make -fmodules-codegen and -fmodules-debuginfo work also with PCHsAllow to build PCH's (with -building-pch-with-obj and the extra .o file)with -fmodules-codegen -fmodules-debuginfo to allow emitti
make -fmodules-codegen and -fmodules-debuginfo work also with PCHsAllow to build PCH's (with -building-pch-with-obj and the extra .o file)with -fmodules-codegen -fmodules-debuginfo to allow emitting shared codeinto the extra .o file, similarly to how it works with modules. A bit ofa misnomer, but the underlying functionality is the same. This saves upto 20% of build time here. The patch is fairly simple, it basically justduplicates -fmodules checks to also alternatively check-building-pch-with-obj.This already got committed as cbc9d22e49b434b6ceb2eb94b67079d02e0a7b74,but then got reverted in 7ea9a6e0220da36ff2fd1fbc29c2755be23e5166because of PR44953, as discussed in D74846. This is a corrected versionwhich does not include two places for the PCH case that aren't includedin the modules -fmodules-codegen path either.Differential Revision: https://reviews.llvm.org/D69778
show more ...
Revert "make -fmodules-codegen and -fmodules-debuginfo work also with PCHs"This caused PR44953. See also the discussion on D74846.This reverts commit cbc9d22e49b434b6ceb2eb94b67079d02e0a7b74.
make -fmodules-codegen and -fmodules-debuginfo work also with PCHsAllow to build PCH's (with -building-pch-with-obj and the extra .o file)with -fmodules-codegen -fmodules-debuginfo to allow emitting shared codeinto the extra .o file, similarly to how it works with modules. A bit ofa misnomer, but the underlying functionality is the same. This saves upto 20% of build time here.Differential Revision: https://reviews.llvm.org/D69778
Modular Codegen: Separate flags for function and debug info supportThis allows using and testing these two features separately. (noteably,debug info is, so far as I know, always a win (basically).
Modular Codegen: Separate flags for function and debug info supportThis allows using and testing these two features separately. (noteably,debug info is, so far as I know, always a win (basically). But functionmodular codegen is currently a loss for highly optimized code - wheremost of the linkonce_odr definitions are optimized away, so providingweak_odr definitions is only overhead)llvm-svn: 300104