December 2006 - Darwin-dev - lists.apple.com

launchd and kext dependency
by Bhavesh Davda 11 Dec '06

11 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=mb+sQXR/drqUV35e7b4zy1AnYOHjbCNjpEm9JS+PogEZ8CQHMzAXKTxs2K7/EArJd4PptC8osRNSABtmKLs6vIVY8M9GPKkLM/b3QHWBk/8DHqLz34GvQ3x+rwKNaIwGAjW1bPyiOn2am9Q0VL+j/3HNzfB6tbwwxpjzGoX2eMo= Is there a way to specify in a launchd.plist that it depends on a kext first being loaded during system boot? I have a daemon that should be launched by launchd only after the network drivers (e.g. com.apple.iokit.AppleYukon on a MacBook Pro) have been loaded. Also, is there a document explaining the system boot process as it relates to the order of such kexts being loaded, versus launchd running system startup tasks? Thanks -- Bhavesh P. Davda _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: MACH_RCV_INTERRUPT
by Jim Magee 08 Dec '06

08 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com --Jim On Dec 6, 2006, at 9:34 AM, Eric Grant wrote: Thanks again, Eric Grant On 12/6/06 9:10 AM, "Dave Zarzycki" <zarzycki(a)apple.com> wrote: More to the point, how did you find yourself in such a scenario where the answer to that question matters? Have you considered using the tool called 'mig' to automate IPC stub generation and demux? davez On Dec 6, 2006, at 9:02 AM, Dave Zarzycki wrote: Why do you ask? davez On Dec 6, 2006, at 8:48 AM, Eric Grant wrote: The mach_msg man page states: http://www.opensource.apple.com/darwinsource/10.4.8.x86/xnu-792.13.8/osfmk/m an/mach_msg.html. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/jmagee%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I try to live by the adage "The only good timeouts are immediate timeouts." ;-) But if you must use a non-zero/non-infinite timeout (and think long and hard as to why), and that timeout is sufficiently large (so that there is a reasonable expectation of it getting interrupted by signals during the wait) then you must provide the MACH_RCV_INTERRUPT and/or MACH_SEND_INTERRUPT options and recalculate the relative timeout yourself (the mach_msg() library routine won't do that for you). I am doing simple IPC with mach messages, and I need a timeout mechanism. The alternatives of which I am aware -- creating the receive port using mk_timer_create(), or waiting on a port set that includes a timer -- are suboptimal in this situation. As for mig, I am aware of it, but it doesn't seem necessary. Everything seems to be working fine (save for this timeout issue). If MACH_RCV_TIMEOUT is used without MACH_RCV_INTERRUPT, then the timeout duration might not be accurate. When the call is interrupted and automatically retried, the original timeout is used. If interrupts occur frequently enough, the timeout interval might never expire. MACH_SEND_TIMEOUT without MACH_SEND_INTERRUPT suffers from the same problem. Is this statement true for Darwin's implementation of mach? This email sent to jmagee(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: MACH_RCV_INTERRUPT
by Eric Grant 06 Dec '06

06 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Microsoft-Entourage/10.1.6.040913.0 I am doing simple IPC with mach messages, and I need a timeout mechanism. The alternatives of which I am aware -- creating the receive port using mk_timer_create(), or waiting on a port set that includes a timer -- are suboptimal in this situation. As for mig, I am aware of it, but it doesn't seem necessary. Everything seems to be working fine (save for this timeout issue). Thanks again, Eric Grant On 12/6/06 9:10 AM, "Dave Zarzycki" <zarzycki(a)apple.com> wrote: > More to the point, how did you find yourself in such a scenario where > the answer to that question matters? > > Have you considered using the tool called 'mig' to automate IPC stub > generation and demux? > > davez > > On Dec 6, 2006, at 9:02 AM, Dave Zarzycki wrote: > >> Why do you ask? >> >> davez >> >> On Dec 6, 2006, at 8:48 AM, Eric Grant wrote: >> >>> The mach_msg man page states: >>> >>> If MACH_RCV_TIMEOUT is used without MACH_RCV_INTERRUPT, then >>> the timeout >>> duration might not be accurate. When the call is interrupted and >>> automatically retried, the original timeout is used. If >>> interrupts occur >>> frequently enough, the timeout interval might never expire. >>> MACH_SEND_TIMEOUT without MACH_SEND_INTERRUPT suffers from the >>> same >>> problem. >>> >>> http://www.opensource.apple.com/darwinsource/10.4.8.x86/xnu-792.13.8/osfmk/m >>> an/mach_msg.html. >>> >>> Is this statement true for Darwin's implementation of mach? _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: MACH_RCV_INTERRUPT
by Dave Zarzycki 06 Dec '06

