September 2002 - Darwin-kernel

Re: Missing lock.h and others
by Peter Lovell 30 Sep '02

30 Sep '02

On Monday, September 30, 2002, at 01:30 PM, Sean McBride wrote: Finder does not search invisible folders like /usr, but your checkout of the source was probably not to an invisible folder. Sherlock does have settings so that you can search for invisible items. Of course, you have to know that they are (or might be) invisible. Regards.....Peter _______________________________________________ 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: Missing lock.h and others
by Sean McBride 30 Sep '02

30 Sep '02

Michael Cashwell (mboards(a)prograde.net) wrote: >I found this header the next day. One contributing factor was that I'd >search my entire disk via a Finder search for any file called lock.h >and it didn't find it. Then, after checking out xnu somehow it could. I >still can't explain this. Finder does not search invisible folders like /usr, but your checkout of the source was probably not to an invisible folder. _______________________________________________ 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: What does "not authentic" mean?
by Stéphane Sudre 30 Sep '02

30 Sep '02

On Monday, September 30, 2002, at 04:55 PM, Shawn Erickson wrote: On Monday, September 30, 2002, at 07:26 AM, Stiphane Sudre wrote: When doing a kextload -v 6 I'm getting this output: kextload: authenticating extension /Users/stephane/Projects/Complete Mess/build/Test.kext kextload: authenticating bundle directory /Users/stephane/Projects/Complete Mess/build/Test.kext kextload: extension /Users/stephane/Projects/Complete Mess/build/Test.kext is not authentic What is it supposed to mean? BTW, I really like this other output which is giving me a Windows-like answer: [coyotte:~/Projects/Complete Mess/build] root# kextload -t Test.kext kextload: can't open /Users/stephane/Complete Mess/build/Test.kext/Contents/MacOS/Test can't add kernel extension Test.kext (file access/permissions) (run kextload on this kext with -t for diagnostic output) Humm... on my system using the -t option I get more informative message about access/permissions (try using sudo kextload -t Test.kext). I am running as root when doing the kextload -t Test.kext call. Anyway it most likely has to do with the addition security verification in Mac OS X... Read file:///Developer/Documentation/ReleaseNotes/KernelExtensions.html I basically use a scripts to copy my kexts over to my test system (into a test directory), setup file permissions correctly, generate symbols (copying back to my dev system), etc. I checked this too. The files access/permission are properly set as you described it. I found the issue. There is a bug in the kextload code or in the shell code. When the absolute path of the file contains a character with an accent (like a ` (or à if you prefer the HTML version)), then the kext is not loaded. Gone filing a bug report. _______________________________________________ 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: What does "not authentic" mean?
by Shawn Erickson 30 Sep '02

30 Sep '02

On Monday, September 30, 2002, at 07:26 AM, Stiphane Sudre wrote: When doing a kextload -v 6 I'm getting this output: kextload: authenticating extension /Users/stephane/Projects/Complete Mess/build/Test.kext kextload: authenticating bundle directory /Users/stephane/Projects/Complete Mess/build/Test.kext kextload: extension /Users/stephane/Projects/Complete Mess/build/Test.kext is not authentic What is it supposed to mean? BTW, I really like this other output which is giving me a Windows-like answer: [coyotte:~/Projects/Complete Mess/build] root# kextload -t Test.kext kextload: can't open /Users/stephane/Complete Mess/build/Test.kext/Contents/MacOS/Test can't add kernel extension Test.kext (file access/permissions) (run kextload on this kext with -t for diagnostic output) Humm... on my system using the -t option I get more informative message about access/permissions (try using sudo kextload -t Test.kext). Anyway it most likely has to do with the addition security verification in Mac OS X... Read file:///Developer/Documentation/ReleaseNotes/KernelExtensions.html I basically use a scripts to copy my kexts over to my test system (into a test directory), setup file permissions correctly, generate symbols (copying back to my dev system), etc. Example script... sudo rm -rf *.kext scp -r shawnce@quicksilver.local:~/Documents/Development/ftsw/qlc_fc_driver/ build/*.kext . sudo /usr/sbin/chown -R root.wheel *.kext sudo find *.kext -type d -exec /bin/chmod 0755 {} \; sudo find *.kext -type f -exec /bin/chmod 0644 {} \; rm -f *.sym ssh shawnce(a)quicksilver.local rm -f /tmp/*.sym sudo kextload -l qlc_fc_driver.kext sudo kextload -s ~/test -A -r ~/test qlc_fc_driver.kext sudo kextload -r ~/test -l qlc*xx.kext sudo kextload -s ~/test -A -r ~/test qlc*xx.kext sudo kextunload qlc*xx.kext sudo kextunload qlc_fc_driver.kext scp *.sym shawnce@quicksilver.local:/tmp/ -Shawn _______________________________________________ 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

