September 2004 - Darwin-dev

swap files question.
by Dieder Bylsma 04 Sep '04

04 Sep '04

Dieder _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. Is there anyway to determine which program/process is paged out to a particular swap file?

1 0

Re: IOPMSchedulePowerEvent kIOReturnNotReady
by Ethan Bold 03 Sep '04

03 Sep '04

Hello - >From IOPMSchedulePowerEvent(): abs_time_to_wake = CFDateGetAbsoluteTime(time_to_wake); if(abs_time_to_wake < (CFAbsoluteTimeGetCurrent() + 30.0)) { ret = kIOReturnNotReady; goto exit; } -ethan On Sep 1, 2004, at 1:46 AM, Matis Schotte wrote: any help would be great! (please excuse my bad english ...) _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. kIOReturnNotReady indicates that you're trying to schedule a wakeup event too soon. There is a 30-second threshold that we enforce, since for wakeup events any less than 30 seconds in the future we cannot guarantee a successful wakeup. The transition from awake to asleep can take up to 30 seconds alone. If you're receiving this error for a date and time significantly in the future (more than a minute past the current time) then that is a bug. hello, i don't know if this post belongs to cocoa, but i'm using it with a cocoa app. i have had a working function that saved a starting time and date with IOPMSchedulePowerEvent to power the machine on, but don't know why, now i get always the error kIOReturnNotReady (0x2d8) - what does it mean ? and why do i get it ? - smat _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored.

1 0

Re: Zombie curiosity
by Paul Ripke 01 Sep '04

01 Sep '04

Last one, I promise! Just checked on a friends 10.3.5 system and the issue with a zombie processes process fields being nulled out (like ppid, uid, etc) has been fixed. Not sure when this happened, but it was somewhere between 10.2.6 (Darwin 6.6) and 10.3.5 (Darwin 7.5). So if you see a zombie process with a PPID of '0', upgrade, and that little issue will go away (ie. it will show the real PID of the parent, not '0'). Cheers, -- stix _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. On Wednesday, Sep 1, 2004, at 23:03 Australia/Sydney, Justin Walker wrote: At the risk of prolonging this thread beyond its useful life, I've cobbled up a small C program that illustrates what I've been talking about (appended to this mail).

1 0

Re: Problems with kqueue
by Jim Magee 01 Sep '04

01 Sep '04

On Aug 31, 2004, at 9:45 AM, Mike Vannorsdel wrote: --Jim _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. I've been trying to use kqueue with the EVFILT_PROC filter but have ran into some problems. One is that once an event fires it will never fire again until re-added to the queue. For instance: Another problem seems to be once any event in the queue fires, the others will never fire. So basically I have to re-add all events after each event I get. This seems to defeat the purpose of having a queue. Is this just how the EVFILT_PROC works, a bug, or programmer error? The term k-"queue" is somewhat of a misnomer (at least in the way that you have interpreted it). Each event source (represented by an individual kevent structure) is treated as a level-triggered event source. Instances of events from a single source are never queued. The "queue" part of the name refers to the ordering of event notifications between the various different sources registered with a given kqueue. For EVFILT_PROC events, we automatically set EV_CLEAR when you register for them. That means that all (watched) events that happen prior to you being notified will be returned at once and atomically cleared so that only future events will be returned. Are you expecting discreet notifications for a series of process events that happened in the past? Or is your test scenario assured that the subsequent events are actually occurring after the first notification is received? If the latter, then there might be a bug. But then I'm not sure why re-registering would fix it - knowing what I do about the implementation (re-registration just alters the existing registration).

1 0

Re: Zombie curiosity
by Justin Walker 01 Sep '04

01 Sep '04

I hope this helps explain what's going on. Regards, Justin ==========================zombie.c=============================== /* * Example to show how zombies are treated */ #include <stdio.h> #include <stdlib.h> #include <errno.h> int SleepTime = 30, WaitForKids=0; main(int argc, char *argv[]) { int pidA, pidB, ReapTime = 0; if (argc > 1) { if (argv[1][0]=='w') WaitForKids=1; else if (argv[1][0] == 'r') { WaitForKids=1; ReapTime = 20; } } if ((pidA = fork()) < 0) { perror("FORK1"); exit(-1); } else if (pidA) /* in Parent */ { printf("Spawned %d\n", pidA); if ((pidB = fork()) < 0) { perror("FORK2"); exit(-1); } else if (pidB) /* in Parent */ { printf("Spawned %d\n", pidB); if (WaitForKids == 0) exit(0); } else /* In child B */ { printf("Child B...");fflush(stdout); sleep(SleepTime); printf("Slept\n"); exit(0); } } else /* In child A */ { printf("Child A...");fflush(stdout); sleep(SleepTime); printf("Slept\n"); exit(0); } ReapEm(ReapTime); } ReapEm(int ReapTime) { int Status, rval; if (ReapTime) { printf("Sleeping...");fflush(stdout); sleep(ReapTime+SleepTime); } while ((rval = wait(&Status)) != -1) printf("WAIT: %d, %x\n", rval, Status); printf("WAIT: %d\n", errno); } -- Justin C. Walker, Curmudgeon-At-Large * Institute for the Enhancement | When LuteFisk is outlawed of the Director's Income | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------* _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. At the risk of prolonging this thread beyond its useful life, I've cobbled up a small C program that illustrates what I've been talking about (appended to this mail). If you run this program with no arguments, it forks two child processes and exits. The children sleep for a specified time, and then exit. In the interim, in another window, run something like "ps -oppid -aux" and you will see the two processes listed as children of init. After they exit, there will be no zombies (because init cleans things up quickly; it's possible you can see them as zombies if you are lucky, but it's unlikely). If you run this program with argument 'w', the parent will wait for the children, and reap them as they exit. Running the above 'ps' while the children sleep will show them as children of the parent. After the parent exits, there will be no zombies. If you run this program with argument 'r', the parent will sleep longer than the children sleep, and then reap them. Running 'ps' with the children sleeping will show them as children of the parent process. Running 'ps' after the children exit, but before the parent wakes will show two zombie processes whose parent is the parent process. Once the parent wakes and reaps the children, there will be no zombies.