06 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com davez On Dec 6, 2006, at 9:02 AM, Dave Zarzycki wrote: Why do you ask? davez On Dec 6, 2006, at 8:48 AM, Eric Grant wrote: The mach_msg man page states: http://www.opensource.apple.com/darwinsource/10.4.8.x86/xnu-792.13.8/osfmk/m an/mach_msg.html. Is this statement true for Darwin's implementation of mach? Thanks, Eric Grant _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/zarzycki%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… More to the point, how did you find yourself in such a scenario where the answer to that question matters? Have you considered using the tool called 'mig' to automate IPC stub generation and demux? If MACH_RCV_TIMEOUT is used without MACH_RCV_INTERRUPT, then the timeout duration might not be accurate. When the call is interrupted and automatically retried, the original timeout is used. If interrupts occur frequently enough, the timeout interval might never expire. MACH_SEND_TIMEOUT without MACH_SEND_INTERRUPT suffers from the same problem. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/zarzycki %40apple.com This email sent to zarzycki(a)apple.com This email sent to zarzycki(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Problem with Technical Note TN2118 Kernel Core Dumps
by Giuliano Gavazzi 06 Dec '06

06 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com $ nvram boot-args="debug=0xd44 _panicd_ip=192.168.63.40" and checking some: $ nvram -p [...] boot-device pci/ACARD,6880M@4/@0:2,\\:tbxi load-base 0x800000 [...] boot-args debug=0xd44 _panicd_ip=192.168.63.40 I then loaded the InstantPanic.kext for testing (see the TN): $ kextload /InstantPanic.kext the machine panicked and the core dump was there: $ file core-xnu-792-192.168.63.37-61964bd2 core-xnu-792-192.168.63.37-61964bd2: Mach-O core ppc but gdb cannot use it: a bt just gives (I even loaded the mach_kernel symbol table): (gdb) bt Unable to read register data from core file #0 0xfeedface in ?? () Note that the file is tiny: 2926592 bytes. Perhaps a problem during the dump? Thank you in advance for any advice you can give me Giuliano _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I followed the advice in the TN2118 to enable a kdumpd server machine on the LAN of a troublesome 10.4.8 machine (the trouble started after applying the 2006-007 security update) called here the client. I set the client as per the TN, and here the first problem appeared. After setting the boot-args: I restarted the machine, but it hung in the startup process with a "no access" road sign.. I then had it restarted with the option pressed to reselect the disk, and it worked (all this at a remote site with a collaborator). I rechecked that the boot-args were still set and they were. $ gdb -c core-xnu-792-192.168.63.37-61964bd2 GNU gdb 6.3.50-20050815 (Apple version gdb-573) (Fri Oct 20 15:54:33 GMT 2006) [...] This GDB was configured as "powerpc-apple-darwin". unable to read unknown load command 0x0 and so on.... Also, since the problem with this machine is more a hung than a panic (I have no kernel panics in /Library/Logs/ and when hung the display just shows the rainbow wheel), should I force a NMI to cause a dump when this happens? This email sent to site_archiver(a)lists.apple.com

