July 2002 - Darwin-kernel - lists.apple.com

Re: ps v. top
by Andrew Pinski 31 Jul '02

31 Jul '02

Have tried to look at the output of ps auxww, it will not print LaunchCFMApp but the application's real name. It looks like ps is just using real executable's name at which the kernel executed it, not at which in turn LaunchCFMApp did. Thanks, Andrew Pinski On Wednesday, July 31, 2002, at 06:42 , Justin C. Walker wrote: On Tuesday, July 30, 2002, at 12:39 PM, David Blanton wrote: a cfm app shows in ps as LaunchCFMApp in top it shows as app_name This may be somehow app-specific, because I don't see it on my system with at least one CFM app running. I'm using 10.1.5, but I don't think 'ps' has changed all that much since the early days... where is the app_name found when uap->fname in execve is ../LaunchCFMApp UTSL! Check top.tproj in the system_cmds project in the Darwin repository (http://developer.apple.com/darwin) Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *--------------------------------------*-------------------------------* _______________________________________________ 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: ps v. top
by Justin C. Walker 31 Jul '02

31 Jul '02

On Tuesday, July 30, 2002, at 12:39 PM, David Blanton wrote: a cfm app shows in ps as LaunchCFMApp in top it shows as app_name This may be somehow app-specific, because I don't see it on my system with at least one CFM app running. I'm using 10.1.5, but I don't think 'ps' has changed all that much since the early days... where is the app_name found when uap->fname in execve is ../LaunchCFMApp UTSL! Check top.tproj in the system_cmds project in the Darwin repository (http://developer.apple.com/darwin) Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *--------------------------------------*-------------------------------* _______________________________________________ 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: ps v. top
by David Blanton 31 Jul '02

31 Jul '02

When a CFM app is launched the path name passed to execve is ../LaunchCFMApp. Yes, ps -auww shows the CFM name. If TOP is running when a CFM app is launched it immediately shows the CFM app name in the 'Command' column. So the question is ... at execve time with a CFM app where is the CFM app name available? TIA & Peace. db On Wednesday, July 31, 2002, at 05:12 PM, Andrew Pinski wrote: Have tried to look at the output of ps auxww, it will not print LaunchCFMApp but the application's real name. It looks like ps is just using real executable's name at which the kernel executed it, not at which in turn LaunchCFMApp did. Thanks, Andrew Pinski On Wednesday, July 31, 2002, at 06:42 , Justin C. Walker wrote: On Tuesday, July 30, 2002, at 12:39 PM, David Blanton wrote: a cfm app shows in ps as LaunchCFMApp in top it shows as app_name This may be somehow app-specific, because I don't see it on my system with at least one CFM app running. I'm using 10.1.5, but I don't think 'ps' has changed all that much since the early days... where is the app_name found when uap->fname in execve is ../LaunchCFMApp UTSL! Check top.tproj in the system_cmds project in the Darwin repository (http://developer.apple.com/darwin) Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *--------------------------------------*-------------------------------* _______________________________________________ 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. David Blanton Elisar Software Corporation Albuquerque _______________________________________________ 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: How to kill an IOKit thread?
by Jim Magee 31 Jul '02

31 Jul '02

On Wednesday, July 31, 2002, at 03:30 PM, Jim Magee wrote: Regardless of that little confusion, it is always best to let the thread terminate itself. Simply wake it up from that block such that it notices it should kill itself. Then just have it call thread_terminate(current_act()). If it truly is a kernel thread, it will not return from this call. However, if it really is user-level thread, the thread will only be marked for termination, and will not terminate until it unwinds out of its current system call. It will stop just short of returning to user-space and then will finish the job of terminating itself. And, as I should have checked before posting this, IOExitThread() encompasses exactly this sentiment. On a more general note, there is no "terminate that other thread" primitive in all the kernel which will work on a kernel thread. The thread_terminate() primitive will fail when called on a kernel thread activation other than the current one. It's that highly discouraged a practice. ;-) Is the problem that you don't know how to wake it from the blocking call it made? Which call is it? Maybe we can make suggestions on how to "roust" your thread such that it notices it should come back and terminate itself(). --Jim PS: An important note: Even though IOCreateThread returns a thread [shuttle] pointer (something it really shouldn't really do for other reasons), it does not return a thread [shuttle] reference. Therefore, the thread [shuttle] pointer is only valid as long as you are positive that it has not already been terminated. PPS: The reason I keep putting "[shuttle]" in these comments is that within the kernel, a thread really consists of two parts, an activation (which represents the state of the thread), and a shuttle (which is the scheduled entity). The activation is what people traditionally call a "thread," and is the handle by which they should be manipulated. The shuttle is an internal construct, separated from the activation itself, to allow for a more efficient implementation of priority inheritance mechanism (known as thread migration). Our goal is to make it so that no entity outside of Mach itself should have direct access to the shuttle. _______________________________________________ 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: How to kill an IOKit thread?
by Brian Bergstrand 31 Jul '02

31 Jul '02

At 8:53 PM +0200 7/31/02, Heiko Panther wrote: I'm using an IOKit thread in my kext. The thread usually blocks in a system call. >From my main workloop, I want to terminate the thread on certain conditions. There is an explanation on how to do so in IMX:Kernel Programming. thread_terminate(getact_thread(thread_t thread)); I think "thread_terminate (current_act());" is the recommended method. It's what I use. Only problem: I can't seem to find the function getact_thread(). Any clues? How do I get rid of the thread? Do not, repeat DO NOT kill a thread from another thread. It is only safe to kill a thread from within itself. The most common way to get around this limitation is to check a flag in the thread that needs to exit and if the flag is true then exit the thread. So in your case, in your main thread you would set the flag, then signal your other thread using a semaphore. If you need to wait for the thread to exit, block on a semaphore in the main thread that the exiting thread signals just before terminating. HTH. -- Brian Bergstrand <http://www.classicalguitar.net/brian/> PGP Key ID: 0xB6C7B6A2 We're the Soprano's in a pickup truck. - Daryll Hammond as Bill Clinton on SNL _______________________________________________ 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: How to kill an IOKit thread?
by Jim Magee 31 Jul '02

