Bump Binutils into 2.44.

[pz9115]

Bump Binutils into 2.44.

[cmuellner]

Thanks for reminding me.
I have a glibc version bump ready as well.

[kito-cheng]

I guess this should not belong this PR?

[pz9115]

Since there exist shallow clone missing commit problem in CI, I'm trying to fix it.
#1669

[TommyMurphyTM1234]

Please also see the more detailed commentary from @mickflemm on the shallow clone problem here:

• git submodule update fails #1654 (comment)

[pz9115]

Ok, thank you Tommy.

[cmuellner]

The question here is, how do we deal with "not advertised" git references?

I don't want a downstream script that checks if we can fetch a commit or need to find out the proper branch first. If the upstream server does not support a simple shallow-clone-workflow, then what's the point in insisting on shallow clones?

I think dropping the shallow = true for servers that make things harder than necessary is reasonable.

We could justify each of these in a comment ("We don't use shallow clones for Binutils, because the git server does not support fetching unadvertised git objects").

[TommyMurphyTM1234]

I think dropping the shallow = true for servers that make things harder than necessary is reasonable

That alone would not be sufficient to stop doing a shallow clone for specific submodules.
This line in Makefile.in would also need appropriate modification:

• riscv-gnu-toolchain/Makefile.in

  Line 350 in a33dac0

  flock `git rev-parse --git-dir`/config git submodule update --progress --depth 1 $(dir $@)

[cmuellner]

#1678