory access interface: removed CPU's `d_bus_lock_o` and `i_bus_lock_o` signal (was always zero anyway); removed top's `wb_lock_o` signal; added *exclusive access request* to Wishbone tag signal `wb_tag_o` (is now one bit wider); added more details to NEORV32.pdf regarding excluisve/atomic memory accesses (interface/protocol) |

| 09.03.2021 | 1.5.2.5 | added bit-manipulation `Zba` sub-extension (shifted-adds: `SH1ADD` `SH2ADD` `SH3ADD`) |

| 07.03.2021 | 1.5.2.4 | :sparkles: added new IO/peripheral module: **Smart LED Interface (NEOLED)** to interface intelligent LEDs (WS2812/WS2811/NeoPixel(c) compatible; supports RGB and RGBW LEDs in *parallel*) with internal TX buffer; new top generics: `IO_NEOLED_EN`: implement NEOLED interface when *true*; new top signals: `neoled_o`: single-wire async. serial data interface; FIFO re-fill interrupt via *fast interrupt request channel 9* `FIRQ9`; added new "NEOLED" section to data sheet; added SW driver library and simple NEOLED example program (`sw/example/demo_neopixel`) |

| 06.03.2021 | 1.5.2.3 | clean-up of CPU control code: fixed minor bug in F-exension's instruction decoding; changed coding style for CSR write access (old version might have caused "inferring latch..." warning in Intel Quartus); fixed default values for CSRs when according extensions are disabled |

| 04.03.2021 | 1.5.2.2 | added two new generics to configure CFS IO conduit sizes (implementing [issue #13](https://github.com/stnolting/neorv32/issues/13)): `IO_CFS_IN_SIZE` - type: `positive`, configures the size of `cfs_in_i` signal; `IO_CFS_OUT_SIZE` - type: `positive`, configures the size of `cfs_out_o` signal; minor edits to floating-point CPU infrastructure |

| 03.03.2021 | 1.5.2.1 | added CPU core infrastructure for *upcoming* single-precision floating-point extension `F`; :warning: **floating-point extension is NOT OPERATIONAL YET!** added new rtl file for the floating-point unit `rtl/core/neorv32_cpu_cp_fpu.vhd` (blank template!) |

| 01.03.2021 | [**:rocket:1.5.2.0**](https://github.com/stnolting/neorv32/releases/tag/v1.5.2.0) | **New release** |

| 27.02.2021 | 1.5.1.11 | :bug: fixed several small bugs in *bitmanipulation extension* instruction decoding (not all `B` instructions triggered and *illegal instruction exception* when B-extension = disabled) |

| 25.02.2021 | 1.5.1.10 | :bug: fixed bugs in UART RTS/CTS hardware control flow - the new setup was verified on real hardware; added double-buffering to UART RX engine |

| 24.02.2021 | 1.5.1.9 | `mcounteren` CSR is hardwired to zero if user mode is not implemented (`CPU_EXTENSION_RISCV_U` = false); added `Zbs` (single-bit operations) sub-extension to bitmanipulation unit |

| 22.02.2021 | 1.5.1.8 | added programmable *RTS/CTS hardware flow control* to UARTs; new top signals: `uart0_rts_o`, `uart0_cts_i`, `uart1_rts_o`, `uart1_cts_i`; UART.TX engine will only start sending (if `CTS` flow control is activated) if `uart*_cts_i` is asserted (low-active); UART.RX engine signals (if `RTS` flow control is activated) via `uart*_rts_o` if it is ready to receive new data (low-active); added hw flow control parameter to uart setup functions `neorv32_uart*_setup()` |

| 20.02.2021 | 1.5.1.7 | removed `err_o` signal from custom functions subsystem `CFS`; processor *SoC fast interrupt input* `soc_firq_i` reduced to 6 channels (was 8) - mapped to CPU's `FIRQ_10` - `FIRQ_15`; added individual fast IRQs for `UART1` "RX complete" and "TX complete" conditions (-> `FIRQ_4` & `FIRQ_5`); changed FIRQ channels of TWI/SPI/GPIO interrupts |

| 18.02.2021 | 1.5.1.6 | added register buffer for enable signals to processor-internal clock generator; :bug: fixed bug in `sw/example/demo_twi` program: TWI clock speed messsage was wrong (factor 1/4 was missing) |

