November 2004 - Darwin-kernel

Re: sysctl
by Quinn 24 Nov '04

24 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com <http://developer.apple.com/bugreporter/> Please email me the bug number so that I can keep an eye on it. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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 16:28 -0700 8/11/04, Matt Jaffa wrote: I thought that setting the flag to CTLFLAG_ANYBODY would allow anyone to be able to set this structure no matter if it is a root app or regular user app. You'd think so, wouldn't you? Alas, it doesn't work, for reasons I'll explain below. Is it because I am branching off of _kern that it will only allow a root app to set my structure. This is the problem. The implementation of sysctl (__sysctl in "xnu/bsd/kern/kern_sysctl.c") has a special-case test for non-super user attempts to set variables in "kern", "hw", "vm", and "vfs", and bounces them with an EPERM error. *sigh* This special case doesn't apply if you hang off other roots. When I changed my test to attached to "net", everything worked fine. I'd be tempted to call this a bug, but I'm not 100% certain because the implementation of sysctl is /way/ complex and there may be side effects that I don't understand. Nevertheless, I'd appreciate it if you could file a bug report, to give kernel engineering the opportunity to make the final decision. And if is this way, must I branch off of something else like _user to allow me to modify it with a non-root app. I have tried to tie my structure to the _user branch, but it doesn't register it with _user so that when I do sysctlbyname it returns ENOENT = 2. "user" is designed for variables that are implemented entirely within user space. The System framework implementation of sysctl captures "user" requests and implements them internally; it never passes them to the kernel. This email sent to site_archiver(a)lists.apple.com

1 0

Re: sysctlbyname returns 1
by Quinn 24 Nov '04

24 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com The following code, when made from within a kernel extension, results in the error variable being set to 1 (EPERM?): [...] SYSCTL_PROC(_sysctl, 3, name2oid, CTLFLAG_RW|CTLFLAG_ANYBODY, 0, 0, sysctl_sysctl_name2oid, "I", ""); S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware kern_return_t sysctlbyname_start(kmod_info_t * ki, void * d) { int err; int numberOfCPUs; size_t size; int oid[2]; oid[0] = CTL_HW; oid[1] = HW_NCPU; numberOfCPUs = -1; size = sizeof(numberOfCPUs); At 12:23 -0600 8/11/04, Geoffrey Schmit wrote: Any ideas? The same code works fine from a user-mode console This is caused by a bug in the kernel. Specifically, the definition of the name2oid sysctl entry in "xnu/bsd/kern/kern_newsysctl.c": should include the CTLFLAG_KERN flag to allow the entry to be used by kernel code. The absence of this flag causes sysctlbyname to fail because it can't translate your name to an OID. See sysctlnametomib (in the same file) for background. This bug has already been fixed in Tiger. Until then you can work around it by using kernel_sysctl instead of sysctlbyname, as illustrated by the code below. err = kernel_sysctl( current_proc(), oid, sizeof(oid) / sizeof(*oid), &numberOfCPUs, &size, NULL, 0 ); if (err == 0) { assert(size == sizeof(numberOfCPUs)); printf("kernel_sysctl: numberOfCPUs = %d\n", numberOfCPUs); } else { printf("kernel_sysctl: failed with error %d\n", err); } return KERN_SUCCESS; } _______________________________________________ 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: detect UFS volumes with getfsent
by Quinn 23 Nov '04

23 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com getmntinfo <x-man-page://3/getmntinfo> S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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 10:21 +0100 23/11/04, Hervé Kergourlay wrote: how can I get it with posix APIs ? This email sent to site_archiver(a)lists.apple.com

1 0

detect UFS volumes with getfsent
by Hervé Kergourlay 23 Nov '04

23 Nov '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 on MAC OS X, this API detects only the nfs mounted volume from another Unix hervé _______________________________________________ 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 can't get information on other Volumes locally mounted, specially my UFS volume, as the "mount" command shows me how can I get it with posix APIs ? This email sent to site_archiver(a)lists.apple.com

