October 2006 - Darwin-dev - lists.apple.com

Re: MacOS X equivalent of swapon/swapoff?
by Jonas Maebe 11 Oct '06

11 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 11 okt 2006, at 14:52, Steve Sisak wrote: This is no longer needed as of Mac OS X 10.4 Jonas _______________________________________________ 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… It's somewhat annoying to need 4GB free on the root volume in order to burn a DVD. This email sent to site_archiver(a)lists.apple.com

1 0

Re: MacOS X equivalent of swapon/swapoff?
by Steve Sisak 11 Oct '06

11 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Ditto for user home directories (although I know how to do that with NetInfo). -Steve _______________________________________________ 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… At 8:31 AM -0400 10/11/06, John Francini wrote: I hadn't thought about it that way; I guess I was still going by the 'old school' idea that there might be times when a process would be completely paged out (in other words, *swapping* rather than paging), and that therefore contiguous disk space would have been a good thing. If you work out the statistics, I don't think there's any need for such a distinction (except in degenerate cases). Another thing to think about is that having a separate swap _partition_ on a system with a single drive would likely decrease performance as the drive has to constantly seek between the system and swap partitions, which are going to be on different parts of the disk. A separate swap drive would be another story, but in the real world that is only likely on servers and possibly high-end workstations. That said, as a developer who has to keep multiple boot partitions (for different OS versions), I do wish Mac OS X provided an easy way to specify an alternate volume to use for all scratch allocations (VM, /tmp, etc.), rather than forcing them to be on the boot volume. It's somewhat annoying to need 4GB free on the root volume in order to burn a DVD. Allowing me to redirect all boot partitions' scratch storage to a single "big disk" would reduce the size of an individual boot partition significantly. This email sent to site_archiver(a)lists.apple.com

1 0

Re: MacOS X equivalent of swapon/swapoff?
by John Francini 11 Oct '06

11 Oct '06

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: Shipping a compiler with applications?
by Graham J Lee 11 Oct '06

11 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 11 Oct 2006, at 05:06, Andrew Miller wrote: How do I go about getting a working build of ld? -- Graham J Lee http://www.thaesofereode.info/ _______________________________________________ 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… 1) Is it true that all OSX boxes come with gcc / ld on them? No - they're part of the developer tools package. Is there no way you could depend on the developer tools? 2) How can I build an ld for maximum forwards and backwards compatibility? I have tried to build cctools-622.3, but it doesn't work just by running make (I get the following error): ../best_arch.c: In function `cpusubtype_combine': ../best_arch.c:478: error: `CPU_TYPE_VEO' undeclared (first use in this function) ../best_arch.c:478: error: (Each undeclared identifier is reported only once ../best_arch.c:478: error: for each function it appears in.) ../best_arch.c:483: error: `CPU_SUBTYPE_VEO_1' undeclared (first use in this function) ../best_arch.c:484: error: `CPU_SUBTYPE_VEO_2' undeclared (first use in this function) grepping in /usr/include for CPU_TYPE_VEO doesn't find anything, and the situation is even worse for the SDKs in /Developer/SDKs/ (I would like to build against the oldest SDK possible). I can't find any mention of how to do this in any past postings to this list. The CPU_TYPE_VEO is definitely in the source code for xnu with 10.4.5 but you *could* just M-x comment-region (we all use emacs, right?) those parts of the switch/case where the VEO is referenced; after all you only really care about ppc, i386, and possibly the 64-bit variants. There's probably still some SPARC, PA_RISC and m68k/m88k code around if you look hard enough ;-) 3) It is okay to just grab the gcc / ld binaries from my system? When I downloaded Xcode, I had to click through a license saying I wouldn't redistribute it, but I have since seen a license on Apple's site which says it can be used on the terms of the Apple Public Source License, which allows me to redistribute the code: "2.1 Unmodified Code. You may use, reproduce, display, perform, internally distribute within Your organization, and Externally Deploy verbatim, unmodified copies of the Original Code, for commercial or non-commercial purposes, provided that in each instance:" ... I'm not from Apple, but the gcc and ld binaries *themselves* are definitely licensed in such a way as to be redistributable. In the case of gcc, you need to also make the source code available. In the case of cctools, you need to include a notice of the licence terms and describe how the source code may be obtained. Xcode of course is not does not have such a licence ;-) This email sent to site_archiver(a)lists.apple.com