| 17.02.2021 | 1.5.1.5 | added a second independent UART: new UART is *secondary UART* `UART0`, the "old" UART is now the *primary UART* `UART0`; by default the **primary UART (UART0) is used for all user interface connection**; reworked *fast interrupt* `FIRQ` assignment/priority list - added UART1 RTX (receive *or* send done) fast interrupt; added hardware driver functions for new `UART1` - the "old" `neorv32_uart_*` function calls will map to the primary UART `UART0` for compatibility; renamed compiler flag to enable UART "simulation mode": `UART_SIM_MODE` -> `UART0_SIM_MODE` for primary UART, `UART1_SIM_MODE` for secondary UART (`UART_SIM_MODE` is still supported for compatibility and maps to `UART0_SIM_MODE`); added second simulation UART receiver for `UART1` to testbench; renamed UART simulation output files: `neorv32.testbench_uart.out` -> `neorv32.testbench_uart0.out` (testbench UART0 receiver), new: `neorv32.testbench_uart1.out` (testbench UART1 receiver), `neorv32.uart.sim_mode.text.out` and `neorv32.uart.sim_mode.data.out` -> `neorv32.uart0.sim_mode.text.out` and `neorv32.uart0.sim_mode.data.out` (for `UART0`), new `neorv32.uart1.sim_mode.text.out` and `neorv32.uart1.sim_mode.data.out` (for `UART1`) |

| 13.02.2021 | 1.5.1.4 | `HW_THREAD_ID` generic is now of type `natural`; `mret` instruction now requires an additional cycle to execute; logic optimization of CPU's control logic -> smaller hardware footprint and higher f_max; updated CPU synthesis results; removed top module's generic initialization using `(others => '0')` (targeting [issue #8](https://github.com/stnolting/neorv32/issues/8)) |

| 09.02.2021 | 1.5.1.3 | modified CPU architecture: now using a "pseudo" ALU co-processor to get the result of a CSR read operation into data path, removing one input from register file input mux -> shorter critical path |

| 08.02.2021 | 1.5.1.2 | added new peripheral/IO module: **Numerically-Controlled Oscillator `NCO`**: three independent channels, 20-bit phase accu, 20-bit tuning word, fixed 50% duty cycle mode or pulsed mode; added according HW drivers and example program |

| 07.02.2021 | [**:rocket:1.5.1.0**](https://github.com/stnolting/neorv32/releases/tag/v1.5.1.0) | **New release** |

| 05.02.2021 | 1.5.0.11 | :bug: fixed error in atomic instruction `LR.W` |

| 05.02.2021 | 1.5.0.10 | CPU now provides 16 fast interrupt request lines (`FIRQ0 .. FIRQ15`) with according `mie`/`mip` CSR bits and `mcause` trap codes; removed IRQ enable flags from SPI, UART & TWI; reworked processor-internal interrupt system - assignment/priority list; UART now features individual IRQs for "RX-done" and "TX-done" conditions; changed bit order in TWI control register |

| 29.01.2021 | 1.5.0.9 | removed custom function units `CFU0` & `CFU1`; :sparkles: replaced them by new *Custom Functions Subsystem `CFS`*, which provides up to 32x32-bit memory-mapped registers; new configuration generics: `IO_CFS_EN`, `IO_CFS_CONFIG`; new top entity signals: `cfs_in_i`, `cfs_out_o`; increased processor's IO area from 128 bytes to 256 bytes, now starting at `0xFFFFFF00` |

| 28.01.2021 | 1.5.0.8 | added *critical limit* for number of implemented PMP regions: When implementing more PMP regions that a certain critical limit an additional register stage is automatically inserted into the CPU’s memory interfaces increasing the latency of instruction fetches and data access by +1 cycle. The critical limit can be adapted for custom use by a constant from the main VHDL package file (rtl/core/neorv32_package.vhd). The default value is 8: `constant pmp_num_regions_critical_c : natural := 8;` |

| 27.01.2021 | 1.5.0.7 | added four additional *fast interrupt* channels `FIRQ4..7`, available via processor's top `soc_firq_i(3:0)` signal for custom platform use; fixed minor error in UART setup function (baud rate prescaler calculation for very high baud rates) |

| 26.01.2021 | 1.5.0.6 | minor logic optimization of CPU's `B` extension co-processor (reducing area); minor logic optimization or `HPM` triggers (reducing area); reworked CPU's co-processor interface; minor logic optimiztation of branch condition check (to shorten critical path) |

| 23.01.2021 | 1.5.0.5 | reworked true random number generator `TRNG`: architecture is now based on several simple ring oscillators with incrementing length; changed control register bits; updated according driver functions and demo program |

