In the CMakeLists.txt
file showing the error is the line
where below that line LIBC_TARGET_OS is put into
the directory string giving the error such that
CMAKE_SYSTEM_NAME is set to 'darwin' (note the lower case change
in the code) by most likely a calling makefile.
in the llvm-project directory the following line
grep -r CMAKE_SYSTEM_NAME *
gives close to 200 lines mostly in makefiles.
Looking down that list it is interesting to see that the
sub-projects to llvm-project have a variety of ways to use the
CMAKE_SYSTEM_NAME string. It is not regular in the sub-projects.
Some distinctions are by OS and others by hardware, x86_64 and
aarch64, with aarch64 identifying ARM processors. But apparently
you have an x86_64 processor.
We do not have the darwin directory for the files
but we have files with the same name in what would be a closely
related directory and that those two related directories are
based on processor type. We might try an experiment, work an
hypothesis. Let's create the needed darwin directory and copy
the two files in from the x86_64
directory and then run the build sequence and see what
If the build fails or the resulting build does
not run as expected, we know that the two files on hand from the
linux directory will not work and someone will need to construct
those files and make them available, which may take a while.
But if the build is successful, then we have
identified an easily correctable bug in the makefiles and you
are making headway.
On 7/26/20 5:05 PM,
Aditya Atluri wrote:
There are only files for linux (x86_64 and
aarch64) there. Don't know why there are no files for
darwin. May be a libc developer care to answer?
On Sun, Jul
26, 2020 at 12:47 AM Neil Nelson via cfe-dev <[hidden email]>
On Ubuntu 20.04 I have both those files
in the directory
Will the following command work in the
find . -name headers.txt
On 7/25/20 10:52 PM, Aditya Atluri via
Not so much
actually, I am seeing build error for libc on MacOS