Add support to build a host GCC for building the toolchain

[cmuellner]

A significant amount of users of this project are using it for toolchain development. These users need to ensure that their code is warning free before submitting it upstream. Let's support this use case by adding a configure flag '--enable-host-gcc', which does exactly that:

• build a host GCC before building other components
• setting PATH such that this new GCC is used to build the cross toolchain
• enable -Werror for the GCC build

This patch was tested on a Fedora 39 machine (GCC 13), with the following modification in a GCC source file:

  #if __GNUC__ == 13
  #error Host compiler in use!
  #endif

This fails when building without the new flag and does not fail when building with latest upstream/master (GCC 14 prerelease).

The '--enable-werror-always' was tested with a warning that showed up recently in upstream GCC.

[TommyMurphyTM1234]

A significant amount of users of this project are using it for toolchain development. These users need to ensure that their code is warning free before submitting it upstream. Let's support this use case by adding a configure flag '--enable-host-gcc', which does exactly that:

• build a host GCC before building other components
• setting PATH such that this new GCC is used to build the cross toolchain

Hi @cmuellner - can you expand on why and when this (building the host GCC compiler from scratch) is necessary? It's not clear to me and it seems very onerous - building the RISC-V toolchain already takes quite a long time in many cases (and a looooong time when it's LLVM!). Why won't just getting a recent (e.g. GCC 13 at the time of writing) host GCC toolchain suffice?

[cmuellner]

For upstream contributions to GCC a bootstrap build and executing all tests of the regression test suite is a must (see https://gcc.gnu.org/contribute.html).

Bootstrapping means building a stage 1 compiler, which then builds the actual GCC. Really performant RISC-V CPUs are still rare (and often used for work other than simple QA checks). Therefore, these builds/tests are usually run on x86_64/QEMU (system or user-mode emulation), but that takes several hours to complete.

Cross-compiling is a solution for that issue because we avoid QEMU for everything except executing the test binaries. And the way this is done in this PR is close to a bootstrap build.

Using a "recent" GCC is insufficient because there might be a change in the warning behavior of GCC in every commit.

The trigger for all this was a patch from somebody that was reviewed and approved by a maintainer, which I picked up for final testing and pushing. I did not run a bootstrap build because I was too impatient (and did not want to block my machine for several hours), but only a make check with this repo. After this showed no regressions, I pushed the patch. A few days later, this patch was identified to trigger a compiler warning. The fix of the warning is trivial, but I want to avoid this from happening again (especially since it has already happened the second time for me). The previous time this happened to me was even more surprising since GCC 13 did indeed not emit the warning that GCC origin/master did.

[TommyMurphyTM1234]

Hi @cmuellner - thanks for the thorough explanation. Much appreciated.

[cmuellner]

@kito-cheng ping

[kito-cheng]

LGTM, That's really useful!