31 Jul '02

On Wednesday, July 31, 2002, at 02:53 PM, Heiko Panther wrote: I'm using an IOKit thread in my kext. The thread usually blocks in a system call. >From my main workloop, I want to terminate the thread on certain conditions. There is an explanation on how to do so in IMX:Kernel Programming. thread_terminate(getact_thread(thread_t thread)); Only problem: I can't seem to find the function getact_thread(). Any clues? How do I get rid of the thread? I'm confused. If it is an IOKit thread created by a KEXT, it can't be blocked in a system call. Regardless of that little confusion, it is always best to let the thread terminate itself. Simply wake it up from that block such that it notices it should kill itself. Then just have it call thread_terminate(current_act()). If it truly is a kernel thread, it will not return from this call. However, if it really is user-level thread, the thread will only be marked for termination, and will not terminate until it unwinds out of its current system call. It will stop just short of returning to user-space and then will finish the job of terminating itself. --Jim _______________________________________________ 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: How to kill an IOKit thread?
by Andrew Gallatin 31 Jul '02

31 Jul '02

Heiko Panther writes: > I'm using an IOKit thread in my kext. The thread usually blocks in a > system call. > From my main workloop, I want to terminate the thread on certain > conditions. There is an explanation on how to do so in IMX:Kernel > Programming. > > thread_terminate(getact_thread(thread_t thread)); > > Only problem: I can't seem to find the function getact_thread(). Any > clues? How do I get rid of the thread? Just get the thread to call IOExitThread (); Eg, I have a thread which wakes up periodically. When it wakes up, one of the things it checks for is a "should_exit" flag: Eg: Module termination code: mydriver_port_reaper_should_exit = 1; while (!mydriver_port_reaper_exited) { IOSleep (1 * MS_PER_SEC); } Thread function: { while (!mydriver_port_reaper_should_exit) { <...> IOSleep (5 * MS_PER_SEC); } mydriver_port_reaper_exited = 1; IOExitThread (); } Cheers, Drew _______________________________________________ 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

How to kill an IOKit thread?
by Heiko Panther 31 Jul '02

31 Jul '02

I'm using an IOKit thread in my kext. The thread usually blocks in a system call. >From my main workloop, I want to terminate the thread on certain conditions. There is an explanation on how to do so in IMX:Kernel Programming. thread_terminate(getact_thread(thread_t thread)); Only problem: I can't seem to find the function getact_thread(). Any clues? How do I get rid of the thread? Heiko Panther _______________________________________________ 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

Unable to find driver for this platform
by Daniel Hazelbaker 31 Jul '02

31 Jul '02

Greetings. I have a 14.1" I book running 10.1.5. A few months ago I wrote a patch to the xnu-9-1 CVS tree of the kernel to allow me to run with the clamshell down. (If you can really still call it a clamshell) The patch worked fine blah blah blah. Since then I have updated the release version twice, I think. It was at 10.1.3 when I did this iir. Anyway, now when I use xnu-9-1 or xnu-201-19 (which is supposedly the latest?), I get a kernel panic immedietly with this message: Unable to find driver for this platform: PowerBook 4,2 I know that message didn't come up before, and I know it was that version of xnu that I used before. Did something in the firmware get changed by an update? Also, is there a version of xnu that WILL run on my computer now? I cannot find anything that defnitely says which version of xnu is being used in 10.1.5 under the apple tree of the cvs server, but there is an OD_APPLE_10_1_5 (or something similiar) in the Open Darwin cvs tree. Please advise. Daniel Hazelbaker P.S., Didn't there used to be a Terminal app in the Apple menu for OSX Installer CD? Pretty sad to have to boot OS9 to rename a file. Even going single user on the CD didn't help, you can't mount anything, ouch! _______________________________________________ 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: How do disk writes work these days?
by Justin C. Walker 30 Jul '02

30 Jul '02

On Tuesday, July 30, 2002, at 02:51 PM, Pelle Johansson wrote: torsdagen den 25 juli 2002 kl 03.15 skrev Wade Tregaskis: Is this still an efficient way of operating, in an increasingly linear [e. g. single-user] OS? I'm just thinking, if you could return from writes faster, the calling process can keep doing other things. My thought is that you could do this without breaking existing apps by performing any error checking [i.e. quickly verifying there is free space, among other things] immediately, returning fail if appropriate, or otherwise returning no error. You could then worry about physically getting the data to disk, seeing as this is the most time intensive part... Side note: I was taught that if you know how large a file you want to write you should use ftruncate() to set the file size before writing. That way the OS can optimize the location of the file on the hard drive. I've noticed Unix application are quite bad at doing this, so perhaps it doesn't apply in some OSes. Most Unices won't take advantage of this 'optimization', because the typical Unix file system allows 'holes' in the allocation. Thus, if you create an empty file, and truncate (or seek to a far corner of the universe), and write a byte, the intervening space will *not* be allocated. This is dependent on the type of file system, of course. The HFS+ file system does not support 'holes'. Regards, Justin PS: Reading "man ftruncate" leads me to guess that if you 'ftruncate' an empty, or short, file to a large size, nothing will happen on the disk (the man page says "at most 'length' bytes"). -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *--------------------------------------*-------------------------------* _______________________________________________ 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