[PATCH] D29050: [zorg] Add AOSP builder

Stephen Hines via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Jan 24 23:10:53 PST 2017


srhines added a comment.

Are you sure that you want to always be syncing AOSP? There is a good chance that the AOSP build could break for reasons beyond the control of the LLVM community. It might be better to pick a particular AOSP release to minimize those kinds of issues. One other potential problem is going to be the increasing presence of stricter warnings/errors from Clang. These can also cause issues with AOSP.



================
Comment at: buildbot/osuosl/master/config/builders.py:686
+                sync=True,
+                patch="aosp-O3-polly-before-vectorizer-unprofitable")}
+    ]
----------------
pzheng wrote:
> kristof.beyls wrote:
> > My understanding is that the 'aosp-O3-polly-before-vectorizer-unprofitable' patch is local to your builder, and needed to make it possible to use a non-AOSP-version of clang to build AOSP.
> > Would it be possible to also share this upstream somehow so that others could also set up AOSP builders, if wanted?
> Yes, that is correct. The pre-generated patches are local to each buildslave and are needed to make it possible to use a non-AOSP-version of clang to build AOSP.
> 
> Sure, I will post the patch I use if you are interested.
I agree. You should be able to point to a different clang installation today (once I submit a patch tonight/tomorrow that makes the version number issue less problematic). https://android-review.googlesource.com/q/topic:clang-3625443+(status:open%20OR%20status:merged) is the topic that I am testing in AOSP gerrit.


================
Comment at: zorg/buildbot/builders/AOSPBuilder.py:101
+    if target_clang:
+        command += " TARGET_CLANG=%s" % target_clang
+    if target_flags:
----------------
You can switch from TARGET_CLANG to LLVM_PREBUILTS_BASE along with LLVM_PREBUILTS_VERSION. You might also need to set/adjust LLVM_RELEASE_VERSION (currently 4.0, but obviously upstream will move past that).


https://reviews.llvm.org/D29050





More information about the llvm-commits mailing list