Overview
The RISC-V tools includes a formatter to provide consistent layout for assembly language (typically .S) files. The formatter includes various settings to define layout columns, conversion of tabs to spaces, automatic removal of whitespace at the end of lines and ability to specify supported RISC-V instruction sets.
Features
Token types
The formatter works by tokenising the content of a file, line by line. Any non blank line within the file contains one of the following primary token types:
- Directive - one of the assembly directives such as
.section, .bss and .data.
- Label - a code label, e.g.
LABEL_1:.
- Instruction - a risc-v op code instruction, e.g.
la, call, addi.
A Comment is a conditional primary token in that it is only a primary token when it is the only non whitespace token on a line, e.g.
# Comment that is a primary token
but not when it follows other primary tokens, e.g.
li a0,R32_RCC_CFGR0 # Comment that is NOT a primary token
Alignment
The formatter works by aligning tokens of a line to defined columns within the line. The are three configurable column types.
Primary Token Column - is the column that the primary token is aligned within the line. A primary token is the main token for a file and is typically a directive (e.g. .section, .bss), label (e.g. LABEL_1:) or an instruction (e.g. la, call, j).
Data Token Column - is the column that any data associated with the primary column is aligned within the line. E.g. for a command the data is the register or immediate values that define the instruction data (the a0,R32_RCC_CFGR0 in the example li a0,R32_RCC_CFGR0).
Comment Column - is the column that any comment is aligned within the line. There are two comment column configuration options:
- When the comment is the only item on a line (no primary token).
- When the comment following a primary token (and any data).
Whitespace
There are three whitespace processing steps, two of which are configurable.
The formatter removed trailing whitespace from the end each line. This option is not currently configurable and is on permanently.
Convert tabs to whitespace using the tabs.replaceTabsWithSpaces configuration option.
Include file trailing blank line (blank line at end of file) with the endOfFileHasBlankLine configuration option.
Configuration file
You can use the formatter with default settings, however you can also change the settings by adding a .criscv JSON file to your workspace. Adding a configuration file to your workspace means that the configuration can be committed to your GIT repository along with your code and then anyone who checkouts your repository will get the same formatter settings. This helps to ensure that pull requests are not polluted with changes relating to formatting inconsistencies across the team.
Adding a .criscv file
A .criscv file can be added to your workspace by opening the Command Palette (Ctrl+Shift+P) and typing RISC-V language tools
Default configuration
{
"meta": {
"version": 1,
"help": "See: https://github.com/mekatrol/vscode-riscv-extension for description of configuration values."
},
"endOfFileHasBlankLine": true,
"commentOnlyLineColumn": 1,
"commentCharacter": "#",
"tabs": {
"replaceTabsWithSpaces": 2,
"tabWidth": 2
},
"directive": {
"primaryColumn": 5,
"dataColumn": 15,
"commentColumn": 40
},
"label": {
"primaryColumn": 5,
"dataColumn": 15,
"commentColumn": 40,
"hasOwnLine": true
},
"instruction": {
"supportedInstructionSets": [
"R32I",
"Pseudo"
],
"primaryColumn": 5,
"dataColumn": 15,
"commentColumn": 40
},
"value": {
"primaryColumn": 5,
"dataColumn": 15,
"commentColumn": 40
}
}
The configuration file contains the meta element. This element provides information to manage a configuration file including the version of the configuration as well as where to obtain help (this page).
Configuration settings
endOfFileHasBlankLine [true | false] True to add blank line at end of file, false for no blank line.
Default is true.
commentOnlyLineColumn [>= 1 | null] The number column to place the comment of a comment only line. No processing alignment of comment tokens is performed if this setting is null.
Default is 1.
commentCharacter [# | ;] The single character to use as the comment starting character. Must be one of '#' or ';'.
Default is #.
tabs.replaceTabsWithSpaces [>= 1 | null]. For null tabs are not replaced with spaces. For [>= 1] tabs are replaced with the number of spaces specified.
Default is 2.
tabs.tabWidth [>= 1] The number of spaces that a tab equates to. Used to determine column numbering when padding out columns (primary, data, comment). e.g. a tabWidth of 1 will add 1 to the column number when a tab is found, whereas a tabWidth of 4 will add 4 to the column number when a tab is found. This setting is redundant if replaceTabsWithSpaces = true (because tabs will be replaced with spaces!).
Default is 2.
directive.primaryColumn [>= 1 | null]. The column to place the directive (e.g. .section). A value of 5 will position the directive in the 5th column (ie 4 spaces then the directive). No processing alignment of directive primary tokens is performed if this column is null.
Default is 5.
directive.dataColumn [>= 1 | null]. The column for any data following the directive. No processing alignment of data tokens is performed if this setting is null.
Default is 15.
directive.commentColumn [>= 1 | null]. The column for any comment following the directive. No processing alignment of non-primary comment tokens is performed if this setting is null.
Default is 40.
label.primaryColumn [>= 1 | null]. The column to place the label (e.g. LABEL_1:). A value of 3 will position the label in the 3rd column (ie 2 spaces then the label). No processing alignment of label primary tokens is performed if this setting is null.
Default is 5.
label.dataColumn [>= 1 | null]. The column for any data following the label. No processing alignment of data tokens is performed if this setting is null.
Default is 15.
label.commentColumn [>= 1 | null]. The column for any comment following the label. No processing alignment of non-primary comment tokens is performed if this setting is null.
Default is 40.
label.hasOwnLine [true | false]. True to force label onto its own line, false to leave label as is (i.e. can have instructions on the same line).
e.g. LABEL_1: li t0,0x01.
Default is true.
instruction.primaryColumn [>= 1 | null]. The column to place the primary instruction (e.g. addi). A value of 4 will position the instruction in the 4th column (ie 3 spaces then the instruction). No processing alignment of instruction primary tokens is performed if this setting is null.
Default is 5.
instruction.dataColumn [>= 1 | null]. The column for any data following the instruction. No processing alignment of data tokens is performed if this setting is null.
Default is 15.
instruction.commentColumn [>= 1 | null]. The column for any data following the instruction. No processing alignment of non-primary comment tokens is performed if this setting is null.
Default is 40.
instruction.supportedInstructionSets Array of supported instruction sets, one of [R32I, R64I, RV32M, RV64M, RV32A, RV64A, RV32F, RV64F, RV32D,RV64D, Pseudo, Compressed].
Default is [R32I, Pseudo].
value.primaryColumn [>= 1 | null]. The column to place the value (e.g. .section). A value of 5 will position the value in the 5th column (ie 4 spaces then the value). No processing alignment of value primary tokens is performed if this column is null.
Default is 5.
value.dataColumn [>= 1 | null]. The column for any data following the value. No processing alignment of data tokens is performed if this setting is null.
Default is 15.
value.commentColumn [>= 1 | null]. The column for any comment following the value. No processing alignment of non-primary comment tokens is performed if this setting is null.
Default is 40.
