| Summary: | [EFL][GTK] Bump up FDE relocation fixing patch to version accepted to LLVM trunk | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Akos Kiss <akiss> | ||||
| Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> | ||||
| Status: | RESOLVED FIXED | ||||||
| Severity: | Normal | CC: | cgarcia, commit-queue, gyuyoung.kim, ossy | ||||
| Priority: | P2 | ||||||
| Version: | 528+ (Nightly build) | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Bug Depends on: | |||||||
| Bug Blocks: | 143605 | ||||||
| Attachments: |
|
||||||
That also means that FTL should work with llvm from git master, right? Maybe we could use a git repo instead of a tarball in jhbuild, using the revision when the patch landed, and we don't need to use any patch. At least until a new version is released. Created attachment 242205 [details]
Proposed patch.
Discussed with KaL on irc that stability should be considered: we trust the 3.5.0 release of LLVM to be reasonably stable and applying 2 patches is not a big overhead. So, instead of grabbing a revision from master, I'd still go with this approach.
Did clean builds (update-webkit*-libs, update-webkit*-libs llvm, build-jsc --ftl-jit) both for EFL and GTK, all jsc tests pass.
Comment on attachment 242205 [details]
Proposed patch.
Thanks!
Comment on attachment 242205 [details] Proposed patch. Clearing flags on attachment: 242205 Committed r176546: <http://trac.webkit.org/changeset/176546> All reviewed patches have been landed. Closing bug. |
After some changes, the patch that was used to make LLVM work correctly in FTL JIT on x86-64/Linux/{EFL,GTK} got accepted to LLVM trunk. http://reviews.llvm.org/rL222760 Seems to be a good idea to replace the original but "unofficial" patch in Tools/{efl,gtk}/patches with the one accepted to LLVM trunk.