Add snprintf implementation and refine printf family funcs

[ChAoSUnItY]

This patch adds snprintf and refines printf family funcs by changing function signature to return an integer to match their original definitions defined by specifications.

The second parameter "size" is expected to accept only zero or positive integer, otherwise as C11 specification states, it would be undefined behavior.

Behavior of current snprintf implementation

Quoting C11, chpater §7.21.6.5, The snprintf() function:

int snprintf(char * restrict s, size_t n,const char * restrict format, ...);

[...] If n is zero, nothing is written, and s may be a null pointer.

If n is negative, it's undefined behavior, and it would likely cause program to crash due to null terminator's assignment.

About reusability

Currently shecc does not support va_list-related functionality, we may want to replace __format_to_buf with variadic-function-variant of printf family functions to furthur enhance support and reusability.

Note on previous Pull Request

The previous Pull Request is outdated for several commits, and it's opened on master branch, so I decides to reopen here and resolve conflicts locally.

Summary by Bito

This pull request introduces a new snprintf function that complies with C11 standards, enhancing size parameter handling to prevent undefined behavior. It also refines the printf family of functions by changing their return types to integers and updates constants for larger data sizes. Comprehensive unit tests have been added to ensure correctness.

Unit tests added: True

Estimated effort to review (1-5, lower is better): 2

[ChAoSUnItY]

I've modified $(BUILD_SESSION) task in MakeFile by replacing $(PRINTF) with $(VECHO), due to a weird stage1 artifact building failure, which is caused by segmentation fault. I am not sure the cause of this since it does not exit abnormally when manually compiling artifact, and it also would generate stage1 artifact even exit abnormally while executing make target.

[jserv]

Generate a compilation of potential code snippets that could benefit from implementing snprintf().

[ChAoSUnItY]

The weird is back as soon as I refactored file inclusion logic with snprintf. It's still not reproducible under gdb environment.

[DrXiao]

@ChAoSUnItY
FYI: I have fetched the commit from icgmilk:snprintf, found the potential cause, and fixed it. (57f61d1)
(All GitHub actions passed after applying my patch.)

After using valgrind, it indicated that the stack size for liveness analysis might be insufficient. I enlarged the default size, and the issue was resolved.

$ valgrind out/shecc -o shecc src/main.c
...
 ==24861== Process terminating with default action of signal 11 (SIGSEGV)
 ==24861==  Access not within mapped region at address 0xFFFFFFFE6DE41EB0
+==24861==    at 0x1144D0: bb_add_killed_var (ssa.c:473)
 ==24861==    by 0x114E67: bb_solve_globals (ssa.c:537)
 ==24861==    by 0x113BCC: bb_forward_traversal (ssa.c:40)
 ==24861==    by 0x114F04: solve_globals (ssa.c:552)
 ==24861==    by 0x116552: ssa_build (ssa.c:1205)
 ==24861==    by 0x11C8D1: main (main.c:91)
 ==24861==  If you believe this happened as a result of a stack
 ==24861==  overflow in your program's main thread (unlikely but
 ==24861==  possible), you can try to increase the size of the
 ==24861==  main thread stack using the --main-stacksize= flag.
 ==24861==  The main thread stack size used in this run was 8388608.

shecc/src/ssa.c

Line 473 in 6196ab1

bb->live_kill[bb->live_kill_idx++] = var;

[ChAoSUnItY]

Thanks for the solution, but for near future development's sake, and as introduction of #184, the memory issue won't be heavily affected by this PR. Thus, I'll leave current changes to hard limits of various data structure as is, since these bugs are either sometimes hard to reproduce or being not so straightforward.

Of course, it would be good to replace all fixed-size data structures with extensible ones, I'll later assign my team member and I address this kind of issue.

[DrXiao]

Because the proposed changes not only implement snprintf() but also refine the implementations of functions such as printf() and sprintf(), I think the git commit messages could be improved to describe the details of these changes.

[jserv]

Thank @ChAoSUnItY for contributing!