1 0

Re: Zombie curiosity
by Justin Walker 01 Sep '04

01 Sep '04

On Aug 31, 2004, at 23:29, Sanjay Mishra wrote: Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | Men are from Earth. | Women are from Earth. | Deal with it. *--------------------------------------*-------------------------------* _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. Though daemon works that way ( calling fork() ),but forking fast enough on Mac 10.3.5 seems to slow it down and also the danger of "Resource temporarily unavailable" error message lurks around. I don't understand the above; can you clarify? Do you mean that if you fork 'n' copies of your process, things slow down? There is always a possibility of "resource unavailable". If you are forking with abandon, then you will reach a point where (a) you have too many processes running; or (b) you don't have enough swap space to accommodate them all. This happens on my G4. Am I doing something nasty somewhere? I just fork() and the child created calls execlp(). My main program wait() for the child to end and return with exit(). If you are doing this just once, then I'm unable to explain the symptoms you are seeing. I certainly don't see them, and since the system does just this (fork/exec) for most operations, it's hard to see how it can happen. You must be putting some kind of excessive burden on the system.

1 0

IOPMSchedulePowerEvent kIOReturnNotReady
by Matis Schotte 01 Sep '04

01 Sep '04

any help would be great! (please excuse my bad english ...) - smat _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. hello, i don't know if this post belongs to cocoa, but i'm using it with a cocoa app. i have had a working function that saved a starting time and date with IOPMSchedulePowerEvent to power the machine on, but don't know why, now i get always the error kIOReturnNotReady (0x2d8) - what does it mean ? and why do i get it ?

1 0

Re: Mounting ISO file system on a hybrid CD
by Shantonu Sen 01 Sep '04

01 Sep '04

On Aug 31, 2004, at 9:30 PM, James W. Walker wrote: $ mkdir /Volumes/Foreign $ sudo mount -t cd9660 /dev/disk1s1 /Volumes/Foreign This is the only way to accomplish this under Mac OS X. $ diskutil list $ disktool -l Shantonu _______________________________________________ darwin-development mailing list | darwin-development(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-development Do not post admin requests to the list. They will be ignored. I'm interested in mounting the ISO/Joliet part of a hybrid CD-ROM. I can sort of get it to work using commands like: Then the volume is shown by mount and seen by Carbon's FSGetVolumeInfo, but it is not shown in the Finder or listed by either of If you do "open /Volumes/Foreign", you'll trigger the Finder process' Carbon File Manager cache to "see" the filesystem, and will show an icon in the sidebar. I have the impression that this has something to do with disk arbitration, not that I fully understand what that means. I tried mounting the volume using the function DADiskMountWithArguments, but apparently it only allows the -o arguments to mount, not the -t arguments. Anyone have any wisdom to share on this? You cannot force a file system, nor can you convince DiskArbitration that there's anything to mount on disk1s1, because of the way it probes media. By the way, mount shows the HFS part of the CD as /dev/disk1s1s2, but the ISO part as /dev/disk1s1. This offends my sense of symmetry. I would expect that the two file systems would be at the same level of the hierarchy. Not that I know what those s's and 1's and 2's really mean. Actually, they are symmetric. The ISO "filesystem" takes up the entire media (disk1s1). disk1 represents the entire CD, and disk1s1 is the first session. On disk1s1, there are two overlapping containers. The ISO "filesystem", and an Apple Partition Map. The Apple Partition Map further publishes several slices corresponding to each partition in the map. That's where disk1s1s2 comes from. The ISO filesystem and APM are peers, and symmetric. Running "ioreg -l" or IORegistryExplorer.app may be instructive. What is your higher level goal? If you want two things to show up in the Finder, you may want a multi-session disk with two Apple Partition Maps, or a single session with a multi-partition map. You cannot do the former with ISO content, though. If you are trying to preview the ISO content of your hybrid, the manual mount is your best bet.

1 0