What does "not authentic" mean?
by Stéphane Sudre 30 Sep '02

30 Sep '02

When doing a kextload -v 6 I'm getting this output: kextload: authenticating extension /Users/stephane/Projects/Complete Mess/build/Test.kext kextload: authenticating bundle directory /Users/stephane/Projects/Complete Mess/build/Test.kext kextload: extension /Users/stephane/Projects/Complete Mess/build/Test.kext is not authentic What is it supposed to mean? BTW, I really like this other output which is giving me a Windows-like answer: [coyotte:~/Projects/Complete Mess/build] root# kextload -t Test.kext kextload: can't open /Users/stephane/Complete Mess/build/Test.kext/Contents/MacOS/Test can't add kernel extension Test.kext (file access/permissions) (run kextload on this kext with -t for diagnostic output) _______________________________________________ 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: Missing lock.h and others
by Michael Cashwell 30 Sep '02

30 Sep '02

On Sunday, September 29, 2002, at 12:39 PM, Jim Magee wrote: Actually, in this particular case, kern/lock.h is included in the Kernel.framework headers. The problem is more likely that Michael was looking for the header in /usr/include. It will not be there. It is a kernel-only header. If you want to look at the header, it will be in: /System/Library/Frameworks/Kernel.framework/Headers/kern/lock.h and the Kernel.framework will be included when you tell Project Builder that you are building a KEXT. Thanks Jim. I found this header the next day. One contributing factor was that I'd search my entire disk via a Finder search for any file called lock.h and it didn't find it. Then, after checking out xnu somehow it could. I still can't explain this. The other factor was that I did tell PB that my new project was to be a kernel module. (I used the "Generic Kernel Extension" assistant.) I would have expected that to have populated the project with whatever paths or frameworks were appropriate for such a product. In any event, I've got the header working now and am again making progress. Thanks! Mike _______________________________________________ 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: Missing lock.h and others
by Jim Magee 29 Sep '02

29 Sep '02

On Thursday, September 26, 2002, at 07:29 PM, Justin C. Walker wrote: On Thursday, September 26, 2002, at 03:23 PM, Michael Cashwell wrote: Greetings all, I'm trying to be good and not include userland headers in my NKE. I'd think that was enlightened self-interest (since doing so would likely cause problems in your NKE and the kernel). Is that not a complete setup for doing NKEs? Do I need to download something else, possibly from the Darwin repository)? If so, what (and what tag) would ensure I get files compatible with Apple's released kernels? Roughly speaking, what you need to do is check out the appropriate version of the kernel (project 'xnu', with a tag "Apple-xxxx" matching what your kernel says (uname -a) in the form "xnu-xxxx"). Then use the headers from that kernel. Your best bet is to get the missing headers from there and put them in a convenient place, and reference them with an additional "-I" flag. If you are building for the system you are running on, it's OK to put them in the "proper" place. Actually, in this particular case, kern/lock.h is included in the Kernel.framework headers. The problem is more likely that Michael was looking for the header in /usr/include. It will not be there. It is a kernel-only header. If you want to look at the header, it will be in: /System/Library/Frameworks/Kernel.framework/Headers/kern/lock.h and the Kernel.framework will be included when you tell Project Builder that you are building a KEXT. --Jim _______________________________________________ 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

