Bug 113482
| Summary: | [webkitpy] infinite loop in _test_patch() retries | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Glenn Adams <glenn> |
| Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
| Status: | NEW | ||
| Severity: | Normal | CC: | dpranke, eric, lforschler, rniwa, thorton |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
Glenn Adams
In webkitpy.tool.bot.patchanalysistask.PatchAnalysisTask._test_patch(), there appear two cases where False is returned which can lead to an infinite retry loop:
(1) when second_results fail a different set of tests than first_results
(2) when the method falls through to the final return False without invoking report_failure()
When False is always returned from this method, the EWS queue will continue to retry testing the patch without stopping unless the patch is obsoleted or r- is marked.
I observed this happening today at http://queues.webkit.org/patch/195207, where 24 attempts were made to retry _test_patch() before I marked the patch as r- and then obsoleted the patch (just to make sure it would fail validation).
The EarlyWarningSystemTask should be modified ensure that some limit is applied to the number of retry attempts.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |