October 2007 - Darwin-kernel

Re: Leopard SDK 10.4 strange phenomena libc
by Terry Lambert 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com $ gcc -mmacosx-version-min=10.4 -fPIC -c src/bundle.c -arch i686 -arch ppc -ansi -Wall -O2 -o build/bundle.o $ nm build/bundle.o 00000000 T _init U _system You say: -mmacosx-version-min=10.4 You say: AFAIK... everything is going well in fact, i was in the same case with a bigger project, but i got an error from ld: $ gcc -c src/obj.c -arch i686 -arch ppc -ansi -Wall -O2 -o build/obj.o $ gcc -bundle build/obj.o -o build/obj.so -mmacosx-version-min=10.4 -arch i686 -arch ppc -Wl,-syslibroot,/Developer/SDKs/MacOSX10.4u.sdk -framework CoreFoundation Undefined symbols for architecture i386: "_system$UNIX2003", referenced from: _method in obj.o .... You say: -mmacosx-version-min=10.4 [ ... ] % nm /usr/lib/libc.dylib | grep "T _printf" 0006efcc T _printf 00115be4 T _printf$LDBL128 000cb720 T _printf$LDBL64 - -- Terry _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… You need to read <sys/cdefs.h>, specifically the section I put in there about compilation environments. Or you need to read the Leopard developer release notes. Or both. On Oct 31, 2007, at 1:47 PM, mm w wrote: Compiler hears: "Hi, I want the legacy compilation environment, rather than a fully UNIX conformant environment." $ gcc -c src/bundle.c -arch i686 -arch ppc -ansi -Wall -O2 -o build/ bundle.o $ nm build/bundle.o 00000000 T _init U _system$UNIX2003 Compiler hears: "Hi, I want the default compilation environment for MacOS X 10.5, which is the UNIX conformant compilation environment" Linker hears: "Hi, I want to link against the 10.4 legacy libraries, even though I may have imported symbols that only exist in the 10.5 libraries; if that happened, can you please give me a link error?" This happens because an SDK includes headers, libraries, and so on, and you've asked it to link against the Tiger libraries. Does this make any sense to you? Yes. Functions which change behaviour because of UNIX conformance have symbol decorations that get attached to them as a result of which compilation environment you are using. You can mix compilation environments, as long as you only use symbols that were also supported in Tiger because of the UNIX conformance work that had been done before Tiger shipped (several of us have been working on UNIX conformance since November 2004 - it was a *very* big project, with not a lot of people on it full time). This is similar to the situation where we added 128 bit long double support to Tiger, and as a result, we decorated functions which had to know the size of long double (such as "printf") with $LDBL128 or $LDBL64, and provided weak symbols for legacy applications that didn't know about the decorations. For example, printf: Bottom line is that you should try to use the same compilation environment for all your code. Basically, pick an SDK you want to target, and target it across all your compilation units and linkages. This email sent to site_archiver(a)lists.apple.com