| 22.01.2021 | 1.5.0.4 | :bug: fixed BUG in bootloader (that caused it to immediately crash after reset if SPI/MTIME/GPIO peripherals were not implemented); reworked watchdog timer `WDT`: removed watchdog access password, added option to lock configuration until next system reset, changed control register bits - updated driver functions and demo/test programs |

| 17.01.2021 | 1.5.0.3 | CPU data register file can now be mapped to a **single** "true dual-port" block RAM by the synthesizer (requiring only 1024 memory bits instead of 2048); :bug: fixed typo error in `sim/rtl_modules/neorv32_imem.vhd`; modified `M` co-processor (due to register file read access modification), reduced switching activity when co-processor is idle; logic/arithmetic operations of `B` extension only require 3 cycles now, reduced switching activity when co-processor is idle |

| 15.01.2021 | 1.5.0.2 | added instruction cache associativity configuration (number of sets); new configuration generic: `ICACHE_ASSOCIATIVITY` -> number of sets (1 = direct mapped, 2 = 2-way set-associative), has to be a power of two; if associativity is > 1 the used replacement policy is *least recently used (LRU)*; :bug: fixed bug in `sw/lib/source/neorv32_cpu.c` PMP.CFG configuration function |

| 14.01.2021 | 1.5.0.1 | added new HPM trigger event: multi-cycle ALU operation wait cycle (`HPMCNT_EVENT_WAIT_MC`); renamed `neorv32_cache.vhd` -> `neorv32_icache.vhd` |

| 10.01.2021 | [**:rocket:1.5.0.0**](https://github.com/stnolting/neorv32/releases/tag/v1.5.0.0) | Renamed configuration generics: `*_USE` -> `*_EN` |

| 10.01.2021 | 1.4.9.10 | :sparkles: Added support for [**bit manipulation extension (`B`)**](https://github.com/riscv/riscv-bitmanip) - base subset `Zbb` only (:warning: RISC-V `B` (sub-)extensions are not officially ratified yet; compatible to version "0.94-draft"); enabled via new configuration constant `CPU_EXTENSION_RISCV_B` (default = false); uported `Zbb` instructions: `CLZ` `CTZ` `CPOP` `SEXT.B` `SEXT.H` `MIN[U]` `MAX[U]` `ANDN` `ORN` `XNOR` `ROL` `ROR` `RORI` `zext`(*pseudo-instruction* for `PACK rd, rs, zero`) `rev8`(*pseudo-instruction* for `GREVI rd, rs, -8`) `orc.b`(*pseudo-instruction* for `GORCI rd, rs, 7`); added `B` flag to `misa` CSR; added `Zbb` flag to `mzext` CSR |

| 03.01.2021 | 1.4.9.8 | Added HPM trigger for instruction issue wait cycle (caused by pipeline flush); all HPM counters do not increment if CPU is sleep mode; fixed CoreMark timer overflow issues; `rtl/core/neorv32_busswitch.vhd`: removed wait states, less load/store wait cycles -> faster execution; updated CoreMark results |

| 02.01.2021 | 1.4.9.7 | :sparkles: added RISC-V hardware performance monitors (`HPM`); new CSRs: `mhpmevent*`(3..31), `[m]hpmcounter*[h]`(3..31), amount configurable via top's generic `HPM_NUM_CNTS`; supported counter events: active cycle, retired instruction, retired compressed instruction, instruction fetcch memory wait cycle, load operation, store operation, load/store memory wait cycle, unconditional jump, conditional branche (all), conditional taken branch, entered trap, illegal instruction exception; PMP can now have up to 64 regions; number of regions configured via top's `PMP_NUM_REGIONS` generic; removed obsolete top's `PMP_USE` generic; removed PMP flag from `mzext` CSR; minimal region granularity (in bytes) configured via top's `PMP_MIN_GRANULARITY` generic, has to be a power of two and >= 8 bytes; :bug: fixed bug in sleep (`wfi`) instruction |

| 29.12.2020 | 1.4.9.5 | New UART features: "frame check" (test if stop bit is set), error indicated via `UART_DATA` reg's `UART_DATA_FERR` flag; configurable parity bit (`UART_CT.UART_CT_PMODE1:UART_CT_PMODE0`, 00=no parity; 10=even parity; 11=odd parity); parity error indicated via `UART_DATA` reg's `UART_DATA_PERR` flag; moved UART's RX overrun flag to `UART_DATA.UART_DATA_OVERR` |

