Re: Maybe we should get rid of ifuncs

[enh <enh@google.com>]

On Wed, Apr 24, 2024 at 6:56 AM Zack Weinberg <zack@owlfolio.org> wrote:
>
> On Tue Apr 23, 2024 at 2:39 PM EDT, enh wrote:
> > one thing i think is interesting (having been looking at ifuncs while
> > adding riscv64 support to Android) is that afaict bionic [Android's
> > libc] is basically the only current user. every library i'm aware of
> > (and certainly every library that's part of the OS) does _not_ use
> > ifuncs, if only because iOS/macOS has no equivalent (and Windows
> > too?), and if you've got to have the manual function pointer
> > manipulation implementation for them...
>
> I assume you meant "every library that's *not* part of the OS"?

no, for the OS [except the already-mentioned exception of libc] there
are _no_ ifuncs. for apps, well, there were none in a _sampling_ of
popular apps i did, but doing a complete survey of all apps ... well,
there are a _lot_ of apps.

> > that said, for llvm at least there's work on function multi-versioning
> > where the compiler basically writes the ifunc resolver. but (a) that's
> > not quite finished yet (?) and (b) i haven't seen anyone _use_ it yet
> > and (c) is at least by definition of being machine-generated pretty
> > regular.
>
> Do you know where to find the llvm work in progress?

the docs are here:
https://github.com/ARM-software/acle/blob/main/main/acle.md#function-multi-versioning

https://github.com/android/ndk/issues/1909 implies that -- despite the
"beta" in the docs -- enough code might be in llvm and then downstream
in the ndk that arm64 FMV is now usable. (but as i said, i haven't
used it yet, and i don't yet know of anyone who does.)

> zw