The union of all files from all check-ins in directory arch [history]
the arch directory of the libk repository contains system-specific definitions and headers, such as error codes and system call tables. none of the files in arch are compiled, but are included from compiled files.
whenever a file is passed to the assembler or C compiler, an include directory is selected based on the chosen target tuple (see /libk.md). thus, when compiling a file that contains the line
#include <system_calls.h> on an x86-32 freebsd system, the compiler will be instructed search (among other directories) the
/arch/x86.fbsd.32 directory for headers.
each subdirectory should be named for a specific target tuple. the top level is reserved for development tools and files of relevance to more than one architecture; for instance, the standardized x86-64 SysV syscall ABI is kept at
/arch/x86.syscall.64.s. these files should not be included directly from any files under the
/mod directory; rather, they should be included from stub files in the appropriate target directory. this provides a consistent header selection mechanism.
arch contains a number of shell scripts. unlike the shell scripts in /global, the shell scripts in /arch are not part of the build process; they are simply developer tools, usually for generating syscall tables for new architectures. the output of the scripts is written directly to the appropriate target directory; these files should be added to the source tree and committed.
mktbl.linux.sh <target> <arch> <ABI>: generates the Linux errno and syscall tables for a given architecture and ABI. requires the kernel source tree to be installed on the host system.
the arch directory also contains C sources for programs that need to be compiled and executed during the build process.
typesize.c: calculates type metadata for the system it is compiled on, producing macros needed for the creation of