| 26.12.2020 | 1.4.9.4 | removed `zicnt_en` option (was used to discard the standard RISC-V counters and timers from implementation); added missing `mcounteren` CSR (to allow read-access from user-level code to `cycle[h]` / `time[h]` / `[m]instret[h]` CSRs); available bits: 0: `CY`, 1: `TM`, 2: `IR`; added missing `mcountinhibit` CSR (to disable auto-increment of `[m]cycle[h]` / `[m]instret[h]` CSRs); available bits: 0: `CY`, 2: `IR`; :warning: renamed CSR bits C-code-aliases: `CPU_*` -> `CSR_*` |

| 25.12.2020 | 1.4.9.3 | Added missing `UBE` flag to `mstatus` CSR, indicates Endianness for load/stores in user mode (always set indicating BIG-endian mode), is a copy of `mstatush.mbe` |

| 23.12.2020 | 1.4.9.2 | :sparkles: added processor-internal instruction cache `rtl/core/neorv32_cache.vhd` (direct mapped); new configuration generics: `ICACHE_USE` (implement cache), `ICACHE_BLOCK_SIZE` (cache block/page/line size), `ICACHE_NUM_BLOCKS` (number of cache blocks); added `SYSINFO_CACHE` register to SYSINFO to check cache configuration by software  |

| 20.12.2020 | 1.4.9.1 | :bug: fixed bug in CPU's instruction fetch engine (alignment_errros/bus_errors were not acknowledged correctly); added `BUS_TIMEOUT` generic to CPU (defines the amount of cycles after which an *unacknowledged* bus access will get terminated and raises a bus access fault exception) |

| 19.12.2020 | [**:rocket:1.4.9.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.9.0) | Testbench: added memory-mapped triggers to trigger core's "machine software & external interrupts"; `sw/example/cpu_test`: removed CFU tests, added `MEI` and `MSI` tests; added **RISC-V-Compliance Test Framework** to repository (`riscv-compliance/`), core passes all `rv32` tests (riscv-compliance v2.1) |

| 18.12.2020 | 1.4.8.13 | Added additional simulation files: simulation-optimized IMEM-ROM (so far, this is only relevant for the *new* NEORV32 RISC-V Compliance test framework v2.0); **:sparkles: Processor now passes all `rv32` tests of the new [RISC-V Compliance Test Framework v2.0](https://github.com/riscv/riscv-compliance/releases/tag/v2.0) :sparkles:** |

| 16.12.2020 | 1.4.8.12 | :warning: fixed (another) bug in `mtval` CSR generation (wrong value for "breakpoint" trap); updated `mtval` value table in data sheet; fixed bug in load/store operation (intoroduced in version 1.4.8.10) |

| 16.12.2020 | 1.4.8.11 | :warning: fixed bug in `mtval` CSR generation (wrong values for some traps); fixed bug in `mip` CSR (writing zero to implemented bits now actually clears pending interrupts); fixed bug in IRQ priority encoding (machine software interrupt `MSI` comes before machine timer interrupt `MTI`) |

| 12.12.2020 | 1.4.8.10 | :warning: fixed wrong `trap_reset_c` encoding (in it's expanded form it should be 0x80000000) and reset logic: hardware `mcause` register is now set to `trap_reset_c` after a hardware reset; crt0.S start-up code now sets `mcause` to `trap_reset_c` after finishing hardware setup |

| 11.12.2020 | 1.4.8.9 | Added option to exclude standard RISC-V performance counters (`[m]cycle[h]` and `[m]instret[h]`) for size-constrained implementations; disabled by setting VHDL package's `zicnt_en_c` constant to false; software can determine state of `zicnt_en_c` via `mzext` CSR's `CPU_MZEXT_ZICNT` bit; added new signal to processor top entity: `mtime_i`, this signal is used for updateting the `time[h]` CSRs if the processor-internal MTIME unit is disabled (via `IO_MTIME_USE` = `false`) |

| 10.12.2020 | 1.4.8.8 | Added missing `mstatush` CSR (only bit `MBE` is implemented yet); added option to configure external bus interface for BIG- or little-endian byte-order, configured via VHDL package `xbus_big_endian_c` constant, default = BIG-endian, software can check endianness of the interface via SYSINFO's `SYSINFO_FEATURES(SYSINFO_FEATURES_MEM_EXT_ENDIAN)` flag; added `mstatush` CSR and endianness information to data sheet |

| 09.12.2020 | 1.4.8.7 | Added missing *environment call from U-mode* exception (via `ecall` instruction in user-mode); added environment call from U-mode to data sheet |

| 09.12.2020 | 1.4.8.6 | :warning: fixed bugs in ALU's co-processor interface: ATOMIC `A` extension could not be used without MULDIV `M` extension, CPU might have permanently stalled when executing an instruction from a disabled ISA extension; :lock: added security feature: illegal user-level CSR read access will always return zero; added new section *Execution Safety* to neorv32.pdf data sheet |