NFS mounts under Darwin 6.0/6.1
by Paul Ripke 29 Sep '02

29 Sep '02

I've been having repeated hanging problems after upgrading to Darwin 6 (well, really MacOS X 10.2.x). I'm wondering if anyone else has experienced hangs when mounting remote NFS filesystems. I've got 100% reproducibility, and 200% frustration. Details: Server is NetBSD 1.6G on i386. Client is Wallstreet PBG3, with MacOS 10.2, then 10.2.1, installed via upgrade. Mount is via "sudo mount -o resvport pc:/arc /arc". 100% repeatable hang will occur shortly after unmounting a mounted disk image in the Finder, where the disk image is on the NFS filesystem (yes, AFAIK that needs MacOS X, but I figure the problem is in the kernel). The Finder is usually the first to lock up, "kill -9" is ignored. Shortly thereafter every process suffers the same fate. I can move the little spinning rainbow cursor around, and it responds to pings, but nothing else. This may be an odd thing to do, but I don't see why it should break. >From one of the many hangs, I managed to find this in my syslog: Sep 28 08:34:31 stix-pbg3 /sbin/autodiskmount[189]: Someone (pid = (402)) isn't responding to a sent disk acknowledgement request. That client has been disconnected. Sep 28 08:34:31 stix-pbg3 /sbin/autodiskmount[189]: (autodiskmount.tproj/DiskArbitrationServerMain.c:4412) mach_msg failed: {0x10004003} (ipc/rcv) timed out Sep 28 08:34:31 stix-pbg3 /sbin/autodiskmount[189]: (autodiskmount.tproj/DiskArbitrationServerMain.c:410) received unrecognized message (id=0x41) on notify port I'm positive the server is OK, the laptop is not the only client, and tcpdump doesn't show any problems. fs_usage does not show anything unusual, and I don't have a second Mac to do remote kernel debugging. I've tried adding soft and intr to the NFS mount options, the hang is the same. >From BSD land, the NFS mount is fine. I've successfully built and installed XFree86 4.2.1 from a NFS mounted filesystem without any problems. Any thoughts? If no one else can reproduce this, I may be able to borrow a second mac to do remote debugging, but I'm pretty sure it's not my specific configuration. For now, I'll use AFP or SMB mounts, I've noticed AFP performance has increased about 10-fold after the upgrades (from NetBSD 1.5Z? and Mac OS X 10.1.5). Not sure which did it, but I'm not complaining! Cheers, -- Paul Ripke Unix/OpenVMS/DBA 101 reasons why you can't find your Sysadmin: 67: It's 9AM and they're not working that late. -- Koos van den Hout _______________________________________________ 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: Missing lock.h and others
by Stéphane Sudre 27 Sep '02

27 Sep '02

On Friday, September 27, 2002, at 01:29 AM, Justin C. Walker wrote: On Thursday, September 26, 2002, at 03:23 PM, Michael Cashwell wrote: Greetings all, I'm trying to be good and not include userland headers in my NKE. I'd think that was enlightened self-interest (since doing so would likely cause problems in your NKE and the kernel). Is that not a complete setup for doing NKEs? Do I need to download something else, possibly from the Darwin repository)? If so, what (and what tag) would ensure I get files compatible with Apple's released kernels? Roughly speaking, what you need to do is check out the appropriate version of the kernel (project 'xnu', with a tag "Apple-xxxx" matching what your kernel says (uname -a) in the form "xnu-xxxx"). Then use the headers from that kernel. Your best bet is to get the missing headers from there and put them in a convenient place, and reference them with an additional "-I" flag. If you are building for the system you are running on, it's OK to put them in the "proper" place. I've also noted that KNE samples (NKEMgr, SharedIP and TCPLogger) also don't compile due to missing system headers. We're just trying to protect you from shooting yourself in the foot with the half-cocked gun we lent you :-}. You can check the archives for discussions regarding missing headers... While speaking of missing headers, does someone know what headers should be in: /System/Library/Frameworks/Kernel.Framework/Versions/A/Headers/netinet and netinet6 I already added the missing headers in /usr/include and now I'm trying to compile Darwin code and it's not working of course. _______________________________________________ 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

Re2: standard library pow function needed from kext
by Frode Isaksen 27 Sep '02

27 Sep '02

>not a static libm on the system. Is this correct? Are there only dynamic >versions of the math and other libs? And if so, can a dynamic library be >linked to a kext? Is there a linker option to specify code from a dynamic >library be pulled into a generated program/kext as would be done with a >static library? Any other suggestions? >Thanks, I have compiled a static version of libm.a for use within the kernel. It works fine... If you want it, I can send the project to you. Make sure that you do not call the floating point function log(), as it collides with a kernel logger function log(). BTW, Apple is using floating point in the kernel in some audio stuff. Hilsen, Frode _______________________________________________ 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