July 2004 - Darwin-kernel - lists.apple.com

Re: Restrictions in kexts
by Quinn 12 Jul '04

12 Jul '04

At 13:54 -0700 6/7/04, Krishna Monian wrote: Is there some other equivalent method of doing what the 3 functions do in a kext? I've answered your original question in another email. However, this is a slightly different question which is worth considering independently. The Core Services File Manager, which implements these routines, is implemented on top of the BSD API. So, in principle, it should be possible to do this work inside the kernel. However, it's not possible to do this in a supported fashion because the mechanism used by Core Services File Manager is private (that is, it's not guaranteed to work that way in the future). There are two mechanism at work here. 1. For volfs volumes, Core Services File Manager works in terms of volfs. You can learn more about volfs in the Q&A. <http://developer.apple.com/qa/qa2001/qa1113.html> However, the take home message of this Q&A is that the volfs file system is not something you should be using directly. It's fine to use it indirectly via CoreServices File Manager, but using it directly will get you into compatibility trouble. 2. For Posix volumes, Core Services File Manager remembers the parent of each node as it traverses the node. That's not something that you can emulate in your KEXT. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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 Space, how much?
by Quinn 12 Jul '04

12 Jul '04

At 16:08 -0700 6/7/04, Godfrey van der Linden wrote: If the amount of data required is large you could try setting up a block of pageable memory inside the kernel. I believe the IOBufferMemoryDescriptor has a pageable option. This may be a good solution for your problem. Also note that the kernel virtual address space is limited (3 GB on 10.3 and later, 1 GB on earlier systems). So, it's not a good idea to allocate too much memory in the kernel, even if it is pageable. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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: Mach Thread API
by Quinn 12 Jul '04

12 Jul '04

At 18:05 +0200 7/7/04, Christoph Ndgeli wrote: 1. Calling pthread_testcancel() during the thread. If I want to kill, call pthread_cancel followed by a broadcast on the condition variable in case the thread is waiting. I think it's a little annoying always calling pthread_testcancel() all the time. There's an even easier solution. Have your game thread poll a global variable. When your main thread wants to quit the game thread, set the global and broadcast the condition (to wake up the game thread if it's blocked). It's fast and safe. You can avoid excessive performance overhead by placing your tests of the global in a part of the code that runs about ten times a second; cancellation has to be responsive, but not /that/ responsive. Another option would be to use per thread signals combined with setjmp/longjmp. However, as Jim said, jumping out of an OS routine will cause you a world of pain. So you should only do this if you can guarantee that that game thread never calls the OS (which is surprisingly difficult to do, consider malloc). 2. Forget about posix threads and work directly with Mach threads (thought they look a little complicated comparing to the pthreads => Is there any example code out there?). Don't do this. pthreads is where Apple places our compatibility stake in the ground. The following quote from Technote 2028 is good advice. User-space processes should not create Mach threads directly. -- <http://developer.apple.com/technotes/tn/tn2028.html#MacOSXThreading> S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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: Getting the full path for a file
by Quinn 12 Jul '04

12 Jul '04

At 9:48 -0700 23/6/04, Krishna Monian wrote: Hi All, I was looking at the MoreFiles samples on the Apple site and the FullPath code in particular. I am trying to modify the fs_usage system call so that it reports the full path. However I am not sure how exactly I should go about using the getFullPath function. I would really appreciate it if someone could guide me on this. I sent the following to Krishna on a private email, but I'm posting it here because I thought others could benefit from the information as well. ------------------------------------------------------------------------- o vnode to full path -- In Mac OS X 10.3 and later, we record the path that the user traversed to access a vnode. You can get this information from user space using the F_GETPATH fcntl. You can get this information from the kernel using the vn_getpath routine. o full path problems -- Even with the above, there are numerous problems associated with using full paths. For example: - race conditions -- There's no guarantee that the full path you get is still valid. Between the time that it's recorded (as the file is being looked up) and the time you see it, someone could have renamed the file, or one of its parents. - uniqueness -- There's no guarantee that the full path is unique. A file with multiple hard links can have multiple valid full paths. The path you get back is the path that the user used to look up the file, which may not be the path you were looking for. - existence -- A common Unix practice is to open a temporary file and then delete the file on disk. The file disappears from the catalogue but is still accessible via the file descriptor. In this case there is *no* valid full path to the file. - volfs -- The Core Services File Manager (which is used by most Carbon applications) does most file access via volfs. You can read about volfs in DTS Q&A 1113 "The '/.vol' directory and 'volfs'". <http://developer.apple.com/qa/qa2001/qa1113.html> As mentioned in the Q&A, Apple explicitly warns against you constructing or interpreting volfs paths; this is an area that's subject to change in the future. Furthermore, there's no efficient way to map from a volfs path to the corresponding Posix path. ------------------------------------------------------------------------- S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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: Finder Get Info not allowing permission change with CXFS
by Quinn 12 Jul '04

