Bug 279637
| Summary: | [iOS 18.1 Beta] WebKit crashes when visiting heavy page | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Dazhen Pan <dazhen> |
| Component: | Layout and Rendering | Assignee: | Nobody <webkit-unassigned> |
| Status: | NEW | ||
| Severity: | Normal | CC: | bfulgham, bracho, rreno, simon.fraser, webkit-bug-importer, zalan |
| Priority: | P2 | Keywords: | InRadar |
| Version: | Safari 18 | ||
| Hardware: | iPhone / iPad | ||
| OS: | iOS 18 | ||
Dazhen Pan
1. Visit https://devblogs.microsoft.com/dotnet/performance-improvements-in-net-9/ which is a rather long page.
2. Safari will try to load it twice then display "A problem repeatedly occurred on".
This is on iPhone 15 Pro Max iOS 18.1
Note if I restart the phone, then this doesn't happen any more, but will repro again if phone is used for a while.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Radar WebKit Bug Importer
<rdar://problem/136330950>
Ryan Reno
I was able to reproduce on my phone as well. This is a highwater jetsam which can occur when the WebContent process crosses a soft limit and the system needs memory due to memory pressure. This is why the crash doesn't reproduce every time; the system won't kill the process if it doesn't need the memory.
From an initial look it seems like it might be due to a high amount of graphics memory being allocated, usually to back compositing layers in the page.
I'll try to get allocation logs to see what is prompting a large peak memory footprint.
Thank you for the report.
bracho
Any updates here? Since last week are the crashes coming more often.
Simon Fraser (smfr)
It's notable that the <html> is composited separately from the main document, and it uses a lot of memory. This might be triggered by a negative z-index somewhere. Is tile coverage working as expected for this layer?
It's also notable that each <pre> seems to take some memory even when outside the viewport, which is worth investigating.