| Summary: | Incorrect V_JITOperation_EJ call implementation in DFG for 32-bit ports. | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Julien Brianceau <jbriance> | ||||
| Component: | JavaScriptCore | Assignee: | Nobody <webkit-unassigned> | ||||
| Status: | RESOLVED FIXED | ||||||
| Severity: | Normal | CC: | commit-queue, ggaren, mark.lam, ossy, rgabor, yong.li.webkit, zherczeg | ||||
| Priority: | P2 | ||||||
| Version: | 528+ (Nightly build) | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Bug Depends on: | |||||||
| Bug Blocks: | 108645 | ||||||
| Attachments: |
|
||||||
|
Description
Julien Brianceau
2014-02-27 05:03:12 PST
Created attachment 225357 [details]
Fix V_JITOperation_EJ call implementation in DFG for 32-bit ports.
Was this a speculative fix? (In reply to comment #2) > Was this a speculative fix? Yes, I only checked compilation. I'm going to retrieve my env to launch run-layout-jsc tests on my arm traditional board. (In reply to comment #3) > Yes, I only checked compilation. I'm going to retrieve my env to launch run-layout-jsc tests on my arm traditional board. run-layout-jsc reports the same results on my board with and without this patch: not better, not worse. > run-layout-jsc reports the same results on my board with and without this patch: not better, not worse.
Interesting. So this patch is needed or not? I suspect this code is not triggered.
(In reply to comment #5) > > run-layout-jsc reports the same results on my board with and without this patch: not better, not worse. > > Interesting. So this patch is needed or not? I suspect this code is not triggered. You're right. If I remove this function, compilation is still ok so it's not even compiled actually :) What's the best choice, removing it or fixing it in case we'd need it later? I think we should trigger it somehow, otherwise we don't know whether the fix is correct :( (In reply to comment #7) > I think we should trigger it somehow, otherwise we don't know whether the fix is correct :( As it's like http://trac.webkit.org/changeset/157797 for the tag/payload swap, I'm pretty sure the fix is correct. However if the code is not compiled, I won't be opposed to just remove it. Comment on attachment 225357 [details]
Fix V_JITOperation_EJ call implementation in DFG for 32-bit ports.
Looks right to me
Comment on attachment 225357 [details]
Fix V_JITOperation_EJ call implementation in DFG for 32-bit ports.
Withdraw r+. A test case should have helped to find this problem.
Comment on attachment 225357 [details]
Fix V_JITOperation_EJ call implementation in DFG for 32-bit ports.
r=me
Comment on attachment 225357 [details] Fix V_JITOperation_EJ call implementation in DFG for 32-bit ports. Clearing flags on attachment: 225357 Committed r164813: <http://trac.webkit.org/changeset/164813> All reviewed patches have been landed. Closing bug. |