1 0

exception chaining and gdb watchpoints
by Brad Post 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcgcBE6djhhnazyTQiqc+vcODLlmyg== Thread-topic: exception chaining and gdb watchpoints _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Leopard SDK 10.4 strange phenomena libc
by mm w 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; bh=vs0GFEfw7sOIx07eFL33IYghiGYvs8e5xjqmN4VyxCY=; b=AjX1fI0+sRG1M9W07QGOjGM+mcnQVb68vPKiHfFkF7D0T4uRTnxcrDsOmXlJNe5GjgTM5H2w6O7hoAAyvGc6xb0/3JS6z6rcxgJnzplurTadZglLXmg+mDIyBOOuD9/u9k+HTU2CBFtuskt0mTwyMj4m4JEWSsLvhADVpTw3ndQ= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=KAdEZ5te3JKdYcPTSRHX4EJcDSSQvHJ6SP2/xhcn/VSsJ+LpHy8XbQusDhI8wP8hOQz65YSyxkeHqzFLbTR4CmF0FFZPI0FaHM0qfDWXbpCX3AUQdCz6nLtEdfI9/9wrIJHFvhQDExgbUmKNk5cPjWGYBANyjhlsDS/ng1JMV2M= Hi all, i provide a simple example: ------------------------------------------------------ bundle.c ------------------------------------------------------ #include <stdio.h> #include <stdlib.h> static int foo(const char *name) { const char *n = name; if(n) system("ls -la /"); return 0; } static int bar(void) { return foo("Finder"); } int init(void) { return bar(); } ------------------------------------------------------ $ gcc -mmacosx-version-min=10.4 -fPIC -c src/bundle.c -arch i686 -arch ppc -ansi -Wall -O2 -o build/bundle.o $ nm build/bundle.o 00000000 T _init U _system $ gcc -c src/bundle.c -arch i686 -arch ppc -ansi -Wall -O2 -o build/bundle.o $ nm build/bundle.o 00000000 T _init U _system$UNIX2003 AFAIK... everything is going well in fact, i was in the same case with a bigger project, but i got an error from ld: $ gcc -c src/obj.c -arch i686 -arch ppc -ansi -Wall -O2 -o build/obj.o $ gcc -bundle build/obj.o -o build/obj.so -mmacosx-version-min=10.4 -arch i686 -arch ppc -Wl,-syslibroot,/Developer/SDKs/MacOSX10.4u.sdk -framework CoreFoundation Undefined symbols for architecture i386: "_system$UNIX2003", referenced from: _method in obj.o .... the selected dylib object was the good one, but my produced object was rotten, a 10.5-libc-stub symbol was catched: $ nm build/obj.o 00000000 T _init U _system$UNIX2003 first i thought it was a option problem, 'maybe?' something was dropped , after many investigations, i discovered the only difference between my simple build test and my code was the order of my 'include' files, ---------------------------------------------------- #include <linkedlib.h> #include <CoreFoundation/CoreFoundation.h> ---------------------------------------------------- then, i decided to make the following tests: ---------------------------------------------------- #include <CoreFoundation/CoreFoundation.h> /* stdio.h stdlib.h */ #include <linkedlib.h> ---------------------------------------------------- ---------------------------------------------------- #include <stdio.h> #include <stdlib.h> #include <linkedlib.h> #include <CoreFoundation/CoreFoundation.h> ---------------------------------------------------- and everything's back to normal, i also made a new test, i removed my standard headers from my simple test file: $ gcc -fPIC -c src/bundle.c -arch i686 -arch ppc -ansi -Wall -O2 -o build/bundle.o src/bundle.c: In function 'foo': src/bundle.c:7: warning: implicit declaration of function 'system' src/bundle.c: In function 'foo': src/bundle.c:7: warning: implicit declaration of function 'system' $ nm build/bundle.o 00000000 T _init U _system Does this make any sense to you? -mmw _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

