Skip to content

README.md

Latest commit

History

History
58 lines (46 loc) · 2.56 KB

README.md

File metadata and controls

58 lines (46 loc) · 2.56 KB
Raw

python-student-support-code

Support code for students (Python version).

The runtime.c file needs to be compiled by doing the following

 gcc -c -g -std=c99 runtime.c

This will produce a file named runtime.o. The -g flag is to tell the compiler to produce debug information that you may need to use the gdb (or lldb) debugger.

On a Mac with an M1 (ARM) processor, use the -arch x86_64 flag to compile the runtime:

 gcc -c -g -std=c99 -arch x86_64 runtime.c

Using the Racket x86 interpreters (starting from the Tuples compiler)

This section describes how you can use racket_interp_x86 and racket_interp_pseudo_x86 interpreters defined in the racket_interp_x86.py file. The x86 interpreter defined in the interp_x86 directory doesn't support the x86-like languages generated by the select instructions pass of every compiler starting from the chapter on tuples. The interpreters in racket_interp_x86.py will help you interpret these x86-like languages along with actual x86 code. As the name suggests, these interpreters run using the x86 interpreter in the Racket version of this repository.

Setup instructions

  1. Install Racket from here.

  2. Clone the Racket version of this repository into the parent directory of this repository. The directory into which you clone the repository is important because racket_interp_x86.py is hardcoded to look for the Racket version of the x86 interpreter in the parent directory.

  3. Add racket_interp_pseudo_x86 interpreter to your interp_dict for passes before assign homes. For the remaining passes, use racket_interp_x86.

Caveats

  • Since your x86 code is interpreted in Racket, any errors occurring during interpretation will show up in the output diff and not as a Python exception.
  • When implementing the languages up to Ltuple, add the num_root_spills field to the X86Program object. Initialize the field num_root_spills to the number of tuple variables that have been spilled into the root stack.
  • When implementing the languages beginning with Lfun and after, add the following fields to every FunctionDef object:
    • num_root_spills: Initialize this field to the number of tuple variables that have been spilled into the root stack.
    • num_params: Initialize this field to the number of parameters that the function had after the limit functions pass but before the select instructions pass.