Assembly Files: Difference between .a .s .asm

In Unix/Linux systems:

• .a is the usual extension for static libraries (aka Archives of multiple .o files, made with ar(1)). Dynamic libraries, aka Shared Objects, use .so.
• .s is used for asm compiler output. (gcc -S foo.c produces asm output, with a default filename of foo.s)
• .S is used for hand-written asm source files. gcc -c foo.S runs it through the C preprocessor (so you can use #include<>, #if, #define, and C-style comments.) Some C headers, like asm/unistd.h only have #defines, and so can be included in a .S to get definitions like __NR_write system call numbers, for example.

In x86, there are two separate versions of asm syntax: AT&T (used by Unix compilers like gcc), and Intel/NASM (with a couple dialects, like MASM vs. NASM itself).

.S is appropriate for asm in GNU as syntax, whether you use any C preprocessor features or not.

In x86, .asm is more often associated with Intel-syntax NASM/YASM, or MASM, source code. Outside of x86, it’s probably a good choice for asm source files that could be assembled by the platform-specific assembler, if it uses different directives than GNU as.

The glibc source tree uses .S for all asm source files.

People with a gcc background might put their MIPS asm into .S or .s files, while people with more NASM/YASM experience (or Windows), might go for .asm.

I’d recommend against .s files, because it’s easy to accidentally overwrite with gcc -S foo.c.