1 0

Re: Darwin-dev Digest, Vol 3, Issue 264
by Michael Smith 11 Oct '06

11 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Oct 10, 2006, at 10:34 AM, Alison Cassidy <cooties(a)mac.com> wrote: Does anyone know if there is a programmatic way of switching VM on and off, similar to the way Linux does it? I've good reason to do this (honest!) and need some way to disable VM for the duration of running my app or at the least, ensure it doesn't flush to disk at the wrong moment. Alison, = Mike _______________________________________________ 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… Let me start by asking the usual question: what is it that you are actually trying to achieve? While we wait for an answer, let's dissect what you asked above as an educational exercise. One does not "switch VM on and off" in Mac OS X or Linux. Virtualisation of the process address space is fundamental behaviour (modulo the no-MMU 'Linux' variants) of both systems; to "switch it off" can be considered impossible. You mention "swapon/swapoff", as though you assume that these tools do this. They do not. In the case of Linux, the operator can manually add and remove disk resources used for backing anonymous memory. Under Mac OS X this process is automatic and invisible to the user. By removing all of these resources, assuming that the system has enough physical memory, you would prevent paging out dirty pages when you need free pages. In this situation you haven't done anything about the need for free pages, so you will cannibalise pages in the system that are not backed by swap; that means program and library text/data pages, and pages mapped from files. For the most part this is a degenerate mode of operation; your pool of free-able pages is small and resident text pages tend to be part of your working set, so if you steal them for some other use some other page will shortly be stolen to back this text again, resulting in another I/O to refill it. This manifests as thrashing, because by making it impossible to clean dirty pages even after they are no longer part of the working set you have effectively reduced the amount of usable physical memory in the system. This can be mitigated to a degree by disabling overcommit (agan, not something that Mac OS X supports), and requiring that all allocations are fully backed by physical pages. You will find that no modern desktop environment will function in this mode. To address the spirit of your request; far too often we hear from developers an argument along the lines of "my app wants total control of the system". Well, sorry, you can't have it. You don't own the system, you cannot own the system, and you never will own the system. 8) It is up to you as a developer to work out how to get your job done while being a responsible partner in the operation of the system as a whole. There are other components that have real, necessary work to do, and you cannot necessarily expect them to either a) know that you are being selfish, or b) care. And you certainly can't expect them to be happy when c) you try to steal away from them the resources that they need and expect. That being said, there are ways to both express your needs, and to meet reasonable realtime targets without the sort of drastic measures that you allude to in your request. If we understand what you're trying to do a little better, it's possible that we can guide you in a useful direction, or at least explain what real constraints you're coming up against. This email sent to site_archiver(a)lists.apple.com

