https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122613
Bug ID: 122613
Summary: Passing --enable-as-accelerator-for= stops creation of
nvtpx-none-cc
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Keywords: build
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: sjames at gcc dot gnu.org
CC: tschwinge at gcc dot gnu.org
Target Milestone: ---
Passing --enable-as-accelerator-for= for nvtpx-none means that `nvptx-none-cc`
etc doesn't get installed.
This is a problem for our packaging for offload compilers because we don't use
a combined tree: we do a stage1 build (no libc) of gcc, then we build newlib,
then we do a stage2 build with openmp etc support.
If we pass --enable-as-accelerator-for= in that stage1 build, we end up with no
`nvptx-none-cc` to build newlib with.
This is also a problem for the final stage2 build of gcc because it means it
can't be used to update itself later on.
I asked about this on IRC a while back and didn't hear back, so filing a bug.
