October 2004 - Darwin-kernel

Re: BSD signal not aborting semaphore_wait() in the kernel on an SMP
by Chandra Khan 20 Oct '04

20 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:references; b=FEgqlQlPPWNQWfJ/PM6M99Qy7lmn7Mj+59jV5hotftJOB+KYeVhMyzi3j2aPxnERmUJ7otfUS20HSfcgQHK/yHcV24cELnHq5EhH5LyU6/paEoWfIug/7Nh5Wig0X8eXC2oU5QQbr8iuP113hVu86s9yYSeC2z6w4nT1iKkw4tM Thanks for the response. Please see my comments below. On Tue, 19 Oct 2004 18:24:25 -0400, Jim Magee wrote: > > Do you have more than one thread in the parent process? The SIGCHLD is > only delivered to one thread, not all of them. So other threads > waiting down in the ioctl() would be unaffected. Both the parent and the child processes are single threaded. > In order to run the signal > handler at user space, the thread that ran it MUST first wind its way > out of its current system call (and then we run the signal handler > instead of returning immediately from the system call). If you are > seeing the signal handler run, that thread DID wind out of its current > system call. And if the system call in question has the thread blocked > in semaphore_wait(), it must have been interrupted out of that wait. Absolutely. This explanation here had escaped my mind when I made the initial post. My post was therefore inconsistent and bogus. To illustrate my setup, here's the pseudocode: Parent: sigaction(SIGCHLD, ...) while(1) { ioctl(mydev, BLOCKING_RECEIVE, &blob); if(blob.msg == PARENT_SIGNALED) { // was interrupted because child got a signal child_was_signaled(); } else { // we were woken up by the child process with interesting info // ...work on blob } } The parent's SIGCHLD handler actually runs when the child gets a signal. However, I find that the parent is indefinitely stuck in the semaphore_wait(), ie, child_was_signaled() doesn't run. I erroneously concluded that the parent was actually blocked in the semaphore_wait() and wasn't woken up. The real reason, upon further reflection, is the classic "missed wakeup" scenario: the signal came and went before the parent process went to sleep unconditionally. I'm thinking of reworking this to handle the real cause of the child's signal in kernelspace. That would do away with the asynchronous signal. I still need to handle child death avoiding the same race condition. I wonder how one gets notification, in the kernel, of child death... > Oh, and some particulars about which version of the OS/kernel you are > running would be helpful. Some of this stuff behaves slightly > differently between releases. It's xnu 517.7.21, darwin kernel 7.5.0. Just curious, what behavior is different? Thanks for your help, Chandra _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: BSD signal not aborting semaphore_wait() in the kernel on an SMP
by Jim Magee 19 Oct '04

19 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com The UNIX process in question does an ioctl(my_dev) and, once in kernel space, sits on a semaphore_wait(). Because this process is ptrace-attached to a child process, and that child process gets a signal (as part of the design), I expect that the parent process will get a SIGCHLD, and in doing so, pop out of the semaphore_wait() with a KERN_ABORTED. It works. Unfortunately, not always. Especially when there is a lot of context switching between the parent and child (as opposed to the parent mostly sleeping), things dont work as I expect. While the SIGCHLD is delivered and the custom SIGCHLD handler *does* run (asynchronously), the parent process never aborts its semaphore_wait(). --Jim _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… On Oct 18, 2004, at 3:40 PM, Chandra Khan wrote: Do you have more than one thread in the parent process? The SIGCHLD is only delivered to one thread, not all of them. So other threads waiting down in the ioctl() would be unaffected. This is the part that confused me (or led me to believe that you had more than one thread in the parent). In order to run the signal handler at user space, the thread that ran it MUST first wind its way out of its current system call (and then we run the signal handler instead of returning immediately from the system call). If you are seeing the signal handler run, that thread DID wind out of its current system call. And if the system call in question has the thread blocked in semaphore_wait(), it must have been interrupted out of that wait. Oh, and some particulars about which version of the OS/kernel you are running would be helpful. Some of this stuff behaves slightly differently between releases. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Remote debug setup problem
by Mark Woollard 19 Oct '04

19 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Duh, was sure I'd tried that, but I guess not, now accepted! Regards Mark On 19 Oct 2004, at 20:59, Peter Lovell wrote: On Oct 19, 2004, at 3:52 PM, Mark Woollard wrote: arp: writing to routing socket: Operation not permitted Thanks Mark got root?? _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I've tried to follow the instructions for setting up kernel debugging using two systems over ethernet, however I can't get the arp -s command to work, it tells me: Internet searches suggest that for linux this is an issue with incorrect subnet mask. I have a 255.255.255.240 subnet and all my networking between machines works fine. I've not found anything specific to OS X. Anyone here have any ideas on this? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Remote debug setup problem
by Peter Lovell 19 Oct '04

19 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Oct 19, 2004, at 3:52 PM, Mark Woollard wrote: arp: writing to routing socket: Operation not permitted Thanks Mark got root?? _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I've tried to follow the instructions for setting up kernel debugging using two systems over ethernet, however I can't get the arp -s command to work, it tells me: Internet searches suggest that for linux this is an issue with incorrect subnet mask. I have a 255.255.255.240 subnet and all my networking between machines works fine. I've not found anything specific to OS X. Anyone here have any ideas on this? This email sent to site_archiver(a)lists.apple.com

