[Magic-dev] Re: Magic problem with MacOSX 10.5

[Michael Doster]

Just so this is documented in the mailing list, I've found the  
solution to why OpenGL isn't compiled in 10.5.0 (10.5.1 as well) even  
when asked for.  LDFLAGS needs to be set to "-Wl,-dylib_file,/System/ 
Library/Frameworks/OpenGL.framework/Versions/A/Libraries/libGL.dylib:/ 
System/Library/Frameworks/OpenGL.framework/Versions/A/Libraries/ 
libGL.dylib".  See the following Apple forum entry for details:

http://lists.apple.com/archives/x11-users/2007/Oct/msg00146.html

This has been tested with the following on an Intel Mac:

magic-7.4.57
Tcl 8.4.16
Tk 8.4.16
BLT 2.4

Also with this compiled setup the I did not see the gdb error.

Michael Doster
On Jan 22, 2008, at 8:18 PM, R. Timothy Edwards wrote:

> Dear Dario,
>
>> I am Dario, a student of Brescia (Italy). I'm using Magic in a
>> Microelectronics course, to design a PLL in a 0.18u technology.
>> I have some problem to run Magic on my old Mac iBook G4, with MacOSX
>> 10.5.1 (Leopard with Service Pack and X11) installed.
>> I install Xcode, the development kit by Apple to compile the source
>> code. I correctly set the "./configure" and I succesfully compile
>> with the argoument "-with-interpreter=tcl" (I have TCL/TK 8.4.7
>> installed by default on my PC) for powerpc-apple-darwin9.1.0.
>> Everything is succesful (apart that it can't find OpenGL, so I use
>> X11 that is enough for me), but when I try to execute magic it give
>> me a "segmentation fault" in the "wish" application. In particular,
>> if I try to debug that error with:
>>
>> gdb /usr/local/lib/magic/tcl/magicexec
>> run
>>
>> it tells me:
>>
>> Program received signal EXC_BAD_ACCESS, Could not access memory.
>> Reason: KERN_INVALID_ADDRESS at address: 0x732e2050
>> 0x0059a49c in XDefaultColormap ()
>
> This is, I think, the third email I've gotten about this problem, two
> related to Magic and one to XCircuit.  The problem seems not to be
> the Magic or XCircuit code itself, but the calls to X11 colormap
> functions.  This problem seems to have started cropping up with
> OS-X 10.5 as people attempt to compile directly.  Previously, I
> think everyone compiled using "fink", although with my lack of
> knowledge about Mac OS-X, I don't know any of the details.
>
> What I understand about this particular problem is that OS-X
> crashes immediately on calls such as XDefaultColormap(),
> XInstallColormap(), and XAllocNamedColor().  In the last
> function listed, used in XCircuit, the colormap is retrieved from
> Tk_Colormap(), which returns a low-integer value (like, say, 4).
> This is very unlike colormap values used by X11, which tend to
> be addresses in upper memory spaced converted to an integer.
> That is not necessarily wrong, but it's really the only thing I
> can find that is remotely unusual in the stack traces.
>
> I poked around with Google for any mention of such a problem but
> came up with nothing.
>
> So I am copying this message to the magic-dev and xcircuit-dev
> mailing lists, hoping that one of the OS-X people monitoring the
> lists knows something about the problem, even (hopefully) how to
> fix it.
>
>                                     Regards,
>                                     Tim
>
> +-------------------------------- 
> +-------------------------------------+
> | Dr. R. Timothy Edwards (Tim)   | email:  
> tim.edwards at multigig.com     |
> | MultiGiG, Inc.                 | web:   http:// 
> www.multigig.com      |
> | 100 Enterprise Way Suite A-3   | phone: (831)  
> 621-3283               |
> | Scotts Valley, CA 95066        | cell:  (240)  
> 401-0616               |
> +-------------------------------- 
> +-------------------------------------+
>
> _______________________________________________
> magic-dev mailing list
> magic-dev at vlsi.cornell.edu
> http://vlsi.csl.cornell.edu/mailman/listinfo/magic-dev