Recent progress on libc

[dpxf]

I thought it might be nice to provide a consolidated update of progress on LLVM’s libc, so here it is.

Math updates:

• Landed:
  • Nearest integer instructions to speed up math implementations:
    • ⚙ D129776 [libc] Add nearest integer instructions to fputil.
    • ⚙ D129916 [libc] Add float type and flag for nearest_integer to enable SSE4.2.
  • Correctly rounded sinf
    • ⚙ D123154 [libc] Implement sinf function that is correctly rounded to all rounding modes.
    • Changed range reduction of sinf to be shared with cosf: ⚙ D130629 [libc] Change sinf range reduction to mod pi/16 to be shared with cosf.
  • Improve expf performance: ⚙ D130498 [libc] Use nearest_integer instructions to improve expf performance.
  • Improve expm1f performance: ⚙ D130502 [libc] Use nearest_integer instructions to improve expm1f performance.
  • Universal exp function for cosh/sinh: ⚙ D129215 [libc][math] Universal exp function for cosh/sinh calculation.

Stdio updates

• Printf Conversion status: Decimal floating point conversions are still in progress.
• Printf compiled size (with size saving enabled) was evaluated, size is approximately 5.8KB in a final binary

Thread library updates

• Working on pthread_setspecific and related API.

Misc updates

• Added POSIX dirent functions opendir, closefir, readdir and dirfd.
• Errno wrapper - ⚙ D130662 [libc] Read and write errno via special getter and setter functions.
• More arm32 entrypoints have been added - ⚙ D129821 [libc][arm32] add string stdlib & math entrypoints

[androm3da]

Sorry to hijack this progress update. I’m trying to understand more about the goals/use case for the LLVM libc.

This does not mean that one can mix llvm-libc with the system-libc. Also, it does not mean that layering is the only way to use llvm-libc. What it means is that, llvm-libc can optionally be packaged in a way that it can delegate parts of the functionality to the system-libc.

Does this mean that the LLVM libc could be used as the system libc?

[sivachandra]

The idea is to make LLVM libc complete enough for it to be a drop-in replacement for your system libc. I don’t think it is complete enough to serve as a drop-in replacement today for everything on your system, but there is enough already to serve the libc needs of simple applications. Stay tuned for more information - we plan to make regular announcements of the progress and milestones we reach.

[androm3da]

@sivachandra

I don’t think it is complete enough to serve as a drop-in replacement today for everything on your system

Ok, thanks for clarifying. Is it a goal to eventually be a libc that could satisfy all the same features we get from something like musl or glibc?

[sivachandra]

Is it a goal to *eventually* be a libc that could satisfy all the same features
we get from something like musl or glibc?

Yes!