More stage1 test coverage

[ghost]

No description provided.

[andrewrk]

Great work. You mind doing one last rebase and then I'll mark this as auto-merge?

[ghost]

I'm pretty sure I didn't change anything that would cause this new failure that previously didn't happen:

zig build-exe maximum_sized_integer_literal Debug x86_64-windows: error: thread 229607 panic: Segmentation fault at address 0xffff37fc3934
/home/ci/actions-runner8/_work/zig/zig/src/link/Coff.zig:1504:29: 0x44d1bb0 in updateDeclExports (zig)
        sym.value = decl_sym.value;
                            ^
/home/ci/actions-runner8/_work/zig/zig/src/link/Coff.zig:1194:34: 0x44c39b3 in updateDecl (zig)
    return self.updateDeclExports(mod, decl_index, mod.getDeclExports(decl_index));
                                 ^
/home/ci/actions-runner8/_work/zig/zig/src/link.zig:551:77: 0x428771b in updateDecl (zig)
            .coff  => return @fieldParentPtr(Coff,  "base", base).updateDecl(module, decl_index),
                                                                            ^

Pretty sure all I did was resolve the conflict and rebase on master/the InternPool changes.
Might be a regression. I added a if (builtin.os.tag == .windows) return; to test/cases/maximum_sized_integer_literal.zig for now.

[andrewrk]

Hmm, I did not take as close a look at the new test cases last time as I should have.

Please use panics instead of illegal behavior (i.e. assert) so that the tests are meaningful in safety-off optimization modes as well.

Also please use behavior tests where possible instead of run test cases.

[ghost]

Oh, that is very true. I must've just taken that idea from other tests without thinking about it enough.

It's also very good to know that behavior tests are in fact preferable.

[ghost]

zig test Debug mips-linux-none: error: Assertion failed: ShiftAmt <= BitWidth && "Invalid shift amount" (/home/andy/dev/zig-bootstrap/llvm/include/llvm/ADT/APInt.h: lshrInPlace: 847)

zig test Debug mips-linux-none: error: the following command terminated unexpectedly:
/home/ci/actions-runner5/_work/zig/zig/build-debug/stage3-debug/bin/zig test /home/ci/actions-runner5/_work/zig/zig/test/behavior.zig --cache-dir /home/ci/actions-runner5/_work/zig/zig/build-debug/zig-local-cache --global-cache-dir /home/ci/actions-runner5/_work/zig/zig/build-debug/zig-global-cache --name test -target mips-linux-none -mcpu mips32 -I /home/ci/actions-runner5/_work/zig/zig/test -L /home/ci/deps/zig+llvm+lld+clang-x86_64-linux-musl-0.11.0-dev.1869+df4cfc2ec/lib -I /home/ci/deps/zig+llvm+lld+clang-x86_64-linux-musl-0.11.0-dev.1869+df4cfc2ec/include --zig-lib-dir /home/ci/actions-runner5/_work/zig/zig/lib --listen=- 
zig test Debug mips-linux-musl: error: Assertion failed: ShiftAmt <= BitWidth && "Invalid shift amount" (/home/andy/dev/zig-bootstrap/llvm/include/llvm/ADT/APInt.h: lshrInPlace: 847)

Interesting. This has also not previously happened before the rebase.

Anyway, AFAIU the zig downloads are broken right now so I'm going to wait till I can get the latest master from the download page.

[ghost]

I'm going to try limiting tests that I think might have to do with the failure to linux-x86_64 for now.

[jacobly0]

I'm going to try limiting tests that I think might have to do with the failure to linux-x86_64 for now.

Vectors are a likely culprit: #13782

[ghost]

Thanks! That helps a lot.