Bug 24546 - [x86_64] JITArithmetic does not build
Summary: [x86_64] JITArithmetic does not build
Status: RESOLVED WORKSFORME
Alias: None
Product: WebKit
Classification: Unclassified
Component: JavaScriptCore (show other bugs)
Version: 528+ (Nightly build)
Hardware: PC Linux
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-03-12 06:03 PDT by Gabor Loki
Modified: 2010-06-12 14:58 PDT (History)
5 users (show)

See Also:

Attachments
Proposed fix for x86_64 (3.54 KB, patch)
2009-03-12 06:06 PDT, Gabor Loki 		ggaren: review-
Details | Formatted Diff | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Gabor Loki 2009-03-12 06:03:55 PDT 
On x86_64 a pointer should not be casted to unsigned int, like in 
' __ cmpl_im(reinterpret_cast<unsigned>(numberStructure), FIELD_OFFSET(JSCell, m_structure), X86::edx)' expressions. uintptr_t should be used instead of unsigned.

Additional, the suffixes or operands are invalid for 'push' and 'pop' in isSSE2Present if the compiler is GCC.

I am going to attach a fix.
Comment 1 Gabor Loki 2009-03-12 06:06:45 PDT 
Created attachment 28527 [details]
Proposed fix for x86_64
Comment 2 Eric Seidel (no email) 2009-05-11 06:01:36 PDT 
This looks very simple to review.  I'm surprised it's sat for 2 months in the queue.

Any JS reviewers?
Comment 3 Geoffrey Garen 2009-05-11 09:53:33 PDT 
Comment on attachment 28527 [details]
Proposed fix for x86_64

There's no harm in this patch, but it won't work. On 64bit, you can't compare a pointer to a 32bit immediate. This code should be #ifdef'd out in a 64bit build.
Comment 4 Alexey Proskuryakov 2010-06-12 14:58:24 PDT 
A build fix from early 2009 is almost certainly obsolete, closing the bug. Please feel free to re-open if you intend to follow up on this.