Lakemont has no x86, no MMX, no SSE and no way of handling any fp-math. In theory LLVM is able to implicitly use the soft-float emulation library calls to legalize any such operation but, given Zig's use of many non-standard features, sometimes we hit a weak spot in the X86 codegen backend.
Consider this as a work-around for this LLVM problem, fixing the problem in LLVM is not so high in my todo list as the target is pretty niche and Intel axed it in '19.
(Commit message by @LemonBoy)
This new option sets a default libc paths file to be used for all
LibExeObjSteps. Setting LibExeObjStep.libc_file overrides this default.
This is required to allow users to cross compile projects linking system
libraries without needing to patch the build.zig.
This breaking change disambiguates between overriding the lib dir when
performing an installation with the Zig Build System, and overriding the
lib dir that the Zig installation itself uses.
This patch adjusts the exit code for a child process to be a u8. Since
the WEXITSTATUS macro returns the lower eight bits, it's safe to assume
that we can truncate the returned u32.
Two bugs in the implementation ported from musl made all the complex
functions relying on ldexp return incorrect results in some cases.
Spotted in #9047
No functional changes are expected, this patch is only moving some code
in order to slim the huge bowl of spaghetti that is debug.zig.
The amount of memory leaked on error is much less than before but not
zero, some further work is required to smooth the edges of this old part
of the stdlib.
* fix a merge conflict discovered upon rebasing latest master
* rename Target.Cpu.Feature.Set.subSet to isSuperSetOf
* convert a comment into an assert
If there is a mismatch of CPU features provided
compared to the whitelist, then will fail the build and
print what the expected CPU model is and the feature
set for the model. Also prints what features need to be
removed.
Before this change, when one or more of name or value are not known at
comptime, build.zig files must allocate and do the concatanation, which can be
cumbersome, and also adds a redundant allocation when name and value are
slices. The new version only does a single allocation directly in the builder's
allocator to concatonate name and value.
The origional behavior is available in defineCMacroRaw, for use in situations
such as parseing c compiler arguments.
Additionally, several places have been updated to use the new funtions.
I've added these three functions to all switches except NetBSD, because I don't know what the function is named there. I've even added it on the .windows switch since all the posix functions seem to be there anyways.
It turns out the code was not ported correctly from C and produced wrong
results for negative input values. As a bonus fix the NaN codepath by
adding yet another missing piece of code.
Spotted in #9047
