Yasm regression test suite
Assembly Python C
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
arch/x86
core
objfmts
parsers
.gitignore
README.md
custom_nextgen.py
custom_yasm.py
nextgen.ini
rtest.py
yasm.ini

README.md

Yasm-Regression

This is the regression test suite for yasm and yasm-nextgen.

The test suite includes a test suite runner, rtest.py, which recursively scans for .asm and .s input files. Each input file is run through the assembler and the output and stderr results compared.

Usage

Basic usage:

% ./rtest.py --config=yasm.ini --yasm=../yasm/yasm .

This will run all tests in subfolders of the current directory (.) with ../yasm/yasm as the assembler executable. Result files (e.g. output files from the assembler) will be written to the ./results directory, and a detailed log file of the test run will be written to ./yasm-rtest.log.

Output

Summary output is written to stdout, while more detailed output is written to the log file. To get more detailed output on stderr, add -v to the command line; to get less output, add one or more -q to the command line.

By default, outputs are written to ./results; this can be overridden with the command line option --output-dir. Similarly, the log file output location can be changed using the command line option --log-file.

Configuration Files

Currently there are two configurations that can be used: yasm.ini and nextgen.ini. The configuration file has a [GLOBAL] section which can be used to load a custom Python module (with the CUSTOM_MODULE option) for customizing error/warning parsing and global test skipping, and an arbitrary number of [foo/bar.asm] sections to customize a specific test's behavior with the following options:

  • ERRWARN: specifies file extension of stderr comparison file (defaults to ew)

  • OUTPUT: specifies file extension of output comparison hex dump file (defaults to hex)

  • BROKEN: if yes, 1, or true, causes test to be skipped

Input Files (.s and .asm)

The first line of the input file may specify more detailed configuration for that particular test. Each option needs to be in square brackets. Note this line is passed to the assembler as part of the input file, so it must be a comment if used.

  • [fail]: the test is expected to fail (error output code from the assembler)

  • [parser xxx]: parser override; results in -p xxx to the assembler. The default is based on the file extension (.s = gas, .asm = nasm)

  • [text]: do a text comparison on the output file (instead of binary). Useful for preprocessor tests.

  • [oformat xxx]: set the output file format; results in -f xxx to the assembler

  • [args x y z]: arbitrary extra command line arguments to the assembler; passed through verbatim. Useful for -Dfoo=bar and similar.

  • [yasm x y z]: full command line override (options like [parser] ignored)

  • [ygas x y z]: full command line override for ygas

Output Comparison

For each test, the assembler output is compared to a "golden" result file. Generally this is done byte-by-byte, but tests can specify the [text] option for a text comparison instead of a binary comparison.

The "golden" results can be specified either with a hex dump file (one hex byte per line) or by specifying the output bytes inline in the input file. This latter method looks for comments in the input file starting with ; out: or # out: (depending on the selected parser), followed by raw values. The raw values are treated as octal (if 3 characters) or hex (if 2 characters).

Additionally, the input file may specify repeated raw values using rep-out instead of out. The first number is the rep count, which may be decimal (no prefix) or hex (0x prefix). Following this are the repeated values, which must be 2-character hex values.