IOHIDSystem.h Question In Leopard
by Thorrin 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding:Message-ID; b=yVfXb6SCQPnmHlOhDHxFaRovtb758AFW4JxzYwZlvZsMaO0HuK9l2HW//ZC7x5Gzhqhkm4pZ8uniLjJhQ6z2R1ZC3pPz749V0iQOpKyzhseWWupKnrtzC9rc4W92rS9K9KevGOmPOAeP1ko9yzuYygnvIMR8ae6FVeUC3nygVOs=; Greetings, I am trying to port some code from a 10.4 system to a 10.5 system. In the old version of the code, we are including: <IOKit/hidsystem/IOHIDSystem.h> everything is grand there. Moving the code over to the new environment, the above line generates errors. (IOHIDEvent.h not found, IOHIDEvent not declared). I took a look in IOHIDSystem.h and it is trying to include: <IOKit/hid/IOHIDEvent.h> the problem is that this file doesn't exist anywhere in the Kernel.framework! Scrolling down inside IOHIDSystem.h, there is a function called dispatchEvent that tries to accept a parameter of IOHIDEvent. This, of course explains the errors, but what to do from here? Commenting out the line to include IOHIDEvent.h and the declaration of dispatchEvent fixes the compile problems, but surely breaks stuff somewhere else. I dont want to go around mucking around with system bundles in that fashion anyway. Any ideas as to what to do next? -Thorrin "Everyone is allergic to giant rocks hitting them in the head!" -Thorrin's elf chick after casting a 'meteor swarm' spell ;) __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Leopard Sources?
by Rick Macklem 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Tue, 30 Oct 2007, Anton Altaparmakov wrote: -rw-r--r-- 1 root wheel 4393120 Jun 24 12:27 mach_kernel -rw-r--r-- 1 root admin 8543400 Sep 24 2006 mach_kernel.orig rick ps: I suspect mach_kernel.orig was actually an older 10.4 that 10.4.9, looking at the date on it. Yea, it had crossed my mind that mach_kernel might be a fat binary. Thanks for correcting. I will point out that it does mean that I do use the kernel sources though (related to other messages in the thread) and I do have a rather large patch that adds nfsv4 support up anonymous ftp. I have been reluctant to post patches on this mailing list, since it is awkward to include the APSL notice in something like a source diff. When I've encountered issues that I believe to be bugs in the kernel sources, I have emailed Apple engineers about it. (In other words, Apple is getting some feedback w.r.t. their open sourced kernel. Whether they find this useful is another story, but...:-) rick _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Your comparison is flawed. You are comparing a single arch kernel (mach_kernel) with a bi-arch kernel (mach_kernel.org) This email sent to site_archiver(a)lists.apple.com

1 0

Re: Leopard Sources?
by Arvind Dalvi 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:references; bh=Vf0xd/MTIceODNOs+HjXqyYQjhUx1RP1io0WA09ZZ/s=; b=bhtCupYGh9IR4JNn3Cg5A8wvymwxCnlweGg6LjzSJF+22R4lnWKh9w4455sWuJrjc8CWe7EZ+2jFpprRG65RiuNDO0MStO/zJ2TfeBVBfsoLX0WG6pcAeqciiOBSPBwp8MQDkpHRvx3yK9k5w/0DhvqK79itehn9AFwT+Sx6Wd4= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:references; b=eMhaaSOFSRlIAJlpx9mAFVD+upApBEG+4HfBT+KyBOsuaaZURongswV5Rz4HtzX/nkym3XWUPYv4fjzOObRqXjBuLWMVk/nH7KsmkG/JQPHZOCIU13VeorHxGW0wMpTrHRc7XChMzqpXtdUDG0p4UppuebkN8CotU3ZMS5Zr6UI= _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Leopard Sources?
by Chuck Remes 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Oct 31, 2007, at 4:15 AM, Andreas Fink wrote: I can't resist at least one more comment. Great! I look forward to helping review your kernel patch. :-) cr _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… If Andreas wants to add SCTP support to the kernel, I have a proposal. When the leopard sources appear (in a few days, weeks or months), do the hard work to create the KPI infrastructure you need to support this protocol. Write the code, create tests if possible along with a test harness (hard work for kernel code), plus documentation and submit it to Apple as a patch. Publish this patch far and wide, take feedback from other interested kernel hackers and *maintain* the patch as new Leopard releases ship. Make it your mission to get the patch accepted in 10.5.x or whatever the next major release is in 20 months. Don't wait for Apple to do this work for you (once you have the sources). Do the work, create the patch and be its champion. I think Apple folks would be surprised. Thus far the level of "community" support for kernel patches barely rises above *apathy*. Anyone who created a kernel patch and tried to support it this way would probably garner some sub-rosa engineering support within Apple simply because they'd be impressed with the effort. Interesting idea. I'll definitively bite into that one... but it basically means no benefits because it basically means doing the same thing with every release. So to make a public API for just one release. HOWEVER: I've just seen the source has been posted... so back to WORK... This email sent to site_archiver(a)lists.apple.com

1 0

Re: Leopard Sources? (Radars)
by Andreas Fink 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Leopard Sources?
by Andreas Fink 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Leopard Sources?
by Andreas Fink 31 Oct '07

31 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0