1 0

Re: MacOS X equivalent of swapon/swapoff?
by Tim Seufert 11 Oct '06

11 Oct '06

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

Shipping a compiler with applications?
by Andrew Miller 11 Oct '06

11 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Thunderbird 1.5.0.4 (X11/20060516) Hi, How do I go about getting a working build of ld? Thanks and Best regards, Andrew _______________________________________________ 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… My application generates C source code, compiles and links it to a shared object, and then loads it into the core. On non-OSX platforms, this is supported by compiling gcc and GNU binutils, and shipping the minimum set of stripped binaries needed for the compiler to work in my packages. However, GNU binutils doesn't seem to support OSX properly, so I assume I need to use cctool from Apple. A few questions: 1) Is it true that all OSX boxes come with gcc / ld on them? Can I just assume that it will be there, or is it likely to be missing on non-developer systems? Is there a list of what files are on a pristine install of OSX for each version anywhere? 2) How can I build an ld for maximum forwards and backwards compatibility? I have tried to build cctools-622.3, but it doesn't work just by running make (I get the following error): Building on: Darwin <hostname> 8.7.0 Darwin Kernel Version 8.7.0: Fri May 26 15:20:53 PDT 2006; root:xnu-792.6.76.obj~1/RELEASE_PPC Power Macintosh powerpc with Xcode 2.4 cc -O -g -I../../include -Wall -Wno-long-double -no-cpp-precomp -fno-builtin-round -D_MACH_I386_THREAD_STATUS_FPSTATE_LEGACY_FIELD_NAMES_ -D_ARCHITECTURE_I386_FPU_FPSTATE_LEGACY_FIELD_NAMES_ -static -DRLD -DSA_RLD -fno-builtin -finline -fno-keep-inline-functions -fsigned-bitfields -force_cpusubtype_ALL -msoft-float -nostdinc -I/System/Library/Frameworks/Kernel.framework/PrivateHeaders/libsa -I/System/Library/Frameworks/Kernel.framework/PrivateHeaders -I/System/Library/Frameworks/Kernel.framework/Headers/libsa -I/System/Library/Frameworks/Kernel.framework/Headers/bsd -I/System/Library/Frameworks/Kernel.framework/Headers -DKERNEL -DKERNEL_PRIVATE -DAPPLE -DNeXT -DLIBSA_PRIVATE -c \ -MD -dependency-file ./get_toc_byte_sex.d \ -o ./get_toc_byte_sex.o ../get_toc_byte_sex.c ../get_toc_byte_sex.c:25:16: No include path in which to find ar.h I do have an ar.h, but it is in /usr/include rather than in any of the directories given by the -I command lines. I can get past the error by using: make CFLAGS="-I/usr/include -I`pwd`/include" but it fails again soon after... cc -I/usr/include -I/Users/cmiss/Build/cctools-622.3/include -static -DRLD -DSA_RLD -fno-builtin -finline -fno-keep-inline-functions -fsigned-bitfields -force_cpusubtype_ALL -msoft-float -nostdinc -I/System/Library/Frameworks/Kernel.framework/PrivateHeaders/libsa -I/System/Library/Frameworks/Kernel.framework/PrivateHeaders -I/System/Library/Frameworks/Kernel.framework/Headers/libsa -I/System/Library/Frameworks/Kernel.framework/Headers/bsd -I/System/Library/Frameworks/Kernel.framework/Headers -DKERNEL -DKERNEL_PRIVATE -DAPPLE -DNeXT -DLIBSA_PRIVATE -c \ -MD -dependency-file ./best_arch.d \ -o ./best_arch.o ../best_arch.c ../best_arch.c: In function `cpusubtype_combine': ../best_arch.c:478: error: `CPU_TYPE_VEO' undeclared (first use in this function) ../best_arch.c:478: error: (Each undeclared identifier is reported only once ../best_arch.c:478: error: for each function it appears in.) ../best_arch.c:483: error: `CPU_SUBTYPE_VEO_1' undeclared (first use in this function) ../best_arch.c:484: error: `CPU_SUBTYPE_VEO_2' undeclared (first use in this function) grepping in /usr/include for CPU_TYPE_VEO doesn't find anything, and the situation is even worse for the SDKs in /Developer/SDKs/ (I would like to build against the oldest SDK possible). I can't find any mention of how to do this in any past postings to this list. 3) It is okay to just grab the gcc / ld binaries from my system? When I downloaded Xcode, I had to click through a license saying I wouldn't redistribute it, but I have since seen a license on Apple's site which says it can be used on the terms of the Apple Public Source License, which allows me to redistribute the code: "2.1 Unmodified Code. You may use, reproduce, display, perform, internally distribute within Your organization, and Externally Deploy verbatim, unmodified copies of the Original Code, for commercial or non-commercial purposes, provided that in each instance:" ... Could someone (from Apple) please clarify which license applies to the gcc and ld binaries from the Xcode 2.4? If anyone knows how portable they are to older systems, that would also be useful. This email sent to site_archiver(a)lists.apple.com

