April 2002 - Darwin-kernel - lists.apple.com

Darwin 6.0 compile problems
by Nathan Hjelm 16 Apr '02

16 Apr '02

I tried to compile darwin 6.0 on my machine and ran into some problems compiling ndrv.c. It seems that struct ndrv_multiaddr is not defined in any of the headers. Is this a problem with the cvs source or am I doing something wrong. heres what i have done: checkout xnu checkout Commands/NeXT/bootstrap_cmds cd xnu and . SETUP/setup.sh make all and make install in bootstrap_cmds cd xnu and make exporthdrs and make all _______________________________________________ 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: installing a new kernel broke carbon
by Dallas De Atley 16 Apr '02

16 Apr '02

Ack, this sounds like Carbon bridge library problem. Swapping out /mach_kernel shouldn't bust this. Did something get installed into /System/Library/CFMSupport? - Dallas On Monday, April 15, 2002, at 08:07 PM, Jim Magee wrote: It sounds to me like he installed the "less functional" Darwin version of CoreFoundation, and Mac OS X wants, no needs, the full version. On Monday, April 15, 2002, at 10:03 PM, Louis Gerbarg wrote: Well, what I was thinking was that if he followed some of the more common build instructions he would have not just replaced the kernel, but also the System, IOKit, and Kernel frameworks. If he then reverted by simply copying the old kernel he might have a kernel/library mismatch. Louis On Monday, April 15, 2002, at 05:26 PM, Justin C. Walker wrote: On Monday, April 15, 2002, at 05:02 PM, Louis Gerbarg wrote: It depends, how did you build and install it, and what version did you build. I don't get the fact that rebooting with the old kernel still gives you the same problem. Did something else change at the same time? Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------* - - _______________________________________________ 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. _______________________________________________ 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. _______________________________________________ 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. _______________________________________________ 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: installing a new kernel broke carbon
by Jim Magee 16 Apr '02

16 Apr '02

It sounds to me like he installed the "less functional" Darwin version of CoreFoundation, and Mac OS X wants, no needs, the full version. On Monday, April 15, 2002, at 10:03 PM, Louis Gerbarg wrote: Well, what I was thinking was that if he followed some of the more common build instructions he would have not just replaced the kernel, but also the System, IOKit, and Kernel frameworks. If he then reverted by simply copying the old kernel he might have a kernel/library mismatch. Louis On Monday, April 15, 2002, at 05:26 PM, Justin C. Walker wrote: On Monday, April 15, 2002, at 05:02 PM, Louis Gerbarg wrote: It depends, how did you build and install it, and what version did you build. I don't get the fact that rebooting with the old kernel still gives you the same problem. Did something else change at the same time? Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------* - - _______________________________________________ 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. _______________________________________________ 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. _______________________________________________ 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: installing a new kernel broke carbon
by Louis Gerbarg 16 Apr '02

16 Apr '02

Well, what I was thinking was that if he followed some of the more common build instructions he would have not just replaced the kernel, but also the System, IOKit, and Kernel frameworks. If he then reverted by simply copying the old kernel he might have a kernel/library mismatch. Louis On Monday, April 15, 2002, at 05:26 PM, Justin C. Walker wrote: On Monday, April 15, 2002, at 05:02 PM, Louis Gerbarg wrote: It depends, how did you build and install it, and what version did you build. I don't get the fact that rebooting with the old kernel still gives you the same problem. Did something else change at the same time? Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------*- - _______________________________________________ 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. _______________________________________________ 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: installing a new kernel broke carbon
by Justin C. Walker 16 Apr '02

16 Apr '02

On Monday, April 15, 2002, at 05:02 PM, Louis Gerbarg wrote: It depends, how did you build and install it, and what version did you build. I don't get the fact that rebooting with the old kernel still gives you the same problem. Did something else change at the same time? Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------*-- _______________________________________________ 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

Proposed new behavior for NKE's sf_soconnect
by Vincent Lubet 16 Apr '02

16 Apr '02

Folks, The socket connect intercept calls are current invoked after the protocol connect handler. This is a problem for at least 2 reasons: - One cannot modify the destination address; - The connection is not necessary complete when the sf_soconnect are called. The fix is as simple as calling the NKE sf_soconnect handlers before calling the protocol connect handler. This is a long standing problem tracked in Radar 2655804. Several NKE developers have requested this problem to be fixed but before I commit the change I would like to make sure this change is not going to break existing NKE that rely on the current quirky behavior. Please let me know if your NKE is expecting for the sf_soconnect to be called after the protocol handler. Thanks, Vincent _______________________________________________ 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: installing a new kernel broke carbon
by Louis Gerbarg 16 Apr '02

