• MacTech Network:
  • Tech Support
  • |
  • MacForge.net
  • |
  • Apple News
  • |
  • Register Domains
  • |
  • SSL Certificates
  • |
  • iPod Deals
  • |
  • Mac Deals
  • |
  • Mac Book Shelf

MAC TECH

  • Home
  • Magazine
    • About MacTech in Print
    • Issue Table of Contents
    • Subscribe
    • Risk Free Sample
    • Back Issues
    • MacTech DVD
  • Archives
    • MacTech Print Archives
    • MacMod
    • MacTutor
    • FrameWorks
    • develop
  • Forums
  • News
    • MacTech News
    • MacTech Blog
    • MacTech Reviews and KoolTools
    • Whitepapers, Screencasts, Videos and Books
    • News Scanner
    • Rumors Scanner
    • Documentation Scanner
    • Submit News or PR
    • MacTech News List
  • Store
  • Apple Expo
    • by Category
    • by Company
    • by Product
  • Job Board
  • Editorial
    • Submit News or PR
    • Writer's Kit
    • Editorial Staff
    • Editorial Calendar
  • Advertising
    • Benefits of MacTech
    • Mechanicals and Submission
    • Dates and Deadlines
    • Submit Apple Expo Entry
  • User
    • Register for Ongoing Raffles
    • Register new user
    • Edit User Settings
    • Logout
  • Contact
    • Customer Service
    • Webmaster Feedback
    • Submit News or PR
    • Suggest an article
  • Connect Tools
    • MacTech Live Podcast
    • RSS Feeds
    • Twitter

ADVERTISEMENT
Volume Number:11
Issue Number:5
Column Tag:Compiler Settings

The Devil’s In The Details

Don’t get surprised by the Universal Headers

By Bill Karsh, billKarsh@aol.com

Note: Source code files accompanying article are located on MacTech CD-ROM or source code disks.

There is a particularly insidious problem in Apple’s Universal Headers. Whether you are affected at all, and how devastating it is, depends upon how you use their headers in your projects. Fortunately, the problem only affects 68K code generation. Here are the details, and what you can do as a work-around until new headers become available.

Which Header Versions?

The problem appears in the headers I got with Think C/C++ version 7, although these headers themselves do not seem to have a version number. The problem persists through the latest headers I got with MetroWerks CodeWarrior 5. These are version 2.0a3. Undoubtedly, you are using an affected version (I think they all are).

Structure Alignment Problem

The problem is a structure alignment conflict. Most compilers today let you apply some options regarding how structures are defined, throughout the whole of a project, or while certain pragmas are in effect. For example, you find a pop-up menu in CodeWarrior’s Processor Preferences panel, that let’s you select options called “68K,” “68K 4-byte,” or “PowerPC.” You can also use pragma statements like these:

#pragma options align=mac68k
#pragma options align=mac68k4byte
#pragma options align=native
#pragma options align=power
#pragma options align=reset

These can be used to override a globally set option. I believe that the last option stays in effect until either the next alignment statement, or the end-of-file is encountered.

In Symantec C++, the Compiler Settings page has a radio button group called “Struct Field Alignment,” that sets alignment options for a project.

Generally, the chosen alignment determines how much padding (extra space) is inserted into structures between fields, and how much padding is appended to the end of structures. Deciding which option you would want is a matter of trading-off between compactness of the data (less space) and higher performance (fields aligned on natural addresses for the field’s data-type). We won’t go into those decisions here. What is immediately essential is understanding two things.

First, these options are available to you (settable in an options page or by pragma statements) independently of whether you are generating PowerPC (ppc) code or 68K code.

Second, structures defined by the Mac OS, that is, defined in Apple’s headers, must have no extra padding in them. When the compiler comes across the definition of any Mac OS structure, like GrafPort, BitMap, or whatever, the current alignment option must be mac68k. If the current setting were anything else, your code, and the system software, would disagree on the offsets to various fields in a GrafPort, for example. That’s a disaster.