1 0

thread_policy_set() error
by Arrigo Benedetti 23 Nov '04

23 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Mozilla Thunderbird 0.9 (Macintosh/20041103) I am passing thread_policy_set() these parameters: ttcpolicy.period: 1562500000 ttcpolicy.computation: 937500000 ttcpolicy.constraint: 1250000000 thanks! -Arrigo _______________________________________________ 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.… and the function fails with error code 4 (invalid parameters). I am running the code on a Powermac dual G5, where the bus speed is 1250000000. According to thread_policy.h, period, computation and constraint should be specified in Absolute Time units, and I read on an old thread on this list that in a second there should be 1250000000 divided by 4 Absolute Time units. Any ideas? This email sent to site_archiver(a)lists.apple.com

1 0

Re: kern_control questions
by Stéphane Sudre 18 Nov '04

18 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 18, 2004, at 20:02, Joshua Graessley wrote: When you register the kernel control, are you specifying a unit number? I must confess I'm making my test on a currently unknown cat specie. On Nov 18, 2004, at 7:39 AM, Stephane Sudre wrote: 2 questions related to kern_control in OS X 10.3.0 or later: XXXX is the same value both in the Kernel Extension and UserLand. _______________________________________________ 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.… -josh Yes, 0. I tried after finding a post from Vincent Lubet to set the unit number again in the user code (it's already set via the bzero AFAIK) after addr.sc_id= 'XXXX'. But the result was the same as it could be guessed. * Can registered IDs used for PF_NKE be used for a kern_control too? Or is a new registration required? * I'm using the equivalent of the sample code in the NKE documentation (without the CTL_FLAG_PRIVILEGED) and the following code in UserLand: bzero(&addr, sizeof(addr)); // sets the sc_unit field to 0 addr.sc_len = sizeof(addr); addr.sc_family = AF_SYSTEM; addr.ss_sysaddr = AF_SYS_CONTROL; addr.sc_id = 'XXXX'; // should be unique - use a registered Creator ID here fd = socket(PF_SYSTEM, SOCK_DGRAM, SYSPROTO_CONTROL); if (fd) { result = connect(fd, (struct sockaddr *)&addr, sizeof(addr)); if (result) fprintf(stderr, "connect failed %d\n", errno); } else fprintf(stderr, "failed to open socket\n"); if (!result) { /*result = setsockopt( fd, SYSPROTO_CONTROL, kEPCommand1, NULL, 0); if (result) fprintf(stderr, "setsockopt failed on kEPCommand1 call - result was %d\n", result); */ } I'm getting a errno 2 (No such file or directory) on the connect call. Is there something to change in the code showed in the documentation to make it work? The ctl_register call is saying everything is fine in the Kernel Extenstion. This email sent to site_archiver(a)lists.apple.com

1 0

Kernal Extension
by Setaro, James 18 Nov '04

18 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com I am writing a kernal extension for 10.1.5. I am trying to call kmodstat from inside the of the MoreAuthSample(setuid(0) for admin) tool using pipes(to check the output) and execlp to make sure that my extension is loaded at run-time. I have tried everything I can think of but it won't execute. I think that I am not setting up the enviroment correctly and exec can't find the the kmodstat command, but I don't know where to look for answers on setting this a exec enviroment or if it is even possible to execute kmodstat from execl. Right now I am trying to excute this: execlp( "/usr/sbin/kmodstat", "/usr/sbin/kmodstat", NULL ); but it seems to just break my pipe. Signal status: 10 kextload and kextunload seem to work fine. Any ideas? Thanks, James **************************************************************************** This email may contain confidential material. If you were not an intended recipient, Please notify the sender and delete all copies. We may monitor email to and from our network. **************************************************************************** _______________________________________________ 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: kern_control questions
by Joshua Graessley 18 Nov '04

