Merge lp://staging/~rengolin/cbuild/llvm into lp://staging/cbuild
Proposed by
Renato Golin
Status: | Merged |
---|---|
Approved by: | Matthew Gretton-Dann |
Approved revision: | no longer in the source branch. |
Merged at revision: | 456 |
Proposed branch: | lp://staging/~rengolin/cbuild/llvm |
Merge into: | lp://staging/cbuild |
Diff against target: |
36 lines (+15/-1) 2 files modified
lib/llvm.mk (+8/-1) tools/up_branch.sh (+7/-0) |
To merge this branch: | bzr merge lp://staging/~rengolin/cbuild/llvm |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Matthew Gretton-Dann | Approve | ||
Review via email: mp+167769@code.staging.launchpad.net |
Description of the change
This is a silly merge, using the default options for ARM builds on buildbots and test-suites.
There might have a way of getting the current CPU in CBuild, but if not, I'd rather force A9 than let the default A8, since Clang/LLVM still can't figure out the actual CPU and use it instead of the default.
To post a comment you must log in.
So this looks fine as is (its better than what is currently there).
A better way may be to use the GCCCONFARGS variable. This is set by the various build config Makefiles in cbuild/config (one for each queue type). This means that you can change the target build depending on which queue you use.
However, the GCCCONFARGS may not contain stuff that LLVM understands, so some rework of that or mangling of it may be necessary.
So this patch is an improvement on the current state (and so I'll approve it) - but there may be an even better way ahead.