| 07.12.2020 | 1.4.8.5 | :warning: fixed bug in next-PC logic (introduced with version 1.4.8.1) that caused instruction fetch from memories with more than 1 cycle latency to fail |

| 05.12.2020 | 1.4.8.4 | :warning: fixed bug in physical memory protection (PMP): region size configuration was incorrect; removed `PMP_NUM_REGIONS` and `PMP_GRANULARITY` CPU/processor generics (PMP configuration now via package constants); reworked section *2.4. Instruction Sets and CPU Extensions* of neorv32.pdf |

| 04.12.2020 | 1.4.8.2 | Added PMA (physical memory attribute) to processor-internal IO region: `NO EXECUTE`; added *3.3.Address Space/Physical Memory Attributes (PMAs)* section to neorv32.pdf |

| 03.12.2020 | 1.4.8.1 | Optimized CPU program counter (PC) update logic and "next PC" computation (shortend critical path); updated bootloader (configuration option for direct-boot-from-SPI-flash only) and *customization* text in neorv32.pdf |

| 01.12.2020 | [**:rocket:1.4.8.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.8.0) | :warning: fixed bug in CPU-internal co-processor interface; optimized multiplier unit (~1 faster); added CPU `A` (atomic) extension support (only `lr.w` and `sc.w` instructions yet); added `lock` signal to CPU and processor's external bus interface |

| 28.11.2020 | 1.4.7.6 | Split ALU core operations: shortened  critical path - replaced ALU output 8:1 mux by a 4:1 mux |

| 26.11.2020 | 1.4.7.5 | Minor rtl clean-up; CSR access instructions are one cycle faster now (3 cycles now); system/environemnt instructions (`ecall` `ebreak` `mret` `wfi`) need one additional cycle (4 cycles now) |

| 25.11.2020 | 1.4.7.4 | :warning: fixed bug in `FENCE.I` instruction that corrupted instruction fetch when executing code from processor-external memory; default testbench (`sim/neorv32_tb.vhd`) now features external IMEM, external DMEM and external IO connected via external bus interface; simulation now allows CPU to execute code using external memories only (no internal IMEM/DMEM); optimized CPU's instruction fetch interface (no more unnecessary transfer cancel requests) |

| 20.11.2020 | 1.4.7.2 | :warning: fixed bug in CPU bus unit that caused a memory exception after reset in some cases; added second simulated external (Wishbone) memory to testbench (one memory for simulating an external IMEM, one memory for simulating external memory-mapped IO); external bus interface (`wishbone`) now makes sure that a canceled bus transfer is really understood by the accessed peripheral |

| 20.11.2020 | 1.4.7.1 | Removed deprecated "update_enable signal" from IMEM |

| 11.11.2020 | [**:rocket:1.4.7.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.7.0) | Further optimized pipeline front-end: Jumps and branches are one cycle faster (+5% coremark performance); updated synthesis results; updated performance results; added `hello_world` example program |

| 07.11.2020 | 1.4.6.7 | Updated bootloader (size optimization) and changed processor version output; added project logo; minor data sheet edits |

| 03.11.2020 | 1.4.6.6 | Removed SPI module's *buggy* "LSB-first mode", SPI module now always sends data MSB-first; removed SPI.CTRL `SPI_CT_DIR` bit; modfied bit order in SPI CTRL register; updated SPI SW library |

| 02.11.2020 | 1.4.6.5 | :warning: Fixed bug in CPU's illegal instruction detection logic; CPU rtl code optimizations - further reduced hardware footprint; rtl code clean-ups |

| 01.11.2020 | 1.4.6.4 | :warning: Fixed bug in `[m]instret[h]` and `[m]cycle[h]` carry logic; CPU hardware optimizations (area reduction, shortend critical path) |

| 29.10.2020 | 1.4.6.3 | rtl code clean-up; made preparations for additional co-processors |

| 25.10.2020 | 1.4.6.2 | Added tag signal (`wb_tag_o`) to processor's Wishbone bus; removed processors's `priv_o` - privilege level is now encoded in Wishbone *tag* signal; added a more sophisticated **FreeRTOS** example ("full_demo") |

| 24.10.2020 | [**:rocket:1.4.6.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.6.0) | Completely reworked external memory interface (WISHBONE), removed now-obsolete processor generic `MEM_EXT_REG_STAGES`; added processor wrapper with **AXI4-Lite master interface** |