Inside Mac Headers

If you look inside virtually any Apple header, you will see one of the following examples near the top of the file:

#if defined(powerc) || defined (__powerc)
#pragma options align=mac68k
#endif

or

#if GENERATINGPOWERPC
#pragma options align=mac68k
#endif

Near the bottom you will see one of:

#if defined(powerc) || defined (__powerc)
#pragma options align=reset
#endif

or

#if GENERATINGPOWERPC
#pragma options align=reset
#endif

The intent here is to force the alignment type to mac68k while any Mac OS structures are defined. Then, align=reset puts the alignment back to whatever was in effect before. Can you see the problem? The pragma statements are made conditional upon what type of code is being generated, ppc or 68K. If you are generating ppc code only, there is no problem. If you are generating 68K code, the pragmas are not executed. You’ve got a bad problem in 68K code if, when the headers are compiled, the current alignment option is not mac68k. The pragma statements in the Apple headers were not supposed to be conditional upon anything. I’m guessing here, but it looks like a clear case of macro-mania-a-go-go.

Why Didn’t I See That?

I know, right now you are saying “how can any of my code work at all if this is true? Why didn’t I catch this immediately!?” There are several sneaky ways this problem might have escaped your notice thus far. For one thing, 68K compilers typically ship with mac68k as the default alignment option. If you never changed that, you would have been spared. Another possibility involves precompiled headers.

Your environment was shipped with precompiled headers. That is, many, but not all, of the Apple headers were precompiled at the factory into a giant header called MacHeaders68K (CW) or MacHeaders (Think), or something similar. Such a file is usually included in a global preferences dialog. That would be the Language panel in CW, or the Prefix page in Think’s dialog. There may be separate precompiled headers for C and C++. Again, chances are that you never removed it. Back at the factory, they compiled these precompiled files using the mac68k setting. Therefore, all the structures defined in there are fine. However, not all of the Apple headers are typically included in these things, only the most commonly used. This keeps symbol table sizes down, and speeds compile times.

Any time you find a need to explicitly include an Apple header in your sources, and it’s not among those precompiled at the factory, you are letting yourself in for it. You have to do something to make sure that when the compiler reads that included header, the alignment is mac68k. You have to take these steps yourself because the provision to do this automatically in the headers themselves is incorrectly coded.

What Do I Do?

If you understand what’s happening, you can choose whatever method you like to adjust for this; whatever works and is convenient. Let’s suppose you need Timer.h, and it’s not currently precompiled. Here are some possibilities:

Re-precompile MacHeaders68K, or whichever variant your project needs. Uncomment Timer.h in the source file that generates the precompiled header. Make sure the project’s global option is “68K” (CW) or “Align to 1 byte Boundary” (Think). Precompile a new MacHeaders, and include it in your working project’s preferences dialog.

Include Timer.h in a source file like this:

#pragma options align=mac68k
#include <Timer.h>
#pragma options align=reset

Repair the Apple header directly. Change this:

#if GENERATINGPOWERPC
#pragma options align=mac68k
#endif

to this:

#pragma options align=mac68k

Make the alignment options unconditional, as they were supposed to be. Do this for the reset lines too!

I know this will save you days of debugging and thousands of dollars of development costs. Could you send me just one dollar as a thank you? Worth a try.

 
MacTech Only Search:
Community Search:

 
 
 

 
 
 
 
 
  • SPREAD THE WORD:
  • Slashdot
  • Digg
  • Del.icio.us
  • Reddit
  • Newsvine
  • Generate a short URL for this page:



MacTech Magazine. www.mactech.com
Toll Free 877-MACTECH, Outside US/Canada: 805-494-9797
MacTech is a registered trademark of Xplain Corporation. Xplain, "The journal of Apple technology", Apple Expo, Explain It, MacDev, MacDev-1, THINK Reference, NetProfessional, Apple Expo, MacTech Central, MacTech Domains, MacNews, MacForge, and the MacTutorMan are trademarks or service marks of Xplain Corporation. Sprocket is a registered trademark of eSprocket Corporation. Other trademarks and copyrights appearing in this printing or software remain the property of their respective holders.
All contents are Copyright 1984-2010 by Xplain Corporation. All rights reserved. Theme designed by Icreon.
 
Nov. 20: Take Control of Syncing Data in Sow Leopard' released
Nov. 19: Cocktail 4.5 (Leopard Edition) released
Nov. 19: macProVideo offers new Cubase tutorials
Nov. 18: S Stardom anounces Safe Capsule, a companion piece for Apple's
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live
Nov. 17: Ableton releases Max for Live