February 2004 - Darwin-kernel

Re: Getting BSD process information
by Eric Long 12 Feb '04

12 Feb '04

> No. But that's because it can't be done (reliably). > > For some classes of applications, you can find them indirectly via > Launch Services, but in the general case the information simply isn't > available with any degree of reliability. > > Usually there's a reason for wanting to know this, and often a better > way to go about what you're trying to do. Perhaps we can help in that > area? I need a reliable way of uniquely identifying processes from run to run (or boot to boot) that stays consistent and is understandable to my users. Something like /System/Library/CoreServices/WindowServer is a pretty clear and reliable reference if I can connect it with the WindowServer process from its process id. I don't know what other kind of reference I could use that would uniquely identify a process in such a way that I can reliably identify it and communicate to users what the reference refers to. Eric _______________________________________________ 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: Deadlock in My KEXT with NKE
by Vincent Lubet 12 Feb '04

12 Feb '04

I do not think that using sysctl as a private communication mechanism between a KEXT and a daemon is a good idea. There are many alternative to choose from,: - local sockets - device - SYSPROTO_CONTROL sockets Because you seem to be using a NKE, I would recommend using a local socket or a SYSPROTO_CONTROL socket. SYSPROTO_CONTROL sockets they have be designed to provide a control mechanism between user space and NKE and you won't have to switch funnels (looking back into the archive that was first suggesed by Ollie 2 weeks ago). There is no man page and no documentation about using SYSPROTO_CONTROL sockets but you can learn how to use them by looking at the ppp project. The header file is <sys/sys_domain.h> Vincent PS: Always keep in mind the current NKE API is unstable and subject to change in the future. On Feb 12, 2004, at 11:04 AM, matt jaffa wrote: Right now I am having my Daemon do a sysctlbyname and setting a value in the KEXT to make it aware that it is running, does this sound like an ok way to do it. On Thursday, February 12, 2004, at 11:20 AM, Vincent Lubet wrote: On Feb 11, 2004, at 6:25 PM, Matt Jaffa wrote: Also is there a way for the KEXT to detect if a daemon is running? This is done by having the daemon to initiate some kind of communication channel to the KEXT. This way the daemon becomes a kind of client of the KEXT. If there are no communication channel, there is no daemon running. _______________________________________________ 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: Deadlock in My KEXT with NKE
by matt jaffa 12 Feb '04

12 Feb '04

So what type of 'channel' is this? I have heard of sockets communicating between the two, but someone suggested it is not a good idea. And to connect up the two sockets through AF_UNIX type is too hard. Also heard of sysctlbyname, but as you pointed out, what if the daemon dies. Also Heard of communicating through the IOKit way, where it gets a mach_port and can do it that way, but your KEXT needs to be an IOKit , while mine is a NKE. Is there a channel I am missing, or is it one of those above and I am missing something? Thanks, Matt On Thursday, February 12, 2004, at 12:24 PM, Justin Walker wrote: On Thursday, February 12, 2004, at 11:04 AM, matt jaffa wrote: Right now I am having my Daemon do a sysctlbyname and setting a value in the KEXT to make it aware that it is running, does this sound like an ok way to do it. Sort of. The problem of course is that there is no way to guarantee that the sysctl variable gets 'reset' when the daemon exits (e.g., when it gets a memory fault). With Vincent's suggestion of a 'channel', if there are "process exit" semantics with it (like file descriptors - always 'closed' on exit), then you don't have to worry too much about failure modes. Cheers, Justin On Thursday, February 12, 2004, at 11:20 AM, Vincent Lubet wrote: On Feb 11, 2004, at 6:25 PM, Matt Jaffa wrote: Also is there a way for the KEXT to detect if a daemon is running? This is done by having the daemon to initiate some kind of communication channel to the KEXT. This way the daemon becomes a kind of client of the KEXT. If there are no communication channel, there is no daemon running. Vincent _______________________________________________ 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. -- /~\ 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: Deadlock in My KEXT with NKE
by Justin Walker 12 Feb '04

12 Feb '04

On Thursday, February 12, 2004, at 11:04 AM, matt jaffa wrote: Right now I am having my Daemon do a sysctlbyname and setting a value in the KEXT to make it aware that it is running, does this sound like an ok way to do it. Sort of. The problem of course is that there is no way to guarantee that the sysctl variable gets 'reset' when the daemon exits (e.g., when it gets a memory fault). With Vincent's suggestion of a 'channel', if there are "process exit" semantics with it (like file descriptors - always 'closed' on exit), then you don't have to worry too much about failure modes. Cheers, Justin On Thursday, February 12, 2004, at 11:20 AM, Vincent Lubet wrote: On Feb 11, 2004, at 6:25 PM, Matt Jaffa wrote: Also is there a way for the KEXT to detect if a daemon is running? This is done by having the daemon to initiate some kind of communication channel to the KEXT. This way the daemon becomes a kind of client of the KEXT. If there are no communication channel, there is no daemon running. Vincent _______________________________________________ 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. -- /~\ 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: Getting BSD process information
by Mike Smith 12 Feb '04