| 22.10.2020 | 1.4.5.11 | TWI: Added new control register flag to enable/disable SCL clock stretching by peripheral devices |

| 22.10.2020 | 1.4.5.10 | Added `i_bus_priv_o` and `d_bus_priv_o` signals to CPU_top and `priv_o` to Processor_top to show privilege level of bus access (from `mstatus` MPP); :warning: Fixed bug in external memory interface [WISHBONE] (non-standard Wishbone components were able to corrupt processor-internal ACK/ERR signal logic) |

| 20.10.2020 | 1.4.5.9 | :warning: Fixed bug in CPU "sleep" instruction (`WFI` - wait for interrupt) |

| 20.10.2020 | 1.4.5.8 | *Machine timer interrupt* is available as processor input pin (`mtime_irq_i`) if internal `MTIME` is not implemented (`IO_MTIME_USE` = false) |

| 18.10.2020 | 1.4.5.7 | Added new IO peripheral/Device: Second CFU (CFU1); renamed old CFU to CFU0; CFU VHDL files: `neorv32_cfu0.vhd` & `neorv32_cfu1.vhd`; removed CFU interrupt |

| 17.10.2020 | 1.4.5.5 | New makefile target `upload` allows to directly upload an executable to the bootloader from the console |

| 17.10.2020 | 1.4.5.4 | Added new CPU/Processor generic `FAST_SHIFT_EN` (default = *false*) to enable implementation of a fast (but large) barrel shifter for accelerating CPU shift instructions; updated CoreMark performance results |

| 16.10.2020 | 1.4.5.2 | Added read-only flag to custom `mzext` CSR to check if physical memory protection (PMP) is implemented; added [C] `mzext` CSR name aliases to neorv32.h |

| 15.10.2020 | 1.4.5.1 | Fixed "unprecise exceptions": `mtval` did not always reflect the correct value according to the instruction that caused the exceptions; fixed bug in RTE: Debug trap handler was not showing the correct `mepc` value |

| 13.10.2020 | [**:rocket:1.4.5.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.5.0) | An official *open-soucre RISC-V architecture ID* was assigned to the project: decimal = `19`, 32-bit hexadecimal = `0x00000013` - software can retrieve the ID from the `marchid` CSR |

| 12.10.2020 | 1.4.4.9 | Added *alignment flags* to makefiles: branch/jump/call targets are forced to be 32-bit aligned -> increases performance when using the `C` extension; added makefile flag listing to NEORV32.pdf; updated performance results for CPUs with `C` extension; `crt0.S` will initialize *all* registers with zero if not using `E` extension and not compiling bootloader |

| 11.10.2020 | 1.4.4.8 | Reworked pipeline frontend: Optimized fetch enginge, added issue engine, faster instruction fetch after taken branches + reduced hardware requirements; updated synthesis and performance results |

| 11.10.2020 | 1.4.4.6 | Added option to configure external memory interface (Wishbone) to either use *standard/classic protocol* (default) or *pipelined protocol* (for better timing): via `wb_pipe_mode_c` constant in VHDL package file (`rtl/core/neorv32_package.vhd`); added help text to NEORV32.pdf section "3.4.4. Processor-External Memory Interface (WISHBONE)" |

| 08.10.2020 | 1.4.4.5 | Removed CPU's `BUS_TIMEOUT` and processor's `MEM_EXT_TIMEOUT` generics; instead, a global configuration `bus_timeout_c` in the VHDL package file is used now |

| 08.10.2020 | 1.4.4.4 | Removed `DEVNULL` device; all simulation output options from this device are now available as `SIM_MODE` in the `UART`; `mcause` CSR can now also be written; FIXED: trying to write a read-only CSR will cause an illegal instruction exception; for compatibility reasons any write access to the misa CSR will be ignored and will NOT cause an exception |

| 07.10.2020 | 1.4.4.2 | Simplified ALU's set of core operations; removed co-processor data mux right after ALU -> shorter critical path; CPU control VHDL code clean-up and CSR write logic optimization; optimized IMEM/DMEM access logic; added note regarding alignment of IMEM/DMEM |

| 05.10.2020 | [**:rocket:1.4.4.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.4.0) | :warning: Fixed bug in external memory interface: Executing code from external memory was causing an instruction fetch stall |

| 02.10.2020 | 1.4.3.9 | `[m]cycleh` and `[m]instreth` CSRs are now 32-bit wide (-> fully RISC-V-compliant) |

| 01.10.2020 | 1.4.3.8 | Added CPU top entity wrapper with resolved port signals `rtl/top_templetes/neorv32_cpu_stdlogic.vhd`; optimized ALU core functions – shorter critical path, less control overhead, reduced HW footprint |

