MinGW and CrossWindows crash if the target is an unsupported target architecture.
Changed it to emit an error message.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Thank you for this fix. Can you please add a summary in the details explaining the motivation for this fix and a link to the github bug report as well.
Sorry, I am not familiar at all with this code. Perhaps @shafik could suggest more appropriate reviewers
I think the code changes make sense; these are trivial to hit by a user, so they shouldn't be llvm_unreachable.
clang/test/Driver/unsupported-target-arch.c | ||
---|---|---|
33 | Separate side note; the suggested option -triple isn't really a clang driver level option, and -arch is only relevant for darwin targets (AFAIK) - so maybe the error message text should be revised? |
LGTM. If you don’t have commit access, please say your preferred git author name for the commit, i.e. Real Name <email@address>.
I don't have commit access.
Please commit at KOMATA Manabu <k-mana@mve.biglobe.ne.jp>.
Looks great!
clang/test/Driver/unsupported-target-arch.c | ||
---|---|---|
33 | The part ", please use -triple or -arch" is almost always wrong. For cc1, the error is due to a specified but unknown -triple, we should not suggest specifying -triple. I just removed the hint from the diagnostic. |
clang/test/Driver/unsupported-target-arch.c | ||
---|---|---|
33 | -arch is a driver option, for darwin targets though. But I agree that it's probably best to leave it out since it's more misleading than helping here (and -triple is indeed not a relevant option here). |
Separate side note; the suggested option -triple isn't really a clang driver level option, and -arch is only relevant for darwin targets (AFAIK) - so maybe the error message text should be revised?