1 0

Remote debug setup problem
by Mark Woollard 19 Oct '04

19 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com arp: writing to routing socket: Operation not permitted Thanks Mark _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I've tried to follow the instructions for setting up kernel debugging using two systems over ethernet, however I can't get the arp -s command to work, it tells me: Internet searches suggest that for linux this is an issue with incorrect subnet mask. I have a 255.255.255.240 subnet and all my networking between machines works fine. I've not found anything specific to OS X. Anyone here have any ideas on this? This email sent to site_archiver(a)lists.apple.com

1 0

Re: ld error
by Hervé Kergourlay 19 Oct '04

19 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.2; fr-FR; rv:1.0.2) Gecko/ 20030208 Netscape/7.02 -- _______________________________________________________________________ Hervé Kergourlay bureau 02 97 68 40 29 Chef du groupe Vos herve.kergourlay(a)atempo.com fax 02 97 68 40 25 Atempo S.A. Batiment PIREN - Rue Henri Becquerel - PIBS 56000 Vannes, France http://www.atempo.com Only those who attempt the absurd can achieve the impossible. _______________________________________________________________________ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: __private_extern__
by Garth Cummings 19 Oct '04

19 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Oct 11, 2004, at 1:04 PM, Godfrey van der Linden wrote: Yep. g++ -v <foo>.cpp will show this on 10.3.5 with Xcode 1.5 installed: -D__private_extern__=extern --gc On Oct 11, , at 8:21, stephane sudre wrote: On Oct 11, 2004, at 5:13 PM, Brian Bergstrand wrote: -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Stephane, Are you assigning an initial value to the var definition in your.c file? If not, then the compiler will treat the definition as a declaration, and no storage will be allocated. .h __private_extern__ int foo; .c __private_extern__ int foo = 0; // Allocate storage for foo. Thanks. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/gvdl%40apple.com This email sent to gvdl(a)apple.com This email sent to gcummings(a)apple.com __________________________________________________________________ Garth Cummings Apple Developer Technical Support gcummings(a)apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com One other thing to note, __private_extern__ does not work for C++ files and headers. Godfrey That's definitely my case (no initial value). I would never have suspected this. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/ gcummings%40apple.com smime.p7s

1 0

swapcontext in OS X ?
by Nicolas Morant 19 Oct '04

19 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hello, I am porting an application which use swapcontext, makecontext, set and get context. Do you know if Darwin will have this in the future ? and if yes, do you know when ? I have seen that NetBSD has src/lib/libc/arch/powerpc/gen/swapcontext.S,... Do you know if I can use these files as-is in order to have the functionality ? Thx a lot Nicolas Vous manquez despace pour stocker vos mails ? Yahoo! Mail vous offre GRATUITEMENT 100 Mo ! Créez votre Yahoo! Mail sur http://fr.benefits.yahoo.com/ Le nouveau Yahoo! Messenger est arrivé ! Découvrez toutes les nouveautés pour dialoguer instantanément avec vos amis. A télécharger gratuitement sur http://fr.messenger.yahoo.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

BSD signal not aborting semaphore_wait() in the kernel on an SMP
by Chandra Khan 18 Oct '04

18 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:references; b=YgYyCdwsCpjj+yACRzIsfM7LojzS8b4LFx/YT0x7zblAzK34hiXzUnTbRawqjuQB3sdcU8yQMIzV1Segs+o3aG/2LirYSK7hOBWsRabbJ9Wdvx6fyOxvFHmhtFDQwb0ykviGN1xROUoR9De+yoKREI09RbqU6vbv/zi1SgRornM Hi: I've done some work with character devices on traditional BSD and linux in the past. Given that, and my relative inexperience with C++ and IOKit, I chose to implement my tutorial "hello world" pseudo character device driver wholly within the BSD subsystem of XNU. And I've run into something and would appreciate some guidance. The UNIX process in question does an ioctl(my_dev) and, once in kernel space, sits on a semaphore_wait(). Because this process is ptrace-attached to a child process, and that child process gets a signal (as part of the design), I expect that the parent process will get a SIGCHLD, and in doing so, pop out of the semaphore_wait() with a KERN_ABORTED. It works. Unfortunately, not always. Especially when there is a lot of context switching between the parent and child (as opposed to the parent mostly sleeping), things dont work as I expect. While the SIGCHLD is delivered and the custom SIGCHLD handler *does* run (asynchronously), the parent process never aborts its semaphore_wait(). Is this an SMP issue? Or something to do with funnels? Or should I be abandoning Mach semaphores in deference to tsleep()/wakeup()? Or should I use thread_block(), etc? I'm somewhat lost, even after having read the relevant sections of the kernel programming guide (which is a very well-written document, by the way!), and combing through the archives of this list. Thanks for any help, Chandra _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: startup context: Same player, do not shoot again?
by Stephane Sudre 18 Oct '04

18 Oct '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 15 oct. 04, at 18:43, Quinn wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… At 15:44 +0200 15/10/04, stephane sudre wrote: Does this mean that as soon as you stop your daemon, you have no way to relaunch it in the startup context? Most of Apple's daemons are BSD based, and thus don't really care what Mach bootstrap port namespace they launch in. Strangely from what I've seen on 10.2, I observed different communication abilities whether you were launching a daemon from the user or startup context. This got fixed in 10.3 IIRC. Which ways are the not recommended to start in the startup context? The one I'd recommend, especially for relaunching a daemon, is "/usr/libexec/StartupItemContext" <x-man-page://8/StartupItemContext>. It looks 99% perfect. The 1% left is that it's not available on Mac OS X 10.1.x but this is not a huge issue IMHO. This email sent to site_archiver(a)lists.apple.com

1 0