| 27.09.2020 | 1.4.3.3 | Further improved ALU and control logic; CSR access instruction require one additional cycle now (to let side effects kick in); updated synthesis results; added CFU hardware driver dummy |

| 26.09.2020 | 1.4.3.2 | :warning: Fixed bug in `CSRRWI` instruction (introduced with version 1.4.3.1); further ALU operand logic optimizations; updated CPU data path figure |

| 25.09.2020 | 1.4.3.1 | Register file's `x0` is now a physical register; this register is initialized by the hardware and locked afterwards; removed "set to zero" stage -> smaller hardware footprint and shorter critical path; added processor top entity wrapper with resolved signals `rtl/top_templetes/neorv32_top_stdlogic.vhd` |

| 16.09.2020 | [**:rocket:1.4.3.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.3.0) | Simplified memory configuration: removed processor top's memory space configuration generics (`MEM_ISPACE_BASE`, `MEM_ISPACE_SIZE`, `MEM_DSPACE_BASE`, `MEM_DSPACE_SIZE`); data/instruction space sizes are irrelevant for hardware; instruction/data space base addresses are fixed (but can be modified in NEORV32 VHDL package file); modified SYSINFO registers; adapted bootloader, crt0 start-up code and linker script; stack configuration is now done via linker script; reworked chapter "address space"; added CFU interrupt -> fast interrupt channel 1 (shared with GPIO) |

| 14.09.2020 | 1.4.2.0 | Removed option to disable CSR counters (via `CSR_COUNTERS_USE` generic) since these counters are mandatory according to the RISC-V specs; added new IO/peripheral device: custom functions unit (`CFU`) for tightly-coupled custom co-processors; improved timing of processor-internal clock generator; fixed wrong labels in address space figure and removed dedicated exception vectors box; added mask register to GPIO unit to specify which input pins can trigger a pin-change interrupt |

| 11.09.2020 | 1.4.0.4 | Reworked `TRNG` architecture and interface; added text regarding fast interrupt channels usage for the NEORV32 processor |

| 02.09.2020 | 1.4.0.2 | :warning: Fixed bugs in external memory interface; added option to define latency of simulated external memory in testbench; hardware configuration sanity checks will now only appear once in console; added more details to data sheet section 3.3. Address Space; fixed typos in MEM_*_BASE and MEM_*_SIZE generic names |

| 01.09.2020 | 1.4.0.1 | Using registers above `x15` when the `E` extensions is enabled will now correctly cause an illegal instruction exception |

| 29.08.2020 | [**:rocket:1.4.0.0**](https://github.com/stnolting/neorv32/releases/tag/v1.4.0.0) | Rearranged and reworked data sheet; added FreeRTOS port, demo & short referencing chapter; removed booloader-specific linker scripts – main linker script is used for both, applications and bootloader; bootloader can now have `.data` and `.bss` sections; improved IMEM and BOOTROM memory initialization – faster synthesis; image generator now constrains init array size to actual executable size; peripheral/IO devices can only be written in full word mode (= 32-bit); GPIO ports are now 32-bit wide |

| 23.08.2020 | 1.3.7.3 | Added custom `mzext` CSR to check for available Z* CPU extensions; multiplier's FAST_MUL mode is one cycle faster now; updated performance data |

| 20.08.2020 | 1.3.7.2 | Removed bootloader-specific crt0 – bootloader now uses std crt0; makefiles now also support asm and cpp files; made linker scripts more general; renamed makefile "compile" (which is still available for compatibility) target into "exe" |

| 14.08.2020 | [**:rocket:1.3.7.0**](https://github.com/stnolting/neorv32/releases/tag/v1.3.7.0) | Simplified CPU fetch engine; added configurable CPU instruction prefetch buffer (ipb) FIFO; optimized CPU execute engine; updated performance data |

| 06.08.2020 | 1.3.6.5 | Added `FAST_MUL_EN` generic to enable mapping of the multiplier core to DSP blocks; ALU.shifter is no more triggered when executing MULDIV operations; added benchmark results for DSP-based multiplier configurations; updated implementation and performance results; simplified makefiles – using implicit libc definition; crt0 only initializes lowest 16 registers |

| 03.08.2020 | [**:rocket:1.3.6.0**](https://github.com/stnolting/neorv32/releases/tag/v1.3.6.0) | Relocated `DEVNULL` (changed base address); minor edits, optimization and clean-ups |

