MaskRay wrote: This PR will address a Linux kernel need. https://lore.kernel.org/all/CAFP8O3JkgQsH-4Lmr2W_teuvLjOCPi1htr9r3CO1O0yLyw=a...@mail.gmail.com/
The code already assembles with `clang -S -fno-integrated-as` and `gcc -c`, and this patch is to make `clang -c` work by removeing an unnecessary restriction in MCObjectStreamer. I plan on making separate, more nuanced changes to MCAsmStreamer in the future, (https://github.com/llvm/llvm-project/issues/62520#issuecomment-2096643143), but I believe it is not worth holding back the MCObjectStreamer improvement. Do you think this description is clear? I can create a Discourse post, but given the specific nature and subtlety of integrated assemblers, there might be limited general interest. BTW: `clang -c -save-temps` using an `.s` immediate file and MCAsmStreamer inherently has some differences with `clang -c` using MCObjectStreamer. For example, I've fixed a debug info issue https://github.com/llvm/llvm-project/pull/75022 and some clangDriver stuff I forget now. https://github.com/llvm/llvm-project/pull/91082 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
