http://d.puremagic.com/issues/show_bug.cgi?id=1230

           Summary: std.math.poly asm segfaults
           Product: DGCC aka GDC
           Version: 0.23
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Phobos
        AssignedTo: dvdfrdmn@users.sf.net
        ReportedBy: larsivar@igesund.net


The std.math.poly function segfaults on Linux. Can be tested by running the unittests for that module. The unittest passes if the asm block is disabled.

The asm works fine with DMD (no segfaults), but Don Clugston says:
"It's clear that the asm code for poly() is buggy - it assumes a frame pointer
in EBP, but the ABI spec does not mandate the use of frame pointers."

Tested with GDC 0.23 and latest svn.

The issue has a corresponding ticket for Tango, http://www.dsource.org/projects/tango/ticket/467


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1230





------- Comment #1 from bugzilla@digitalmars.com  2007-06-28 22:28 -------
I'll adjust the doc to mandate the setting up of EBP, unless 'naked' is used.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1230





------- Comment #2 from larsivar@igesund.net  2008-01-24 16:40 -------
This is presumably due to GDC not following the D calling convention at all, which is a major PITA.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1230


Iain Buclaw <ibuclaw@ubuntu.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


--- Comment #3 from Iain Buclaw <ibuclaw@ubuntu.com> 2012-09-03 08:16:17 PDT ---
D_InlineAsm (and the D Inline Assembler implementation) is no longer part of
GDC.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------