| Summary: | [BlackBerry] Assertions and assumptions in BackingStoreClient around m_frame and m_frame->view() are invalid | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Antonio Gomes <tonikitoo> | ||||
| Component: | WebKit BlackBerry | Assignee: | Antonio Gomes <tonikitoo> | ||||
| Status: | RESOLVED FIXED | ||||||
| Severity: | Normal | CC: | tonikitoo | ||||
| Priority: | P2 | ||||||
| Version: | 528+ (Nightly build) | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Attachments: |
|
||||||
Created attachment 141163 [details] (committed r116657, r=rbuis) Patch Comment on attachment 141163 [details] (committed r116657, r=rbuis) Patch LGTM. Comment on attachment 141163 [details] (committed r116657, r=rbuis) Patch Committed r116657. |
George Staikos 2012-05-06 23:04:05 EDT This is an interesting bug. IntPoint BackingStoreClient::scrollPosition() const { ASSERT(m_frame); return m_frame->view()->scrollPosition() - pointToSize(m_frame->view()->mini mumScrollPosition()); } m_frame->view() is null. m_frame is in the middle of construction and is trying to create its view but malloc is blocked on a mutex. Meanwhile the test driver, which runs in the Plaid Zone in terms of speed, is already pinch zooming the page. Realistically no user can ever hit this. In actuality, the assertions in backingstoreclient are wrong and need to be fixed. at a minimum, m_frame->view() can be null or invalid, and perhaps m_frame can be too. Adam / Antonio please fix in all branches.