<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content="text/html; charset=windows-1255" http-equiv=Content-Type>
<META name=GENERATOR content="MSHTML 8.00.6001.18876"></HEAD>
<BODY dir=ltr bgColor=#ffffff text=#000000>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN class=750090008-09022010>Hi
Shachar,</SPAN></FONT></DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010>Interesting problem. Here are some
thoughts:</SPAN></FONT></DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN class=750090008-09022010>Can
you control the level of optimization used by the customer? Does -O0 create
identical object code?</SPAN></FONT></DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN class=750090008-09022010>What
are the differences in code? Perhaps they're related to the different paths? (If
the lengths of the paths are different, and they're stored somewhere in the
object code, that'll change the results)</SPAN></FONT></DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN class=750090008-09022010>You
might also want to compare the assembly-language output (-S) to narrow down the
problem.</SPAN></FONT></DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN class=750090008-09022010>If the
customer needs cygwin for internal use, I'd suggest using a virtualization
solution such as VirtualBox instead - the integration with Windows is quite
impressive.</SPAN></FONT></DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010>Cheers,</SPAN></FONT></DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN
class=750090008-09022010></SPAN></FONT> </DIV>
<DIV><FONT color=#0000ff size=2 face=Arial><SPAN class=750090008-09022010>
Rony</SPAN></FONT></DIV><FONT color=#0000ff size=2 face=Arial></FONT><BR>
<DIV dir=ltr lang=en-us class=OutlookMessageHeader align=left>
<HR tabIndex=-1>
<FONT size=2 face=Tahoma><B>From:</B> linux-il-bounces@cs.huji.ac.il
[mailto:linux-il-bounces@cs.huji.ac.il] <B>On Behalf Of </B>Shachar
Shemesh<BR><B>Sent:</B> Tuesday, February 09, 2010 8:37 AM<BR><B>To:</B>
linux-il<BR><B>Subject:</B> toolchain's output depends on toolchain used to
build the compiler?<BR></FONT><BR></DIV>
<DIV></DIV>Hi all,<BR><BR>I'm having a weird problem at a client's. They need a
toolchain for both Windows (cygwin) and Linux. What's more, then need the two
toolchains to match exactly - given identical source code, the toolchains must
produce the exact same machine code (tested by doing objdump -d and diffing the
outputs). The toolchains are for bare metal power-pc (powerpc-eabi target).<BR><PRE class=moz-signature cols="72"><SPAN class=750090008-09022010><FONT color=#0000ff size=2 face=Arial> [...] </FONT></SPAN></PRE></BODY></HTML>