1 0

Re: MACH_RCV_INTERRUPT
by Dave Zarzycki 06 Dec '06

06 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Why do you ask? davez On Dec 6, 2006, at 8:48 AM, Eric Grant wrote: The mach_msg man page states: http://www.opensource.apple.com/darwinsource/10.4.8.x86/xnu-792.13.8/osfmk/m an/mach_msg.html. Is this statement true for Darwin's implementation of mach? Thanks, Eric Grant _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/zarzycki%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… If MACH_RCV_TIMEOUT is used without MACH_RCV_INTERRUPT, then the timeout duration might not be accurate. When the call is interrupted and automatically retried, the original timeout is used. If interrupts occur frequently enough, the timeout interval might never expire. MACH_SEND_TIMEOUT without MACH_SEND_INTERRUPT suffers from the same problem. This email sent to zarzycki(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

MACH_RCV_INTERRUPT
by Eric Grant 06 Dec '06

06 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Microsoft-Entourage/10.1.6.040913.0 The mach_msg man page states: If MACH_RCV_TIMEOUT is used without MACH_RCV_INTERRUPT, then the timeout duration might not be accurate. When the call is interrupted and automatically retried, the original timeout is used. If interrupts occur frequently enough, the timeout interval might never expire. MACH_SEND_TIMEOUT without MACH_SEND_INTERRUPT suffers from the same problem. http://www.opensource.apple.com/darwinsource/10.4.8.x86/xnu-792.13.8/osfmk/m an/mach_msg.html. Is this statement true for Darwin's implementation of mach? Thanks, Eric Grant _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: altivec issue..
by Andrew Gallatin 04 Dec '06

04 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Andrew Gallatin writes: > > I'm using altive asm code to copy 64 bytes of memory to a PCI device > which my driver has mapped in to userspace. Once in a very blue moon, > we'll see 0x7fffdead copied, rather than what we expected. > We seem to have forgotten VRSAVE. Whoops.. Drew _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

altivec issue..
by Andrew Gallatin 04 Dec '06

04 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I'm using altive asm code to copy 64 bytes of memory to a PCI device which my driver has mapped in to userspace. Once in a very blue moon, we'll see 0x7fffdead copied, rather than what we expected. Grepping around in the kernel sources reveals that the kernel somehow uses 0x7fffdead to mark altivec registers which it thinks are unused. Is there some system call or instruction we need to call in order to inform the kernel we're using altivec registers v0..v3? The copy routine looks like this: #define LVX "lvx v" #define STVX "stvx v" static inline void my_copy_64_bytes_with_altivec(void *to, uint64_t *from) { __asm__ __volatile__( " sync\n" LVX "0, 0, %4\n" LVX "1, 0, %5\n" LVX "2, 0, %6\n" LVX "3, 0, %7\n" STVX "0, 0, %0\n" STVX "1, 0, %1\n" STVX "2, 0, %2\n" STVX "3, 0, %3\n" " sync\n" : : "b"(to), "b"(((char*)to)+16), "b"(((char*)to)+32), "b"(((char*)to)+48), "b"(from), "b"(((char*)from)+16), "b"(((char*)from)+32), "b"(((char*)from)+48) }; } Thanks, Drew _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Remapping MacBook Pro eject key
by Terry Simons 03 Dec '06

03 Dec '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com - Terry On Dec 3, 2006, at 11:19 AM, Andre-John Mas wrote: Hi, Andre _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/galimore%40mac.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… It requires a driver in Windows, so it's definitely not hard-wired to the DVD drive. I vaguely remember some discussion of the sound and eject keys on the Mac Pro keyboard being USB buttons rather than standard keyboard keys... maybe this is the case for the laptop eject key as well. Every so often I will catch the eject key on my MacBook Pro. I have looked around for existing solutions, but I haven't found anything to date. Does anyone know whether this is something that is doable, or is hit hardwired to the DVD drive? This email sent to galimore(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0