| Summary: | [JSC] Upstream iOS Stack bound checking | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Benjamin Poulain <benjamin> | ||||
| Component: | New Bugs | Assignee: | Benjamin Poulain <benjamin> | ||||
| Status: | RESOLVED INVALID | ||||||
| Severity: | Normal | CC: | fpizlo, ggaren, oliver | ||||
| Priority: | P2 | ||||||
| Version: | 528+ (Nightly build) | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Bug Depends on: | 110858 | ||||||
| Bug Blocks: | |||||||
| Attachments: |
|
||||||
|
Description
Benjamin Poulain
2013-02-25 15:18:46 PST
Created attachment 190137 [details]
Patch
Comment on attachment 190137 [details] Patch Clearing flags on attachment: 190137 Committed r144004: <http://trac.webkit.org/changeset/144004> All reviewed patches have been landed. Closing bug. I'm sure the stack bounds used to be thread local, i'm not sure why that can't just be the rule.. I must be missing something obvious, but I don't understand why the Parser and BytecodeGenerator can't cache the stack bounds. Those are temporary, stack-local objects. Even though WebKit in general may run on the main or web thread, we shouldn't every start parsing on one thread and then switch to another, right? (In reply to comment #5) > I must be missing something obvious, but I don't understand why the Parser and BytecodeGenerator can't cache the stack bounds. Those are temporary, stack-local objects. Even though WebKit in general may run on the main or web thread, we shouldn't every start parsing on one thread and then switch to another, right? That was my thought. Many moons ago that's essentially what they did. Then gavin refactored, then the thread local disappeared. I'm not sure why exactly. (In reply to comment #5) > I must be missing something obvious, but I don't understand why the Parser and BytecodeGenerator can't cache the stack bounds. Those are temporary, stack-local objects. Even though WebKit in general may run on the main or web thread, we shouldn't every start parsing on one thread and then switch to another, right? (Also workers _do_ only run on a single thread) Sorry, I do not have a clue, I did not work on the original bug. I did this change to avoid an horrible workaround in StackBounds. I'll try to find the original radar tomorrow. After looking at why that was needed in the first place, I think this is wrong. Can you guys confirm we never enter the event loop or a WebKit callback from BytecodeGenerator and Parser? If that is the case, I will remove the iOS changes on both side. > Can you guys confirm we never enter the event loop or a WebKit callback from BytecodeGenerator and Parser?
Confirmed.
And done. Thank you for double checking this. |