December 2002 - Darwin-kernel

Re: Kernel Extensions
by Stéphane Sudre 03 Dec '02

03 Dec '02

On mardi, dic 3, 2002, at 02:22 Europe/Paris, Justin C. Walker wrote: Libc and similar libraries are *not* available in the kernel (for one thing, the execution environment is much different than that for user mode). A much-reduced subset is available (cf. libcc_kext.a). Memcmp() is not in there, so it may just be that no one needed it when it was written (memcpy() is, for example). memcmp is in the kernel AFAIK. It's implemented via the IOKit. _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

testing for open files or volumes
by Alex Reynolds 03 Dec '02

03 Dec '02

I was wondering if there are ways to test whether a file is open by another process under Darwin (and by extension Mac OS X)? I have been testing out methods that use the functions: -- flock() -- lockf() -- fcntl() and Carbon methods that look at a catalog's nodeFlags against the kFSNodeLockedMask constant. None of these seem to return any positive indication that a file is open elsewhere in the OS. I was also curious about whether there is a way to determine if a volume is busy. There is a DiskImages framework that man pages for hdiutil hint at, but this framework appears to be private and so I don't know how to call any functions it might use to return whether this condition is true or not. Has anyone run into this sort of thing before and found a solution? Thanks, Alex _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Kernel Extensions
by Justin C. Walker 03 Dec '02

03 Dec '02

On Monday, December 2, 2002, at 01:38 PM, Eric Long wrote: I am discovering the world of kernel extensions and using Project Builder. I have some quick questions that I hope someone can easily answer. 1 - In documentation I've been looking at it indicates class methods should be declared static. It seems to imply this is for performance reasons. However, in playing with it, it seems as if methods have to be static period. I say this because my extension fails loading due to undefined symbol problems if I don't declare all methods static. The rules of engagement for C++ in the kernel are somewhat akin to the limitations for Embedded C++. All compilation for kernel extensions, whether IOKit gizmos or "just C", must be compiled for static linking. I can't say whether this requires static declarations of methods, but you should be able to verify whether that is the norm by checking a few IOKit projects in the Darwin repository. 2 - I would like to use exceptions. Sorry; see above (Embedded C++). Exceptions aren't possible for kernel operation. I also find it strange that many standard ANSI functions, like memcmp(), are unavailable for use in kernel extensions. I don't understand the reasoning behind that. Can I even use things like the new and delete operators? Libc and similar libraries are *not* available in the kernel (for one thing, the execution environment is much different than that for user mode). A much-reduced subset is available (cf. libcc_kext.a). Memcmp() is not in there, so it may just be that no one needed it when it was written (memcpy() is, for example). It's worth looking through the code in the darwin repository, while awaiting further answers, to see what current code does. Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | It's not whether you win or lose... | It's whether *I* win or lose. *--------------------------------------*-------------------------------* _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Kernel Extensions
by Eric Long 02 Dec '02

02 Dec '02

I am discovering the world of kernel extensions and using Project Builder. I have some quick questions that I hope someone can easily answer. 1 - In documentation I've been looking at it indicates class methods should be declared static. It seems to imply this is for performance reasons. However, in playing with it, it seems as if methods have to be static period. I say this because my extension fails loading due to undefined symbol problems if I don't declare all methods static. Maybe I just haven't investigated this enough. It is interesting that if I include methods in a struct, I don't have to declare them as static. Perhaps that is a given. I don't think a struct can be sub-structed, like a class can be sub-classed, but maybe I'm ignorant about that. 2 - I would like to use exceptions. PB tells me when I try to build that I need to use -fexceptions because exceptions are disabled. I tried adding that to the "Other Compiler Flags" section of the GCC compiler settings for the target. That didn't help. The docs show usage of -fhandle-exceptions, but I believe that is deprecated in favor of -fexceptions. In any case, it says I'm supposed to link against System.framework to use those. Can I do that? It appears that I can only link against Kernel.framework or particular modules within it for a kernel extension. I also find it strange that many standard ANSI functions, like memcmp(), are unavailable for use in kernel extensions. I don't understand the reasoning behind that. Can I even use things like the new and delete operators? I'm just trying to gain some perspective on this. It looks like C++ compilation is supported in a kernel extension, but in practice it is almost like being limited to C functionality for the most part. I am I off-track? Thanks for anyone's help! Eric Long ___________________________________________________________________________ Macintosh Software Engineer Aladdin Systems, Inc. http://www.aladdinsys.com 245 Westridge Drive, Watsonville, CA 95076 (831) 761-6200 x628 _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: problem building kernel on 10.2.2
by Justin C. Walker 02 Dec '02