18 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com When you register the kernel control, are you specifying a unit number? -josh On Nov 18, 2004, at 7:39 AM, Stephane Sudre wrote: 2 questions related to kern_control in OS X 10.3.0 or later: XXXX is the same value both in the Kernel Extension and UserLand. _______________________________________________ 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 * Can registered IDs used for PF_NKE be used for a kern_control too? Or is a new registration required? * I'm using the equivalent of the sample code in the NKE documentation (without the CTL_FLAG_PRIVILEGED) and the following code in UserLand: bzero(&addr, sizeof(addr)); // sets the sc_unit field to 0 addr.sc_len = sizeof(addr); addr.sc_family = AF_SYSTEM; addr.ss_sysaddr = AF_SYS_CONTROL; addr.sc_id = 'XXXX'; // should be unique - use a registered Creator ID here fd = socket(PF_SYSTEM, SOCK_DGRAM, SYSPROTO_CONTROL); if (fd) { result = connect(fd, (struct sockaddr *)&addr, sizeof(addr)); if (result) fprintf(stderr, "connect failed %d\n", errno); } else fprintf(stderr, "failed to open socket\n"); if (!result) { /*result = setsockopt( fd, SYSPROTO_CONTROL, kEPCommand1, NULL, 0); if (result) fprintf(stderr, "setsockopt failed on kEPCommand1 call - result was %d\n", result); */ } I'm getting a errno 2 (No such file or directory) on the connect call. Is there something to change in the code showed in the documentation to make it work? The ctl_register call is saying everything is fine in the Kernel Extenstion. _______________________________________________ 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/jgraessley% 40apple.com This email sent to jgraessley(a)apple.com smime.p7s

1 0

kern_control questions
by Stephane Sudre 18 Nov '04

18 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com 2 questions related to kern_control in OS X 10.3.0 or later: XXXX is the same value both in the Kernel Extension and UserLand. _______________________________________________ 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.… * Can registered IDs used for PF_NKE be used for a kern_control too? Or is a new registration required? * I'm using the equivalent of the sample code in the NKE documentation (without the CTL_FLAG_PRIVILEGED) and the following code in UserLand: bzero(&addr, sizeof(addr)); // sets the sc_unit field to 0 addr.sc_len = sizeof(addr); addr.sc_family = AF_SYSTEM; addr.ss_sysaddr = AF_SYS_CONTROL; addr.sc_id = 'XXXX'; // should be unique - use a registered Creator ID here fd = socket(PF_SYSTEM, SOCK_DGRAM, SYSPROTO_CONTROL); if (fd) { result = connect(fd, (struct sockaddr *)&addr, sizeof(addr)); if (result) fprintf(stderr, "connect failed %d\n", errno); } else fprintf(stderr, "failed to open socket\n"); if (!result) { /*result = setsockopt( fd, SYSPROTO_CONTROL, kEPCommand1, NULL, 0); if (result) fprintf(stderr, "setsockopt failed on kEPCommand1 call - result was %d\n", result); */ } I'm getting a errno 2 (No such file or directory) on the connect call. Is there something to change in the code showed in the documentation to make it work? The ctl_register call is saying everything is fine in the Kernel Extenstion. This email sent to site_archiver(a)lists.apple.com

1 0

Re: kextload error: "not a bundle"
by Arrigo Benedetti 18 Nov '04

18 Nov '04

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com User-agent: Mozilla Thunderbird 0.9 (Windows/20041103) Brian Jorgage wrote: Does anyone know how to fix this ? Did you check ownerships and permissions? You can try this: /usr/sbin/chown -R root.wheel $(SRC)/SimpleDataQueue.kext find $(SRC) SimpleDataQueue.kext -type d -exec /bin/chmod 0755 {} \; find $(SRC) SimpleDataQueue.kext -type f -exec /bin/chmod 0644 {} \; -Arrigo _______________________________________________ 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 am following the "Hello Kernel" example under Darwin documentation. When I try to load the kext using "sudo kextload -v HelloKernel.kext" I get the following response: "can't load kernel extension HelloKernel.kext (not a bundle)...". I've looked through the documentation but nothing makes any mention of this error or how to fix it. This email sent to site_archiver(a)lists.apple.com

1 0