12 Jul '04

At 17:00 +1000 23/6/04, Nick Blievers wrote: Using 10.3.3 the Finder 'Get Info' file/dir option (on a CXFS volume) only shows a read only summary of permissions. eg it says under Ownership & Permissions: "You can read and write" It does not show any combo boxes or the 'details' drop down section for changing owner and group info. This works with 10.2.8, using the same code. So, any ideas what has changed in Finder? And what we can do about it? What does the "Format" item in the Info window show? It seems that Finder has a special case for certain volume formats (UFS, NFS, AppleShare, HFS Plus, HFS Plus Journaled) and only shows the detailed privileged view for those formats. I'd encourage you to file a bug to request a generic interface for this. <http://developer.apple.com/bugreporter/> As to why this worked on 10.2.x, I don't really know. However, looking at the source it seems that there was a major rework of the dialog between 10.2.x and 10.3. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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

why does nfs use it's own buffer cache?
by Nikita Danilov 10 Jul '04

10 Jul '04

Hello, what are the reasons of XNU kernel using separate buffer cache (storing struct nfsbuf's) for NFS? It seems that all buffer cache operations are going through file system VOP_STRATEGY() at which point NFS could initiate networking IO. There should be some important reason I am missing, because in the current state of affairs there is significant code duplication (hashing, async IO daemon, whole nfs_bio.c), NFS doesn't enjoys advantages of vfs_cluster.c, etc. Nikita. _______________________________________________ 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: pf in 10.4?
by Justin Walker 08 Jul '04

08 Jul '04

On Jul 8, 2004, at 12:45, OpenMacNews wrote: i *am* a customer. direct/indirect purchasing control of ~2200 machines, including ~ 200 Macs. i've had LENGTHY, repeated discussion about THIS issue with apple numerous times. sales, support, etc, etc. nothing for ~ two years. given that response, which is wholly within apple's right to choose/control ... there's a really clean solution for *us*. we won't use the macs for anything that requires enterprise-class firewall/gateway functionality. I can't really address the above, but this: and mention your sparkling attitude every time our apple sales person calls ... I am at a loss to understand why my sparkling wit would be an issue for you, in your relations with Apple, or why they would care. Regards, Justin -- /~\ The ASCII Justin C. Walker, Curmudgeon-at-Large \ / Ribbon Campaign X Help cure HTML Email / \ _______________________________________________ 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: pf in 10.4?
by Markus Hitter 08 Jul '04

08 Jul '04

Am 09.07.2004 um 00:18 schrieb perdurabo: Although I can truly sympathize with your plights, I'm not sure that Justin is the one you should be venting to. Who that is at Apple, I'm not sure, but. A strong vote to make the inclusion of any open source product happen whould be to port it yourself, then file enhancement requests at <http://bugreport.apple.com>. Those reports go to the responsible engineer at Apple, definitely the right direction. Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ 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: pf in 10.4?
by perdurabo 08 Jul '04

08 Jul '04

What exactly are you trying to do? You should be able to do most anything you can do in pf with ipfw. pf's advantage over ipfw (in my mind, anyways) is that the rules syntax is quite a bit cleaner, pf is a little faster, and it has a few leading edge features (like rules based on OS fingerprinting, etc). Although I can truly sympathize with your plights, I'm not sure that Justin is the one you should be venting to. Who that is at Apple, I'm not sure, but. ~ per On Thu, 08 Jul 2004 12:45:57 -0700, OpenMacNews <darwin-kernel.20.openmacnews(a)spamgourmet.com> wrote: > i *am* a customer. direct/indirect purchasing control of ~2200 machines, including ~ 200 Macs. > > i've had LENGTHY, repeated discussion about THIS issue with apple numerous times. sales, support, etc, etc. > > nothing for ~ two years. > > given that response, which is wholly within apple's right to choose/control ... > > there's a really clean solution for *us*. we won't use the macs for anything that requires enterprise-class firewall/gateway functionality. and mention your sparkling attitude every time our apple sales person calls ... > > duh. > > > _______________________________________________ > 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: pf in 10.4?
by OpenMacNews 08 Jul '04

08 Jul '04

i *am* a customer. direct/indirect purchasing control of ~2200 machines, including ~ 200 Macs. i've had LENGTHY, repeated discussion about THIS issue with apple numerous times. sales, support, etc, etc. nothing for ~ two years. given that response, which is wholly within apple's right to choose/control ... there's a really clean solution for *us*. we won't use the macs for anything that requires enterprise-class firewall/gateway functionality. and mention your sparkling attitude every time our apple sales person calls ... duh. _______________________________________________ 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