02 Dec '02

On Monday, December 2, 2002, at 04:11 AM, Heiko Panther wrote: Hi, I'm trying to build the 10.2 kernel on 10.2.2. I'm using the "Building the XNU kernel (for mortals) HOWTO". I can't seem to get past this point... tried with the September as well as the December dev tools. This happens when trying to make RC_OS=macos kld_build the cctools-435. All packages are tarballs from the darwin projects directory page at opensource.apple.com. cc -g -I.. -I../../include -I. -Wall -Wno-precomp -Wno-long-double -no-cpp-precomp -static -DRLD -DKLD -fno-builtin -finline -fno-keep-inline-functions -fsigned-bitfields -force_cpusubtype_ALL -msoft-float -nostdinc -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 -O -c -MD \ -dependency-file ./pass1.d -o ./pass1.o ../pass1.c In file included from ../pass1.c:36: /System/Library/Frameworks/Kernel.framework/Headers/sys/file.h:63:24: No include path in which to find sys/unistd.h I haven't spent much time with the 10.2/6.x kernel source, but at first blush, this looks really gross. There is (AFAIK) no such file as 'sys/unistd.h', so the compiler is quite right to complain. I would look at 'sys/file.h', determine if it is really there, and why, and then remove it. 'unistd.h' is typically a "user-mode code" include file, and (a) should never be included in kernel compiles; and (b) should never be referenced in kernel headers. Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | Men are from Earth. | Women are from Earth. | Deal with it. *--------------------------------------*-------------------------------* _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

problem building kernel on 10.2.2
by Heiko Panther 02 Dec '02

02 Dec '02

Hi, I'm trying to build the 10.2 kernel on 10.2.2. I'm using the "Building the XNU kernel (for mortals) HOWTO". I can't seem to get past this point... tried with the September as well as the December dev tools. This happens when trying to make RC_OS=macos kld_build the cctools-435. All packages are tarballs from the darwin projects directory page at opensource.apple.com. cc -g -I.. -I../../include -I. -Wall -Wno-precomp -Wno-long-double -no-cpp-precomp -static -DRLD -DKLD -fno-builtin -finline -fno-keep-inline-functions -fsigned-bitfields -force_cpusubtype_ALL -msoft-float -nostdinc -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 -O -c -MD \ -dependency-file ./pass1.d -o ./pass1.o ../pass1.c In file included from ../pass1.c:36: /System/Library/Frameworks/Kernel.framework/Headers/sys/file.h:63:24: No include path in which to find sys/unistd.h make[1]: *** [pass1.o] Error 1 make: *** [kld_build] Error 2 Any advice? Thanks a lot. Heiko Panther _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

harddisk suspend/sleep management
by Matthias Kretschmer 01 Dec '02

01 Dec '02

Hello! I was thinking about the noflushd project (http://noflushd.sourceforge.net/) and thought, this should be possible for darwin, too. The logic for suppressing write access to the hdd should be easy if proper asynchron write is implemented. I wouldn't say to port noflushd, but implement some other simple tool which handles this (of course if it is less work to port :) ). So but this needs fundamental support from the kernel. I don't really know where to begin to look at. Or is something like this already implemented? -- Greetings Matthias Kretschmer _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0