1 0

Re: cp -f option
by seebs＠plethora.net 10 Oct '06

10 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <C15179E2.49BE8%jellis(a)dhnet.us>, Jeffrey Ellis writes: >Sorry for being dense. Not sure I understand exactly what this does. Can >someone explain this...? > > -f For each existing destination pathname, remove it and create a new > file, without prompting for confirmation regardless of its >permissions. > >Doesn't cp do this by default unless you use -i or -n ...which this >overrides? No. $ cd /tmp $ cp /dev/null t $ chmod 600 t $ cp /dev/null t $ chmod 0 t $ cp /dev/null t cp: t: Permission denied $ cp -f /dev/null t $ -s _______________________________________________ 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

cp -f option
by Jeffrey Ellis 10 Oct '06

10 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: Acbsw1Gxj/oyrVi2EdulngANk0QgEg== Thread-topic: cp -f option User-agent: Microsoft-Entourage/11.2.5.060620 Sorry for being dense. Not sure I understand exactly what this does. Can someone explain this...? -f For each existing destination pathname, remove it and create a new file, without prompting for confirmation regardless of its permissions. Doesn't cp do this by default unless you use -i or -n ...which this overrides? Thanks :) All My Best, Jeffrey _______________________________________________ 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: per user agents/login items
by Philip Lukidis 10 Oct '06

10 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcbsvLlr0vqyZeStR4GC8glaEYIsfAAAYneg Thread-topic: per user agents/login items Hello, thank you for your answer. For now I have found that LoginHook seem to be what I am looking for (http://developer.apple.com/documentation/MacOSX/Conceptual/BPSystemStartup/…) A LoginHook executes a script as root, so my script simply uses su to run my executable as the logged in user. Since LoginHook works for 10.3 and up, this seems OK. As I execute under the current user session, I can use the CF locale calls. If anyone knows another way to start a per user executable on login (for 10.3 and 10.4), please let me know. thanks, Philip Lukidis > -----Original Message----- > From: Stéphane Sudre [mailto:ssudre2@wanadoo.fr] > Sent: Tuesday, October 10, 2006 6:33 PM > To: Philip Lukidis > Cc: darwin-dev(a)lists.apple.com > Subject: Re: per user agents/login items > > > > On mardi, octobre 10, 2006, at 11:02 PM, Philip Lukidis wrote: > > > Hello. I have a daemon which provides access to a certain device. > > However, the device needs to have text displayed on its LCD > which fits > > the current locale. Unfortunately, I don't know of a way > to query the > > current user's locale in a daemon (assuming there is a > current user), > > as it runs under the root user. Any solution must work in 10.3 and > > 10.4... > > > > Therefore, is there a way to: > > a) Let the daemon be notified of user logon, and query the > locale of > > that user? > > b) Split user dependent functionality out of the daemon and > run it as > > a per user item. However, the daemon would need to > communicate with > > this item and get the locale, and I don't know how to create a per > > user startupitem. > > c) Could the daemon be notified of user login, and spawn a > helper app > > in that user session which would return the locale? > > 1. When no user is logged in, you can probably assume that the system > language is OK, you can find this information in the > /var/log/CDIS.custom file. > > 2. You can be informed of user login/logout via the > SystemConfiguration > Framework. Just take in count that Fast User Switching can pretend > there is a user logged in while it's only the login window being > displayed. > > 3. You can spawn a helper app and communicate with your daemon using > Message Ports, sockets, pipe, etc... I'm a big fan of > CFMessagePort but > YMMV. > > 4. If you use a helper app, you should probablynot look for > the locale > but use the CoreFoundation API to access localized strings. > > > _______________________________________________ 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