Summary: | New string formatting triggers crashes in compositing logging code | ||
---|---|---|---|
Product: | WebKit | Reporter: | Simon Fraser (smfr) <simon.fraser> |
Component: | WebKit Misc. | Assignee: | Darin Adler <darin> |
Status: | RESOLVED FIXED | ||
Severity: | Normal | CC: | darin, simon.fraser, tsavell, webkit-bot-watchers-bugzilla, webkit-bug-importer |
Priority: | P2 | Keywords: | InRadar |
Version: | WebKit Nightly Build | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
See Also: | https://bugs.webkit.org/show_bug.cgi?id=195210 | ||
Bug Depends on: | |||
Bug Blocks: | 194893 |
Description
Simon Fraser (smfr)
2019-02-27 20:13:46 PST
It's weird that characters is "0.000" which is obviously one of the other float-formatted strings. The caller is notable in that it takes a lot of Strings things. Looks like a bug in how "pad" combines with "hex". I’ll add unit tests for that, get it working, and re-land the String::format removal patch. My rollout caused storage tests to crash: https://build.webkit.org/results/Apple%20High%20Sierra%20Debug%20WK1%20(Tests)/r242199%20(8015)/results.html Do we need to roll out more? Tracking the crashing tests in https://bugs.webkit.org/show_bug.cgi?id=195210 I talked with sihui and she found that https://trac.webkit.org/changeset/242075/webkit is calling a function from r242014 and is crashing due to the rollout. Looks like we will need to roll this change out too to resolve the crashing tests. The pad function had an object lifetime bug. Now fixed and checked in. <https://trac.webkit.org/changeset/242308> I don’t understand how the other problem was caused by the rollout, so not sure it’s fixed. |