16 Apr '02

It depends, how did you build and install it, and what version did you build. Louis On Monday, April 15, 2002, at 02:44 PM, Nathan T. Hjelm wrote: I built and installed the darwin kernel on my OS 10.1.3 machine and now i get messages like this one when I try to run any carbon based application: The application "Mozilla" could not be launched because of a shared library error: "&<Mozilla><CarbonLib><CFMPriv_System><>" The problem remained when I put the old kernel back. Is there a way to correct this problem without reinstalling macos 10.1 from cd? _______________________________________________ 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. _______________________________________________ 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

installing a new kernel broke carbon
by Nathan T. Hjelm 15 Apr '02

15 Apr '02

I built and installed the darwin kernel on my OS 10.1.3 machine and now i get messages like this one when I try to run any carbon based application: The application "Mozilla" could not be launched because of a shared library error: "&<Mozilla><CarbonLib><CFMPriv_System><>" The problem remained when I put the old kernel back. Is there a way to correct this problem without reinstalling macos 10.1 from cd? _______________________________________________ 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: Enabling IEEE 754 floating-point exceptions: API?
by Peter Andersen 08 Apr '02

08 Apr '02

In http://lists.apple.com/mhonarc/unix-porting/msg00225.html I wrote: > Does an API for enabling IEEE 754 floating-point exceptions in OSX exists, > or will I have to do this by assembly code directly manipulating the PowerPC > FPSCR and MSR registers? > > In classic MacOS development under MPW there was an API in <fenv.h>, notably > the feraiseexcept() function. I do not know if this is available as part of > carbon, but I prefer a direct interface instead of Carbon. I got no answers, but I guess the answer is no. I tried doing it myself in assembler as such: /* See * http://developer.apple.com/techpubs/mac/PPCNumerics/PPCNumerics-154.html * and table 5-17 in PowerPC 601 ref. manual */ #define FPU_ZERODIVISION (1L << 27) /* ZE bit */ #define FPU_INVALID (1L << 24) /* VE bit */ #define FPU_OVERFLOW (1L << 25) /* OE bit */ #define FPU_UNDERFLOW (1L << 26) /* UE bit */ #define FPU_PRECISIONLOST (1L << 28) /* XE bit */ static void EnableIEEE754Exceptions(mask) { #if 0 register unsigned long msr; /* Need to set MSR bits too, but mfmsr/mtmsr are supervisor only!*/ /* The stuff below is like get_msr()/set_msr() in <architecture/ppc_basic_regs.h> */ __asm__ __volatile("mfmsr %0" : "=r" (msr)); msr |= (1L<<(31-23)) /*FE1*/| (1L<<(31-20)) /*FE0*/; __asm__ __volatile("mtmsr %0" : : "r" (msr)); #endif if (mask & FPU_ZERODIVISION) __asm__ __volatile("mtfsb1 27"); if (mask & FPU_INVALID) __asm__ __volatile("mtfsb1 24"); if (mask & FPU_OVERFLOW) __asm__ __volatile("mtfsb1 25"); if (mask & FPU_UNDERFLOW) __asm__ __volatile("mtfsb1 26"); if (mask & FPU_PRECISIONLOST) __asm__ __volatile("mtfsb1 28"); } As you can see from the code, i have a problem with the MSR register: I need to set two bits in this register to enable exceptions (as also mentioned in the Numerics manual), but the assembler instructions for doing this are for supervisor mode, and will fail with Illegal Instruction if I use them. So my original question now boils down to whether there exists a system call which allows me to set the exception enable bits in the MSR register. I have cross-posted to darwin-kernel and darwin-development in the hope of an answer. -- Sincerely, Peter Andersen, Mjolner Informatics A/S <pa(a)mjolner.com> and University of Aarhus, Denmark <datpete(a)daimi.au.dk> _______________________________________________ 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: XNU Kernel compiling problem
by Daniel Hazelbaker 06 Apr '02

06 Apr '02

Okay, I still cannot get the CVS version of the kernel to compile. However, I found the problem with the distribution version of xnu on the darwin site. One of the .cpp files is mis-named .cp only after extraction by StuffIt Expander (the default expander that is run when you download a .tar.gz file). The reason for this is that the filename is longer than 31 chars and thus truncated, so make all cannot find the file and aborts. I was able to compile the kernel with my modifications and am using it now with no problems. surprisingly the ibook stays VERY cool with the lid closed. I ran the computer under a heavy work load for 5 hours and after I opened the lid it was no hotter than normal. ./iokit/Kernel/IOFilterInterruptEventSource.cp <- is the culprit. I also, BTW, got the error with cvs mentioned below. Never seen that before. Daniel Hazelbaker On Saturday, April 6, 2002, at 06:44 AM, John Clark wrote: Am Donnerstag den, 4. April 2002, um 20:08, schrieb Daniel J. Luke: Huh? Both xnu-201.19 and xnu-222 built _very_ easily for me (on my Mac OS X 10.1.3 system). Well from the the tag listed in MacOSX-CURRENT found in the Darwin 'module', I get: [localhost:/Volumes/FFS/MacOSX_CURRENT] root# cvs -z3 co -r xnu-233.1 xnu Protocol error: uncounted data discarded This occurs for a number of tags listed in that file. Such as the tag listed on the offical darwin source pages of the opensource web site: cvs -z3 co -r xnu-201-9.1 xnu Protocol error: uncounted data discarded [localhost:/Volumes/FFS/MacOSX_CURRENT] root# About the only version of xnu that I can get is with no tag indication at all. _______________________________________________ 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. _______________________________________________ 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