September 2005 - Darwin-dev

Re: What is different about the way Apple builds GCC ?
by Sam Hart 26 Sep '05

26 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I wrote : S a m _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… during Apple's build process it copies the front-end ( e.g. gcc-3.3 ) and a number of other files from the currently installed version. So as you can see it's a bit of a chicken and egg situation. This opens up a new problem of how I am going to compile 3.5 and 4.0. to answer my own question, this appears not to be the case in 3.5 and 4.0. This email sent to site_archiver(a)lists.apple.com

1 0

RE: Prioritizing SBP2 daemon
by priti.rathi＠wipro.com 26 Sep '05

26 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcXCCgRDVRX//cVeT66a7AFI7zpk9wAd444u Thread-topic: Prioritizing SBP2 daemon _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: What is different about the way Apple builds GCC ?
by Sam Hart 26 Sep '05

26 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 23 Sep 2005, at 23:43, Stan Shebs wrote: After some digging around I think I have found some answers. I hit on a few things quite by accident. My patch will compile if I build the Apple way as root : % mkdir -p build/obj build/dst build/sym # gnumake install RC_OS=macos RC_ARCHS=ppc TARGETS=ppc \ SRCROOT=`pwd` OBJROOT=`pwd`/build/obj \ DSTROOT=`pwd`/build/dst SYMROOT=`pwd`/build/sym # ditto build/dst / when the build broke with : ld: Undefined symbols: ___guard ___stack_smash_handler So as you can see it's a bit of a chicken and egg situation. This opens up a new problem of how I am going to compile 3.5 and 4.0. S a m _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I'm somewhat out of touch with the specifics of your issue (my first inclination would be to double check system vs compiler versions, not all combos work), but I thought I'd explain the theory; Apple GCC has a special build process because we want to support multiple architectures and fat^Wuniversal binaries. The "FSF way" only ever builds one compiler, while the "Apple way" builds n-squared compilers for n architectures, with appropriate hosts and targets, bootstrapping the native versions, and lipo-ing together the target libraries. The Apple way also takes care of some smaller issues, such as some pathname trickery. it was a bit of a red herring, there was a problem with a file permission previous to this. Now the last section where I said that it builds the back-end properly ( i.e. cc1 etc. ) well the reason for that is that you need a compiler of the same version installed ( in this case 3.3 ) during Apple's build process it copies the front-end ( e.g. gcc-3.3 ) and a number of other files from the currently installed version. My proposed solution is to build with an old version of 3.3 that I can patch and build the fsf way ( because 1809 b0rks, ) then install it so that I can build a more recent patched version. I feel that I should explain that my goal is to build Darwin with protection added where possible, thus I have been working on compiler patches, lib patches and kernel patches and these are at various stages. I am also reworking libssp. I am currently building so if anyone is interested I'll let you know how it goes. My thanks go to Shantonu Sen and Kevin Van Vechten for providing the DarwinBuild environment. This email sent to site_archiver(a)lists.apple.com

1 0

Re: What is different about the way Apple builds GCC ?
by Sam Hart 26 Sep '05

26 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 24 Sep 2005, at 00:08, Boyd Waters wrote: hence the patch. that is not the case mainline ? or the Apple branch ? Of course it depends what you compare it to. S a m _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I sent a mail to the list a couple of days ago but it seems to have disappeared into the void ... I didn't know that SSP was enabled under Darwin, particularly Darwin PPC. The patch from reference 4 of your email turns SSP on by default I see also that GCC 4.1 stage 2 include a "Reimplementation of IBM's Pro Police Stack Detector" but I can't find any discussion of such. Richard Henderson of RedHat reworked SSP in May of this year. A lot of work followed that to make it work on platforms other than x86. In any event, GCC CVS HEAD seems to be rather unstable at the moment on PPC Darwin? certainly HEAD can be unstable from one moment to the next as it is a moving target. Or is that unfair? I have been working with the latest checkouts from both mainline and Apple's branch and they are both fine. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Corefoundation && Linux
by Alexander v. Below 26 Sep '05

26 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Bye Alex Am 25.09.2005 um 14:15 schrieb Nicola Vitacolonna: I am attempting to compile CF-Lite on Linux using mainline GCC 4.0.2. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Some updated info on the developer site on this subject would be great, because apparently many people seem to be having problems with compiling CoreFoundation under Linux (including me). I have been able to compile CF-299.35 - but not CF-368.1 - under Linux (I don't what current version is). I used gcc 3.3.5 (it didn't compile with gcc 2.95) ... This email sent to site_archiver(a)lists.apple.com

1 0

Re: Prioritizing SBP2 daemon
by Mike Smith 25 Sep '05

25 Sep '05

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

1 0

Re: Corefoundation && Linux
by Nicola Vitacolonna 25 Sep '05

25 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I am attempting to compile CF-Lite on Linux using mainline GCC 4.0.2. http://lists.apple.com/archives/Darwin-dev/2005/May/msg00221.html Best regards Nicola _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I have been able to compile CF-299.35 - but not CF-368.1 - under Linux (I don't what current version is). I used gcc 3.3.5 (it didn't compile with gcc 2.95). Here is my post about that: I have essentially followed the instructions at http:// developer.apple.com/darwin/cflite.html (applying the CF-299.linux.diff patch is crucial!), but in addition I had to replace the strlcpy and strlcat functions with strncpy and strncat anywhere in the source code. I hadn't played much with that, however. I was especially interested is the plug-in architecture provided by CoreFoundation, but I hadn't clue about how a CFBundleExecutable should look like in Linux. After several attempts and searches on the web, I had to give up. After browsing the source code, I suspect that only dyld and dll under Windows are supported by that version of CoreFoundation. This email sent to site_archiver(a)lists.apple.com

1 0

RE: Prioritizing SBP2 daemon
by priti.rathi＠wipro.com 25 Sep '05

25 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcXBnC6QEfAWL/aHTj6xauFuVwCPIQAK1/nE Thread-topic: Prioritizing SBP2 daemon _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

RE: Prioritizing SBP2 daemon
by Kevin Elliott 25 Sep '05

25 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com At 10:14 +0530 on 9/25/05, priti.rathi(a)wipro.com wrote: Yes that makes sense. How to lad the driver from utility? Is it possible that while loading the drive i can launch the application..??? Not really possible. The connection between IOKit and user space is not really designed for drivers to "push" stuff into user space. The normal pattern is for user space to connect to drivers (and for a lot of reasons that's a really good thing). Their was some talk of allowing drives to have secondary app for configuration and the like, but that was a long time ago and I think the general consensus was that it was a bad idea. In any case, it's not something the system supports today. I have to go back to my original question- what exactly are you trying to do? If your trying to prevent IOKit from doing something, you really need to be in the kernel. Based on what you've posted so far the technique your trying to use isn't going to work. The system simply isn't designed to work that way. -- ______________________________________________________ Arguing with an engineer is like wrestling with a pig in mud. After a while, you realize the pig is enjoying it. ______________________________________________________ Kevin Elliott <mailto:kelliott@mac.com> AIM/iChatAV: kelliott(a)mac.com (video chat available) ______________________________________________________ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

RE: Prioritizing SBP2 daemon
by priti.rathi＠wipro.com 25 Sep '05

25 Sep '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcXBSWf151Crb6yLRCeXZ15xcwvO/QAQmFto Thread-topic: Prioritizing SBP2 daemon _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0