12 Feb '04

On Feb 12, 2004, at 10:36 AM, Eric Long wrote: I would like to be able to identify the file system location of the executable associated with a pid_t value. Can anyone explain to me how that is done? No. But that's because it can't be done (reliably). For some classes of applications, you can find them indirectly via Launch Services, but in the general case the information simply isn't available with any degree of reliability. Usually there's a reason for wanting to know this, and often a better way to go about what you're trying to do. Perhaps we can help in that area? = 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: Deadlock in My KEXT with NKE
by matt jaffa 12 Feb '04

12 Feb '04

Right now I am having my Daemon do a sysctlbyname and setting a value in the KEXT to make it aware that it is running, does this sound like an ok way to do it. On Thursday, February 12, 2004, at 11:20 AM, Vincent Lubet wrote: On Feb 11, 2004, at 6:25 PM, Matt Jaffa wrote: Also is there a way for the KEXT to detect if a daemon is running? This is done by having the daemon to initiate some kind of communication channel to the KEXT. This way the daemon becomes a kind of client of the KEXT. If there are no communication channel, there is no daemon running. Vincent _______________________________________________ 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

Getting BSD process information
by Eric Long 12 Feb '04

12 Feb '04

Hi, I found information about getting a list of all processes on OS X at this url: <http://developer.apple.com/qa/qa2001/qa1123.html> This is nice, but not being all that UNIX savvy, I don't know how to identify the information I want for daemon processes that the Mac Process Manager won't give me a process serial number for from their process IDs. I would like to be able to identify the file system location of the executable associated with a pid_t value. Can anyone explain to me how that is done? Thanks, Eric _______________________________________________ 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: Deadlock in My KEXT with NKE
by Vincent Lubet 12 Feb '04

12 Feb '04

On Feb 11, 2004, at 6:25 PM, Matt Jaffa wrote: Also is there a way for the KEXT to detect if a daemon is running? This is done by having the daemon to initiate some kind of communication channel to the KEXT. This way the daemon becomes a kind of client of the KEXT. If there are no communication channel, there is no daemon running. Vincent _______________________________________________ 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: Deadlock in My KEXT with NKE
by Justin Walker 12 Feb '04

12 Feb '04

On Wednesday, February 11, 2004, at 06:51 PM, Justin Walker wrote: On Wednesday, February 11, 2004, at 06:25 PM, Matt Jaffa wrote: [snip] Reading my reply again, it may be a bit cryptic. If so, your best bet may be to spring for a copy of Stevens's "TCP/IP Illustrated, V2" (Addison Wesley, 1995). This is a full code walkthrough of the 4.2 BSD networking stack. The code, of course, is out of date with respect to Darwin, but the fundamentals have not changed that much (spl() is used in place of funnels, ...). This will show you how the pieces of this stack work together, from the mbufs to the socket layer, to the protocol implementations. NKEs are naturally not covered. And of course, you can ask for clarification here. Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | "Weaseling out of things is what | separates us from the animals. | Well, except the weasel." | - Homer J Simpson *--------------------------------------*-------------------------------* _______________________________________________ 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: Deadlock in My KEXT with NKE
by Justin Walker 12 Feb '04

12 Feb '04

On Wednesday, February 11, 2004, at 06:25 PM, Matt Jaffa wrote: Ok I tend to not explain what I am doing real well, sorry about that. Well, your explanation below seems pretty good. Maybe you just need to write the first message, throw it away, and then write the second one to send :-} I don't want to hold the funnel while it is sleeping, but keeping the state as it was before it slept. But the most important thing I am trying to do here is make it so that the Thread is sleeping while work is being done by the daemon so that it doesn't do a while loop without sleeping in the kernel, here is my while loop: [snip] If I understand what you want correctly, the more-or-less standard way to do this in the kernel is with a combo of 'sleep' and 'wakeup'. One thread sets a variable to indicate that it is sleeping. Then it sleeps. A second thread comes along, does something, sets the variable to indicate that the reason for sleeping has past, and then does a wakeup (sleep has a 'wait' channel; wakeup uses that as the 'target' of the wakeup). The first thread now takes off running. It should, by rights, check the variable to assure that it was awakened for the right reason (spurious wakeups can occur). If you look at the networking code in the kernel, you will see calls like sbwait() and sbwakeup(). These implement that kind of scheme, using sleep and wakeup on socketbuffers. You should also be sure to use your own channels (generally, an address), so as not to interfere with other threads. Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------* _______________________________________________ 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