patches for additional libc backend

[ahgamut]

(WIP) related to oracle/graal#12865

[olpaw]

@ahgamut thanks for providing this PR.

Many changes of this PR can and should be avoided by providing additional stub implementations for https://github.com/jart/cosmopolitan, just like ones you already have in https://github.com/ahgamut/cosmopolitan/tree/header-stubs

For the sake of labsjdk maintainability using the stubs header approach is much more reasonable.

For the remaining changes, every modification that is specific to cosmo-support must be guarded with ifdefs (or m4 macro name) to be acceptable. No hacks like e.g.
0da77ca#diff-229b30d66ee3168bdb7b0cadec57c5b39145f5b08a1c3416248721ecc1d901c0
0da77ca#diff-83ba4f513f316e2b1bd706bba84a9eb39515b17c5a0685ff671d16b37f01f776
...

In other words, if I preprocess all sources with __COSMOPOLITAN__ undefined (and the equivalent for m4) I must end up with quasi-identical source state to current master.

Once there is a cosmopolitan release with all the necessary stubs, and the above issues are resolved we can take another look.

cc @lewurm @christianhaeubl @wirthi

[olpaw]

@ahgamut I saw you pushed some commits last week.

Is this ready for another round or do you plan more changes before I have another look?

[ahgamut]

I think I've fixed all the ifdef guards discussed earlier. Also expect 0d70347 is not required, but have not been able to fix that one yet.

I was trying a couple of separate changes, but if this one is ok, we can look at those later.

[olpaw]

Thanks for the update.

Where are you on getting your Cosmo changes from
https://github.com/ahgamut/cosmopolitan/tree/header-stubs
merged upstream? Any progress there?