Bug 74244

Summary: v8 benchmark takes 12-13 million function call slow paths due to extra arguments
Product: WebKit Reporter: Geoffrey Garen <ggaren>
Component: New BugsAssignee: Geoffrey Garen <ggaren>
Status: RESOLVED FIXED    
Severity: Normal    
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Attachments:
Description Flags
Patch fpizlo: review+

Geoffrey Garen
Reported 2011-12-10 15:20:12 PST
v8 benchmark takes 12-13 million function call slow paths due to extra arguments
Attachments
Patch (139.56 KB, patch)
2011-12-11 14:17 PST, Geoffrey Garen
fpizlo: review+
Geoffrey Garen
Comment 1 2011-12-11 14:17:30 PST
Oliver Hunt
Comment 2 2011-12-11 15:01:34 PST
Comment on attachment 118713 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=118713&action=review r=me, though it would be good to verify kraken + sunspider are okay > Source/JavaScriptCore/interpreter/Interpreter.h:88 > - enum { MaxLargeThreadReentryDepth = 93, MaxSmallThreadReentryDepth = 32 }; > + enum { MaxLargeThreadReentryDepth = 93, MaxSmallThreadReentryDepth = 16 }; > #else > - enum { MaxLargeThreadReentryDepth = 256, MaxSmallThreadReentryDepth = 32 }; > + enum { MaxLargeThreadReentryDepth = 256, MaxSmallThreadReentryDepth = 16 }; Sadness -- we really should switch to stack size limits rather than reentry shenanigans :-/
Geoffrey Garen
Comment 3 2011-12-11 16:29:27 PST
> r=me, though it would be good to verify kraken + sunspider are okay Neutral on Kraken+SunSpider. > Sadness -- we really should switch to stack size limits rather than reentry shenanigans :-/ Yeah. I think we should also find the giant object(s) we're putting on the stack and put them on the heap instead.
Geoffrey Garen
Comment 4 2011-12-11 16:35:57 PST
Note You need to log in before you can comment on or make changes to this bug.