| 30.07.2020 | 1.3.5.2 | Added register stage to PMP mask generation to shorten critical path; removed automatic IRQ enable/disable from RTE install/uninstall functions |

| 30.07.2020 | 1.3.5.1 | :warning: Fixed bug(s) in PMP mask generation; `misa.Z` flag is not yet defined by the RISC-V specs., hence it is read-only and read as zero |

| 29.07.2020 | 1.3.5.0 | Added user privilege level, enabled via new `CPU_EXTENSION_RISCV_U` generic; :warning: fixed error in `mstatus(mpie)` logic; implemented RISC-V spec.-compliant Physical Memory Protection (PMP); allows up to 8 regions but only NAPOT mode is supported yet |

| 25.07.2020 | 1.3.0.0 | `mcause` CSR is read-only now!; removed `CLIC`, added 4 fast IRQ channels to CPU with according flags in `mie` and `mip` and trap IDs; updated core libraries; updated NEORV32 RTE; highly reworked data sheet; updated synthesis and performance results |

| 21.07.2020 | 1.2.0.6 | Added doc section regarding the CPU's data and instruction interfaces; optimized CPU fetch engine; updated iCE40 synthesis results |

| 20.07.2020 | 1.2.0.5 | Less penalty for taken branches and jumps (2 cycles faster) |

| 19.07.2020 | 1.2.0.0 | CPU bus unit now has independent busses for instruction fetch and data access – merged into single processor bus via new bus switch unit; doubled speed of ALU shifter unit again; all bits of `mcause` CSR can now be modified by application program (full RISC-V-compliant); performance counters CSRs `[m]cycleh` and `[m]instreth` are only 20-bit wide; removed NEORV32-specific custom CSRs – all processor-related information can be obtained from the new `SYSINFO` IO module (CPU is now more independent from processor configuration); changed IO address of `DEVNULL`; fixed bug in bootloader's trap handler; added `USER_CODE` generic to assign a custom user code that can be read by software (from `SYSINFO`) |

| 14.07.2020 | 1.1.0.0 | Added `fence_o` and `fencei_o` signals to top entity to show if a `fence` or `fencei` instruction is executed; added `mvendorid` and `marchid` CSRs (both are always zero); ALU shift unit is faster now; two lowest bits of `mtvec` are always zero; fixed wrong instruction exception priority; removed `HART_ID` generic – `mhartid` CSR is always read as zero; performance counters (`[m]cycle[h]`, `[m]instret[h]` and `time[h]`) are also available in embedded mode – but can be explicitly disabled via the `CSR_COUNTERS_USE` generic; mcause CSR only allows write access to bit 31 and bits 3:0; updated synthesis reports |

| 10.07.2020 | 1.0.6.0 | Non-taken branches are now 1 cycle faster; the `time[h]` CSR now correctly reflects the system time from the MTIME unit; fixed WFI instruction permanently stalling the CPU; `[m]cycle[h]` counters now stop counting when CPU is in sleep mode; `minstret[h]` and `mcycle[h]` now also allow write-access |

| 09.07.2020 | 1.0.5.0 | `X` flag of `misa` CSR is zero now; the default SPI flash boot address of the bootloader is now `0x0080000`; new exemplary FPGA utilization results for Intel, Lattice and Xilinx; `misa` CSR is read-only again, switching compressed extension on/off is pretty bad for the fetch engine; `mtval` and `mcause` CSRs now allow write accesses and are finally RISC-V-compliant; time low and high registers of `MTIME` peripheral can now also be written by user; `MTIME` registers only allow full-word write accesses |

| 06.07.2020 | 1.0.1.0 | Added missing `fence` instruction; added new generic to enable optional Zifencei CPU extension for instruction stream synchronization |

| 05.07.2020 | 1.0.0.0 | New CPU architecture: Fetch and execute engines; increased CPI; timer and counter CSRs are now all 64-bit wide; :warning: fixed CSR access errors; fixed `C.LW` decompression logic; `misa` flags `C` and `M` are now r/w – compressed mode and multiplier/divider support can be switched on/off during runtime; PC(0) is now always zero; :warning: fixed bug in multiplier/divider co-processor; renamed SPI signals; added RISC-V compliance check information – processor now passes the official RISC-V compliance tests |

| 25.06.2020 | 0.0.2.5 | Added `DEVNULL` device; added chapter regarding processor simulation; fixed/added links; fixed typos; added FPGA implementation results for iCE40 UP |

| 23.06.2020 | [**:rocket:0.0.2.3**](https://github.com/stnolting/neorv32/releases/tag/v1.2.0.5) | Publication |